Say, what are "day chips"? are they smaller, less greasy than regular chips, more like french fries perhaps?
Oh, you mean that you have to equalise the frequency response of the DUT based on this reference file produced through Diffmaker before you run the "real test" in Diffmaker?. So this requires running the new test through DUT into an equaliser? Can't see how this is automatic or am I missing something?
I have never seen this being suggested in the user guide or done in practise - is it detailed somewhere?
Last edited:
I went back and reread the test methodology and to me it was clear. The reference file is compared to an AD/DAC loop the frequency response of the loop is part of the diff. There is no indication at all of an explicit attempt to remove all the "linear" differences and show only the THD/IMD and other added distortions. I saw mention of timing and overall gain, in fact the gain correction for the whole file is given in the results. As for timing it would make sense to align the beginning and end and do a SRC to make them meet.
Last edited:
No. Because the reference file has been through the loop, it now is equal to the source file multiplied by the frequency response and distortion of the loop. If we neglect the loop distortion, then reference file is equal to the source file multiplied only by the loop frequency response. This is what diffmaker assumes.
If we then set the reference file aside for later, we can go ahead and make a test file for some change to the system. Suppose we connect and test amplifier between the D/A and A/D and play the source file though it. We record the result and call that the test file for the amplifier.
Then in the next step, we have diffmaker compare the test file we just made to the the reference file we made earlier. Since both the test file and the reference file were recorded through the loop, if we subtract one from the other (again ignoring any distortion in the reference file, and assume the only distortion is in any of the files is from the test amplifier), the difference we find should be equal to the test amplifier frequency response and distortion.
That's all fine if the loop distortion is small enough to be neglected.
EDIT: I wrote this before knowing that diffmaker creates EQ files so that the reference file doesn't have to be used for the comparison. Still, reading the above may help with some intuition about how differences in principle could be measured.
Last edited:
Ah, I was missing something - I found this in the help file:
So, let me see if I understand this - if we want to eliminate/mitigate linear frequency differences we need to run a two stage Diffmaker test:
- stage 1: run Diffmaker with a chirp signal or equivalent & capture the eq file
- stage 2: I'm still not sure how this eq file is used in a test of AD/DAC using a music file as the input?
You posted while I was writing my post
I'm afraid I still don't get it - I can't see the connect between doing a freq analysis in Diffmaker with a chirp signal to produce an eq file recording & how this eq file can be subsequently used in Diffmaker when we want to use a music track as input?
I get the gain adjustments & time shift adjustments attempted by Diffmaker but can't see how eq files can be used in the scenario published in Gearslutz i.e analyzing the A/D D/A loop itself? Maybe that is the disconnect I have in understanding?
The chirp has to go through the same DAC loop to capture it's FR, this clearly was not mentioned. So the process would entail timing with a gain/eq. Which still leaves me a question of minimum phase vs linear phase eq. the digital filters in DAC's and AD's can have elements of both. Analog signal chains like amplifiers will rarely (if ever) deviate from minimum phase.
EDIT - To clarify an inverse eq is generated and applied to the DAC loop to make it flat with frequency. It that clearer?
EDIT EDIT - Here's a discussion of deriving the impulse response from chirps. http://www.thinksrs.com/downloads/PDFs/ApplicationNotes/SR1_SweptSine.pdf
Last edited:
His name IS angrypat
Yes but self deprecation in one's handle is usually a sign of a good sense of humor!
By using EQ compensation instead of using the reference file for comparison, that would make it possible to use different source files for testing the test amplifier, relative to the process described in my earlier post above. If would also try to remove loop distortion imprinted on the reference file from affecting any difference calculation if using a new, different source file for testing the amplifier. There is still an assumption that loop distortion is negligible relative to test amplifier distortion anyway.
Last edited:
Yes the chirp uses the same DAC loop to produce an eq file but what is then done with the eq file that can automatically adjust Diffmaker's analysis when running a music sample through the DAC loop - I can't see anywhere that this eq file can be used as an ancilliary input along with the input music sample to apply this linear correction?
This is what I thought was being claimed?
Last edited:
Yes but again I can't see how this works to correct for a music file as input file being used for testing which is what I thought was being claimed?
Where the confusion seems to arise is that we are talking about two separate scenarios:
- one can analyse the A/D D/A loop with a chirp signal for a particular type of analysis
- or one can analyse the A/D D/A loop with a music track for another type of analysis
but it seems the two results are independent & it's not possible to use the first freq analysis as an automatic correction to be applied in the second music track analysis?
Last edited:
Using a chirp signal would guarantee that all frequencies would get excited and thus measured when acquiring the FR of the data converter loop. Once the loop FR is known, it's effect can be removed from any test amplifier test files.
Again, diffmaker is intended to be used to measure changes to the loop, such as the effects of a test amplifier inserted into the loop, not the baseline performance of the loop itself. Diffmaker also only tries to compensate for baseline loop FR, not loop distortion when trying to measure some change to the loop.
Last edited:
This is the same thing I just said.
Chirp -> DAC/AD -> eq file
Reference file -> DAC/AD -> apply eq - > timing and gain - > diff w/o FR error
I see your problem the FR is assumed stationary and the same for any input this is reasonable (not sure to to the -120dB level)
Last edited:
Well, I have two issues with the above
- yes, I have a issue with the FR being the same for any input - does the dynamic signal that is music have the same interaction with DAC & digital filters as a chirp signal?
- my second issue is the highlighted step above "apply eq" I can't see it being applied automatically in Diffmaker when music file being used as input ("reference track")
I think Markw has confirmed what I said earlier, Diffmaker may not be that suitable for analysing the DAC loop itself !
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.