Pages

February 9, 2011

Headphone & Amp Impedance

INTRO: The output Impedance of headphone sources is one of the most common reasons the same headphones can sound different depending on what they’re plugged into. This important parameter is rarely specified by manufactures but can make a huge difference in sound quality and headphone compatibility.

HEADPHONE IMPEDANCE MOVED: This article used to be about both output impedance and headphone impedance. But, in the interest of shorter articles, I’ve split them. So if you’re looking for info headphones, please see:

THE SHORT VERSION: All you really need to know is most headphones work best when the output impedance is less than 1/8th the headphone impedance. So, for example, with 32 ohm Grados the output impedance can be, at most, 32/8 = 4 ohms. The Etymotic HF5s are 16 ohms so the max output impedance is 16/8 = 2 ohms. If you want to be assured a source will work well with just about any headphone, simply make sure the output impedance is under 2 ohms.

WHY DOES OUTPUT IMPEDANCE MATTER? It matters for at least three reasons:

  • The greater the output impedance the greater the voltage drop with lower impedance loads. This drop can be large to enough to prevent driving low impedance headphones to sufficiently loud levels. A real world example is the Behringer UCA202 with a 50 ohm output impedance. It struggles with some 16 - 32 ohm headphones.
  • Headphone impedance changes with frequency. If the output impedance is much above zero this means the voltage delivered to the headphones will also change with frequency. The greater the output impedance, the greater the frequency response deviations. Different headphones will interact in different, and typically unpredictable, ways with the source. Sometimes these variations can be large and plainly audible.
  • As output impedance increases electrical damping is reduced. The bass performance of the headphones, as designed by the manufacture, may be audibly compromised if there’s insufficient damping. The bass might become more “boomy” and less controlled. The transient response becomes worse and the deep bass performance is compromised (the headphones will roll off sooner at low frequencies). A few, such as those who like a very warm “tube like” sound, might enjoy this sort of under damped bass. But it’s almost always less accurate compared to using a low impedance source.

THE 1/8th RULE: To minimize all three of the above problems, it’s only necessary to keep the output impedance less than 1/8th the headphone impedance. Or, put another way, just divide the headphone impedance by 8 to get the maximum output impedance without potential audible degradation.

IS THERE A STANDARD FOR OUTPUT IMPEDANCE? The only standard I’m aware of is IEC 61938 from 1996. It specifies an output impedance of 120 ohms. There are numerous reasons why this is standard is way out of data and a really bad idea. In a Stereophile article about headphones, they said of the 120 ohm standard:

“Whoever wrote that must live in a fantasy world.”

I have to agree with Stereophile. The 120 ohm standard might have been (barely!) tolerable before the iPod and other portable music sources became immensely popular, but it’s not any more. Most headphones are designed very differently today.

PSUEDO STANDARDS: A lot of professional gear has a 20 – 50 ohm headphone output impedance. I’m not aware of any that follows the 120 ohm IEC standard. Consumer gear tends to be in the range of 0 – 20 ohms and, with the exception of tube and certain other esoteric designs, most high-end audiophile headphone sources are well under 2 ohms.

THE iPOD INFLUENCE: Since the 120 ohm standard was published in 1996, music players advanced from lo-fi cassette tape and skipping portable CD players to the massive iPod craze. Apple helped take high quality audio portable and there are at least half a billion portable digital players in circulation not including phones. Nearly all portable music/media players now run from a single cell Li-Ion battery. These batteries only produce a bit over 3 volts which means you typically get less than 1 volt RMS of audio output driving typical headphones (sometimes much less). If you add 120 ohms to the output, and use typical portable headphones (nearly all of which are in the range of 16 –32 ohms) the headphones usually won’t play loud enough. And most of the battery power is wasted as heat in the 120 ohm resistor. Only a small fraction of the power makes it to the headphones. That’s a big problem in portable audio where getting the best battery life from ever smaller devices is critical. It’s much more efficient to deliver all the power to headphones.

HEADPHONE DESIGN: So what output impedance do headphone manufactures design for? As of 2009 well over 220 million iPods had been sold. The iPod, and similar portable players, are the 800 pound gorillas in the headphone market. So, not surprisingly, most manufactures started designing many or all of their headphones to work well with the iPod. That means they’re designed to work with an output impedance under 10 ohms. And higher-end full size cans are most often designed for sources that follow the 1/8th Rule or have a near zero output impedance. I’m not aware of any current audiophile headphones intended for home use designed to the ancient 120 ohm standard.

THE BEST HEADPHONES ARE DESIGNED FOR THE BEST SOURCES: If you do a quick survey of the most well reviewed high-end headphone amps and DACs, they nearly all have very low output impedances. Examples are products from Grace Designs, Benchmark Media, HeadAmp, HeadRoom, Violectric, etc. It only stands to reason that most high-end headphones are designed to be at their best with similar products. Some of the most highly regarded headphones have relatively low impedances including several models from Denon, AKG, Etymotic, Ultimate Ears, Westone, HiFiMAN and Audeze. All of these, as far as I know, were designed to be used with low (ideally near zero) impedance sources. I’ve also had a Sennheiser representative tell me they design their audiophile and portable headphones for zero ohm sources.

THE FREQUENCY RESPONSE PROBLEM: If the output impedance is more more than 1/8th the headphone impedance there will be variations in the frequency response. With some headphones, especially balanced armature or multi driver designs, these variations can be rather extreme. Here’s what 43 ohms of output impedance does to the Ultimate Ears SuperFi 5’s frequency response—a total, and very audible, variation of 12 dB:

FiiO-E9-Frequency-Response-Ultimate-[1]

 

10 OHM OUTPUT IMPEDANCE: Some might look at the above example and think it’s extreme with a 43 ohm source. But plenty of sources have around a 10 ohm output impedance. Here’s the same headphones with a 10 ohm source—there’s still a very audible 6 dB of variation. This sort of curve creates weaker bass, a “glaring” midrange emphasis, muted high frequencies, and odd phase characteristics due to the sharp “notch” at 10 khz that can alter spatial perceptions:

FiiO-E9-Frequency-Response-Large-Jac

 

FULL SIZE SENNHEISERS: Here are the full size, higher impedance, Sennheiser HD590 cans with the same 10 ohm output impedance. Now the variation is only a bit over 1 dB above 20 hz. While 1 dB isn’t that much, it’s right in the most “boomy” bass region which is the last place most want any sort of emphasis:

FiiO-E9-Frequency-Response-Sennhesie[2]

 

DAMPING EXPLAINED: Any dynamic driver, in a speaker or headphone, moves back and forth with the music. That’s how it creates sound and they all have moving mass. The laws of physics say an object in motion tends to stay in motion. Damping is used to help avoid unwanted motion. Without going into too many details, if a speaker is under-damped, it keeps moving after it should have stopped. And if it’s over-damped (rare) its ability to accurately follow the signal is compromised—imagine a speaker trying to operate submersed in maple syrup. There are only two ways to damp a driver—mechanically and electrically.

BOUNCING CARS: Mechanical damping is much like the shock absorbers on a car. They add resistance so when you hit a bump the car doesn’t keep bouncing up and down long after the bump. But they also add harshness because they reduce the suspension’s ability to accurately follow the road. They’re a compromise—soft shocks give a softer but more bouncy ride and stiff shocks control the bouncing better but make the ride harsher. Mechanical damping is always a compromise.

ELECTRICAL IS BETTER: There’s a better option to control unwanted motion of headphone drivers and it’s called electrical damping. The voice coil and magnet of the driver work with the amplifier to control the motion of the driver. This kind of damping has fewer negative side effects and allows headphone designers to create headphones with less distortion and better sound. Just like a car suspension that can better follow the road, an optimally damped headphone driver can better follow the audio signal. But, and this is the critical part, electrical damping is only effective when the output impedance of the amplifier is much lower than the impedance of the headphones. If you plug 16 ohm headphones into an amp with a 50 ohm output impedance, there will be almost no electrical damping. That means when the driver is supposed to stop moving it might not. The headphone is more like a car with worn shock absorbers. If the 1/8th Rule is followed, however, there will be sufficient electrical damping.

A SPEAKER ANALOGY: Back in the day, before my time, speakers were mostly driven by amplifiers that used tubes instead of transistors. Tubes are high impedance devices that operate at high voltages so nearly all tube amps use output transformers. Without going into all the details, tube amps had widely varying output impedances that were often significant and violated the 1/8th Rule. Speaker manufactures couldn’t rely on amplifiers having a low enough impedance to provide much electrical damping. This compromised speaker design much like headphone design is compromised today if a headphone designer can’t rely on a low impedance source for proper electrical damping.

ACOUSTIC SUSPENSION: In the 1970’s the situation changed as solid state amplifiers became popular. Almost all solid state amps easily pass the 1/8th Rule. In fact, most pass a 1/50th Rule—their output impedance is generally below about 0.16 ohms—known as a damping factor of 50. Suddenly speaker manufactures were free to design better speakers that could take advantage of these much lower output impedances. And the first really good acoustic suspension sealed box speakers like the original AR's, Large Advents, etc. were developed. They had deeper and better bass than any of their tube-powered predecessors could manage from a similar box size. It was a big milestone in "hi-fi" to rely on lots of electrical damping from the amplifier. It’s too bad many headphone sources are 40+ years behind.

WHAT OUTPUT IMPEDANCE DOES MY SOURCE HAVE? Some manufactures make it clear they strive for a low output impedance (such as Benchmark), while others specify the actual output impedance of their products (such as Behringer does with the UCA202 at 50 ohms). And most, sadly, keep it a total mystery. Some product reviews, such as the ones on this blog, include measurements of the output impedance as it’s critical to the sound of the device with various different headphones.

WHY DO SO MANY SOURCES HAVE A HIGHER OUTPUT IMPEDANCE? The most common reasons are:

  • Headphone Protection - More powerful sources with a low output impedance might be capable of delivering too much power into low impedance headphones. To help protect such headphones, some designers raise the output impedance. This is a compromise to try and have the amp adapt to the load used. But it comes at a big price with many headphones. A better solution is offering two gain options The low gain setting can lower the maximum output voltage when using low impedance headphones. And, in addition, active current limiting can be used so the source will automatically restrict the maximum output into lower impedance headphones even if the wrong gain setting is used. 
  • To Be Different - Some manufactures raise the output impedance on purpose claiming it makes their source sound better. Sometimes “different sells” as it’s a way to differentiate the sound of their product from their competitors. But, in this case, the particular “different sound” you get is entirely dependent on which headphones are used. With some it might be an improvement and with others it’s more likely a big step backwards. The odds greatly favor degrading the sound.
  • It’s Cheap – A higher output impedance is a band-aid for many inexpensive headphone sources. It’s a cheap way to achieve stability, a crude form of short circuit protection, and it can allow using an otherwise substandard op amp or output device that would be unable to drive 16 or even 32 ohm headphones directly. By adding some series resistance to the output all these things get “fixed” with a $0.01 part. But the cheap “fix” comes at a substantial price in the sound quality with many headphones.

EXCEPTIONS TO THE RULE: There are a few headphones supposedly designed for significantly higher output impedances. I do wonder if this might be more myth than reality these days in terms of audiophile and consumer headphones as I’m not aware for any specific examples. But it’s certainly possible. If so, using these headphones on a low impedance source might cause under-damped bass performance and a different frequency response than the manufacture intended. This might explain some of the “synergy” claims when certain headphones are mated with a certain source. But those “synergies” are entirely subjective—one man’s “bright and detailed” is another man’s “harsh”. The only way to get consistent performance is to use a low impedance source and follow the 1/8th Rule.

A CHEAP TEST: If you’re wondering if your current source is compromising the sound quality because of an unknown output impedance, consider buying the $19 FiiO E5 amp. It has a near zero ohm output impedance and has enough output for most many headphones under 100 ohms. If it obviously improves the sound, it’s likely your source has an output impedance that’s too high.

BOTTOM LINE: Unless you know your particular headphones sound better with a specific higher output impedance, it’s best to always use a source with an output impedance no higher than 1/8th the impedance of your headphones. Or, to make it even simpler, an output impedance of 2 ohms or less.


TECH SECTION:


IMPEDANCE VS RESISTANCE: These two terms are used somewhat interchangeably, but technically there are some significant differences. Electrical resistance is represented by the letter “R” and has the same value at all frequencies. Electrical impedance is more complex and its value typically changes with frequency. It’s represented by the letter “Z”. For the purposes of this article, the unit of measure for both is Ohms.

OUTPUT IMPEDANCE DIAGRAM: The diagram below shows the effect of output impedance. The blue circle on the left above represents a “perfect source”, the blue resistor (zig zag line) in the middle represents the output impedance. And the resistor on the right represents the load impedance (the headphones). If the output impedance is not zero, the voltage produced by the source will be reduced when a load is connected. The higher the output impedance, the greater the drop in voltage at the load. This drop is given by the formula: Load Voltage = Source Voltage * ( Zload / ( Zload + Zout) ). For more information see Wikipedia Voltage Divider.

output impedance

VOLTAGE AND CURRENT: It’s important to have at least some understanding of voltage and current to understand impedance and this article. Voltage is analogous to water pressure (i.e. PSI) while current is analogous to the volume of water (i.e. gallons per minute). If you let water run out of the end of your garden hose with nothing attached you get a lot of flow (current) and can fill a bucket quickly but the pressure at the end of the hose is near zero. If you put a small nozzle on the hose the pressure (voltage) is much higher but volume of water is reduced (it takes longer to fill the same bucket). The two are typically inversely related. High pressure usually means low flow and visa versa. The same is true of voltage and current. The relationship between voltage, current, and resistance (and for the purposes of this article, impedance) is defined by Ohm’s Law. Substitute Z for R.

WHERE DOES THE 1/8th RULE COME FROM? The smallest audible difference most can hear is about 1 dB. For the output impedance to create a -1 dB change, you have antilog(-1/20) = 0.89. Using the divider formula from above, when the output impedance is 1/8 the load impedance you get 0.89 or a 1 dB drop. Headphone impedance can vary by a factor of 10 or more over the audio band. The SuperFi 5 is rated at 21 ohms but varies from 10 ohms to 90 ohms. So the 1/8 Rule gives a max output impedance of 2.6 ohms. Assuming a 1 volt source we get:

  • Headphone Voltage at 21 Ohm Nominal Impedance = 21 (21+2.6) = 0.89 volts
  • Headphone Voltage at 10 Ohm Minimal Impedance = 10 (10+2.6) = 0.79 volts
  • Headphone Voltage at 90 Ohm Maximum Impedance = 90 (90+2.6) = 0.97 volts
  • Frequency Response Variation = 20*LOG(.97/.89) = 0.75 dB (under the 1 dB goal)

MEASURING OUTPUT IMPEDANCE: As seen in the diagram above the output resistance forms a voltage divider. By measuring the output voltage with no load, and with a known load, you can calculate the output impedance. This online calculator makes it easy. The no load voltage is the “Input Voltage”, R2 is the known load resistance (don’t use headphones), the Output Voltage is the loaded voltage. Click Compute and R1 is the calculated output impedance. This can be done using a 60 hz sine wave file (Audacity can create such a file), a Digital Multi Meter (DMM), and a 15 – 33 ohm resistor. Most DMMs are only accurate around 60 hz. Play the 60 hz sine wave file and adjust the volume for about 0.5 volts. Then attach the resistor and note the new voltage. For example, 0.5 volts with no load, and 0.38 volts with a 33 ohm load gives an output impedance of about 10 ohms. The math is: Zout = (Rload * (Vnoload - Vload)) / Vload

REACTIVE LOADS: Few headphones represent a purely resistive load that’s constant over the audio band. Instead, they’re reactive loads and represent a complex impedance. Because of the capacitive and inductive elements in headphones their impedance changes with frequency. For example, here is the Super Fi 5’s impedance (yellow) and phase (white). The impedance is only 21 ohms below about 200 hz. Above 200 hz it climbs to nearly 90 ohms at 1200 hz and then drops down below 10 ohms at 10 Khz:

UE-superfi5-free-air-impedance-phase

FULL SIZE CANS: Some are not interested in IEMs like the Super Fi 5’s so here the impedance and phase for the popular Sennheiser HD590. It still varies from about 95 ohms to nearly 200 ohms—a range of 2X:

Sennheiser-HD590-Impedance-and-Phase

THE MATH: Earlier a graph was shown demonstrating about 12 dB of frequency response variation for the SuperFi 5’s driven from a 43 ohm source. If we take their rated impedance of 21 ohms as the reference level, and assume a 1 volt source, the voltage at the headphones will be given by:

  • Reference Level: 21 / (43 + 21) = 0.33 V and we’ll call that 0 dB.
  • At their minimum impedance of about 9 ohms it’s 9 / (9 + 43) = 0.17 V = – 5.6 dB
  • At their maximum impedance of 90 ohms it’s 90 / (90 + 43) = 0.68 V = +6.2 dB
  • Total Variation = 6.2 + 5.6 = 11.8 dB

DAMPING LEVELS: The damping of a headphone driver, as explained earlier, is either entirely mechanical damping (Qms) or a combination of electrical (Qes) and mechanical damping. The total damping is known as Qts. How these parameters interact at low frequencies is explained by Thiele Small modeling. Damping can be generalized into three categories:

  • Critically Damped (Qts = 0.7) - This is widely considered ideal as it provides the deepest bass extension without any frequency response variations or excessive "ringing" (uncontrolled driver motion). The bass from a critically damped driver is often described as “tight”, “quick”, and “clean”. Q of 0.7 gives what most consider the ideal transient response.
  • Over-Damped (Qts < 0.7) - This keeps even tighter control over the driver but at the expense of less deep bass (the response rolls off sooner). So manufactures rarely intentionally over damp their products.
  • Under-Damped (Qts > 0.7) - This trades off some low bass extension for a peak at higher bass frequencies. The driver is also no longer well controlled and exhibits excessive "ringing" (i.e. it doesn't stop soon enough when the audio signal stops). Under-damping creates frequency response variations, less deep bass, poor transient response and an upper bass peak. Under-damping is a cheap way to provide the illusion of more bass at the expense of the quality of the bass. It's frequently used in cheap headphones and speakers to provide "fake bass". Under –damped headphones/speakers are frequently described as having "boomy" or "sloppy" bass. If your headphones were designed for electrical damping, and you use them with a source impedance greater than 1/8th their impedance, you will get under-damped bass.

TYPES OF DAMPING: There are three ways to damp the driver and control resonance:

  • Electrical Damping – This is known is Qes and it’s something like regenerative braking on on hybrid or electric car. When you hit the brakes, the electric motor slows the car by turning into a generator and sending the energy back to the battery. A driver in a headphone (or speaker) can do the same thing. But as the output impedance of the amplifier goes up, the braking effect is greatly diminished—hence the 1/8th Rule.
  • Mechanical Damping – This is known as Qms and, as explained earlier, it’s more like the shock absorbers on a car. As you add mechanical damping to a driver, it resists the musical signal driving it, and becomes more non-linear. This increases the distortion and degrades the sound quality.
  • Enclosure Damping – The enclosure can provide damping but this usually requires either a sealed enclosure, one with a tuned port, or one with a controlled restriction. Many of the best headphones, however, usually are open backed. This largely eliminates the headphone designer’s option of using the enclosure to provide damping as is done with speakers.

EAR CUP LOADING: For headphones that form a fairly consistent seal, such a fully circumaural over the ear headphones with earpads that fit snuggly against the head, the designer can somewhat rely on the “enclosure” formed by the ear cup to possibly provide some damping. But the shape of heads, ears, type of hair, headphone positioning, eyeglasses and other factors make it highly variable. And this option isn’t available at all for all the supra-aural (on the ear) headphones. Here are two graphs of the Sennheiser HD650 impedance. Note the open air bass resonance peaks at about 530 ohms but drops to 500 ohms on a simulated head. This is due to damping provided by the ear cup enclosure and the ear pad.Sennheiser HD650 Impedance & Phase New Open Air

Sennheiser HD650 Impedance & Phase New Simulated Head

FINAL WORDS: Hopefully I’ve made it clear the only way to get consistent performance between headphones and their source is to follow the 1/8th Rule. While some may prefer the sound using a higher output impedance, that’s very specific to each particular headphone, the particular output impedance, and the person’s own subjective tastes. Ideally a new standard should be developed and manufactures should be encouraged to design headphone sources with an output impedance below 2 ohms.

180 comments:

  1. What happens if you try to drive high impedance headphones from a portable device that doesn't have a high enough output voltage? Can you ruin your headphones or device?

    ReplyDelete
  2. Good question. I added the answer to the article. It's almost impossible to damage any source device by using the wrong headphones. Some devices might have so much output they could damage certain headphones, but only if you crank up the volume way too high.

    ReplyDelete
  3. Awesome article again.
    Just curious , if the triplefi 5 has such a crszy effect on a regular ipod what will the triple fi 10 do ? would it be better to have a portable amp like the Fiio E7 coonnected between the ipod and the triple fi 10 ? the Fiio E7 amp would have a high fixed impedance and allow the ipod amp to work well without having to deal with the crazy tripfi 10(3 armatures) directly ?

    ReplyDelete
  4. Glad you liked that article. Three things to help answer your questions:

    1 - If you just want an amp for an iPod, consider the much cheaper FiiO E5 rather than the E7 which is also a DAC (and much bigger).

    2 - I'm hopefully going to be testing the FiiO E7 soon. Assuming it has a near zero output impedance like the cheaper FiiO E5 (and no other major problems) it should be a good choice for any of the TripleFi IEMs.

    3 - I haven't tested any, but I've been told the 4th generation iPod Touch and iPhone4 have a much lower output impedance (< 2 ohms) than my 3rd gen Touch's 7 ohms. If that's true, they can drive any of the TripleFi's just fine directly. I also don't know about the Nano, Classic, etc.

    ReplyDelete
  5. thanks for the reply..
    I will assume that the Line out of the IPod is a safer option than a headphone out.
    Now for this line out again there are some options out there and Im not sure if it worth 330$ to get an Imod(from Red wine audio) and corresponding dock(from ALO) ?
    Just atleast theoretically by looking at the advertised changes for an IMOD do you think its worth it ?
    http://redwineaudio.com/products/imod
    I know we wouldn't know exactly abt the performace till you actually measure it ..but still any opinions from a technical aspect would be great as I cant find any technical reviews of this..

    ReplyDelete
  6. In theory, the iPod's line out may have an advantage. In practice, for a variety of reasons, I doubt it does.

    First, the iPod is powered by a single 3.7 volt Li-ion battery. This is the main limitation for the headphone output and the line output. To do a line output correctly you need a higher voltage power supply, and ideally a "split rail" dual power supply (i.e. plus and minus 9 volts like in a cmoy amp).

    Second, the headphone output already measures very well--at least on my 3G Touch. For driving an external amp it should work just as well as the line output.

    Finally, if you really want to spend lots of money to extract the best sound from an iPod, I'd suggest an external DAC solution like the Wadia iTransport, HRT iStreamer, etc.

    ReplyDelete
  7. Hi ,looking forward to the E7 review .wonder how this small guy measures technically ..
    Also have you hear of the Schiit Asgard headphone amp ? do you think a Fiio E9 could be as good for 600 Ohm headphones ? they cost nearly the same..200$ for fiio E9 and 245$ for the Asgard
    I also noticed the Behringer Amp800 headphone amp for around 120$ any thoughts if this Behringer can be as good or even better than these 2 ?

    ReplyDelete
  8. I don't know much about the Asgard besides it looks nice. Single ended designs tend to have rather high distortion and other problems including a relatively high output impedance. Zero feedback makes both of those problems even worse. So my guess is the Asgard is more a "novelty" product that's designed to sound very different on purpose--in this case "different" being significantly less accurate than even a $20 FiiO E5.

    I haven't tested the E9, but the specs look very good and I've been impressed with FiiO's other products.

    The Behringer I'd stay away from. It's not really made for this application and has a 47 ohm output impedance. It's designed for live monitoring while performing which isn't a very critical application.

    ReplyDelete
  9. -wow, how much people will pay for a colored amp instead of just using the equalisers to get a diff sound :-)
    -Ok, I kind of understood the issues that accompany 47 Ohm output impedance..and behringer being intended for a diff purpose etc.

    Another popular desktop amp I keep hearing about is the Headroom desktop Amp for 349$
    does that "theoretically/design wise" seem like a neutral sounding amp as compared to the Fiio E9 for hi impedance cans ?

    ReplyDelete
  10. There are an almost endless number of desktop headphone amps out there. It might not weigh 10 pounds, but products like the FiiO E9 can, at least in theory, approach the point of diminishing returns in terms of accuracy and sound quality. I haven't tested the E9, but assuming they didn't get anything seriously wrong, it should work well with nearly all headphones--high impedance or otherwise. It has a lot more output than the E7.

    ReplyDelete
  11. "I haven't tested the E9"

    I hope you will test it soon.

    BTW great blog, love it and thank you for your hard work!

    ReplyDelete
  12. Hi NwAvGuy, great write-up like always.
    I'm trying to compare the output power of my Xonar D1 soundcard to my FiiO E7, however I'm encountering a problem:

    ---Xonar D1---
    Spec says: Output/input Full-scale Voltage - 2Vrms (5.65 Vp-p)
    Thus, using P = (V^2)/R for a 300Ohm load (e.g. my HD600's) = 13mW
    50 Ohm Load = 80mW
    etc.

    ---FiiO E7---
    Spec says:
    16 Ohm Load = 150mW
    300 Ohm Load = 16mW
    Thus using Voltage = sqrt(power*impedence)
    16 Ohm Load ~ 1.55V
    300 Ohm Load ~ 2.19V
    Therefore, Voltage ~ 1.9Vrms ~ 2Vrms

    Hence, Xonar D1 ~ FiiO E7

    Is this in theory correct?

    Many thanks!

    ReplyDelete
  13. I suspect you're comparing the Xonar's line output specs with the E7's headphone output specs. That's not a valid comparison.

    Line Outputs typically drive 10,000 ohms or more. So most get really unhappy trying to drive say 300 ohm Sennheisers. Put another way, it's highly unlikely the Xonar can manage 2 volts RMS into 300 ohms via the line outputs unless they included a low impedance, high output, headphone amp on the board.

    Well designed headphone outputs can also server as a decent line out but it costs a fair amount of money to implement one that way so few manufactures bother. The reverse, line outputs trying to drive headphones, almost never works very well. Even the M-Audio Transit, which claims to do both via a single jack, does a poor job with headphones.

    There's a graph in my FiiO E7 review (about half way down the page) that shows the output voltage vs distortion into 15 and 150 ohms. It shows 1.87 V RMS at the 1% THD point (clipping). You could expect about the same 1.9 volts into your Sennheisers. For some, that's enough power for the HD600's, for others it's not enough. The FiiO E9 would be a much better match for the HD600's.

    If you can point me to a link for the full Xonar D1 specs, I might be able to better answer your question? Feel free to contact me privately if you want.

    ReplyDelete
  14. Hi NwAvGuy,

    You may be right, however, I am still not sure. Here is a link to the Xonar D1 manual: http://bit.ly/jqLTlW
    It has the audio processor + converters, but I couldn't find the op amp there.

    It is said that the op amp for the headphones is the Texas Instruments R4850, whilst the JRC NJM5532 handles the other outputs, in this overclockers review: http://bit.ly/ikoiMm

    I couldn't find the R4850 on google or Texas Instruments' website however.

    As for getting the E9:- I am tempted as it provides more power for the HD600's, but I haven't found a song where I need more volume. I am wary of the claims that amps can increase the soundstage and so forth. Would the E9 provide a discernible increase in sound quality in your opinion?

    Thanks!

    ReplyDelete
  15. NullZero, Thanks for the link. I don't see anything in the Xonar specs that implies they did anything special for driving headphones directly.

    The PC hardware reviewers are great for running CPU benchmarks, torturing video cards, etc. but they're generally not electrical engineers. Most don't understand the finer points of things like op amps, output impedance, etc. Few even realize all the problems with using RMAA for measurements.

    The "R4850" might be the JRC 4850 but it's hard to say. They probably read the number off the top of the part which, due to space limitations, are often incomplete. There are few op amps that make really good all around headphone amps. Many, correctly implemented, will do a decent job with higher impedance headphones (as in the better Cmoy designs).

    As for the Xonar vs the E9 that's hard to answer as I don't have personal experience with either one. An op-amp-based headphone output that's "working too hard" can certainly mess up the soundstage, add distortion, etc. And a higher output impedance (which the Xonar almost certainly has) will give your HD600's at least a slight bass peak around 100 hz and provide less damping possibly making the bass more "warm" and "boomy".

    The other issue is clipping. The average volume of your HD600's might seem loud enough but the peaks might be distorted by the Xonar--especially if it's not really designed to directly drive headphones.

    But without testing a Xonar I can't say for sure. And, while the E9 looks really good online, and I've been impressed with the E5 and E7, I can't say for sure it doesn't have significant problems. I've had a lot of requests to test an E9 so perhaps one will find its way onto my bench soon?

    My general rule applies here. If you think your HD600's sound great now. I wouldn't spend a bunch of money to upgrade your set up. The E9 is cheap by audiophile headphone amp standards, but if you're already happy it might be a waste of money to upgrade. Normally I'd suggest the E5 might be a cheap way to find out but it's also a bit marginal for something like the HD600's.

    ReplyDelete
  16. Hi NvAvGuy,

    I suppose we won't be able to pinpoint the design of the circuit. You're very perceptive regarding the peak at 100Hz. It would certainly explain why I am getting more of an impact in that frequency response area.

    I will probably get the E9 just to make sure that the signal is coming through cleanly for my HD600's. It'll also tell me whether it really makes a difference as many people say.

    I look forward to a potential E9 review, and many thanks for kindly helping me!

    ReplyDelete
    Replies
    1. It is a very late reply, but in case anyone is interested, the Xonar D1 and DX have an output impedance of 100 Ohms. The maximum output voltage is slightly less than 2 Vrms. The card can produce the maximum output at 0 dBFS with at most about 0.0015% distortion (based on loopback tests only, so this may be inaccurate) in the 20 Hz to 20 kHz range when loaded with 220 Ohm resistors. With the same resistors, a 20 Hz tone is attenuated by 0.16-0.17 dB relative to 1 kHz (0.07-0.08 without a load), but this roll-off is partly due to the AC coupled input in the loopback test. So, apparently at least some effort was made to make the output usable with headphone loads, but the 100 Ohm impedance is generally too much. Unfortunately, these cards are often recommended for use with low impedance headphones, without knowing or understanding the output impedance issues.

      Delete
    2. If the Xonar D1 and DX use the TI 6120 headphone amp (like the STX, etc.) I would be very surprised if the distortion is only 0.0015% at 20 hz. I have tested three different headphone DACs/amps using the TI chip (see my review of the FiiO E9 and QRV amp for two of them) and the 6120 seems to have a significant problem with low frequency distortion.

      If you're quoting RMAA results, then all bets are off as RMAA often has significant issues properly calculating distortion vs frequency.

      But, regardless, Asus either doesn't understand, or doesn't care about, the benefits of a low headphone output impedance. The TI chip requires at least 10 ohms in series with the output so I can somewhat understand 10 ohms, but 100 ohms causes big problems with far too many headphones. It's especially odd when Asus also provides a software selectable "limiter" feature which removes the main argument for a higher output impedance.

      In my opinion, it's really poor engineering--especially for an allegedly "high end" premium product. Despite Xonar products being criticized in reviews by Stereophile and others, the continue to churn out new products that still have output impedance problems. They either don't get it or they don't care. But, in fairness, audio isn't their core business. If Emotiva started selling laptops they would probably have some serious design flaws as well.

      Delete
    3. I do not know what chip is on the front output of the lower Xonars (D1, D2, etc.), but it is perhaps not the TPA6120. It might be just some kind of op-amp with relatively high current output capability.

      While the tests I have done are loopback based, I did not use RMAA, and I did look at the spectrum of the signal for the level of the distortion products, rather than just quoting calculated values.

      The high output impedance is indeed an annoying flaw, since the cards are otherwise quite good for the price. Although specs do not advertise a headphone amplifier, and these seem to be geared more towards multichannel speaker setups.

      Delete
    4. The front output on the Xonar D1 is apparently driven by a JRC 5532 op amp, through a 100 Ohm serial resistor and 220 uF capacitor per channel. Other than the use of large capacitors, this looks more like a line output buffer than a headphone amplifier. The other similar cards might use the same output stage.

      Delete
  17. Hey, awesome article. It's nice to get an actual explanation of the stuff we often end up paying so much for.

    An earlier commenter asked about the headphone output vs the line output on the bottom of the iPod; while there may not be a difference when it comes down to driving an external amp, is there a difference in sound quality between the iPod's line output and its headphone out? I've read that the headphone output on the iPod may be 'colored' by its internal amp, and that you'd get a much better sound listening through the line output. Is this true?

    Thanks!

    ReplyDelete
  18. Drew, you're welcome. I'm going to be testing the LOD (line out dock) of at least the Touch 3G and 1G iPad and write up a short article on it. Not having done any LOD testing yet, I can only say the Touch 3G doesn't add much "color" and I suspect in blind listening you would be hard pressed to detect any. Apple did an amazingly good job with the audio hardware and supposedly the Touch 4G and iPhone 4 are even better.

    ReplyDelete
  19. Hi, great article! Just wanted to ask you a question:

    I'm going to buy Shure SRH-840 headphones (44 ohm impedance; 102 dB/mW sensitivity).

    I just wanted to know if I'm going to need an amp or not. I did some research but I'm not sure how to read all this data and I need a hand.

    Sources will be:

    -Cellphone
    Apparently it's got a wolfson wm8990 audio chip. official webpage reads: "40mW output power into 16 ohm at 3.3V".

    -PC (integrated card on an asus motherboard; ALC892)
    Manual reads: Output Impedance:
    Ampified output:2 ohm.
    Non-amplified output: 20 ohm.

    -Sansa Clip+ (maybe, haven't got it yet)

    -Zoom H4n (data here: http://www.zoom.co.jp/english/products/h4n/)

    Thanks in advance, sorry if I overwhelmed you!

    ReplyDelete
  20. Nick, you've done your research. For a variety of reasons chips often don't meet all their specs depending on how they're implemented. So with your phone and motherboard it's hard to say what the real world performance will be. Both might have a higher output impedance and/or audible noise for example.

    I can say the Clip+ should drive the Shures fairly well unless you like to listen well into hearing damage territory. I looked it up, and the Zoom is rated at 20 mW into 32 ohms. That might be a bit marginal but is likely OK--especially if you're mostly using it for recording.

    My basic advice applies to all of the above. If you really want the Shures, I'd get the headphones and try them out with your devices and see if you're happy. If you want a "reference device", buy a 2 GB $29 Clip+ and compare the sound of your other gear to the Clip.

    ReplyDelete
  21. Would i get a linear frequency response feeding a typical dynamic iem with an ipod whose z=5 ohms?

    For ba iems i use a headphone amp but i was wondering if with dynamic iems i could get rid of the amp and still get a linear fr fed up into the iems.

    ReplyDelete
  22. Dynamic IEMs have a typically much more "tame" impedance than the balanced armature variety. See "WHAT ABOUT MORE TYPICAL HEADPHONES" above for the impedance plot of some dynamic Sony IEMs. So yes, you will probably have fairly flat response with Z=5 ohms from your iPod. I say "probably" because it really depends on the exact headphones.

    ReplyDelete
  23. Hey nice blog man, very informative and clear. 'Bookmarked'

    ReplyDelete
  24. Great article! A question - have you evaluated the effectiveness of cheap headphone attenuators such as the Radio Shack #42-2559? In theory, a low-impedance set of cans could be used with these, which have a 0-470 ohm dual pot inside. Dialing up a few ohms of impedance may work wonders, and allow even the Behringer UCA202 to function better as a headphone amp. I have used it with good results, although I am not able to evaluate it critically.

    The question is: is the attenuator accurate with respect to frequency (many speaker-level volume controls are not), does it add distortion, and/or does it have inherent channel balance issues? At less than $10, it is a good bargain if they are reasonably transparent. Perhaps your test gear could decide? :-)

    ReplyDelete
  25. Devin, any sort of attenuator plugged into a headphone jack, like the one from Radio Shack, generally will increase the output impedance of the source. That's usually not a good thing for all the reasons I mention in this article.

    While a few headphones might sound better to some listeners with more resistance between the headphones and the amp, most just sound worse. The higher the impedance of the source, the less controlled the bass, the more boomy it is, the less accurate the frequency response.

    With extremely efficient headphones, and a source with a really high output impedance (like the UCA202), it might be possible to use an attenuator with low values of resistance that actually lowers the effective output impedance. But the resistance the headphones are connected to would have to be less than the amp's output impedance (47 ohms for the Behringer). Those circumstances are relatively rare. The better solution is to use a proper low impedance source.

    ReplyDelete
  26. Thanks - I guess it makes sense that the increased impedance applies to the source, not the destination (i.e., there is no free lunch...) I am running Grado SR60's, which are 32 ohm nominal, and have the UCA202 coming in the mail, so it is probably not exactly a match made in heaven. I also have a PA2V2 to run off of the line-outs of the UCA202 and then into the Grado cans, which will likely be more satisfactory. Thanks for the info!
    Kevin

    ReplyDelete
  27. It might be nice to have a few suggestions of well designed zero impedence output designs. I own a pair of DT 880 600 ohm headphones and based on my reading and experience these should (and do seem to) pair well with my Creek Evolution integrated amp and it's 220ohm output impedence for the headphone section as well as my Schiit Valhalla amp.

    ReplyDelete
  28. I've been trying, but having a hard time finding headphone amps that I can universally recommend. Many work well with some headphones but far from all headphones. So I took on the challenge of designing my own. You can read the first article here:

    O2 Headphone Amp

    ReplyDelete
  29. Thanks again for the info above re: in-line resistive attenuators. I set up a spreadsheet using the SuperFi 5 data, and the dB deviation (between the 9 and 85 ohm freqs) only increases from 17.5 dB as I add more resistance in series with the headphone load.

    With that in mind, what do you make of this:
    http://andreasodegard.com/2009/10/tutorial-how-to-make-a-resisitor-adapter/
    which has an RMAA graph which shows a smoothing out of the bass response by using a couple 40-ohm resistors in the same configuration (both in series between source and load) as with the Radio Shack attenuator? Here the SPL differences are tamed by adding series resistance.

    Methinks there must be something else going on here? Thanks for any insights/help - Kevin

    ReplyDelete
  30. There IS something else going on in the link above. He's using a substandard MP3 player with a capacitor coupled output. The output capacitor forms a high-pass filter with the headphone impedance.

    I'm guessing his player has 220 uF output capacitors which, with 16 ohms, is - 3dB at 45hz. If you add 40 ohms for a total of 56 ohms, it drops to - 3 dB at 13 hz (and you also lose most of the output power). See:

    http://www.muzique.com/schem/filter.htm

    Modern decent players, like newer iPods, the Clip+, etc. don't have capacitor coupled outputs so his entire article doesn't apply. It's a classic case of something helping in one instance so it's easy to assume it will help everywhere. But the opposite is true here. The 40 ohm resistors will hurt more than they help.

    ReplyDelete
  31. Thanks - this really helps! To get a sense for what my situation really is, I found the impedance curve for my Grado SR60's on Stereophile. As it turns out, the min/max range of impedance over the audible range is somewhat narrow: 30-37 ohms (it peaks at about 150 hz). Using your equations, with the Behringer DAC's 50-ohm output impedance, this would cause a 1.1 dB peak at 150 hz - I think I can live with that, especially since Foobar allows a tweak down (if the subjectivist in me actually thinks I hear it...).

    Corect me if I’m wrong, but this indicates that if you really, really want to use low-impedance headphones, make sure the impedance swings are minimal, and you’ll be OK, as long as the amp can still drive them. Otherwise, as with the Super-Fi’s, the frequency response can really get skewed.

    Do you know if the Behringer’s headphone output has the volume pot on the output? If so, that would make things worse by adding to the output impedance. In that case, I presume I should run the DAC’s headphone volume wide open (pot = 0 ohms), and volume can be controlled on my laptop (either on Foobar or with the laptop’s main volume control).

    ReplyDelete
  32. You're welcome. Volume pots don't work well on outputs (right before the headphones). The UCA202 has fixed 47 ohm series resistors in the output and the volume, like nearly all, is before the op amp.

    You're partly correct about low impedance headphones and wanting a flat impedance curve. But the other big thing mentioned in this article, that has little to do with impedance curves, is the bass damping. If the output impedance is more than about 1/8 of the headphone impedance (also known as a damping factor of less than 8) the bass performance of the headphones will very likely suffer unless they're the rare headphones designed for a high output impedance.

    With a higher output impedance the amplifier has much less "control" over the headphone drivers at low frequencies. This is especially an issue around the resonance frequency of the driver where the driver needs the most control. The electrical damping of the driver is negligible with say a 50 ohm output impedance and 16 headphones (a damping factor of only 0.32 when you want >8).

    ReplyDelete
  33. Gotcha. One other possible avenue (and then I promise I will quit...): as you also mentioned above, tube gear has output transformers to step down the impedance from the tube (>1kohm?) to the speaker (4-16 ohm ?) range. Have you heard of such an animal used for headphones? A 5:1 winding (25:1 impedance) ratio would require no power and could be quite small, and would potentially make a very positive difference. I would imagine, though, that the transformer itself may impart some problems of its own?

    ReplyDelete
  34. Clever segue :-) I look forward to the next O2 article!

    ReplyDelete
  35. Thanks for this - it helps explain why my laptop sounds so poor through my low impedance IEMs. I wish manufacturers would consistently publish the output impedance spec. I'm an intermediate noob - can this be measured with a sufficiently sensitive DMM that reads into the mV range, doing something like this?
    - run a test tone or two, or white noise from an empty part of the FM dial, into the amp
    - measure the unloaded output voltage (assumes a DMM has very high input impedance?)
    - put a (say) 20-ohm resistor on the amp's output
    - measure this loaded voltage across the resistor
    - then do the math

    I don't want to damage the amp or DMM, but it seems I wouldn't. Might this work?

    ReplyDelete
  36. Yeah, that might at least get you in the ballpark but the white noise is going to cause the DMM fits and the reading will not be very stable and likely very inaccurate. But as long as it's consistently inaccurate you should get a relative difference which is all you need.

    A better strategy is to download, or make, a 100 hz test tone WAV file, play it, and measure that. The free Audacity software will let you create such a WAV file. And DMMs are fairly accurate and stable at 100 hz. That's also around the resonance frequency of many headphones so it's where bass damping, and hence output impedance, is most critical.

    ReplyDelete
  37. Thanks for your guidance on this! I finally got my buddy's Fluke auto-ranging DMM and got to work on measuring the PA2V2, which I purchased from Electric Avenues about 15 months ago.

    To first check the set-up, I measured the output of a Behringer UCA202 DAC/amp, which is known to have a 50 ohm output impedance. Using a 100 hz WAV test tone on my laptop as the test signal, and with a half-watt 100 ohm resistor as the load (98.6 ohms, to be precise), the Behringer measured at 50.4 ohms. Sweet.

    Next, I sent the Behringer DAC's output to the the PA2V2 (operating on its charger), and with the 100-ohm resistor load, the output impedance of the PA2V2 was 2.52 ohms. Scaling down to a 11-ohm resistor, which being closer to the PA2V2's value likely made it a little more accurate, I got 3.29 ohms.

    So, it looks like the PA2V2 has about 3.3 ohm output impedance: not terrible, but higher than I was hoping for. Using your 8-times rule, headphones with the PA2V2 should therefore probably be at least 25 ohms or so to get good bass damping and a nice flat frequency response.

    ReplyDelete
  38. Hello.

    It's absolutley obvios that high amp's output impedance affect frequency response curve. In theory and in the prictice. But how can you proof that dumping factor affect sound quality in headphones? Even for loudspeakers is questionable subject.

    For example:
    http://gilmore2.chem.northwestern.edu/faqs.htm
    http://www.audioholics.com/education/amplifier-technology/damping-factor-effects-on-system-response/damping-factor-effects-on-system-response-page-2

    So did you have any proofs or disproof for this subject?

    ReplyDelete
  39. Alexander, in the links you provided the authors are mainly talking about speaker damping. It's true there's not much difference between a 0.01 and 0.1 ohm output impedance (damping factors of 800 and 80 respectively with 8 ohm speakers) like some amp manufactures want people to believe. But when the output impedance starts to approach the speaker, or headphone impedance, damping factor can matter.

    I'm only suggesting a damping factor of 8 or better with headphones. This is consistent with what's in the references you linked which talk about the DC resistance of the driver vs the output impedance. When the output impedance starts to approach to the headphone impedance the electrical damping is greatly reduced. So what happens when I plug my 25 ohm Denons into my 50 ohm UCA202 output? The bass sounds bad.

    From what I understand, old school headphones mainly used mechanical damping as mentioned in the Gilmore reference. But that was back when most headphone source impedances were very high so they were forced to use mostly mechanical damping. It's exactly analogous to the old accordion suspension woofers used with high output impedance tube amps. But solid state amps came along and suddenly speaker designers could depend on the amplifier for damping. That same transition happened in headphones when output impedances dropped with the popularity of the iPod a decade ago.

    Look up Qes at: http://en.wikipedia.org/wiki/Thiele/Small That's the electrical damping component of any dynamic driver--including the ones in headphones.

    The guys who work for the headphone companies have told me they design their headphones to use electrical damping to varying degrees because it allows for better performance. They have also said they would go even further if they could count on more sources having a near zero ohm impedance.

    The old 120 ohm standard has been widely criticized including in the Stereophile reference I provided.

    It's difficult to measure the effect of electrical damping because there's usually also frequency response variations due to a non-flat impedance. It would, however, be interesting to compare impulse response from a low and high impedance source with a microphone and test set up that had good impulse fidelity. I might work on that in the future.

    ReplyDelete
  40. >I'm only suggesting a damping factor of 8 or better with headphones.

    Why exactly 8?

    >So what happens when I plug my 25 ohm Denons into my 50 ohm UCA202 output? The bass sounds bad.

    Can you describe what exactly is bad? What about other frequencies? And what Denons you use?

    >It's exactly analogous to the old accordion suspension woofers used with high output impedance tube amps. But solid state amps came along and suddenly speaker designers could depend on the amplifier for damping. That same transition happened in headphones when output impedances dropped with the popularity of the iPod a decade ago.

    But, for example, professioanl headphones like Sennheiser HD280PRO and others must be designed for high output impedance amps. Mixers and other devices in pro sphere have high output impedance - from 50 to 120 Ohm. And it's strange if my Sennheiser HD595 not designed for 50 Ohm output imedance of my source.

    ReplyDelete
  41. The Damping factor of 8 is somewhat convenient as that also yields less than 1 dB of response variation with typical impedance curves. And 1 dB is around the threshold of audibility. I've seen headphone damping numbers in the range of 5 to 10 from others.

    If you do the math on the Theile/Small models you'll see the "point of diminishing returns" is in that range of 5 - 10.

    I suspect the HD280 is mainly designed for the iPod bunch despite it being a "pro" headphone. And Sennheiser is one of the companies that told me (and Tyll Hertsens) they rely on electrical damping. Tyll (of HeadRoom and InnerFidelity fame) and I agree, by the way, on output impedance and damping.

    Your last point is exactly the problem. There's only one standard in the iPod age that makes any sense for output impedance and that's under about 2 ohms. But headphone designers don't really know what their headphones will be plugged into. It's a mess. And they don't want to tell you what they designed for in the specs because that would limit sales and shine a light on the mess.

    If you look at the "reference standard" headphone amps on the market--the really expensive ones--they nearly all have zero ohm outputs. Check out Violectric (which list damping factors in their specs), the HeadAmp GS-1, Benchmark DAC1, Grace Designs, Centrance, Adenio D1, etc. They all have near zero ohm outputs.

    So if you're a high-end headphone manufacture, it stands to reason, you'll want your headphones to perform their best with the kinds of high-end sources listed above. Not some $2 cheapo headphone output on something like the UCA202.

    ReplyDelete
  42. >Check out Violectric (which list damping factors in their specs), the HeadAmp GS-1, Benchmark DAC1, Grace Designs, Centrance, Adenio D1, etc. They all have near zero ohm outputs.

    As I know, this devices not used in pro/studio sphere. Headphones plugs directly to mixers, DAC etc. which frequently have high impedance outputs. So this is another reason why this situation is little strange for me.

    >So what happens when I plug my 25 ohm Denons into my 50 ohm UCA202 output? The bass sounds bad.

    Can you describe what exactly is bad? What about other frequencies? And what Denons you use?

    >It would, however, be interesting to compare impulse response from a low and high impedance source with a microphone and test set up that had good impulse fidelity. I might work on that in the future.

    It would be very intresting.

    ReplyDelete
  43. Benchmark Media is a pro studio company so is Violectric. Cheap studio gear has high impedance headphone outputs because they're much cheaper to make and "good enough" for typical use--like a musician monitoring the mix while laying down a track. Those devices are typically not used for critical listening.

    I don't think companies like Beyerdynamic, AKG, Sennheiser, etc. are designing their $300+ headphones to be plugged into cheap studio gear. They're assuming something more like the Benchmark, Violectric, etc. Check out Gearslutz.com for what the studio guys use for mastering recordings. It's nearly all "zero ohm" gear as they're after accuracy.

    My Denons are the AH-D2000. And the bass sounds significantly more boomy and less defined when used with a high impedance source. See the write up of the AH-D2000 on the HeadRoom site. They say basically the same thing. And in my Beyer DT770 review I talk about the same sloppy bass when the Beyers are used from a high impedance source (which I think is partly responsible for their reputation for uncontrolled bass).

    ReplyDelete
  44. I just got my Xonar U3 from Amazon. Using my DMM and 100 hz test tone, the output impedance is about 25.5 ohms, so quite a bit higher than I was expecting/hoping for. They claim it is suitable for 32 ohm cans, but apparently they have different criteria. This site has some RMAA measurements for the U3:
    http://techreport.com/articles.x/21256/2
    So, it's not the ultimate headphone amp, but the Dolby surround sure is nice. At any rate, I am going to run it through its paces and see how it performs.

    ReplyDelete
  45. I wasn't expecting much from the headphone output of the U3. My hope is it will make a decent inexpensive DAC for driving the O2 Headphone Amp. Output impedance doesn't matter for that.

    ReplyDelete
  46. Since I listen mostly to classical music, under-damped bass is typically not a problem for me, and sending the output of the U3 to the PA2V2 doesn't seem to make any noticeable difference on my 32-ohm Grados. I found/stole a pair of refurb Beyer DT-770 Pro-80's on Beyerdynamic USA's eBay factory outlet store ("bd-usa") for $90 shipped, and the higher impedance will presumably allow for more confident forays into the lower octaves.

    After a few hours of playing around with the U3, the Dolby 7.1 capability strikes me as one huge reason to buy one. Not sure what other players or processors have it, but it is a definite treat. I have the full selection of Dolby Headphone plug-ins set up on Foobar, which is IMHO far better than crossfeed. However, the addition of the 7.1 plug-in on the U3 (on which one can toggle back and forth between regular Dolby and 7.1-enhanced mode) makes a huge difference in terms of focusing the music in front of me, but not putting it at arm's length. For example, listening to applause at a live concert, the 7.1 makes me feel like I'm in the crowd, while the straight Dolby plug-in makes it seem as if I'm watching a video of some people clapping. It's as close as I've come yet to feeling like I am in front of/in the middle of floor-standing speakers.

    I look forward to your review and comments regarding using it to feed the O2!

    ReplyDelete
  47. Thanks Kevin. Some of the better DSP effects can be rather addictive to many.

    I do have a question. Do you know if the U3 is plug-and-play with the native drivers built into Windows/OS X? The Dolby DSP requires special drivers but will it work as a standard audio device without the drivers?

    ReplyDelete
  48. Yeah, they are addictive, especially for music that needs help with the ping-pong from extreme stereo. The U3 allows one to simulate how far the seven speakers are from you, and even lets you move around within the "arena" created within them (backwards, forwards, towards the left rear, etc.).

    As for the U3 plug-n-play, I first plugged the U3 in, and after a few moments the Windows balloon popped up to inform me that my new USB device was ready to go, which I thought was interesting. However, I didn't listen to it, but went right into installing the driver package off of the CD. Alas, I didn't have music running, so I don't know if it took over as the sound card. So, it could be that the basic DAC/amp without all the DSP toys is ready to roll in Windows (I have Windows 7 home edition on a little Asus netbook). Not sure how usable the U3 is without the user interface, though, since there are no volume or any other controls on the U3 chassis itself. Maybe you can give the barefoot approach a go when you get yours?

    ReplyDelete
  49. If windows didn't pop up a further dialog that means it has native support which is good (and expected). I'll be testing it that way.

    As for the UI, the regular Windows controls should work fine in the usual way.

    ReplyDelete
  50. I just tried the U3 on our Mac OSx machine, and it played just fine without installing driver, although I had to go into System Preferences and select it, since it didn't automatically take over from the internal sound card.

    I then convinced my kid to let me plug it into their Vista laptop. The "USB Device Ready" balloon came up, and I heard a click over the headphones as if the connection was made, but nothing would play, even after going into the Control Panel to make sure it was enabled. I tried the speaker test, and that worked fine (almost blew my ears out), but nothing from my kid's Itunes that just kept playing on the internal sound card. I'm not totally up on how to force Vista to select a sound card, but didn't see anything that would work. Suggestions welcome...

    ReplyDelete
  51. There's a link on the upper right side of the blog about computer audio that has some useful info. Vista has the capability to play different sources over different audio hardware which is a nice feature to keep all those windows sounds out of your music. It's improved in Win7. Sometimes it's selectable in the player (such as Foobar). With iTunes for Windows I have no idea as it won't even index my music collection without crashing.

    ReplyDelete
  52. Thanks - upon plugging the U3 in again this morning, I immediately got the standard "ba-bump" system sound over the U3 that indicated something had been plugged into the computer. Still, nothing from Itunes, Youtube, etc.

    However, I went into Control Panel-Sound and randomly selected 48 khz DVD-audio, and the audio from the Youtube video that was playing was suddenly co-opted by the U3, which tells me that Vista knew the U3 was there, but needed some sort of positive message to start sending it anything but system sounds. At any rate, it is certainly natively supported as a basic DAC with standard Windows audio controls available.

    ReplyDelete
  53. Hi NwAvGuy!Congratulate to your blog!One of the best what I have ever ridden! Anyway I would like to ask you about this impedance stuff. I am a mechanical engineer, but I have a great interest on electronics, I am also doing some radioamature stuff. As far as I know, the highest power can only be taken out when the output and the load impedance mach. For example if I have a Sennheiser headphone with 62 Ohm load, the output of the amp should be 62 Ohm too. Is the 1/8 rule and the dampind is more important than the impedance mach?

    ReplyDelete
  54. Thanks for the comments. Daniel. The "impedance matching" you refer to is not applicable at audio frequencies with direct coupled audio circuits. What your talking about is true at RF frequencies where you get reflections, standing waves, etc. from improper termination. It can also be true where coupling transformers are involved. But modern push-pull solid state audio amps with NFB, by themselves, typically have an output impedance of a few ohms or less and behave as a voltage source. If you add series resistance to the output, as many manufactures do for stability reasons, short circuit protection, or headphone protection, it only reduces the power to the load.

    With your 62 ohm headphones, if I add 62 ohms in series that cuts the voltage to the headphones in half which cuts the power delivered by a factor of 4. So you only get 1/4 the power versus a "zero ohm" source.

    ReplyDelete
  55. I don't get the part about matching the impedances for best volume performance. In engineering classes we're taught about maximum power transfer and that the output and load impedance should be similar in order to satisfy the theorem. On the otherhand if headphones respond to voltage changes, shouldn't the output impedance be very low, and the headphone's impedance be very high? This is very confusing.

    ReplyDelete
  56. See the previous comment Anon. All you need to know about output impedance is it should be less than 2 ohms. Or if you want to be more exact, less than 1/8 the headphone impedance (i.e. 32 ohm Grados need 4 ohms or less output impedance).

    Another example: Most anything designed to drive high fidelity speakers (amps, receivers, etc.) have an output impedance less than 0.2 ohms. Yet the speakers are nearly all around 8 ohms. If it was better to match the impedances don't you think those designing the amps, etc. would? But that's not how it works with audio amps and loads.

    ReplyDelete
  57. "Higher impedance headphones also require less current to drive and that typically reduces distortion in multiple ways."

    That contradicts Ohm's law though?

    ReplyDelete
  58. In what way does that contradict Ohm's law? I = V/R... as R increases, I decreases.

    And, as all my reviews show, virtually all amplifiers produce less distortion into higher impedance loads.

    ReplyDelete
  59. Hi,
    Great Blog...

    And about litle dot products, u know de output impedance of Little Dot MK1 and MKV?

    ReplyDelete
  60. I have no idea. From what I've seen of Little Dot products they don't follow any consistent design philosophy. They seem like a company trying to profit from the latest myths, fads, and flavor of the month.

    ReplyDelete
  61. I have a question about output impedance. This is about my Little Dot MK IV and my HD 650.

    Output impedance of MK IV: 600Ohm (not confirmed by measurement)
    Output voltage of MK IV: ~36V
    I found this by finding the voltage for which a load impedance vs. power best fit the manufactures specs. (i.e. 500mW into 300/600Ohm, etc.) I also conformed that this voltage was more or less correct. I measured a maximum of 46 volts, but that was with the volume pot of my DAC at max.

    If we then define the function for power as f(Z,R) = (36*Z/(Z+R))^2/Z.
    Where Z is load impedance and R is output impedance

    Then finding the range of load impedances of the HD 650 as frequency changes. This is roughly 300-500Ohm.

    If we then compute log(f(300,600) / f(500,600))*10, we get 0.48dB. Which is the maximum variance of power with frequency. Note that this is an inaudible difference.

    If the amplifier would have zero output impedance: log(f(300,600) / f(500,600))*10 = 2.2dB, which is plainly audible!

    Is it just me, or would a large output impedance be better?
    Or is that these headphones are designed to have this larger variance in power output? I would imagine constant power across the entire frequency band is desired.

    Help would be appreciated :)

    ReplyDelete
  62. Sorry, but your analysis is flawed and I suspect some of your measurements/assumptions might be as well.

    First of all, by definition, a zero ohm output impedance will always have perfectly flat frequency response. There's no resistance to form a voltage divider so the voltage cannot change with frequency.

    Second, if the Little Dot really does have a downright awful 600 ohm output impedance, if it's putting out 1 volt (to make the math easy) you get:

    300 / (300 + 600) = 0.33 volts at 300 ohms

    500 / (500 + 600) = 0.45 volts at 500 ohm peak

    20*LOG(.45/.33) = 2.8 dB change which is plainly audible.

    ReplyDelete
  63. Hmmm, I figured it couldn't be right.

    But one thing though, you just calculated voltage the same way I did, i.e. Vh = Vo*Zh(Zh+Zo)
    Where 'h' indicates the headphone, and 'o' the output of the amplifier.

    But to get power shouldn't you then do P = V^2/Z?
    Isn't it true that volume depends on power? Or does it depend on voltage?

    I would argue that it depends on the rate of change of current (F = B*i*l sin(theta)), but isn't increasing the power analogous to increasing the current with the same factor? I'm confused.

    I don't think my measurements are wrong though; they seem to fit the power vs. impedance data given by the manufacturer perfectly.
    I do realize that I'm assuming the amplifier can deliver infinite current (which it obviously can't), as well as ignoring the 1% THD mark and just calculating theoretical power output.

    ReplyDelete
  64. In this case, for a given impedance at least, power and voltage are effectively the same. Frequency response is nearly always measured as voltage. And that's what we're talking about here.

    I suspect your measurements are off because the HD650 is near the threshold of pain at about 2.5 Vrms so your claimed measured voltages of 36 - 46 volts, even with a 600 ohm output impedance, still mean 12 - 15 Vrms to your 300 ohm headphones. And it would be 18 - 23 Vrms to 600 ohms headphones. Those are dangerous values and you would never get the volume control past about 20% with your Sennheisers which are not especially sensitive.

    Even a 100 watt per channel home audio amp can't manage 36 Vrms of output! If the Little Dot really has that much output that's another reason to avoid it besides the high output impedance.

    ReplyDelete
  65. I measured it without a load at max volume. i.e. I connected a male-male 3.5mm TRS cable and I measured the AC voltage of a 60Hz sine wave between the ground and the left channel.
    On normal listening levels (9 - 10'O clock) it was about 2.6V. I would never ever listen to this amplifier at maximum volume, as I'm pretty much guarantied that the drivers would blow.

    I have about six 75ohm resistors available for sacrifice.
    Should I connect an e.g. 150ohm load to the jack and then measure it again?
    And how would you do that? Make a 150ohm connection between each channel and the ground? Or would only one channel suffice?
    It doesn't matter if the resistors or cable die in the process, as long as my amplifier is spared.

    Thanks for your help.
    -Tilpo

    ReplyDelete
  66. I've seen DIY cables with inline resisters which only add output resistence if I've read everything (including comments) correctly.

    So what happens if I add the resistance only to the common/ground wire of my headphone cable? I feel like it's putting something at risk, but I can't say for sure. Will it have the desired effect of driver suspension?

    ReplyDelete
  67. Adding resistance between the headphones and jack won't hurt anything except the sound. If you add resistance to the common wire it will severely degrade the crosstalk. With 32 ohm headphones, using a 32 ohm resistor will degrade the crosstalk to only -6 dB (when it should be more like -60 dB).

    ReplyDelete
  68. Simple question: Is it possible for the output impedance to increase with the volume? As in, the higher the volume, the higher the output impedance?

    Thank you very much.

    ReplyDelete
  69. Soviet123 not in any way I'm aware of but there are some pretty weird tube and single ended designs out there that barely behave like audio amplifiers. So I suppose anything is possible.

    ReplyDelete
  70. So Basically I have a Westone 4. and the impedance is 32. and I'm using my phone as source however i cannot find info on the output impedance. should I just assume that it's ok? or should I buy a new source if yes what do you suggest? price is anything below 300. and since it's an IEM most people told me I dont need amps is that true?

    ReplyDelete
  71. @Anon, without someone testing your phone's output impedance there's no way to know. The $20 FiiO E5 (or likely the new E6--stay tuned for a review) would be an inxepensive way to find out if an amp can improve the sound of the Westone's. Being a B.A. design they really need a low output impedance to perform correctly.

    ReplyDelete
  72. Hi i was the anon before that posted, I had an E6 it didn't really improve anything because I was using it with the headphone jack and I was told that's bad because its amping the already amp'd signals or something.

    my phone is Sidekick4g and I can't really find out the output impedance, do you think something like the clip+ is good for W4s?

    ReplyDelete
  73. The Clip+ (and likely the new Clip I'll be reviewing soon) would work great with your W4s.

    ReplyDelete
  74. Could you please tell me how to measure the output impedance of a device? I have: a DMM, 75 and 82 ohm resistors (the 82 is more precise with a tolerance of 1%) and all kinds of bulk wires and plugs/jacks.

    Thank you very much.

    ReplyDelete
  75. @soviet123, please see the paragraph above Measuring Output Impedance (towards the end of the article).

    ReplyDelete
  76. Great Write up.....

    Got one question just to be sure

    I have 2 sound cards
    1 rme rpm head phone output impedance: 30 Ohm
    1 rme multiface phone output impedance: 75 Ohm

    THe headphones i have are sony MDR-7506 with
    Impedance of 63 Ohms

    Both the above sound cards are not good for these headphone are they.....

    Can you advise what my options are to correct so i get the right ohms out for the headphones.

    Kind regards Harn

    ReplyDelete
  77. A small amp with a low output impedance like the FiiO E5 or E6 should correct your problem. Most of the pro sound interfaces have 22 - 50 ohm output impedances. Few are really designed (despite some marketing hype) to be high quality headphone DACs. Most are mainly intended for recording rather than playback.

    ReplyDelete
  78. Hi, thanks for a well written blog post.
    I have a pair of denon 2000's which are 25ohms.
    The E6 would probably do the job with these, what do you think of the musical fidelity v can, it is supposed to have output impedance of 5 ohms (which is a bit more than 3(25/8))?

    Do you have any more recommendations in the 100-200$ price range for the denons?

    ReplyDelete
  79. The V-Can would probably be OK. If you already have it I wouldn't replace it. If you're shopping, however, I'd find a lower output impedance. Keep an eye on this blog for other options :)

    ReplyDelete
  80. hi, haven't got the v cans yet. looks like the fiio e7 is one that you might recommend...any plans on testing the new e10 in the near future?

    ReplyDelete
  81. Yes, I have said several times I plan to test the e10 as soon as it's available from a reliable USA dealer.

    ReplyDelete
  82. Just wondering what the ideal amplification for the Audeze LCD-2 rev2 would be? They have a nominal impedance of 60 ohms, and apparently a flat impedance curve. Would an impedance mismatch (say 10.7 ohms output impedance) have any affect on their sound quality?

    ReplyDelete
  83. @DRUB, I'm not sure if the LCD-2 was designed with electrical damping from the amplifier in mind, but if so, it will have tighter more controlled bass when driven with a low output impedance (under 2 ohms).

    As for voltage/power, I'd have to look up the numbers and do the math but I think they need at least around 3 Vrms (150 mW at 60 ohms, 560 mW at 16 ohms or 15 mW at 600 ohms) to play fairly loud. That's more than you can get from a lot of portable amps which is one reason I designed the O2 amp. It can drive the LCD-2 with ease and has an output impedance around 0.5 ohms.

    ReplyDelete
  84. Thank you for the response! I'll be getting your O2 amp in 3 weeks time (someone's building it for me as I'm a buffoon when it comes to DIY circuits/soldering), I'm looking forward to blind testing it against the Asus Essence ST which I'm currently using (it's got the 10.7 ohm output impedance figure I quoted earlier).

    Thanks again for the help.

    ReplyDelete
  85. I've been on the fence on building an O2, but then out of curiosity I decided to measure the output impedance of the source I've been using till now, an Marantz PM-80mkII integrated stereo amplifier in my opinion, one of the last good mainstream amplifiers made in the mid-90s before the surround craze, and the Chinese outsourcing (it's made in Japan). http://www.dutchaudioclassics.nl/Marantz_pm80-mkII_class_A_amplifier/

    I was pleasantly surprised to find out that the output impedance was a very low 4Ohms, and since i'm very satisfied with the power and sound it can produce into the can I own, I'll keep using it, even if I don't receive "audiophile cred".

    Out of curiosity I did the same measurement on a relatively modern Marantz 6.1 surround receiver my father owns, and I was disappointed to find the output impedance measured 90Ohms. That amplifier suffers from other problems too (like large distortion problems when driving a pair of speakers with a 87dBspl/w/m sensitivity at moderate levels, despite the claimed 80W RMS power, as well as a sub-par tuner).

    Even other manufacturers have dropped the quality of their products, and good stereo amplifiers (with all the conveniences like multiple inputs and the like) are hard to find, and I'm puzzled.

    ReplyDelete
  86. @Freongr, as long as you use headphones of at least 24 ohms or so and you're happy with the sound, you're correct you don't need an O2. Sadly, integrated amps and receivers are all over the map when it comes to how they implement their headphone outputs. Headphone jacks are mostly treated as a "checklist item" on such gear and it's rare they spend the money to provide a decent internal dedicated headphone amp when they're often trying shave every cent possible out of their designs.

    These days if you want a good 2 channel amp you're best turning to some of the mainstream but still reasonably priced brands like Cambridge, NAD, and Rotel. But even some of those may cut corners with their headphone jacks.

    I usually recommend staying completely away from the "boutique" brands like NuForce, Peachtree, Vincent, etc. as their products are often designed around rather questionable priorities--like using tubes or intentionally high levels of distortion.

    ReplyDelete
  87. So what would happen if you ran a low impedance headphone on a high output amp? How would this effect the sound?
    eg Hifiman HE-300 on Bottlehead Crack amp.

    ReplyDelete
  88. @pulleyking, the Crack would be very unhappy and have very high distortion. But that's because it's a transformerless tube amp. They hate low impedance headphones.

    The Hifiman planars have a fairly flat impedance curve with frequency (unlike dynamic and especially balanced armature headphones). With a low distortion high impedance solid state amp with a higher output impedance, like say the FiiO E9, they'll sound reasonably OK but the higher output impedance may still degrade the bass damping and limit the maximum volume to be too low (Hifiman planars are mostly really power hungry).

    The Hifiman planars require a serious amp that's not only low impedance but has lots of voltage and current output. Not many amps I know of can do it with suitably low distortion--especially ones under $200 or so. That's one reason I designed the O2 Amplifier.

    The Bottlehead Crack is best with 150 ohms as the minimum and 300+ ohms (like the HD600/650) is even is better. But, like most single ended amps, it's still going to have relatively high amounts of distortion and you'll always be listening to your amp as well as the music instead of just the music.

    ReplyDelete
  89. The HE300 is actually a dynamic, but its low impedance and not much more efficient than their planars so the Crack won't be happy with them either.

    ReplyDelete
  90. According to this thread: http://rockgrotto.proboards.com/index.cgi?board=review&action=print&thread=2565 Beyerdynamic design their headphones for a 120 ohm output impedance - there's a quoted reply in there directly from Beyerdynamic themselves which states this. Further, their A1 amp has a 100 ohm output impedance. In the specs for their DT48, it actually says "rated source impedance: 120 ohms". It's a shame they don't state the source impedance for their other models too!

    ReplyDelete
  91. @Sullivang, I was told by a Beyer employee they design their lower impedance (less than 250 ohm) headphones for a zero ohm source. I can attest my DT770-Pro 80's sound much better with zero ohms than even 47 ohms let alone 120. With their 600 ohm models I doubt it makes much difference. With the 250 ohm models it might if they're really designed for a 120 ohm source. Very few companies offer amps with a 120 ohm output impedance so it seems like a poor design decision and they also lose the ability to take advantage of electrical damping.

    ReplyDelete
  92. I´ve got a rather silly cuestion (I dont know much about electronics) about two pairs of headphones conected in parallel. Say you use an 80ohm and 32ohm headphones at the same time. In my particular example with an Macbook, wich i suppose has about 32ohm output at least. What would one expect to happen. Would there be any changes in output ohms or drop in voltage? What happens exactly. I do notice that there is less volume in both of them, so i suppose voltage suply is divided between the two; but in equal measure, or it depends on relative ohms between the headphones?

    Oh by the way your blog is one of those "internet treasure chests". Im very happy to stumble upon it while in search of a really "expensive" "monitor" headphones and amplifiers for them, without know anything about how do they work(wich is essential). So yeah I wish everyone knew and cared about how their stuff works and why it does so. So there would be less BS marketing.
    Thank you very much.

    ReplyDelete
  93. Parallel headphones is a good question. If you use a "Y" cable both headphones receive the same voltage and the amp sees an impedance of (Z1 * Z2) / (Z1 + Z2). For your example it's (32*80) / (32+80) = 23 ohms.

    Anything over 10 ohms or so is unlikely to cause any harm but most headphone outputs produce more distortion as the impedance drops. The 1/8th rule still applies. For a 23 ohm load you want an output impedance under 3 ohms.

    The big problem paralleling different headphones is they will usually have different sensitivities and impedances making the odds fairly good one will be louder than the other--sometimes much louder. That creates a fight over the best volume setting when two different people are listening.

    Finally, while a Macbook may be designed to drive 32 ohm headphones (i.e. Apple's cheapo earbuds) the output impedance is likely under 10 ohms. I'm going to be testing the latest Macbook Air.

    ReplyDelete
  94. headphones should be durable, because most of the time i buy headphones then their either of the ear stops working or they start producing noise which is very irritating. Best headphones are those which should tackle these problems for long time.

    ReplyDelete
  95. Can I use a EQ(digital or analog) before the headphone output to compensate for excess or insuficient damping? Or is it purely electric/mechanical issue?

    ReplyDelete
  96. @Artyom M, That's a good question! The short answer is no, you can't correct for a lack of damping with EQ. It's a bass quality issue as well as a frequency response issue. EQ can try to correct the F.R. issue, but not the bass quality issue.

    Picture a car with worn out shock absorbers bouncing excessively on a bumpy road. The car doesn't stop bouncing as quickly as it should after hitting a bump. That's what happens to the diaphragm in a headphone (or speaker) driver when there's not enough damping. You get unwanted motion in addition to the music and no amount of EQ can prevent that.

    ReplyDelete
  97. Thank you for your response.
    I thought that electrical damping achieves less "diaphragm exitement" on certain frequencies. So if you feed less signal that makes the headphone resonate most, you effectively make it more damped.
    That what I thought, but I assumed wrong. It has to do with magnetic forces driving the diaphragm and at the same time making it stay in place, isnt it? Resistance?
    I should have to remind myself that ohms are used for measuring "resistance" so I guess I have to start reading about how electricity works.
    This is very interesting.

    ReplyDelete
  98. Just a question on impedance & power specs - when I calculate the open circuit voltage of the FiiO E9, using the power specification @16 ohms load and taking into account the voltage drop across the output impedance, and then use that result to calculate the power into 600 ohms, I get a different result to the specified power at 600 ohms. (I get 68mW, they specify 80mW). I would like to understand why this is.
    (I haven't contacted FiiO yet) It looks like the Graham Slee Solo has some kind of load sensing, so of course I realise my calculation will never work for that kind of amp. However, in the case of the E9, my calculation is just a "bit" off, but I don't know why. \Greg.

    ReplyDelete
  99. @SullivanG, have you looked at my review of the E9 (Junje) an also of the E6 (Nov)? FiiO's specs are sometimes rather, um, optimistic. They make claims for the E6 that are well beyond what TI claims the chip inside can do even in the best of circumstances. Or put another way, they seem to generously like to "round up" with some of their power specs. Normally what you suggest is a safe calculation to make. Going the opposite direction (extrapolating down from a higher impedance) is not.

    ReplyDelete
  100. Thanks, but even when I plug your actual measurements in, I need to use an output impedance of about 12 ohms for the measurements to be consistent with an ideal voltage source with a constant output impedance. This is no big deal for me - I'm just curious to understand that's all.

    ReplyDelete
  101. A very timely article-I have Audeze LCD-2s on order. Your commnent about the Schiit products confused me. What is it about being single-ended that equates to high distortion? By all accounts, the Schiit Lyr is the value leader when it comes to mating with the LCD-2s, with the Violectric V200 being the next step up at close to $1,000. And then things get very scarces unless one goes to something very pricey like a Cavelli Liquid Fire at $3,250! That's if you can wait in line for the next production run. One thing does seem to be clear-the last two years has seen a paradigm shift from conventional voice coil drivers to orthodynamic/planars thanks to HiFi-Man and Audeze and now the amp manufacturers have to play catch-up.

    ReplyDelete
  102. Great article thank you. I have a pair of Audeze LCD-2s on order. They appear to be enigmatic. You can drive them reasonably well with an iPod and yet on the high-end, very few amps interact optimally with them. Among the few (by report on the various boards) are the Schiit Lyr, the Violectric V200, and the Cavelli Liquid Fire. I would love to hear specific recommendations from you but I understand if you don't have personal experience with these cans.

    ReplyDelete
  103. First, I'm not sure which definition of "single ended" you're referring to. Some use that phrase to describe unbalanced (i.e. RCA line level inputs) versus balanced (i.e. XLR inputs/outputs). But, in this case, it refers to a very ancient amplifier architecture that dates back to the beginnings of AM radio. You can look it up on Wikipedia if you're curious.

    Single ended amps only have a small range of relatively linear operation--and even that range produces more distortion than a more modern push-pull design. And when you only use limited, or no, negative feedback as claimed by Schiit, the distortion is even worse.

    I'm not sure where your "by all accounts" comes from, but if it's mainly Head-Fi, you have to remember Schiit is a major sponsor there and there's a strong trend on Head-Fi to have lots of favorable posts about sponsor products--even when they have serious problems such as Schiit, NuForce and AudioGD. Those who speak out against sponsor products, like me, end up censored and/or banned. See: Banned at Head-Fi

    There's are very few reasons to spend more than a few hundred dollars on a headphone amp for 98% of the headphones out there. The O2 Amplifier explains why and demonstrates it's entirely possible. I've also put up a $500 challenge to compare the O2 to any amp that measures reasonably well in a blind comparison, including any of the Violectrics, but nobody has come forward.

    Bottom line: The O2, and the upcoming ODA, will make a pair of LCD-2s sound every bit as amazing as any Violectric for a fraction of the cost. If you want to listen to your amp, by a Schiit or Bottlehead. If you want to listen to the music as the recording engineer intended, buy an accurate amp with a low output impedance like the O2, a Violectric, Grace, Centrance, Adenio, Headamp, Headroom, etc.

    ReplyDelete
  104. Your explanation of "single ended" jogs my memory on that issue. Thanks. I must confess that I am largely ignorant about 1) all things electronic from an engineering standpoint and 2) the history of the headphone boards. I was not aware that you had been banned by Head-Fi. I only recently started perusing the various boards about headphones after deciding to purchase the Audeze phones. For better or worse, I compulsively made the decision to buy the phones before doing my homework at to the optimum type of amp. My current (pun) headphone amp is a Little Dot Mark IV SE which I used to drive AKG701s and Denon AHD-2000s. I was never that thrilled with the Little Dot and don't intend to purchase a Chinese piece of gear again any time soon. The rest of my system is decidely high end-A VPI Classic, Benz Glider, Simaudio LP5.3, Audioprism Mantissa tubed pre-amp, ARC VS 110 amp, Acoustic Zen Adagio floorstanders, and high quality cabling. Head-Fi is the most active forum for headphones(as you likely know), but there are others and I have been lurking there too. It is amusing to me that on Head-Fi only four or five people seem to account for 90% of the posts about headphone amps. It is as if they lack daytime jobs.
    But back to the point at hand. It seems to me that once you start talking about headphones, objectivism is inherently compromised. As just the most obvious example, headphones don't and can't portray soundstage accurately in the manner captured by the recording engineer. Sound directed directly into the ear is never going to sound "real". Hence the camp that tries for cross-feed like Headroom and SPL. But here is the question that lingers in my EE-deprived noggin; how can a battery powered portable amp like your design possibly provide the degree of CURRENT that a pair of cans like the Audeze LCD-2s need to show their best? Perhaps you disagree with the premise of my question. Nonetheless, my instinctive bias is that I need some "heavy iron", an amp with a large transformer and large capacitors that can provide not only a low output impedacne but also plenty of current. I trust you can teach me something here (and I am earnest as can be). In fact, I wonder if I would not be better off purchasing a used 10-15 watt speaker amp and having a tech who knows what he's doing modify it for me for headphone usage.

    ReplyDelete
  105. @Fsonicsmith, the answer to your LCD-2 current question is in this and my other impedance article. Current requirements are just simple math and not magic.

    According to real world tests (see InnerFidelity, etc.,) or the Audeze specs, the LCD-2 needs somewhere around 1.1 volts RMS to hit peaks of 110 dB SPL (as loud as most sane people ever care to listen). That's about 1.6 volts peak which, at a measured impedance of 47 ohms, means they require an ABSOLUTE MAX of 1.6/47 = 0.034 amps of current (34 mA). Even if you want to really hammer them with say 3 volts peak and your pair is only 32 ohms (apparently they vary), that's still only 0.094 amps of current.

    The O2 can produce nearly SIX TIMES the current needed for 110 dB SPL with the LCD-2 (it maxes out around 200 mA or 0.2 amps). So, in reality the O2 will be loafing along not even working up a sweat while you're damaging your hearing with the LCD-2s. A much bigger challenge is the HiFiMan HE5 which needs about 150 mA but that's still comfortably withing the O2's abilities. See:

    O2 Design Requirements

    Your instinctive bias about heavy power supplies etc. is 100% wrong and the O2 is the proof. Speaker amps tend to have excessive noise to use as headphone amps. They can also easily damage your expensive headphones. See also: More Power

    ReplyDelete
  106. I will closely monitor how things go with your ODA and I hope to try one out one day with the Audeze LCD-2s. If it comes to pass that I get an ODA, chances are good that I will be comparing it to a Violectric V200. If nothing else, your articles have instilled in me confidence as to Lake People/Violectric-at least they publish the wide array of meaningful specs that virtually all other head amp manufacturers don't. It seems to me that if they bother to publish the important specs, it's because they understand how critical the key perameters are.

    ReplyDelete
  107. @Fosonicsmith, you're right on the money regarding Violectric. If you have a reasonably generous budget, and judging from the components you listed you probably do, I would encourage you to buy a Violectric amp. I've not tested any but I have been in contact with one of their engineers and, like you, I have a lot of confidence in their designs. The build quality of their products likely exceeds what most will manage with the ODA. And they offer other features in their higher-end models you might be interested in as well. Any of their amps would be a great match for the LCD-2.

    You're welcome to compare the ODA to the V200 but it needs to be done under proper level matched blind conditions for it to be meaningful. If you doubt that, at least consider following a few of the links in Subjective vs Objective.

    There's all sorts of well documented evidence when we know what we're listening to, we all hear differences that don't exist (try the Mcgurk Effect video link for just one example).

    ReplyDelete
  108. hi im sorry but i just dont have a clue here and was looking for some help please.
    i have just purchased the shure se535 earphones and at the moment i am using the with my iphone 4s which to be fair they seem to sound quite good, i was just wondering though what would be the best portable player to match up to these earphones i am only intrested in music quality and dont need fancy features.
    i have been looking at the cowon j3 but i dont know if its going to actually be a step up from the iphone 4s or not?
    any help is much appreciated

    ReplyDelete
  109. @Adrian, Sansa Clip Zip (or Clip+). The Clip measures very well, and with a low output impedance, is a great match for your SE535 balanced armature IEMs. The Cowon players, in my opinion, are overpriced, have poor user interfaces, and the Cowon i7 I measured is outperformed by the Clip in some areas.

    It's OK to keep using your phone as well or the iPod Touch 4G. One of these days I might test the Apple Nano as well. I'll be testing the Sansa Clip Zip soon but word is it's very similar to the Clip+.

    ReplyDelete
  110. Thank you for the quick reply
    am i actually going to notice any difference in sound quality or will it be that small i just wont notice at all?
    i know i probably shouldnt think like this but it seems so cheap can it really offer amazing sound quality after all i paid a fair amount for my earphones the se535's are not the cheapest around so i would like i player that is going to extract the best from them and if that isn't going to be any different from my iphone4s which i currently use then ill just stick with that in the knowledge that what im hearing is the best im likely to hear :)
    thanks again

    ReplyDelete
  111. @Adrian, if you did a proper blind tests (i.e. they levels were exactly matched, no EQ on either device, and you don't know which you're listening to) I would bet you would have a very hard time telling your iPhone 4S from the Sansa Clip+/Zip or Cowon.

    Some Cowon players have higher output than many players so they're popular with heapdhones that need extra power. But Shure IEMs need very little power. But they do need a reasonably quiet source if you don't like hiss. The iPhone/iTouch/Sansas are relatively quiet and I doubt the J3 is significantly quieter.

    Yeah the Clip is cheap, but honestly it sounds very good--check out the reviews. I have all sorts of portable players, but my Clip by far gets the most use. I clip it to my shirt and forget it's even there at the gym. It's way easier than using an armband with a huge iPhone or iTouch hanging off it (or in your pocket).

    ReplyDelete
  112. :0 that made me laugh a bit, again thank you very much just one more question and im done
    with Christmas just around the corner i have about £500 to invest.

    what is the best set up considering i have the se535 already and there staying, im thinking of getting custom ear molds for definite (any suggestions)

    i have been looking at headphone amps and dac... im not very clued up on all this i know the minimum details about them, but just wondering if you was me what would you go for? or would you just grab the custom molds and stick the rest of the money back in the bank because its not going to make much difference either way

    thanks again

    ReplyDelete
  113. Your Shures just need a source with a low output impedance, flat frequency response, and low noise. The Sansa, and your iPhone, already meet the criteria.

    Adding an amp to a portable device is not needed in your case and will probably make things worse noise-wise. I know it's hard for some to believe a $30 player can sound as good as a $300 player, but for the Clip+ with IEMs, it's true. If you still want to spend your money on something expensive that won't sound any better, feel free.

    I can't comment on ear molds. That's a very subjective, and individual thing. It's also way off topic for this article.

    ReplyDelete
  114. thanks for your views, im sticking with my ipod then and my shure se535, thank you for saving me quite a bit of money :)

    ReplyDelete
  115. Many manufactures recommend certain headphone impedances that only go as far as matching the output impedance. For example, the Little Dot MKIII recommends load impedance of 32-600 ohm, but then states it's lowest output impedance is 32 ohm. It seems most manufacturer's recommended load impedances don't take into account the 1/8th rule... Should their recommendations not be trusted or are they making other attempts to resolve voltage drop and electrical damping?

    ReplyDelete
  116. @Adrian, you're welcome

    @TheGsus, they should not be trusted. See my recent Winter Solstice article for a link to yet another argument against high output impedance. It's much cheaper to make a tube amp without the output transformers needed for low impedance loads. So the company behind the Little Dot (and many others) left them out but still wants to sell as many amps as they can.

    The more manufactures restrict the headphones for their amps, the fewer amps they will sell. So they stretch to the truth as far as they dare. The Asian vendors tend to be especially bold about it. iBasso recently got caught, for example, quoting bogus power output specs.

    There's no "magic". Manufactures either need to spend more money on their amps to lower the output impedance and increase the output current, or the performance will suffer with many headphones. Sadly, many of them cut corners anyway and rely on subjective hype, snake oil, and myth to sell their amps. I'm sure many hope nobody will ever measure their products and publish the results.

    ReplyDelete
  117. http://gilmore2.chem.northwestern.edu/faqs.htm

    Is an amplifier's damping factor important to headphone performance?
    With loudspeakers, the lower the amplifier's output impedance, the higher the damping factor into the rated load. Damping factor is given as the ratio of loudspeaker impedance to the amplifier's output impedance. As the theory goes, the higher the damping factor, the better the amplifier's ability to control a loudspeaker's low frequency response (when the motional reactance of the system is at maximum), because the low output impedance of the amplifier allows any back-emf generated by the loudspeaker to be absorbed by the amplifier. That theory has been discharged by members of the audio community as unsubstantiated.
    However, even if the theory were correct for loudspeakers, its applicability to headphones is suspect. John Woodgate, a contributor to The Loudspeaker and Headphone Handbook (1988), had the following to say about the effect of damping factor on headphone performance:

    Headphone transducers are resistance-controlled, not mass-controlled like loudspeaker drivers above the main resonance. In any case 'damping factor' is largely nonsense - most of the resistance in the circuit is the voice-coil resistance and reducing the amplifier source impedance to infinitesimal proportions has an exactly corresponding effect on damping - infinitesimal.
    However, the source impedance affects the *frequency response* of a loudspeaker because the motional impedance varies with frequency, and thus so does the voltage drop across the source impedance. This means that the source impedance (including the cable) should be less than about one-twentieth (not one two-hundredth or less!) of the rated impedance of the loudspeaker, to give a *worst-possible change* in frequency response from true voltage-drive of 0.5 dB.

    The motional impedance of headphone transducers varies very little (or should vary very little - someone can always do it wrong!) with frequency, so the source impedance can be high with no ill effect.

    The IEC 61938 international standard specifies that headphones should be driven by a 120 ohm source - regardless of the impedance of the headphones themselves. If the headphones were designed to this standard, then an amplifier's high output impedance should have little effect on the sound of the headphones. In general, headphones with a flat impedance curve over the audio range will not be affected by high output impedance. For example, in May 1995, Stereo Review published a review of the Grado SR125 headphones. The impedance curve of the SR125s, which have a nominal impedance of 32 ohms, varied from 31 to 36 ohms over the entire 20Hz to 20kHz spectrum. Not all headphones may be as well behaved as the Grados, but neither do they usually have the roller-coaster impedance runs of a loudspeaker. Tube amplifiers (with their high output impedances), it should be noted, have very low damping factors.




    It seems headphone and speaker are different.Some headphones have a flat Impedance Versus Frequency and low Impedance.For example, Denon D2000.http://www.innerfidelity.com/images/DenonAHD2000.pdf

    The playback of Korg MR-2 supports SACD ultra-high quality 1-bit DSD@2.8224 MHz and multi-bit PCM formats up to 24-bit@192 kHz.But its output Impedance is 10 Ohms.And Phone Maximum level is just 20 mW+20 mW @16 ohms. In terms of theory,is Denon D2000 appropriate for Korg MR-2?

    ReplyDelete
  118. Hey, I've been looking at audio interfaces. I have AKG k 240 MkII which says it's rated impedence is 55ohm. The audio interface i've been looking to get is this http://www.roland.com/products/en/QUAD-CAPTURE/
    do you think the headphone out will be able to power the headphones properly?

    ReplyDelete
  119. @Adrian, you're reciting a lot of information. Some of it is accurate and some is pure myth. I'm also guessing you didn't read this entire article. The Stereophile link, for example, helps confirm just how much of a train wreck the IEC 120 ohm output impedance really is. I do the same in the article. The evidence is huge.

    Some newer links (that I really should add to the article) support differences even among lower output impedance headphone sources. Check these out:

    Do Specifications Lie?

    Sonic Advantages of Low Impedance Headphone Amps

    The above articles clearly show output impedance makes a big difference, and not just in frequency response.

    You talk about 1/20th as being a good ratio for output impedance to load impedance (i.e. a damping factor of 20) and I agree. My article says you can probably go as low as 1/8th. But Anon's 55 ohm AKGs on the 22 ohm Roland fail even the 1/8th rule. So do the Denon's on your 10 ohm Korg. And an IEC 120 ohm source would only be suitable 960+ ohm headphones.

    Finally, I can attest even a 10 ohm output impedance degrades the performance of many lower impedance headphones. It's true for my Denon D2000's, it's true for my DT-770 Pro 80's, and it's especially true for all my balanced armature IEMs. I can pick all of those out in a level matched blind test from a 0.5 ohm vs 10 ohm source.

    @Anon, see above. It's a poor match. I have a Quad Capture.

    ReplyDelete
  120. Rock Grotto has a thread
    http://rockgrotto.proboards.com/index.cgi?board=b&action=display&thread=7589
    where owners of the AKG K701 (with bass-shy response) intentionally use high output impedance (100+ ohms) to take advantage of the K701's impedance hump, and resultant bass hump, around 100 hertz:
    http://www.soundstagemagazine.com/measurements/headphones/akg_k701/impedance2.gif
    Damping is out the window, but apparently it smooths out the response.

    You've probably read that thread, which mentions you directly: worried about the slow slew rate on the O2's op-amps, etc. I reckon they didn't read your design article :-)

    ReplyDelete
  121. @Kevin, what little I've read at Rock Grotto has been even worse than much of the misguided, and often just plain wrong, mythology being passed around at Head-Fi. But I suspect if you're willing to wade past all the "noise" there's probably some good information there somewhere.

    To each their own, but most people (me included) find boosting bass around 100 hz to be unpleasant as that's more "boom" than real bass. Lots of small cheap speakers and headphones have a peak around 100 hz and it greatly contributes to making them sound small and cheap. So, to many, using a headphone like the K701 with a higher impedance source would make them sound worse. If their resonance was an octave lower down around 50 hz that would be very different but most full size headphones have a much higher resonance frequency.

    And yeah, those being critical of the O2's 3+ V/uS slew rates need to step up and objectively defend their position with either credible science or blind listening tests. The problem is they can't. I have many very solid references and basic math supporting my arguments on slew rate. They have nothing but their own sighted biased listening, false myths, and personal choices in op amps/gear to defend. Their earth is flat and they like it that way.

    ReplyDelete
    Replies
    1. I agree - anything over 70-80 hertz just makes it sound bloated.

      One thing that Rock Grotto does have is an interesting set of mods for the Superlux HD681, which places a parallel RLC group in series with the driver to partially notch out the pronounced peak at about 7500 hz.
      http://www.rock-grotto.co.uk/HD6812.pdf
      With an amp output impedance of less than 10 ohms (as with my O2 - thank you!), the parallel shunt (with the RLC in series) is not needed. I have done the mod, and it makes the cans quite nice sounding: still bright, which I like, but not piercing.

      Some may recoil in horror at this mod, placing these passive components in the audio path. The objectivist in me thinks that with all of the op-amps, resistors and other components that have gone between the musician's voice and the final CD, these additional passive components wouldn't making any real difference. This is on top of the RLC circuit that makes up the headphone drivers themselves. The type of ferrite material in the inductor might be a topic for discussion, but given that it operates only between 5000 and 9000 hz, I doubt there would be much impact even from this. I am using poly film caps and 1/2-watt wire-wound resistors.

      Your thoughts???

      Delete
  122. Hey, Nwavguy, I've been wondering about a situation I've always wondered about, and reading this post, nobody has addressed it yet, so here goes:

    When you use a DAC/amp combo, and use the headphone output of such a device to run it to an external amp, what ends up being the output impedance of the "amp" in this situation? Example: If you use the original DACport to run it to an external amp, are you essentially "double-amping" or running it through 2 amplifiers? How do you calculate the output impedance in this situation, to make sure it's appropriate for your headphone of choice?

    Centrance says to use the DACport as a DAC/preamp, you should turn the volume knob on maximum, and control the volume/gain from your external amplifier. But is this sending out an actual line-level output, or is just maintaining a certain level output? (sorry if im displaying some ignorance here, dont really know the difference)

    Common sense says its ideal to use a line-out in these situations, but it seems to me that a lot of DAC/amp combos just have a headphone out. I feel like this is a common issue so i'm surprised nobody's asked this question yet (as far as i know)

    ReplyDelete
  123. The reason I inquire about this, is because I'm planning on using the DACport with an ODA in the future. I'm trying to figure out if the ODA would effectively "cancel out" the DACport's 10 ohm impedance, to allow me to use virtually every headphone on the market.

    ReplyDelete
  124. @Luke, the answer is the first amp doesn't make any difference in the final output impedance. It's only whatever the headphones are plugged into that matters.

    Yes, the ODA (or O2) will effectively drop the standard DACport's 10 ohm impedance to near zero. They both also have much more output than the DACport. The DACport is available in a "line out" version if you want to avoid the extra headphone amp in the signal path if you don't mind always using it with an amp.

    ReplyDelete
  125. (this is Luke, from above)

    Okay, great! From what I've seen browsing around the internet, there's some typical audiophile worrying about "double-amping" when you plug a DAC/amp into an external amplifier through the headphone out. So is that just another common audiophile misconception?....Or is there something else other than the impedance that I should worry about, something else that could actually affect the sound quality?

    yeah, I know about the DACport's line-out version (the LX), but since I was able to get the DACport at $350, I figured it was worth the extra $50 to be able to have a portable DACport. DACport LX wouldnt offer the same portability (would need to carry around an amp with it as well).

    again, thanks for the time to respond to my questions, and MANY thanks for this great blog! you've improved the lives of every music-lover who stumbles across this blog.

    ReplyDelete
  126. In general headphone amps may have more noise and distortion than a line output. How much more, and if it's audible, depends on the specific device. The headphone amp in the FiiO E7 DAC, for example, has some potentially significant problems. The one in the FiiO E10 is much better. In my testing the difference between the line out and headphone out on the E10 is very likely inaudible.

    If you already have a DACport, I wouldn't worry about it. The guys at Centrance seem to know what they're doing and the DACport's headphone output measured well when Stereophile tested it.

    ReplyDelete
  127. @Kevin, Blogger seems to have swallowed your comment regarding the RLC network. The short answer is you often do have to worry about ferrites in the audio path but in that applications it's probably somewhat less critical. When I tried to remove the 10 ohm output resistor from the TDA6120 used in the FiiO E9 and other amps and replace it with a ferrite inductor or bead I ran into all sorts of disappointing results due to the non-linearities of the ferrite.

    ReplyDelete
    Replies
    1. Thanks - Blogger probably didn't like the link I inserted. At any rate, I think you're right - the ferrite non-linearity in the octave between 5K and 10K hertz is probably negligible. Foobar's extended graphic EQ (home) and the Clip+/Rockbox parametric EQ (mobile) could take care of the 7K peak as well, but fiddling with gear is fun too :-)

      Delete
  128. Hello NwAvGuy, first of all thank you for your really great work, you're helping newbies like me to better understand this world.

    I'd like to ask you a question if you don't mind:
    I purchased a CMoy from ebay and after reading several of your articles talking about the importance of output impedance, I asked the seller about this measurement.
    He told me he had put an output resistance to eliminate hiss and that the output jack was 56Ohms. Is he referring to the output impedance? I know there's a difference between resistance and impedance but it isn't clear for me.
    The CMoy comes also with a jumper for low and high gain. Does it modify the output impedance? I also asked it to the seller but he didn't clarified me...

    Thank you in advance! Regards.

    ReplyDelete
    Replies
    1. @Anon, it's hard to know. Changing the gain will not significantly change output impedance by enough to worry about for headphone use (i.e. it might change from 0.4 ohms to 0.5 ohms).

      In a basic Cmoy there's usually no output resistor (resistor connected between the op amp output and headphone jack) so they tend to have fairly low output impedance (but many have very limited output current due to the op amp used).

      To get a 56 ohm output impedance the designer would have to add a 56 ohm series resistor to the output. That's a really bad idea for the sound quality of many headphones. As this article explains, you want the output impedance under 2 ohms.

      And, just to be clear, output impedance isn't directly related to hiss and noise. What the designer is probably talking about is a 56 ohm output will reduce the overall level into 16 and 32 ohm headphones (as well as making most of them sound worse). Because the level is reduced the perceived hiss when nothing is playing is also reduced. But the actual Signal to Noise ratio of the Cmoy (the way hiss is measured) doesn't change much because the signal is reduced just as much as the noise forcing you to turn the volume up higher.

      Delete
  129. I've read some (many, actually) comments on head-fi asserting that the "1/8th rule" doesn't apply to planar drivers - only dynamic drivers. On the surface this seems to make sense since the driver doesn't move, thus no damping is required. On the other hand, the amplifier still sees a 50 ohm load so wouldn't there still be an electrical interaction of some kind that could affect frequency response? So how does it work with planars?

    ReplyDelete
    Replies
    1. A driver that "doesn't move" can't produce any sound. Sorry, but you can thank the laws of physics for that. Planars do tend to have much flatter impedance vs frequency than dynamic headphones. So that does reduce the frequency response issues with a higher output impedance. But the issue of damping and distortion are much less clear. I would strongly suggest you have those arguing "it doesn't matter" to read this white paper:

      Sonic Advantages of Low Impedance Headphone Amps

      If they want to dispute the above, ask them to post their measurements documenting their claims.

      Delete
    2. I was one of the people he was asking about that. I know planars don't need to worry about FR changes and they always require mechanical and acoustic damping which will mitigate any loss in electrical damping but that still leaves the article you just linked.

      Do you know the actual cause of that distortion? Comparing the distortion graphs to the impedance curves of the headphones in question shows some similarities but the correlation isn't perfect.

      Delete
    3. It's something I plan to explore further. I suspect some of it is related to out-of-band reactive components of the load. Even a headphone load that is relatively resistive below 20 Khz becomes more reactive at higher frequencies and that may cause some (likely subtle) changes in the distortion performance under 20 Khz.

      Delete
    4. @maverickronin: I cannot say for sure if this is correct, but I think the distortion with high output impedance is mostly just the transducer's own distortion appearing (partly) in the current it draws. This is related to the fact that it also works as a microphone, and if the movement of the voice coil in a permanent magnetic field is affected by non-linear distortion, then it will induce voltage/current that also reflects the non-linear distortion. Electrical damping basically works like a form of (weak) negative feedback, and reduces the distortion, in addition to flattening any resonances. However, unless the damping factor is really poor, the difference is not necessarily large, it might be something like 10% (i.e. one tenth of the driver's distortion at a particular frequency, not 10% absolute THD) with a damping factor of 2.5, but it also depends on the headphones. The Benchmark article shows tenths of a percent bass distortion with an underdamped MDR-V6, but the same headphone has multiple percents of distortion in its acoustic output with loud bass, according to the Inner Fidelity graph.
      Again, all the above is only my guessing, so it may not have much scientific value.

      Delete
  130. Hi Maverick and NwAvGuy!

    I was wondering about this because I tried out a new planar headphone and I heard bloated bass and recessed treble. Now it could just be the way the headphone sounds (can't trust those "reviews"), but it made me curious to know if my amp was a factor.

    By the way, I have an O2 at home now and I like it a lot!

    ReplyDelete
  131. NwAvGuy
    Thanks upfront for any help,
    If there is anyone out there who could help me make a decision I would really appreciate it. I purchased a pair of headphones last month which turned out to be faulty, on returning them to the store they could only offer me a refund as the model was out of stock. A member of staff recommended a pair to me for a similar price around £35, as it turns out I didn't have time to try them, if some one out their have experience of this model Philips SHL5003 I would be thankful

    Paul R
    United Kingdom

    P.S I am new to 'e' correspondence, please forgive any lack of internet etiquette.

    ReplyDelete
  132. I noticed something interesting with my Toshiba NB300 netbook - the output impedance changes, depending on whether it is in headphone mode, or line-out mode. Circa 50 ohms for headphone mode, and circa 170 ohms in line-out mode. When I plug something into it, it pops up a message, asking what type of device has been plugged in. I didn't expect this to actually do anything - I'm suitably impressed! \Greg.

    ReplyDelete
  133. So I'm trying to figure out how to measure the output impedance of a particular device I have since I can't find it anywhere. How do you measure your devices in your reviews?
    Thanks,
    Boyd

    ReplyDelete
    Replies
    1. See the paragraph that starts out, in bold, with "MEASURING OUTPUT IMPEDANCE". I use the dScope but it can be done with a DMM and a test file as described in the article above.

      Delete
    2. You can even do it without a multimeter (although getting one is not a bad idea anyway), using only a sound card and audio editing/analysis software, since only the relative levels matter for measuring output impedance. But the line input of the sound card might have relatively low impedance itself (a few kiloohms), so it can alter the results slightly, as if it was connected in parallel with the output impedance. If you do not even have resistors, then a headphone is also usable as a test load at a frequency where its reactance is low, for getting an approximate output impedance.

      Delete
    3. RE: the input impedance of soundcards, if your soundcard has any inputs that are classed as "instrument" inputs, try to use them, because these inputs typically have a much higher impedance. For example, the instrument inputs on my M-Audio Fast Track Ultra have an impedance of 1M ohm, whereas the line inputs are 10K. \Greg.

      Delete
  134. Do you think a 5 ohms output impedance is low enough to preserve the tonal balance of not wild balanced armatures (By not wild i mean armatures whose lowest impedance is higher than 20 Ohms)

    ReplyDelete
    Replies
    1. It really depends on the headphones. If Headroom, InnerFidelity, etc. has tested them and published an impedance curve, you can get a much better idea of what sort of frequency response problems you might get. Generally it's best to still follow the "1/8th rule" meaning you should use headphones 40 ohms or greater with a 5 ohm source. But if the impedance curve is really flat you could probably get by with 32 ohm headphones or even a bit lower. But there's more to worry about than just frequency response, see:

      Sonic Advantages of Low Impedance Headphone Amps

      Delete
  135. Thanks for the reply. I already went thru that interesting article but for some odd reason i thought that electrical damping was not as hard to achieve on balanced armature transducers because of the technology being different from typical dynamic drivers.

    ReplyDelete
  136. And doing my math what i get is as follows:

    Driving my Shure SE420's out of the 5 Ohm Z source i get an overall 1.01db variation in the frequency response (from DC to 10kHz)the signal fed to the iem.

    Driving my Westone UM3x out of the 5 Ohm Z source i get an overall 1.32db variation in the frequency response fed to the iem.

    Subjectively the deviation is more audible on the SE420 because of it flat frequency response against the U shaped response of the UM3x in which the bump in the mids still leaves them way below the bass end treble levels.

    The Shure SE420 brings a reference sound and will sound as the disc does. It's a loudness war crap detector ;) I think you would love this iem for subjective testing.

    ReplyDelete
  137. Dear nwavguy,
    I've translated your article to a different language:
    http://rinchoi.blogspot.com/2012/03/blog-post.html

    I will delete the translated version if do not approve it. Thank you.

    ReplyDelete
  138. Hi there NwAvGuy,

    I have a question regarding impedance mismatch between headphone Amps. I want to get a discrete Amp for my AKG K702s, and I am considering the Burson HA-160. I would like to pair the Burson with the DAC section of Centrance's Dacport, but I don't know if it is necessary to consider the impedance of both Amps in order to get the best match.

    My dilemma is basically if I should buy the Dacport and get a great portable Amp/Dac for only a 100 bucks more, or if should just go for the Dacport LX and avoid problems.

    A side question I also have, is regarding ALO audio's AKG 702 SXC 22 headphone cable, and its supposed advantages. I have read some comments that you've made in some threads regarding headphone cables and balanced headphones, in the sense of these kind of implementations being mainly marketing hype and snake oil, and as products that don't really produce the improvements they promise and claim.
    I have considered giving the cable a try, but since they cost more than the headphones themselves, I prefer to go for a second opinion before spending the money.

    ReplyDelete
    Replies
    1. I would get the DACmini and forget the Burson entirely but that doesn't solve your portable DAC issue. It's tough to properly drive the K702 from a USB powered device. But the DACport by itself, if you get the special 1 ohm version, might do a semi-respectable job. You could ask Centrance for their opinion.

      I'm not a big fan of Burson because they've been caught misrepresenting their products and most of their success seems to be related to the usual commercial shilling and subjective hype machine on Head-Fi. That said, I haven't personally tested any of their gear. For the prices they charge they should provide measurements and better document the performance of their products.

      The headphone amp in the regular DACport will slightly degrade its performance feeding another amp compared to the LX but I doubt it would be audible.

      Headphone cables are 99.9% snake oil. They're mainly promoted the same way as Burson products. See above. :) Blind cable tests consistently demonstrate people can't hear any difference between a $5 cable and a $500 one when they don't know which one they're listening to.

      Delete
  139. Thanks for your impressively quick answer NwAvGuy. My main interest is really for a desktop Amp that can really drive the K702s. Getting a decent portable Amp for just a hundred more than the LX, is of course convenient, but not really my objective. I have listened to headphones on the go for some 21/2 decades now, but doing it at home is simply much more satisfying, in my experience.

    My research -limited to other people's internet references, of course- led me to the Burson as the ultimate Amp to drive the K702s, and even Srajan Ebaen, from 6moons (who reviewed and awarded the HA-160) suggested -when I asked him- the Burson driving the K702s along with the LX, was a spot on idea. But then again, 6moons doesn't publish any measurements or test results, and they also defend the improvements of ALO audio's cables on several headphones. (Here's Srajan's review of the Burson HA-160: http://www.6moons.com/audioreviews/burson3/burson.html)

    My initial dilemma was between the Dacmini and the Burson + LX, but I found no one that would defend the Dacmini as sufficiently capable or particularly apt to drive current hungry K702s. My primary dilemma is of course that it is impossible for me to get to test any of these components, let alone their combination at any store. But alike you, I prefer hearing the truth -and absolutely defend telling it no matter the consequences- when it somes to spending my money.

    Thanks again for your sincere advice.

    ReplyDelete
    Replies
    1. If you haven't read my Subjective vs Objective article you might want to check it out. The DACmini was recently tested by Stereophile and it measured very well. It's a very solid design that's big on real engineering and low on snake oil hype.

      I also would suggest the same commercial interests are at play behind the scenes at 6moons as at Head-Fi. Look at their sponsors, vested interests, etc. They're not providing unbiased reviews. They're obligated to, as a rule, claim more expensive gear sounds better--even when it doesn't. To do otherwise would harm their sponsors and sources of free gear and income. See the Follow The Money section of the Subjective article.

      It's easy to get wrapped up in the hype. But it's good to understand sighted listening bias (check out the Mcgurk Effect video in the Subjective article), the commercial interests at work behind the scenes, and to compare gear using credible independent measurements instead of unavoidably biased opinions.

      Figuring out what amps can drive your K702s well isn't a subjective art like wine tasting. It's simple math. See my More Power article for more on that.

      I created the O2 Amplifier to be a good match for 98% of current headphones (electrostatics excluded). And if you read the many reviews from happy O2 customers with difficult headphones like the K702, I seem to have succeeded. It's not rocket science, magic, or voodoo like many at Head-Fi, 6moons, etc. would like everyone to believe. It's just solid engineering.

      Delete
  140. I will certainly give your mentioned contributions a read-through. I thank you again for the kind and sincere effort and advice. It's truly hard to come across serious people who aren't trying to dig into your pockets without any scruples. I have read about your O2 and ODAC designs, but only partially so far (not that I have the knowledge to follow you along every line), I will continue reading.

    To me, your concept of proving -with facts and measurements- that it is possible to build an DAC/Amp capable of the very same results of respected and well built gear for a mere fraction of the cost, is highly valuable, or more likely invaluable. I would hope to see this tendency emulated by manufacturers in the future, and show interest in REALLY doing their best, instead of getting rich REALLY quick.

    I know you haven't concluded your ODAC design yet, but I would be willing to wait for that, and the eventual availability of it, from JDS Labs or any other manufacturer brave enough to pick up the challenge (and especially if you give your approval of their final product). For now I'm eager to navigate through the pleasures of my large collection, and I would consider getting one of JDS Lab's O2s (or any other manufacturer you would recommend), if you could recommend a DAC, like for example the Dacport LX to pair it with. I have a growing collection of 24 Bit recordings, and I really want to get my ears on them.

    Thanks again for sharing your in-valuable work and sincerity.

    ReplyDelete
  141. In the related article 'Headphone Impedance Explained', you briefly mentioned Resistance R, and mentioned how it is measured with a DC current. Normally, resistance does not receive much attention, so I'm wondering why. Is resistance important at all when it comes to amplifiers? Is there a desirable amount of resistance for a given audio application? Also, what is the significance of R-load when measuring a device? In the Titanium HD measurements that I showed you, the Audio Precision SYS2722 used an R-load of 10,000 Ohms, that's why I'm wondering.

    Sigh, I really need to get a beginner's book in audio engineering...

    And congratulations on hitting the bulls-eye for the second time with the ODAC. Marking the threshold for transparency with a green line is very helpful for me.

    ReplyDelete
    Replies
    1. You might want to spend some time at Wikipedia if my two articles didn't entirely make sense. In electronics "R" is technically DC resistance and "Z" is AC impedance. For some things, like say a lightbulb or resistor, they're roughly the same at audio frequencies. For a pair of headphones, however, they're often quite different.

      The load is most important when you're testing headphone amps, or speaker power amps. Both of those have performance that can change dramatically with different loads.

      For line-level gear, like a pure DAC (no headphone jack), the load typically is between 10K and 100K. Within that range there shouldn't make much difference if the source gear is properly designed. Some tube and "NFB" (no feedback) gear, however, is very fussy about the line level impedance it has to drive.

      If the Titanium HD's LINE OUTPUT was tested at 10,000 ohms, that's fine. If the HEADPHONE OUTPUT was tested at 10,000 ohms, that's a huge mistake as there are no headphones even close (the highest I know of are 600 ohms). Headphone outputs should be tested at multiple impedances, but if you have to pick just one, 32 ohms is probably the best bet. Amps to drive speakers are usually tested at 8 ohms.

      Delete
  142. hi,
    looking to get a o2 amp from jds i recently purchased a some hd650' and am driving them with a fiio e17 at 6 db gain and they sound good, the specs for the fiio e17 are around 34mw for 300 ohms and the hd650 specifies a max of 500 mw.

    ReplyDelete
  143. im more curious about the mw claims, is that quoted according to output impedance which isn't stated? just have a selection of low/high impedance stuff and only really want the one amp.

    ReplyDelete
    Replies
    1. The 500 mW rating is the damage limit for the HD650, not how much power they need. The HD650 needs about 2.5 Vrms for 110 dB peak SPL. That's 25 mW at 250 ohms. So an amp that can (really) produce 34 mW at an even higher 300 ohms is fine.

      Delete
  144. I've been having trouble with measuring output impedance. I'm trying to objectively determine whether a friend's Ipod Nano 2nd Gen is affecting his 16 Ohm headphones (can't find any info on this factoid on the web). It seems to weaken the bass through my Ety HF5's (testing by using the O2 to A/B) but I suspect expectation bias.

    Following your instructions I get a max voltage at highest volume(using a 60Hz tone generated by Audacity) of .248. When I then put the resistor on the voltage doesn't change?!? Nano volume limiting is switched off. The DMM itself rates itself as accurate down to 2V. Maybe that's the issue. But looking around I see that even £70 DMMs claim accuracy down to only 1V. Is this an accuracy issue? Should any cheap DMM do the job? I should mention that it's auto-ranging and a cheap chinese product. I used it with great success to test my own build of your marvellous O2. But on this one test I am getting nowhere.

    I also tried it on my laptop's headphone out which did reach .5V but again, no change with the resistor. I tried a couple of other resistors also, of slightly differing numbers. Same.

    Any clues?

    ReplyDelete
    Replies
    1. You're probably not reading the DMM specs correctly. It might be the lowest range is only 2 Volts. Or perhaps it has 2 milliVolt (mV) resolution. Sometimes accuracy is also rated in "digits".

      If you are genuinely adding a 16 ohm resistor properly in parallel, and the voltage reading of a 60hz signal does not change, it's likely it has a very low output impedance. But I'm wondering if your resistor is the wrong value? Try reading just the resistor with the DMM. If it's much more than 50 ohms it's not going to have much effect on many outputs.

      If you have some junk headphones you can also try using them as a load with a "Y" cable with the other output going to the DMM.

      Delete
    2. Thanks for the help, NwAvGuy. I've been trying to get to the bottom of this for a while.

      I did test the resistor, at 33 ohms and lately also at 10 ohms (lost the 33 ohm resistor). (The 16 ohms is in reference to my pal's headphones which wasn't part of the test). I also tested my clip+ with the same result. Maybe the clue is 'in parallel'; I don't know what that means. I have the resistor hanging off one of the the 3.5 jack's stereo outputs, then my red probe goes on the end of the resistor and the black probe to 3.5 jack's ground. Pic here : http://www.flickr.com/photos/midnighttoadstool/7114003381/ .That doesn't look very parallel to me. But I don't know what you mean by parallel in this case. Should I be measuring the voltage across the resistor with the resistor attached to ground? I think I'll go try that right now.

      Delete
    3. Parallel means one end of the resistor to one of the 3.5mm output channels and the other end of the resistor to ground. The DMM should be connect to the same two points. See: http://en.wikipedia.org/wiki/Parallel_circuit#Parallel_circuits

      Delete
    4. Great, yes that did it. Thanks, NwAvGuy. I double checked against the O2 and got .55. So now I know my laptop has a .6 output impedance. It's an ancient Dell D430 that did impress me with how good it is for audio. I've not detected a difference between it and the clip+. Now to get my friend back over to test the Nano.

      Delete
  145. How much output impedance does the FiiO E11 have? I didn't see it on their site.

    ReplyDelete
    Replies
    1. It's below 1 ohm, see my FiiO E11 review from last November.

      Delete
  146. NwAvGuy-

    I'm hoping you can offer me some guidance.

    In my project studio I have two options for headphone outs: a Presonus Monitor Station which outputs 150mW @ 60 ohms, and a MOTU 828 MkIII which outputs 200mW @ 30.5 ohms.

    Neither device seems to drive my new Sennheiser HD600 300ohm cans optimally. The Presonus gets loud enough, but seems a bit distorted and wooly in the upper mids and highs. The MOTU headphone out sounds a little cleaner, but will barely push the Sennheisers to a workable level, even when maxed out.

    Given the specs of the two devices, is there an optimal impedance/SPL I should look for in a headphone? I will be using the headphones for critical listening and mixing tracks.

    Any guidance or recommendation would be appreciated.

    Thanks, GeorgeO

    ReplyDelete
    Replies
    1. See my More Power article. Basically you need relatively sensitive, but ideally relatively high impedance, headphones for most pro sound audio interfaces. Unfortunately most pro sound interfaces have a 20 - 50 ohm output impedance which causes problems with many low impedance headphones (see Sonic Advantages of Low Impedance Headphone Amps). Many also have limited output. Nearly all of them just use a marginal op amp with a 22, 33, or 47 ohm resistor in series with the output which is a very cheap but compromised way to drive headphones.

      A better solution might be to use a real headphone amp like the O2 Amplifier or even an inexpensive FiiO E5 or E6. For your 300 ohm HD600's the FiiO E9 should also work well (or the O2).

      But if you don't want to hassle with a headphone amp, you might try some other headphones. Even though they're low impedance, the inexpensive Grado SR-60/80/etc might yield better results if you're OK with an open headphone. If you need closed cans, the ATH-M50 and Shure SRH series might be worth considering.

      Delete
  147. Thanks so much. I appreciate your guidance.

    On the desktop headphone amp issue, a unit with a line out/through would be best for my project studio setup. Have you had the opportunity to assess the newish FiiO E9k? Would be interesting to see if it addresses some of the issues you found with the older version.

    ReplyDelete
    Replies
    1. I have checked out the revised E9. If it still uses the same TI headphone IC it probably still has the same excessively high output impedance.

      Delete
  148. Hi, awesome stuff about impedances and frequency responses.
    I was wondering about the power transfer theorem. Why do you want
    1/8 of the load impedance as output impedance and not equal?

    ReplyDelete
    Replies
    1. I've answered this before. It's a long answer. Everything, including links to more info, is in the article. The short answer is equal source and load impedance is for RF transmission over a controlled impedance line, not audio. The two are completely unrelated. Stereo power amplifiers typically have an output impedance around 0.05 ohms to drive 8 ohms speakers. You want something similar with headphones but rarely get it from a lot of sources and ill-designed gear.

      Delete
  149. Hi, I submitted a question earlier which is not yet approved but you can disregard that one, i found the answer hidden among all the questions :P

    My new question is, asumming I have a guitar preamp with output load of 36kohm
    what happens if I put in a headphone with relatively low impedance like 20ohm?
    Will that be like a short? and both equipments gets destroyed or just the headphones, assuming the volume is turned up high enough.

    Thanks for taking time to reply to all these questions, awesome stuff!

    ReplyDelete
  150. Hi,
    Can you make files with these differences?

    ReplyDelete
    Replies
    1. I plan to try and make some downloadable files when I have time of various examples of audible differences but I need to get caught up with my articles.

      Delete

PLEASE NOTE! If you see "LOAD MORE..." OR "LOADING..." just above this paragraph, you need to click that link to load the newest comments. It may take several seconds. When all the comments are displayed the Load More/Loading link disappears.

COMMENT APPROVAL: Comments may take anywhere from a few minutes to a few days to be approved. You might want to save a copy (highlight Ctrl-C) of your comments before submitting in case something goes wrong. The approval process is necessary to prevent obscenity, etc. Please keep comments on topic. Comments promoting commercial interests will not be approved and may be reported as SPAM.