• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Master thread on building an affordable and high performing audio analyzer utilizing the E1DA Cosmos ADCiso (crowd sourced)

CleanSound

Major Contributor
Joined
Apr 30, 2023
Messages
1,654
Likes
2,588
[Admin/Moderator - If there are any changes you would like to see, please let me know.]
[Last Updated (mm/dd/yyyy): 2/23/2024]
[Iteration of this post: 2]


Hi All,

If you are a regular ASR member, you will notice that many members here uses the E1DA Cosmos ADCiso as part of a high precision audio analyzer setup (if you don't know what an audio analyzer is, then this topic is probably not relevant to you at this time). Because there really isn't a single location on ASR nor on the internet that's a one-stop-shop to get step-by-step information on using the Cosmos ADCiso as an audio analyzer, the goal here is to create one.

Before we begin, please note:
1) As more and more people contribute to this thread, I will edit this first post to consolidate all the information. I will try my very best to keep it as updated as possible, please use the Last Updated date stamp at the top of this post; please do not use the post's last edit timestamp as I will often make grammatical fixes and clarification which may not contain any new information.
2) I recently purchased a Cosmos ADCiso (Grade A) myself, so whatever information that other members share, I would like to try it myself first to (a) confirm and more importantly (b) fully understand the nuance step-by-step so when I consolidate that info into this post, I can include very detail step-by-step.
3) I am not doing this full time of course, so this will be incremental work-in-progress and there will be times where you see radio silence from me for weeks. Also, I don't have unlimited funds, so I can't just go out and buy parts/gear to experiment.
4) At the initial start, the information may be limited, and it will be incomplete because (a) this is a learning experience for me so it will be incremental (b) I am relying on community's crowd sourcing to get the information, so please contribute and when you do, PLEASE BE VERY SPECIFICIC WITH STEP-BY-STEP INSTRUCTIONS AND EXPLATION.
5) I will from time to time get things wrong as this is a learning experience for me, SO PLEASE DO NOT BLAST ME; just correct me and I will make the corrections.
6) If you post anything on this thread for me to consolidate into this first post, please @mention me, do not just make a post that is not even a reply to my posts as I may not even see your post.
7) Please stay on topic, the more off topic posts we have, the harder it is for me to consolidate the information accurately and succinctly.
8) While there are very few places where this information is somewhat consolidated on the internet, there are few which I will list here for your reference. They are:
-----(a) https://www.openaudiolab.com/measuring-amps/
-----(b) https://forum.quantasylum.com/
-----(c) I will list more as I come across more.


Background
Today's gold standard of audio analyzers is the Audio Precision APx555B (which is what Amir has), but at $35k+ USD this is not something everyone can afford. Recently AP came out with a more affordable APx516B but the performance is not nearly as good as the APx555B and still costs $6k+ USD. And these AP price tags does not even include the advance automation software from AP.

Then there is the QuantAsylum at a much much affordable price, but I believe the performance is on par with the APx516B, so you won't get the performance of the APx555B; though the software for QuantAsylum is perpetually free and the ease of use is similar to the AP as it is purpose built. But thankfully we have Ivan of E1DA who is making very high performance and low cost ADC, combined with other hardware and software, it can be used as a high performing audio analyzer, possibly close to or on par with the APx555B.

Now, let's understand how audio analyzers work: An audio analyzer at its core is essentially a high performing DAC and ADC; it uses a DAC in conjunction with a software to generate an input signal to the DUT (Device Under Test) and compares such input signal with the DUT's output signal, such difference is essentially the noise and distortion.

Because the software by definition can only do such comparison and analysis in the digital domain, the signal is generated by the software is in digital, it then gets converted to analog by a DAC as the DUT only accepts analog signals (with the exception if the DUT is a DAC, more on this later), the analog signal is then sent to the DUT, the DUT will generate an analog output which is then sent to the ADC, which converts the analog signal to digital, and sends it back to the software to do the analysis. See Figure 1 below.

Figure 1.
1708628598765.png


Based on Figure 1 above, this implies that you will need a DAC and the Cosmos ADCiso is the ADC for such setup. So wait, the Cosmos ADCiso is just a ADC? Why is it such a big fuss then? Because according to some sources that I have read (one of these sources I have read being QuantAsylum), the lowest performance between the DAC and the ADC needs to be at least 10dB SINAD or better than the DUT in order to get accurate results, while there are plenty of DACs that has 120+dB SINAD (you can easily find these DAC reviewed here on ASR), but (to the best of my knowledge) there are no ADC on the market with the kind of SINAD and SNR performance as the Cosmos ADCiso, regardless of price point.

About the E1DA Cosmos ADCiso
The Cosmos ADCiso is a very high performing ADC, with a SINAD > 123dB for grade A; to the best of my knowledge, the best performing ADC in the consumer market regardless of price point. It is a low-cost ADC (I paid $255.78 shipped directly from the E1DA store on AliExpress). There are different “grades” of this Cosmos ADCiso; the higher the grade, the better the performance in terms of SNR and SINAD. Please see E1DA’s website for more information on their grading. I suspect the grading is a result of better parts that so happens to land up in each unit as every part that goes into an electronic circuit board has a tolerance.

For the purpose of building out an audio analyzer, you want the highest grade that you can afford or what suites your use case, I personally got grade A.

There are other products from E1DA under the “Cosmos” brand, namely the APU and Scaler. Based on what I pieced together, the APU is a notch filter and a preamp combined. From what I gathered, the preamp feature of the APU is not applicable for the purpose of setting up an audio analyzer, but the notch filter will give you a better performing measurements, down to -132dB SINAD, but only for 1kHz and 10kHz, so you can't do a sweep. I believe the way how the notch filter on the APU works is that, it filters out all frequency band aside from the frequency to be measured (in this case, 1kHz and 10kHz) from the output of your DAC, which inherently also filters out all the noise and distortion that's generated by your DAC outside of the frequency to be measured (someone please fact check me). However for amplifiers, I think the APU is not so useful because most amplifiers' SINAD are frequency dependent, where the higher frequencies (above 10kHz) are usually poorer in performance.

But in any regards, I got a question for the experts out there: If I can use the APU to get SINAD down to -132dB, but the ADCiso is still only -123dB, wouldn't my measurements still be capped at the poorest performance in the chain, which is the ADCiso at -123dB?

I don’t have a very clear understanding of what exactly the scaler does, except that it is a variable gain control. Someone please chime in on this.
__________________________________________


That’s all I got for you on this second iteration of this post. In the meantime, I need the community to crowd source the following information in order for me to write the third iteration (again please provide clear and detail explanations, not just an one-liner and then mic drop please):

1) For the gain setting on the bottom of the ADCiso (see Figure 2 below); can someone first explain or link me to the explanation of gain matching, what is it used for, how (if any) is this related to the scaler?

Figure 2.
U895ca86d8c75432bb929a54474f986a8q.jpg


2) Is using 1% tolerance, non-inductive dummy load mandatory for amp measurements? If so, why and can you link me to US websites where I can buy the required 8 ohm and 4 ohm loads?
3) What exactly is the Cosmos APU and Scaler, what is it used for?
4) The ADCiso XLR input has a voltage limit of 10v and there is a 2.5mm, 4 pole TRRS input with a voltage limit of 43v; can someone please explain the purpose of this voltage limit and what does that mean in terms of what I can measure?


__________________________________________
Topics for future iterations to this post:
1) Selecting a high performing DAC to build out your audio analyzer
2) Accessories, probes, wires, dummy loads, etc. needed to build out your audio analyzer
3) Setting up and configuring the Cosmos ADCiso
4) About audio analyzer software (REW vs Multitone) and how to use them with step-by-step instructions
5) How to create a setup to perform DAC measurements
6) How to create a setup to perform power amplifier measurements for:
-----(a) All amp class
-----(b) Class D
7) How to create a setup to perform pre-amp measurements
8) Stretch topic: How to create a setup to perform headphone amp measurement
9) Stretch topic: How to create a setup to perform phono stage measurements
 
Last edited:
Any issues with using the SMSL D-6S as the DAC to build out this audio analyzer? According to Amir's measurement, there's no ESS hump and SINAD is 121dB. And it's super cheap.
 
Any issues with using the SMSL D-6S as the DAC to build out this audio analyzer? According to Amir's measurement, there's no ESS hump and SINAD is 121dB. And it's super cheap.
I'm definitely interested in this. I would use my D70S or the SMSL SU-8S since I have both of those. I am looking at ordering a Cosmos E1DA.
 
I'm definitely interested in this. I would use my D70S or the SMSL SU-8S since I have both of those. I am looking at ordering a Cosmos E1DA.
Cool, it can be a learning experience for the both of us. You have to make sure that the DAC has at lease 120dB SINAD and does not have the ESS hump by checking the IMD distortion graph.
 
"Then there is the QuantAsylum at a much much affordable price, but I believe the performance is on par with the APx516B, so you won't get the performance of the APx555B; though the software for QuantAsylum is perpetually free and the ease of use is similar to the AP as it is purpose built."

First, full disclosure: I own and use a Cosmos ADCiso ("0" spec, in fact) along with the Cosmos APU and Cosmos Scaler. I use this with a Topping D50s as a source in conjunction with both REW and Multitone. In addition, I also own a QA401, which is an earlier and less highly spec'd version of the latest QuantAsylum device. I have not seen or used a new QA403 and only know what can be read on the QA site.

OK, that out of the way...

I think the QA403 is probably very similar in performance to the Cosmos ADCiso. Of course, that all depends on which grade of ADCiso you're comparing to and how each QA403 grades out due to production variances. After all, they use the same ADC device and both designers are very competent. Of course, the QA403 is integrated with a built-in DAC and a common clock. That has some benefits and some limitations. Everybody will have their favorite that meets their needs. I paid just about the same for my Cosmos ADCiso, a Cosmos Scaler, and the D50s as it would cost for a QA403.

I understand why you are putting this together, since there's a lot more variations of possibilities with the Cosmos setup along with a separate DAC with all of the available software solutions. I'm just trying to point out that both are capable and which is better probably depends more on whether you want a plug'n'play solution or one that has more permutations. And, if you like BNC connectors or not.

One other point... The APx555B is certainly regarded as the industry standard. You won't get much questioning of your results when the Ap logo is showing up in the corner. But, with some care you can get similar or better results with the Cosmos combination. If you're hobbyist or doing R&D, that extra layer of detail is probably OK. For any kind of production or even service work, the plug'n'play capability of the QA and AP devices is hard to beat.

With help from forum members, I was able to get these results a while back:

Some Test Results
 
"Then there is the QuantAsylum at a much much affordable price, but I believe the performance is on par with the APx516B, so you won't get the performance of the APx555B; though the software for QuantAsylum is perpetually free and the ease of use is similar to the AP as it is purpose built."

First, full disclosure: I own and use a Cosmos ADCiso ("0" spec, in fact) along with the Cosmos APU and Cosmos Scaler. I use this with a Topping D50s as a source in conjunction with both REW and Multitone. In addition, I also own a QA401, which is an earlier and less highly spec'd version of the latest QuantAsylum device. I have not seen or used a new QA403 and only know what can be read on the QA site.

OK, that out of the way...

I think the QA403 is probably very similar in performance to the Cosmos ADCiso. Of course, that all depends on which grade of ADCiso you're comparing to and how each QA403 grades out due to production variances. After all, they use the same ADC device and both designers are very competent. Of course, the QA403 is integrated with a built-in DAC and a common clock. That has some benefits and some limitations. Everybody will have their favorite that meets their needs. I paid just about the same for my Cosmos ADCiso, a Cosmos Scaler, and the D50s as it would cost for a QA403.

I understand why you are putting this together, since there's a lot more variations of possibilities with the Cosmos setup along with a separate DAC with all of the available software solutions. I'm just trying to point out that both are capable and which is better probably depends more on whether you want a plug'n'play solution or one that has more permutations. And, if you like BNC connectors or not.

One other point... The APx555B is certainly regarded as the industry standard. You won't get much questioning of your results when the Ap logo is showing up in the corner. But, with some care you can get similar or better results with the Cosmos combination. If you're hobbyist or doing R&D, that extra layer of detail is probably OK. For any kind of production or even service work, the plug'n'play capability of the QA and AP devices is hard to beat.

With help from forum members, I was able to get these results a while back:

Some Test Results
Awesome, thank you for participating.

Few questions I was hoping you can help me with:
1) What electronics have you measured so far?
2) What is the scaler used for? How do you know what gain to set the scaler on?
3) What is the APU used for?
4) How do you know what gain to set the ADCiso to?
 
Awesome, thank you for participating.

Few questions I was hoping you can help me with:
1) What electronics have you measured so far?
2) What is the scaler used for? How do you know what gain to set the scaler on?
3) What is the APU used for?
4) How do you know what gain to set the ADCiso to?

I've measured DACs and preamps. Plus, I used it to test the performance of my USB isolation device.

The other topics are pretty well explained on E1DA's web site and by Ivan himself in the thread I referenced before. I can't do a better job explaining than he did.
 
I've measured DACs and preamps. Plus, I used it to test the performance of my USB isolation device.

The other topics are pretty well explained on E1DA's web site and by Ivan himself in the thread I referenced before. I can't do a better job explaining than he did.
Thanks! I'll give it a read.
 
I actually spent like 3 days after a few weeks working with the ADCiso
and wrote up a detailed testing SOP with notes and photos and howto and what REW options change scaling for this and that and got prettymuch everything on the cosmos working properly.

except being out of spec by 1db and also the dBA rating trashing the -129dB @ 20-1000 and 10k+ hz.
the it uses the best ESS chip on the market.
anything else need adjustments and a massive investment and yearly benchmarks to ensure its within spec.

even a 6000w passively cooled dummy load for the amps, so 6000+Watt active cooled.

you have to look at the REW scope and view the frequency fluctuations to see what's occurring outside of dBA FR (Fourier Transform)

actually, i sent my tutorial as a diagnostic in an email to E1DA Cosmos.
 
actually i really want someone to benchmark the sound devices

833 Portable Compact Mixer-Recorder
the noise floor is supposed to be lower than the ADCiso, but the THD ratings from sound devices is not listed.
they just list, its lower than the audible region at like less than 0.005% max

notice the dBV ratings that really freaking good!!!!! if my math is correct.


  • Frequency Response​

    • 10 Hz to 80 kHz ± 0.5 dB (192 kHz sample rate, re 1 kHz)
  • THD + Noise
    • 0.005% max (mic in, 1 kHz, 22 Hz–22 kHz BW, trim at 20, fader at 0, -10 dBu in)
  • Equivalent Input Noise
    • -131 dBV (-129 dBu) max (mic in, A-weighting, 76 dB gain, 150 ohm source impedance)
 
I actually spent like 3 days after a few weeks working with the ADCiso
and wrote up a detailed testing SOP with notes and photos and howto and what REW options change scaling for this and that and got prettymuch everything on the cosmos working properly.
.
.
actually, i sent my tutorial as a diagnostic in an email to E1DA Cosmos.
Do you have this tutorial to share with us here?
 
As an adjunct to this great thread; should some kind of a shared-depository be created?
For a compendium of various audio test signals/tracks/tones/sweeps, in different analog/digital audio formats, w/such files being royalty and license-free.

@restorer-john' post "SOS Magazine Test Files" is a good start.
 
As an adjunct to this great thread; should some kind of a shared-depository be created?
For a compendium of various audio test signals/tracks/tones/sweeps, in different analog/digital audio formats, w/such files being royalty and license-free.
Great idea, I'll create one, once this thread gets more traction and more crowd source input.
Awesome, will check that out!
 
Do you have this tutorial to share with us here?
i hoped to go through it 1 more time, i dont like releasing anything wrong and i had pasted 1-2 images that were inferring i did something wrong because i sniped them before the change, (change volume to 100% and it shows volume at 0%) plus 1 of my xlr cables was mirrored and noticed it at the end, when i was running mono, so i need to verify it didnt corrupt any other measurements, how to sections.
 
i hoped to go through it 1 more time, i dont like releasing anything wrong and i had pasted 1-2 images that were inferring i did something wrong because i sniped them before the change, (change volume to 100% and it shows volume at 0%) plus 1 of my xlr cables was mirrored and noticed it at the end, when i was running mono, so i need to verify it didnt corrupt any other measurements, how to sections.
Amazing, looking forward to it!
 
Cool, it can be a learning experience for the both of us. You have to make sure that the DAC has at lease 120dB SINAD and does not have the ESS hump by checking the IMD distortion graph.
Well I think the D70S has like 119 and it is using dual AKM DAC's so no ESS hump.
 
Back
Top Bottom