• 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!

Best spec ADC Chip currently.... ??

@IVX, understood.

FWIW, just yersterday night I found an unbelievably simple and reliable break-through tweak to get a THD(-N) of -140dB (pure H2) from 2 AK4493's summing the four channels, consistent vs. level and frequency. With a bit of care I might even reach -150dB, APX555 territory. Or get the -140dB with only two channels....
Will post an article once I have it fully fleshed out.
JohnPM, can REW FFT work with 2ch simultaneously? I believe, I saw some screenshots with two FFTs but can't get any clue with google search yet.
The multi-input capture feature of the Pro upgrade does FFT on multiple channels but only shows the RMS average. Showing at least 2 channels simultaneously is on the todo list for the Pro upgrade.

Sounds very interesting. Do you get this THD performance even at levels close to full scale?
@JensH, the core of the idea is using digital DC-offsets and signal polarities and then a summing tends to cancel "everthing", even order (easy to understand) but also odd order, of both the DAC itself and the first filter stage, after which I currently tap off the summing. My new "0dBFS" is actually -6dBFS as I apply 50% DC, and this is the level I'm testing with, lower levels quickly become even better (as long as I can find something sticking out of noise floor at all with 4M FFT @ 96kHz).

Most importantly, it works extremely well a low frequencies where the non-zero impedance reference voltage feed (via RC) normally spoils distortion.

Of course my instrumentation has to be put under scrutiny, notably that the ADC (@-30dBFS, not notch used) distortion is not cancelling with the DAC distortion. I'll need to build precision balanced notch filters +LNA at various frequencies. My own AP2322 not reliable below -120...-125dB...
@IVX, I also thought about LPF or bandpass to bring down harmonics and noise, but as you say, only fixed frequency. Same problem with notch filters...
The AK5572 ADC in anti-parallel mode is really good when driven below -30dBFS but I don't know exactly how good. Again, fixed osc with LPF would be good to get more data here...
KSTR, BTW, why did you say that your SYS2322 doesn't really see <-125db harmonics? As I noticed my SYS2522 is fine up to -150db but need to know which range in use, 3rd and higher quite stable matched with my notch filter. 2nd a bit worse, and AP gives me H2 often higher than it is, but up to -135db I can trust 2522 for 100%. 2522 analog gen is worse than its notch, for instance, 3V range has H3 -125db but 4-6V range -145db. So need to know the entire map of errors anyway.
@IVX, see https://www.diyaudio.com/forums/equ...rtion-audio-range-oscillator-post6620324.html
Below the -130dB spec I would not be certain if the results are robust. Also depending on level, etc.
Will need to hook up this AK4493 experiment to the AP and see what I get. Same for any tests with additional LPF.
I will assume the 2322 and 2522 are close to identical in the notch etc, so I trust your word here but -150dB is a statement that sound too good to be true....
I've played with range settings already and there is an influence. On the notch and then feeding the ADC for spectrum, higher was better as long as (power) averaging will still bring out the residual. Sync'ed averaging is a bitch with the AP.
For THD+N, lower was better until clipping distortion kicks in.
An interesting approach!
I tested the AK4493 in my own design some time ago and got a THD of -125dB at -6dBFS. If that can be pushed down to around -140dB it would be fantastic. And by using 2 devices per channel the dynamic range should still be around 123dB.
Looking forward to learn more about the details of your design!
Yes. But every time you double the number of DAC's you gain 3dB, in total 6dB with 4 DAC channels, so we are back at 123dB.
On the bench:
4Channel HD-Cancelling-Bench.jpg

Only hardware mod is connecting all the four reference voltages together but this is not paramount. It helps a bit for cancelling and stabilizing in general.

Some preliminary plots at the new full scale point (which runs the DACs at 50%DC + -6.5dBFS signal). Going down in level first leaves the HD levels as is or a tiny bit higher and then, when 10dB down or so, they start to vanish.

1kHz, measured with AP2322 (top) and with the RME's own ADC (bottom)
1kHz AP.gif

1kHz RME.png

And 30Hz:
30Hz AP.gif

30Hz RME.png

30Hz normally look like this:
30Hz (non-cancelling) RME.png

At very high frequencies (3kHz++) H3 starts to dominate, going -130dB'ish, but that may have a lot of reasons.

In general good accordance, and I'm pretty much convinced at ~ -30dBFS the AK5572 in anti-parallel (to cancel even order) without notch is pretty much as good as the AP with notch. The data might even suggest that the 3rd is even more real than with the AP (as cancelling for odd order is much less likely going to happen than with even order).
Last edited:
JohnPM, it is not about Jawa, I used ASIO4ALL v2 and 20-20k REW BW limit was turned On. I did filter up AP's output with Sallen-Key LPF to get low as possible harmonics level to accurately compensate ADC's distortion, at 1Hz that LPF has quite high impedance. Regarding the copy of an analog notch, I thought last night too. I'm sure that AP had to keep that trick even in a digital domain to avoid a mismatch with pure-analog legacy analyzers. Even if that implementation isn't 100.0% correct, AP has to repeat that notch in all new models. And msmucr's experiment confirms that as well.
PS: I did repeat your settings but after 500 AVGs got -114db, AP reads today(slightly lower level about -1dbfs) -118.8db.

I have been reading elsewhere that the generic ASIO4ALLis did not perform as good as an ASIO that was provided by the manufacturer. In other words: There was a marked difference in performance and the recommendation was to stay away from the ASIO4ALL. Cannot exactly remember the old saying from when I was a child but it was something like this: 12 different professions but master in none. Same, same in software, generic driver never performs the same as a dedicated piece written especially for the application. Just my opinion, peace.
Last edited:
Obviously ASIO4All is only ever going to be as good as the sound drivers' WASAPI exclusive / kernel streaming implementation... this comes with being a wrapper. It'll generally work a treat on anything HDA, while if you have an old Asus Xonar D1 or D2, you may find that the only way to have hardware sample rate switched automatically is via an ASIO driver that'll cut out under high CPU load when using a driver version that'll actually still init the hardware correctly past 48 kHz... yeah. (The joys of botched old driver architecture.)
@IVX, understood.

FWIW, just yersterday night I found an unbelievably simple and reliable break-through tweak to get a THD(-N) of -140dB (pure H2) from 2 AK4493's summing the four channels, consistent vs. level and frequency. With a bit of care I might even reach -150dB, APX555 territory. Or get the -140dB with only two channels....
Will post an article once I have it fully fleshed out.
Short update... As to be expected, things at these precision levels are not so easily handled.

It turned out my initial measurements were better than real life due to a damn lucky combination of cancelling mechanisms at work. It also turned out (again not unexpected) that this is very sample dependent, with some chips I got excellent results, with others not so excellent ones (I have only two AK4493 and two AK4490 to test with, as incorporated in the RME ADI-2 Pro FSR and FSR models, resp.). Only two channels are needed, using two chips does not give that much improvement and just opens to many variables for optimization.

At least I now know the limits of my AP better than before and ways to improve resolution. @IVX, rather than using the limited resolution built-in FFT of the AP I'm now using the monitor out for FFT and set the input range to about 10x the auto level detect would normally select, in order to reduce internal voltages as much as possible. Input set to DC to avoid that cap's influence as well. By this I think I can get valid results down to -150dB for 3rd and -140dB for 2nd. 2nd still has variation depending on signal polarity but now it is within small bounds so that I can use the median value as an estimate.

From what I've seen now I think -135dBc is realistic with a "typical" chip, at -6dBFS. 3rd dominating, it only reduces a little by paralleling channels with different DC bias (not necessarily 50%). Even order products are supressed quite well, though.

All this will be valid to the specific circuit and layout of the RME devices (with the only tweak of connection the Vrefs together, atm). In different implementations one might get different results.

I've learned a lot about the specifics of the chips wrt t Vref and Vcm characteristics and after a heavy web search I now found a key document, a presentation from AKM that allows insight why the chips are so damn sensitive to VrefH/L supply voltages and impedances, same for Vcm pin impedance. It's a bit dated and dealing with the previous generation AK439x models but I'm pretty sure the basics of the Switched Capacitor DAC core have not changed much:
... you really can see that ADC board in this year, see attached ;) In that proto I did not solder XLRs yet but you can see a 2.5mm jack for an AUX 1/4 gain inputs to investigate amps up to 40Vrms.View attachment 118145

really a great project !

After reading your other messages, I see you may developp a Pro version later. Do you think it could include a clock output to sync with a DAC for example ?
If the performance is there, it could really be an interesting device to record the stereo mix coming from analog external hardware (mastering or recording a analog mix for example). Ideally, it would need balanced inputs too.

capslock, it could be relatively expensive, about $200 or so, and I afraid if under my brand E1DA it will be popular. 100-150 range should be fine I believe.

I can be popular if all measurements are great ;-)
Adding DAC in the case of audio pro device, two ouputs would allow some kind of work like when you need to work on a stereo signal, and more (8 at least) would allow external mixing.
The goal in this case would not to get the best measurements who you can get, but also highest transparency in a audio loop, which doesn't always match measurements like THD+N measurements.
Top Bottom