True that.
Which Cosmos ADC did you get? The "0 grade" with the lowest THD or one of the other ones? Just curious. It doesn't matter greatly, I don't think, but it would be nice to know for completeness.
That comment was directed at anyone reading these pages and not directly at you. Sorry I stepped on your toes.
Tom
Your assumption is incorrect. The DAC used has lowest distortion at 0dBFS output level, and the same applies to ADC input. There is no knee and turning point with distortion rising, up to the clipping. We have made direct comparison with QA403 and QA403 noise and distortion results are inferior. Yes, it has big advantage of autoranging and may be operated by (almost) lay person.
Cosmos ADC noise limit is about -124dBFS (flat 22kHz) with highest sensitivity of 1.7V/0dBFS. This clearly defines the intrinsic noise of the ADC part.
Last edited:
Sounds like the system is limited by thermal noise then. That's not a bad thing, but it does indicate that there could be room to push further.
Who's "we"?
I can compare numbers too, ya know... 🙂 It is clear that the Topping+Cosmos combo performs better at its sweet spot than the QA403. It is further possible that the sweet spot of the Topping+Cosmos combo is wider than that of the QA403.
All I'm saying - to anyone and everyone who reads these pages now or years from now - is: Be careful when comparing the graphs as 0 dBV may or may not be equal to 0 dBFS. This makes it impossible to tell exactly how much better the Topping+Cosmos combo is. That is all.
The QA403 doesn't do auto-ranging very well. Unlike the AP, you can't just plug any signal in and have it go. You have to fiddle with the attenuator settings manually. It does offer a form of "poor man's auto-ranging" in the automated test mode, but that only applies to sweeps.
Have you tried Cosmos' own DAC? They have one based on the same ES9039Q2M DAC chip as the Topping D10s. $139 from Cosmos as I recall. So a few bucks more than the Topping.
Tom
Those are distortion plots (up to which harmonic? Does that make a difference?) And often the plots published are THD+N which are not compariable with THD, so potentially more confusion. Understanding noise also gets confusing, since higher resolution (number of samples in the measurement) makes for apparently lower noise, its the RMS sum of those samples in the specified bandwidth thats is the noise measurement.
Personally, even though I'm a founding member of the THDnut brigade, I think THD much below 100 dB is not a fruitful pursuit. Other aspects may be more important and more rewarding to address.
Personally, even though I'm a founding member of the THDnut brigade, I think THD much below 100 dB is not a fruitful pursuit. Other aspects may be more important and more rewarding to address.
It should be easy enough to properly compare the dac/cosmo to the quantumA generator , by measuring the output and calibrating with a voltmeter.
I do this, it is my first step, check for calibration of my measuring instrument before I use it. Otherwise all the numbers are wrong.
I do this, it is my first step, check for calibration of my measuring instrument before I use it. Otherwise all the numbers are wrong.
Depends on which fruits you enjoy pursuing. I enjoy operating at the intersection between engineering, physics, and material science. That's where the -140 dBc THD and below is found. But I agree that much past -100 dBc is largely a number's game. A fun number's game. 🙂
Tom
What I see from the whole cosmos stack is that they require too much hassle to make it work, plus need USB isolation, attenuator, filters those tiny 2.5mm connectors are something else, at current prices we talking about $1000 for the whole thing, with real chances that they can blow if one gets distracted.
(Photo from archimago's website)
Or just grab QA403 that is designed for the task for $600, and test right away ( I have the feeling that the software will get better and better) with more options starting to appear.
That said, designing a buffer/attenuator with real connectors sounds like fun. (not only thinking on the Cosmos ADC)
(Photo from archimago's website)
Or just grab QA403 that is designed for the task for $600, and test right away ( I have the feeling that the software will get better and better) with more options starting to appear.
That said, designing a buffer/attenuator with real connectors sounds like fun. (not only thinking on the Cosmos ADC)
and you have an extra room to purchase a good oscilloscope/wave generator.
I use a lot the computer to generate sound/complex waveforms to test things, I always have a control unit to know that what I send is not exactly as designed, to isolate the component under test.
I use a lot the computer to generate sound/complex waveforms to test things, I always have a control unit to know that what I send is not exactly as designed, to isolate the component under test.
As far as I know, the only currently available product that offers everything in one box is made by Audio Precision. At least on this performance level.
~~~
I think Hipocrates assessment is inaccurate in some of the facts.
First, in the E1DA Cosmos line the only 2.5mm connectors are for the attenuated high voltage input to the ADC. Everything else in the analog domain is either 3.5mm or XLR. That might be to your taste, or not. Personally, I am agnostic with regard to whether 3.5mm is better or BNC as found on the QA gear I have. Unbalanced connections are about equally convenient, while balanced using BNC is kind of awkward. YMMV etc. There no standards in this. My AutoRanger has an even different mix of connections. I just had to make or buy some more cables to match.
Second, that photo shows two APU units. That is not comparable to a QA403 system. The APUs are for auxiliary functions, like a notch filter for better ADC dynamic range and some extra gain stages. QA offered the QA480 in the past for the notch filter capability (and low distortion 1 KHz generator functions) as well as the QA472 for extra low noise gain.
So, because of that, a comparable Cosmos system to the QA403 would be a DAC (E1DA sourced or otherwise), a Cosmos Scaler, and a Cosmos ADCiso. Note that both the QA403 and the ADCiso provide isolation, so no external box is needed for that.
If you use a Topping E50 DAC as your source (I bought one because it is reasonably low distortion and offers both unbalanced and balanced outputs) combined with a Cosmos Scaler and Cosmos ADCiso, the total comes to about $600.
That sounds comparable to me.
Note - both can be blown up. The AP system is close to bullet-proof, but if you look around DIYAudio you can find a number of comments about AP's gaps in reliability. (I have no experience with AP - I'm a hobbyist - so I can't comment on that.)
There's also the software question. I find that I like both Multitone and REW with the Cosmos/Topping setup. Personal preference.
If you are primarily concerned with 1 KHz distortion and all that's associated with that, a really good solution is the Scarlett Solo that Tom pointed to. Combine that with a low distortion test generator from DIYAudio member Victor (Victor's Generator) and a Cosmos APU ($55 without a box) and you can get close to the capability of measuring to -150 dBc for loopback. Yeah, it's not all in one box. But, it's not so expensive. Note that in concept, that approximates the internal functions of a current AP test system. There's a low distortion analog oscillator and a notch in front of the ADC portion.
All of that should be plenty good enough for a hobbyist. If you're doing QA for manufacture, of course it isn't good enough, mostly because of traceability and overall ruggedness.
There's plenty of other measurements that are actually relevant to audio reproduction quality that largely get ignored. So, although it can be fun chasing those last few dB of distortion, that might not be the entire solution to better sound quality.
~~~
I think Hipocrates assessment is inaccurate in some of the facts.
First, in the E1DA Cosmos line the only 2.5mm connectors are for the attenuated high voltage input to the ADC. Everything else in the analog domain is either 3.5mm or XLR. That might be to your taste, or not. Personally, I am agnostic with regard to whether 3.5mm is better or BNC as found on the QA gear I have. Unbalanced connections are about equally convenient, while balanced using BNC is kind of awkward. YMMV etc. There no standards in this. My AutoRanger has an even different mix of connections. I just had to make or buy some more cables to match.
Second, that photo shows two APU units. That is not comparable to a QA403 system. The APUs are for auxiliary functions, like a notch filter for better ADC dynamic range and some extra gain stages. QA offered the QA480 in the past for the notch filter capability (and low distortion 1 KHz generator functions) as well as the QA472 for extra low noise gain.
So, because of that, a comparable Cosmos system to the QA403 would be a DAC (E1DA sourced or otherwise), a Cosmos Scaler, and a Cosmos ADCiso. Note that both the QA403 and the ADCiso provide isolation, so no external box is needed for that.
If you use a Topping E50 DAC as your source (I bought one because it is reasonably low distortion and offers both unbalanced and balanced outputs) combined with a Cosmos Scaler and Cosmos ADCiso, the total comes to about $600.
That sounds comparable to me.
Note - both can be blown up. The AP system is close to bullet-proof, but if you look around DIYAudio you can find a number of comments about AP's gaps in reliability. (I have no experience with AP - I'm a hobbyist - so I can't comment on that.)
There's also the software question. I find that I like both Multitone and REW with the Cosmos/Topping setup. Personal preference.
If you are primarily concerned with 1 KHz distortion and all that's associated with that, a really good solution is the Scarlett Solo that Tom pointed to. Combine that with a low distortion test generator from DIYAudio member Victor (Victor's Generator) and a Cosmos APU ($55 without a box) and you can get close to the capability of measuring to -150 dBc for loopback. Yeah, it's not all in one box. But, it's not so expensive. Note that in concept, that approximates the internal functions of a current AP test system. There's a low distortion analog oscillator and a notch in front of the ADC portion.
All of that should be plenty good enough for a hobbyist. If you're doing QA for manufacture, of course it isn't good enough, mostly because of traceability and overall ruggedness.
There's plenty of other measurements that are actually relevant to audio reproduction quality that largely get ignored. So, although it can be fun chasing those last few dB of distortion, that might not be the entire solution to better sound quality.
It would be beneficial to have the ADC and the DAC on the same clock. You get lower spectral leakage that way. That's why they need to be in the same chassis. That's one advantage of the QA403 over the Cosmos stack.
That said, the Cosmos is quite a bit better at its sweet spot.
Tom
That said, the Cosmos is quite a bit better at its sweet spot.
Tom
The QA403 is better than most equipment you will be measuring. For most people it would be better than fantastic.
I can see your points.
That cosmos ''ecosystem'' seems to be fun to play with. I specially like that ADC a lot, It would fun to build something with them with connectors that I like. I can see now that I can get the boards via Hong Kong, (not available via US), with a massive difference in price. Thanks!
No argument from me on that point. I wonder how much difference that makes in practice.
There's still the software aspect. The QA software seems to be aimed at small, or even larger, factory production testing and QA.
The older QA401 probably fits that description, too. Not as cool, though. This is why a Victor generator slash APU slash Scarlett Solo might be the best deal.
Gee, this is the very first time in all my years on DIYAudio that somebody didn't violently disagree with me! 🙂
I considered buying a Cosmos ADC, a Cosmos DAC, building a bunch of switchable notch filters for the ADC side and bandpass filters for the DAC side, putting them all in a nice box with my preferred connectors, and adding a built-in properly isolated power supply and USB hub with isolation. I even went so far as to design a few of the pieces.
Then I realized that it would be better for me to finish some preamp and amp projects that I could enjoy with my wife. These are tools to support my other audio hobby activities; I never gave much consideration to building my own diagonal cutters, so why go here? The existing products do what I want them to, and do it pretty reliably at an acceptable price.
I'm glad I didn't pursue the project, because ESS already has a better ADC chip on the market that I'd be wanting to build into my home grown test system. So, I'd be designing my own ADC board. The trail never ends.
Still, I truly get the appeal of trying to do better on your own.
There's a fair argument that -100 dB performance with multitone testing, especially "pink" shaped tone spectral amplitude, is way below audibility in just about everybody's room, no matter how good your ears may be. At least at the levels you'd listen to. (Above there, so many things distort, like your ears.) Of course, there's other sources of masking like the USB common mode noise discussed elsewhere, that routinely get glossed over and ignored.
Example: USB Isolation Effects
That's just internet people getting emotional.
This is just fun for me, you have some fair points that led me to the idea to build something and how to get better prices and boards, and that's a good thing.
Cheers