Best spec ADC Chip currently.... ??

[IVX]

Blumlein 88, REW gives you different results of THD+N depends on FFT window type and I didn't see more or less correct numbers there in a range of my interest i.e. THD+N -117db and lower. Maybe 1% is correct I didn't check. Only one FFT app looks similar to AP results is Arta but this one is paid.

 

[adriano93]

How many channels do you plan? Looking for something like that
https://focusrite.com/de/audio-interface/scarlett/scarlett-18i20
with at least 10 balanced channel outputs and a microphone input.

Thanks a lot

 

[IVX]

adriano93, why 10ch needed? I guess it could be DIY measurement ADC. Regarding a mic preamp, I like the idea to use external adapters see attached above, a low-noise preamp could take such a role.

 

[adriano93]

I want to create a digital XO for 4 speakers plus 1 subwoofer with a digital XO like described here

https://audiophilestyle.com/ca/ca-a...e-alignment-driver-linearization-walkthrough/

So i need for each speaker 2 outputs and a microphone input with the same clock. Thats the reason because i dont want to use different devices.

 

[boXem]

What kind of over-stress protection are you planning?

 

[KSTR]

chris719,
that's why I pushing my coder to start Windows FFT project to provide precision measurement Cosmos users
Yes, we can use a few free FFT apps for Windows but no one of them matched vs AP results and each other too. I believe I tried all, and no one correctly calculates noise or THD+N.

Hhm, maybe a year or so ago I did a short sanity check comparing REW with my SYS2322 and the results were absolutely comparable in comparable settings. In fact I find that REW has one of the best implementations of a real-time (and offline) FFT analyzer I've ever seen (and in general is a killer software), and that includes some $$$ software/hardware like AP, PrismSound, R&S, ListenInc, ...
Once you have the scaling right noise measurements are absolutely spot on with REW, that's for sure.
EDIT: enable 64bit FFT processing in REW when doing stuff in -120dB realms. I'v also checked the analyzer with some analytically generated test files and the results are 100% correct. So I would assume there are setup issues.

 

[mansr]

AP probably hasn't had to think about hardware too much until recently. They satisfy most of their market with the capabilities of APx555. It's really all about paying for software, convenience, and a known level of performance.

A Matlab licence is practically free compared to an AP and lets you do whatever you want quite easily.

 

[IVX]

boXem | audio, TBD yet. Also, I think about the buffer opamp candidate so far I use the input straight to the differential stage. The dilemma is higher opamps rails(+/-17V) are needed to get 10Vrms range as planned, and my USB power rail is already 590mA, at the same time I see no suitable R2R input and low-noise opamp to take 10Vrms from the input at +/-15V rails.

 

[IVX]

KSTR, ok, I checked that again right now again. The setups is: AP analog gen goes to my filter-board for harmonics elimination, next to the Cosmos input, next by USB to my PC, where it plays thru the SPDIF and goes back to the AP optical input for analysis. AP shows THD+N -118.5db(<0.00012%) steady.
REW on the same PC thru the ASIO4ALL v2 shows THD+N jumping from 0.002% to 0.004% with the Hanning window. Other windows give different THD+N reading but not even close to the APs reading yet.
Seriously, if you know where I do mistake, I would happy to learn, I'm such a kind of guy, I love truth more than myself.

 

[boXem]

boXem | audio, TBD yet. Also, I think about the buffer opamp candidate so far I use the input straight to the differential stage. The dilemma is higher opamps rails(+/-17V) are needed to get 10Vrms range as planned, and my USB power rail is already 590mA, at the same time I see no suitable R2R input and low-noise opamp to take 10Vrms from the input at +/-15V rails.

Just keep in mind that some idiots like me will want to use it with power amps. Not toys, kW ones .

 

[IVX]

BTW, even the fundamental level is wrong, AP reads -.69dbfs, REW -3.9dbfs. Spectraplus -1.12dbfs ))

 

[IVX]

boXem | audio, for such things I prepared AUX input with 40Vrms range.

 

[capslock]

$200 for what kind of package? For a bare ADC it is a lot, and before long, we'll see $35 ADC boards with ES9822 popping up on Alibaba. With gain switching, overvoltage protection and a functional USB interface it becomes reasonable. Include a 9038S grade DAC and maybe software, it becomes a steal. Where can I sign up to buy a COSMOS ADC?

I would be happy to play with an ADC board and add my own front end, but overvoltage protection and gain switching (unless you use mechanical switches or relays) is where you can degrade performance, so a board with a robust and performant front end is very welcome!

Notches as an addon are a great idea. At least passiv notches are not rocket science, but again, a turnkey solution would be welcome to a lot of hobbyists. The better the ADC and DAC, the less need to resort to notches. By the way, Quant Asylum have something like that:
https://quantasylum.com/blogs/news/introducing-the-qa480

As for software, I was going to suggest REW, Arta and Virtins Multi Instrument (but Virtins is more expensive than $200 just for the software). It would be nice if the software could operate the gain switches, but at this point, only Virtins has that for the RTX6001 or Quant Asylum for its own proprietary solution. They also have an integrated software menu for calibrating the notch.

 

[boXem]

boXem | audio, for such things I prepared AUX input with 40Vrms range.

I was more thinking to something in the 90-100 V RMS

 

[capslock]

boXem | audio, TBD yet. Also, I think about the buffer opamp candidate so far I use the input straight to the differential stage. The dilemma is higher opamps rails(+/-17V) are needed to get 10Vrms range as planned, and my USB power rail is already 590mA, at the same time I see no suitable R2R input and low-noise opamp to take 10Vrms from the input at +/-15V rails.

Why the requirement for 10 V Rms? Most DACs we might want to measure have a maximum of 4 V RMS, and for power amps, one needs an attenuator anyway.

If you want to stick to 10 V range, you could add an aux power in and make the rail voltage switchable.

 

[IVX]

boXem | audio, 100Vrms not a big deal too, you'll see a clipping but the differential amp is safe.
capslock, I do design only what I like to have and use. If somebody likes to buy that, good, if somebody thinks it is too expensive, sorry about that and wait alibaba then. I know my opportunities to produce a product, my spends and I feel I can't make such ADC+DAC aka Cosmos io, assembled in the case and calibrated for less than $200. Especially if I'll calibrate each unit, and sell it next for $35? ADC isn't a portable DAC which I can sell thousands of units per year, ok if 500 ADC will be sold. So, I'll look like an idiot if spend on tooling, assembling etc more than I can earn on it.

 

[capslock]

boXem | audio, 100Vrms not a big deal too, you'll see a clipping but the differential amp is safe.
capslock, I do design only what I like to have and use. If somebody likes to buy that, good, if somebody thinks it is too expensive, sorry about that and wait alibaba then. I know my opportunities to produce a product, my spends and I feel I can't make such ADC+DAC aka Cosmos io, assembled in the case and calibrated for less than $200. Especially if I'll calibrate each unit, and sell it next for $35? ADC isn't a portable DAC which I can sell thousands of units per year, ok if 500 ADC will be sold. So, I'll look like an idiot if spend on tooling, assembling etc more than I can earn on it.

I think you misunderstood my post We will see bare ADC boards on Alibaba before long, but they are not something that us measurement or home recording hobbyists can use out of the box. It takes lots of time to add a proper front end, protection, switches.

So an ES9822 ADC, a proper front end with protection and switches and calibration, a USB interface and a decent DAC is really an attractive package for less then < $ 200. That's why I asked where I can sign up. I seriously want to buy one, and I guess several others would too.

The other market that you might want to go for is a better version of the QA401/402. I have the QA400, and while the box and the software are good, the biggest limitation is its single ended inputs. The QA401 was about twice as expensive and strangely appealed to hobbyists as well as pros including manufacturers using it for developement as well as line testing. Hell, there are even people who use it as a music DAC with ASIO drivers. So if you want to get software written, this might be the market niche that would allow you to recoup the cost of software coding and maintenance.

 

[KSTR]

KSTR, ok, I checked that again right now again. The setups is: AP analog gen goes to my filter-board for harmonics elimination, next to the Cosmos input, next by USB to my PC, where it plays thru the SPDIF and goes back to the AP optical input for analysis. AP shows THD+N -118.5db(<0.00012%) steady.
REW on the same PC thru the ASIO4ALL v2 shows THD+N jumping from 0.002% to 0.004% with the Hanning window. Other windows give different THD+N reading but not even close to the APs reading yet.
Seriously, if you know where I do mistake, I would happy to learn, I'm such a kind of guy, I love truth more than myself.

Difficult to go into details from remote.... could you provide a recording of that exact signal?

One (known) issue with REW and many other software is that the THD+N measurements does not discard the skirts of the frequency needles. Everything outside of the main bin is treated as noise which is not entirely correct. The skirt width depends on the windowing (and jitter etc).
I will suggest a correction to the author (@JohnPM) : at least the known skirt shape, depending on chosen window, should always be discarded for the noise part in a THD+N.
I'm using digital gen on exact bin centers 99% of the time so I'm not running into the issue.

For the level thing, REW has two settings for 0dBFS, one is the correct for this application, the other is off by 3dB, maybe you have set the wrong one?

 

[IVX]

KSTR, wait a minute, your AP SYS2322 has the same A_GEN, with the same skirt, can you try REW on your side?

 

[KSTR]

@IVX, Yep, will do, give me some (haven't used that boat anchor for some time and it's currently stowed away).
I don't know if I can replicate your test case as my ADC is not nearly as good as yours, nor is the analog gen.

But I think the root cause of the problem is identified, the skirts introduced by the windowing.