Global Feedback - A huge benefit for audio

[Markw4]

Thank you. I've seen those DNR vs. DC slides before, and I'm asking myself if anybody else ever showed such results. I'm eager to try... though it's unclear what level of DC step size would be needed to expose the "noisy" DC levels, I hope it's not down to full 24 bits.

Then, once I could recreate such a plot for the AK4490 or 4493 I could use a short and low level music signal on top of very slowly moving (or stepped) DC and do subtractive tests. There would be a consistent noise floor difference when choosing proper "operating points", wouldn't it?
Hopefully, it would be not buried in analog and other noise which otherwise could only be reduced by sync'ed averaging which would require a known (found) period size to avoid the averaging taking out the modulation noise as well.

Thanks, see what you mean. Regarding the number of bits to test, from looking at the various noise floor verses DC level plots ESS shared it seems like the pattern can vary a lot with dac design particulars. If a more coarse set of measurements were made first, the results might suggest what sort of general pattern a dac produces. Based on that it might only be needed to investigate a repeating pattern at finer bit resolution one time, not sure. Along with noise floor modulation one may also see spurs sweeping across the audio band as DC level is changed. IIUC those would be 'idle tones,' another undesirable artifact.

 

[bimo]

I agree, there is no much interest in such tests I guess i know why... Without the "solid starting point" story about 3 camps sounds like story of 3 cults/religions.

So, you expect an expert tell you which is the best and without try it yourself and then you just believe it?

If you learn physics or have an EE, just explore what make sense to you and learn what the best for you.

I learn from other people including syn08 who I do not like his personality. But if his opinion make sense I will try to prove it myself. I design many amplifiers from very cheap and relative expensive. I have many friends who give me a chance to try their expensive speakers with my amplifiers.

 

[KSTR]

Thanks, see what you mean. Regarding the number of bits to test, from looking at the various noise floor verses DC level plots ESS shared it seems like the pattern can vary a lot with dac design particulars. If a more coarse set of measurements were made first, the results might suggest what sort of general pattern a dac produces. Based on that it might only be needed to investigate a repeating pattern at finer bit resolution one time, not sure. Along with noise floor modulation one may also see spurs sweeping across the audio band as DC level is changed. IIUC those would be 'idle tones,' another undesirable artifact.

In a first test with AK4493 I don't see anything, no noise increase (rms, peak sample, FFT), and no birdies. Ramping up from 0 to 10% DC in 10bit resolution, each step being 9seconds, 128k FFT. Will try smaller steps over a larger DC span but don't have to much hope finding something.

 

[Markw4]

KSTR,
Just curiously, as a double check is there confirmed stepped DC offset at the output of the dac as the PCM codes are stepped?

 

[bogdan_borko]

So, you expect an expert tell you which is the best and without try it yourself and then you just believe it?

If you learn physics or have an EE, just explore what make sense to you and learn what the best for you.

I learn from other people including syn08 who I do not like his personality. But if his opinion make sense I will try to prove it myself. I design many amplifiers from very cheap and relative expensive. I have many friends who give me a chance to try their expensive speakers with my amplifiers.

Oh man, you know what I want and what I have tried.. You're one of the clairvoyant members to..

 

[KSTR]

KSTR,
Just curiously, as a double check is there confirmed stepped DC offset at the output of the dac as the PCM codes are stepped?

Yes, the two DAC sections of the ADI-2 Pro are fully DC-coupled. I sometimes use the secondary DAC for +-12V supply of small circuits.

It could well be the AKM is the one competitor DAC shown in that ESS slide that has constant noise vs DC also, albeit at slightly higher level (-120dB vs -130dB). -120dBFS is what I actually measured ballpark using the ADC at +15dBr gain vs. the DAC, showing -105dBrms, 50Hz...20kHz bandlimited.