I have been lurking around here for a while, commenting on other peoples designs. Well, it's time to put my own head on the block...
This is a DAC based on PCM1704 with a AD1896. It also has five Jung/Didden regulators. I have taken some measurements of this DAC and a TDA1541 with some interesting results. Instead of posting everything here, I made a website for it.
Let the flames begin!
This is a DAC based on PCM1704 with a AD1896. It also has five Jung/Didden regulators. I have taken some measurements of this DAC and a TDA1541 with some interesting results. Instead of posting everything here, I made a website for it.
Let the flames begin!
tschrama said:
I made some test signals in Cooledit and burned them to a CD. The output of the CD-player and my DAC was hooked up to my M-Audio Audiophile 2496 through a simple 10kohm 0.5 voltage divider. I used the Spectralab software.
The jitter-test signal is the same that Stereophile uses for their tests.
I also tried sending the 1kHz signal through a notch filter to eliminate distortion in the soundcard ADC, but this did not change the results (the harmonics did not change), so I dropped that idea.
All in all, the Audiophile 2496 is a pretty decent card for doing measurements.
Set DIR1703 to 24bit right-justified
You can set DIR1703 to 24bit right-justified output formart, then it can compatible with PCM1704. The max frequency of AD1896 is 30Mhz, when use it for 192khz, you need to set it to slave mode(see AD1896 Datasheet). If you use SRC4192 instead of AD1896, the result must be much better and it supports 192KHz without set it to slave mode.
ojg said:
You can set DIR1703 to 24bit right-justified output formart, then it can compatible with PCM1704. The max frequency of AD1896 is 30Mhz, when use it for 192khz, you need to set it to slave mode(see AD1896 Datasheet). If you use SRC4192 instead of AD1896, the result must be much better and it supports 192KHz without set it to slave mode.
Yes, the DIR1703 would have to be set for right-justified, but that would only work for the left channel. What about the right channel? The data would need to delayed by 32 BCLK's to be in sync with the left channel, no?
You are correct about the AD1896 not needing the 49.152MHz though if you run it in slave mode (which I don't), but you would still have the same interface problem as for the DIR1703 above.
You are correct about the AD1896 not needing the 49.152MHz though if you run it in slave mode (which I don't), but you would still have the same interface problem as for the DIR1703 above.
Very interesting.. and I will be awaiting some of your other tests!
The 1704 is a DAC chip I am thinking of building something around, and a friend of mine has an old Cambridge CD2 (?) that we've slowly been modding up.. and it uses TDA1541's, actually 4 of them all summed together..
Only done a few things so far, and it's improved nicely.. Especially interested in the new digital filter being used with it.. as well as trying the AD797's!! (got a few samples to try but haven't got to it yet)
I'd like to know a little more about how you're generating that jitter test tone in CE.. looks like a possible easy way to test modding results!
The 1704 is a DAC chip I am thinking of building something around, and a friend of mine has an old Cambridge CD2 (?) that we've slowly been modding up.. and it uses TDA1541's, actually 4 of them all summed together..
Only done a few things so far, and it's improved nicely.. Especially interested in the new digital filter being used with it.. as well as trying the AD797's!! (got a few samples to try but haven't got to it yet)
I'd like to know a little more about how you're generating that jitter test tone in CE.. looks like a possible easy way to test modding results!
Since there have been some interest in how I did the jitter measurements I updated my webpage. You can also download the wav-file I made for the test.
I calculated the numerical results from the plots and:
SAA7220-TDA1541-NE5532: ~480ps
DF1704-PCM1704-OPA627: ~183ps
AD1896-DF1704-PCM1704-OPA627: ~58ps
Thanks again for your comments!
I calculated the numerical results from the plots and:
SAA7220-TDA1541-NE5532: ~480ps
DF1704-PCM1704-OPA627: ~183ps
AD1896-DF1704-PCM1704-OPA627: ~58ps
Thanks again for your comments!
ojg said:Since there have been some interest in how I did the jitter measurements I updated my webpage. You can also download the wav-file I made for the test.
I calculated the numerical results from the plots and:
SAA7220-TDA1541-NE5532: ~480ps
DF1704-PCM1704-OPA627: ~183ps
AD1896-DF1704-PCM1704-OPA627: ~58ps
Thanks again for your comments!
Hi,
Most interesting web page and measurements
With the 1896 the masterclock is important. Also consider jitter ripple through by the data
You may look at
http://members.chello.nl/~m.heijligers/DAChtml/dactop.htm
for reclocking and low jitter, typically 8ps (measured at DAC clock)
best regards
-
Guido Tent said:
I am not quite sure what you mean by this, could you please explain?
I have visited your webpage in the past and I am most impressed by your PLL-VCXO design. I think this is better than an ASRC, and I will probably put something similar in my next design. I have headers on my DAC-board so that I can easily hook into the signal-path at different places if need be.
ojg said:
Hi
Thanks for feedback, especially on the PLL. I think the AD1896 is the first decent ASRC, the other ones hurt at least my ears, have to try the 1896.
Data jitter affects on-chip clock hence produces distortion at D to A conversion (because of on-chip crosstalk).
We did the experiments, group of 4 people, ABX testing, data input of PCM63 reclocked or not: 80% score.......
Ciao
OK folks, another update to the project webpage.
I have access to an Audio Precision analyzer through work, so I brought my DAC in to take some more measurements. My PCM1704 DAC got -100dB THD+N while the TDA1541 got a still decent -93dB THD+N.
What is interesting though is how much the numbers changed with measurement bandwidth...
I have access to an Audio Precision analyzer through work, so I brought my DAC in to take some more measurements. My PCM1704 DAC got -100dB THD+N while the TDA1541 got a still decent -93dB THD+N.
What is interesting though is how much the numbers changed with measurement bandwidth...
5th element said:
I have been waiting for someone's bubble to burst and finally ask this question!The short answer is, they sound the same! Now that is obviously not a satisfying answer so here is the long one.
I do believe in double-blind listening tests. I also believe that some people have trained their mind to listen for and discern subtle differences in sound despite less-than-ideal conditions. I do not have the equipment for ABX testing, and I don't trust my brain to be objective, especially since one of the subjects is one that I have invested so much time, effort and money (~$500) into.
But what I did hear is this: On "Great Gig in the Sky" from the CD-layer of the latest reissue of "Dark Side..." there was one point where my DAC produced a shrieking sound that made me switch back-and-forth between the Rotel, and I believe it was less pronounced on the Rotel. You all know how bad the mastering of that CD-layer is, but it was an interesting observation nonetheless. Maybe I am too afraid of being biased toward my own DAC that I am overcompensating and finding faults that aren't really there? Lots of psychology involved here unfortunately
This DAC is still a work in progress, and I started this project not because I thought I could make something that was "night and day" when compared with the Rotel, but because I wanted to prove to myself that I could build a DAC according to engineering principles that I feel are the most important, and still measure as well as the high-end commercial products out there.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.