WARNING (update 4/25): This article was published April 1st which, in the USA at least, is a day when journalists enjoy publishing fictitious articles. Even The Economist joins in the fun. This article was my attempt but I think The Economist did a better job.
BLOG COMMENTS: With my infrequent recent articles, and ever increasing traffic (over 1 million unique visitors with 900K in the last 8 months), some articles have hit 200+ comments each. Unfortunately Blogger stops displaying new comments past 200. Hopefully Google will eventually correct the limitation, but until then, I can only make room for new comments by deleting old ones, or closing the articles to new comments. So if I deleted your comment, or you can’t post, don’t take it personally. Talk to Google.
FUTURE ARTICLES: I hope to publish articles more frequently to keep things flowing better and distribute the volume of comments across more articles. That should help prevent running into the 200+ limit. Here are some partially completed I hope to get online soon:
- Timex vs Rolex – As the release of the ODAC grows near, I think it’s important to shine more light on how how sound quality and audio performance relate to price. This article attempts to further explain what demonstrably matters versus what’s usually marketing hype and snake oil. I’ve received several ODAC-related questions about balanced configurations, 24/192 support, S/PDIF vs USB, NOS DAC chips, UAC 2, pre vs post filter ringing, paralleling DAC chips, etc. I also have several new and interesting references to include. And I’ve learned more myself since I wrote my Subjective vs Objective Debate last May. The goal is to roll all that up into a general interest article. Or, if it gets too long, perhaps two articles.
- ODAC Measurements – Just when I think the ODAC is ready for production, I find something else or discover a new measurement technique that reveals a new aspect to refine. Some tweaks have worked out for the better, some made no difference, and some were a big step backwards. For example, power supply noise can be a source of jitter. So I added additional filtering to the digital power supply. The filtering, to my surprise, seriously increased the jitter. Theories are nice but measurements are reality. I’ve reverted back to the previous design at least as often as I’ve moved forward. Once the design is finalized, Real Soon Now, I’ll be publishing a full set of measurements. I’ll include some comparative results from other devices like the Benchmark DAC1, FiiO E10, a Chinese headphone DAC that’s had some good reviews on Head-Fi, and perhaps even a pro sound USB interface.
- ODAC Details – The details article will cover the design choices, components used, unveil my partner in crime, document who will be offering the ODAC for sale, enclosure details, how the ODAC will integrate into the ODA headphone amp, etc.
- ODAC Blind Testing – Once the cat is fully out of the bag more blind listening is on the agenda. I plan to expand my preliminary efforts comparing the ODAC to the Benchmark DAC1. Then I hope to arrange for others to participate in blind tests perhaps against other well regarded DACs. Once there’s a reasonable amount of data, I’ll publish an article outlining the results. If audible differences are found, I hope to put some of my new measurements to good use in search of an objective explanation.
- ODAC Future Offerings – Depending partly on the outcome of the Cheap and Cheerful ODAC in blind testing, and user feedback, there may be future DACs that will plug into the ODA headphone amp just like the ODAC. One possibility is an open source DIY version. And, if there’s a need for added performance to stand up to best DACs in blind listening tests, we already have some ideas for a less cheap but even even more cheerful (higher performing) version that will hopefully be plug-n-play compatible with the ODA.
- ODA Initial Details – The ODAC continues to take priority but that’s about to change as soon as it’s released to production and work on the open source ODA desktop amp resumes.
SOME ODAC NUMBERS: We’re still finalizing the design, but here are some recent preliminary measurements for the current ODAC prototype versus the FiiO E10 and the Benchmark DAC1 Pre. Some ODAC results are even at the measurement limit of the dScope (marked “dS limit”). The geeks in the audience may also want to check out the Tech Section at the end of this article.
| Measurement | ODAC | FiiO E10 | DAC1 Pre |
| Freq. Response 10 hz – 20 khz | +/- 0.01 dB | +/- 0.1 dB | +/- 0.1 dB |
| THD+N 100 hz 0 dBFS | 0.0003% | 0.005% | 0.0009% |
| THD+N 20 hz –1 dBFS | 0.0002% (dS limit) | 0.004% | 0.0009% |
| THD+N 10 Khz –1 dBFS | 0.0002% (dS limit) | 0.004% | 0.0007% |
| IMD CCIF 19/20 Khz –3 dBFS | 0.0002% (dS limit) | 0.013% | 0.0005% |
| IMD SMPTE –1 dBFS | 0.0002% (dS limit) | 0.004% | 0.0004% |
| Noise A-Weighted dBu 24/44 | -117.2 dB (dS limit) | -98.3 dB | -105.4 dB |
| Dynamic Range –60 dBFS Ref 2V | -127.5 dB (dS limit) | -97.6 dB | -110.9 dB |
| Linearity Error -90 dBFS 24/44 | 0.0 dB | 0.0 dB | 0.2 dB |
| Crosstalk 0 dBFS Line Out 100K | -128 dB (dS limit) | N/A | -106 dB |
| USB Jitter 11025 hz J-test 24/44 | Undetectable (dS limit) | Very Good | Excellent |
| Maximum Output Line Out 100K | 8.4 Vrms | 1.65 Vrms | 2.5 Vrms (1) |
NOTE 1: Maximum output of DAC1 is configured with internal jumpers
NOTE 2: DAC1 measured from unbalanced RCA line outputs with the volume set for 0 dBFS = 2 Vrms
NOTE 3: Measurements marked “dS limit” are beyond the dScope’s limits see: dScope Specs
NOTE 3a: Above ODAC measurements made using techniques originally developed by AMB Labs
NOTE 3b: All ODAC prototype measurements are preliminary as of April 1st 2012 and subject to change
TIME FOR A DAC2? Benchmark seems to be resting on their impressive laurels. The ODAC’s performance documented above is evidence they might want to sweep the dust off their workbench, blow the cobwebs out of their audio analyzer, and raise the bar with the rumored DAC2. For now, however, it seems the ODAC may be a more credible USB DAC “benchmark” than the legendary DAC1 series.
FiiO SHOULD RECONSIDER: Even if they ever manage to supply the E17 to Amazon, FiiO should look at the table above and consider a different line of business. Perhaps they could supply those digital audio players that go inside greeting cards?
TIME FOR A SERIES IV? The ODAC has demonstrated Prism Sound might want to update their formerly impressive dScope Series III. In eight of the twelve measurements above the ODAC exceeded the limits of the current dScope. It’s my understanding, however, I can send my Series III back to the UK and have it rewired with Kimber Kable, Black Gate capacitors, a higher current power supply, and a special Qables silver power cord. These mods, I’ve read at Head-Fi, will vastly improve its measurement performance. So I might try that before upgrading to a fully optioned $50,000 Audio Precision SYS-2722 Platinum Edition.
BOTTOM LINE: I’m pleased the ODAC has not only come close to the Benchmark DAC1 but crushed it in many tests. It’s also encouraging, although frustrating, the ODAC is so good it’s largely beyond the measurement capabilities of even a very capable audio analyzer like the dScope Series III. If I can’t get my hands on an audio analyzer up to challenging the ODAC, we can send a board to Lirpa Labs for full analysis by the famous Dr. Loof Lirpa. Check back for more soon!
TECH SECTION
TECH INFO: As mentioned in the first section above, I’ll be publishing full measurements for the ODAC soon enough. For now, however, here’s a preview. One of the greatest challenges in a 24 bit DAC are residual and spurious sources of noise. This is especially true for a USB powered DAC. So here’s a preliminary comparison between the Benchmark DAC1 and the ODAC.
BENCHMARK DAC1 ABSOLUTE NOISE: First let’s look at the reference-grade DAC1’s performance reproducing a –90 dBFS 1 Khz signal. You can see the overall noise floor is impressively off the bottom of the graph hidden below –140 dB. There are, however, several visible spurious noise components reaching a maximum of –117 dB. When these spikes, and the unseen noise floor, are added together, you get 62 uV of absolute unweighted noise in the audio band. Using A-Weighting it’s –105.4 dBu. This is about -113 dB referenced to 2 volts. This is impressive performance for a USB DAC. The linearity is almost, but not quite, perfect at –90.1 and –90.2 dBFS in each channel for an error of 0.1 and 0.2 dB respectively. This level of noise and the linearity error should be inaudible:
ODAC ABSOLUTE NOISE: Here’s the ODAC’s even more impressive result. Note the absolute noise drops from the DAC1’s respectable 62 microvolts to only 0.04 microvolts or a staggeringly low 40 nanovolts. Put another way the Benchmark DAC1 has 1500 times more absolute noise! The weighted result of –117.2 dBu is the dScope’s native noise floor and measurement limit—not the ODAC. The actual noise of the ODAC is much lower—the math indicates it should be around –168 dBu! Note the total absence of any visible noise or spurious components down to –140 dB. The linearity is perfect at exactly -90.0 dB in both channels. Here the ODAC transcends merely transparent performance and elevates the ODAC into being a genuine oracle of audio revelation:
NOTE: These are preliminary results as of April 1st and are subject to change.
