The DUT for this test is the DAC (objective of OP is to compare DAC performance). So if the ADC that is part the loopback test is fouling up the results then you need to address that. Get a better ADC. Try to correct the ADC error. Or abandon the effort, which is what I think has happened.
-
Welcome to ASR. 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!
DAC measurements using DeltaWave
- Thread starter manisandher
- Start date
MaxwellsEq
Major Contributor
- Joined
- Aug 18, 2020
- Messages
- 3,054
- Likes
- 5,206
As I understand it, this is a "why do some people describe differences when measurements say they can't be detected" thread and using "difference scoring" as an alternative test mechanism.
What has been established is this: what the measuring ADC does at or near 1Hz can result in a big "difference score" between a DC-coupled DAC DUT (device under test) and an AC-coupled DAC DUT. But this big "difference score" is inaudible in practice.
Also, if the "difference score" software is modified to compensate for the ADC 1Hz behaviour, the "difference score" is much smaller.
Then the "difference score" is found to be affected by the anti-imaging filter behaviour of the DAC, and (if properly implemented), these differences are also inaudible.
Personally, I found it interesting and learned a lot.
What has been established is this: what the measuring ADC does at or near 1Hz can result in a big "difference score" between a DC-coupled DAC DUT (device under test) and an AC-coupled DAC DUT. But this big "difference score" is inaudible in practice.
Also, if the "difference score" software is modified to compensate for the ADC 1Hz behaviour, the "difference score" is much smaller.
Then the "difference score" is found to be affected by the anti-imaging filter behaviour of the DAC, and (if properly implemented), these differences are also inaudible.
Personally, I found it interesting and learned a lot.
Not quite what ADC does at 1Hz, but rather the effect of a minimum phase DC filter that results in altered phase response across the bandwidth. This can be removed by DeltaWave now by “linearizing” phase response.
Any idea (and also @KSTR ) why the loopback with the caps bypassed does not give a as good or better null than you can get with your software correction?
MaxwellsEq
Major Contributor
- Joined
- Aug 18, 2020
- Messages
- 3,054
- Likes
- 5,206
Agreed - I was trying to simplify the issue a bit for newcomers*... but oversimplified it too much. I was trying to demonstrate that the ADC's DC filter impact on higher frequencies could be determined through analysis of low frequency phase; and that you had added the ability to "desensitise" the effect.
Edit: * newcomers to this thread not newcomers to electronics.
Last edited:
OP
manisandher
Addicted to Fun and Learning
- Thread Starter
- #286
Not as far as I'm concerned. Just been too busy with other stuff, but very much hoping to continue at some point...
KSTR
Major Contributor
In a direct comparison done in situ, there is absolutely no way a DC-capable ADC could ever fare worse that the same identical ADC AC-coupled and DW-corrected.
Deep null depths are so brutally sensitive to everything making it hard to exactly reproduce results. You learn that quickly if you repeat one and the same test on another day and/or with another specimen of the DAC build/revision. Same goes for the ADC. One typical case is not using always the same anti-aliasing / anti-imaging filters, notably when the ADC doesn't run at a higher rate.
What I've learned from dealing with Null Testing for decades now (taking a huge leap when DW came out, thanks @pkane once again) is the following
- above all, restrict bandwidth in DW
- use DC coupled ADC if possible
- if possible use sample-synced record-while-playback, that is one single clock master for ADC and DAC
- run ADC at 2x or 4x the sample rate of the DAC under test (requires either upsampling the source file or downsampling the result for a compare)
And it is very difficult to compare a recording of DAC to the original file because of the unknown contribution of the ADC which may or may not dominate the residual. That is especially true if the ADC has passband ripple like they most always do. One can try use a "reference" DAC to compare against, or use that reference DAC (if known for very low ripple) to measure ADC ripple so one can de-embed it
Not sure at what results you're looking at...
Here's what I see:
1. Stock ADI-2 Pro FS with software DC filter turned off in the menu produces -64dBFS RMS null.
2. Modded ADI-2 Pro FS with hardware DC filter bypassed and software DC filter turned off produces -72dBFS null
3. Applying DC filter linearization to #1 above results in -72dBFS null
In other words, DW software DC filter linearization has exactly the same effect as the hardware DC filter bypass
Last edited:
Sorry, i thought I remembered you getting a result, using correction for the DC filter, better than -80dB on the original2 file. But I think maybe I am thinking of a different file and possibly a different ADI device.Not sure at what results you're looking at...
Here's what I see:
1. Stock ADI-2 Pro FS with software DC filter turned off in the menu produces -64dBFS RMS null.
2. Modded ADI-2 Pro FS with hardware DC filter bypassed and software DC filter turned off produces -72dBFS null
3. Applying DC filter linearization to #1 above results in -72dBFS null
In other words, DW software DC filter linearization has exactly the same effect as the hardware DC filter bypass
That was for a different converter. ADI-2/4 Pro SE. A different set of filters and hardware implementation.
danadam
Major Contributor
- Joined
- Jan 20, 2017
- Messages
- 1,356
- Likes
- 2,150
Are newcomers supposed to know what DC- and AC-coupled mean?
I probably couldn't explain it correctly. Is it that one has DC offset and the other doesn't?
Are newcomers supposed to know what DC- and AC-coupled mean?
I probably couldn't explain it correctly. Is it that one has DC offset and the other doesn't?
I suspect he was referring to newcomers to the thread and not newcomers to electronics
MaxwellsEq
Major Contributor
- Joined
- Aug 18, 2020
- Messages
- 3,054
- Likes
- 5,206
Exactly!I suspect he was referring to newcomers to the thread and not newcomers to electronics
MaxwellsEq
Major Contributor
- Joined
- Aug 18, 2020
- Messages
- 3,054
- Likes
- 5,206
Newcomers to this thread who don't read from post 1. Not newcomers to the fields of electronics or measurements.Are newcomers supposed to know what DC- and AC-coupled mean?
Understanding DC-coupling and AC-coupling are foundational electronics knowledge matters. I would expect anyone with an education in electronics to be fully aware of this, but not a newcomer to electronics.
TL;DR - DC coupling is risky; AC-coupling is safer but may impact phase.
AC-coupling is safe but can introduce phase changes at low frequencies, especially if a minimum-phase filter is used. These phase changes are probably inaudible, but they can affect the null (the point of this thread) if you compare two different devices one of which is DC-coupled or linear phase filtered, whilst the other is minimum phase filtered.
DC-coupling introduces no phase changes, so is ideal. BUT if the device generates a DC offset, this will not be blocked. At best this will result in splats and thumps when switching, at worst will destroy speakers and damage power amplifiers.
Similar threads
