A NOS 192/24 DAC with the PCM1794 (and WaveIO USB input)

@Zoran

few measurements just to complete the picture why the XY-2 is not so great but also shows the naked truth of a R2R DAC without calibration and without filters at the Output.

Zout
5kOhm.

THD+noise measurement at 0dB signal level:
filter 500kHz 0,65%
filter 30kHz 0,12% (3rd order)
Filter 22kHz 0,06% (5th order)

I also run some FFTs with following observations:

  • Sample frequency occurs at the output pretty heavy with -50dB versus signal level.:oops:
  • Regardless of sample frequency and signal level, there is a lot of spurs and noise between 3kHz and 10kHz at a level round -95dB... these levels just stay regardless of signal
  • Noise above 20kHz is not rising, on the contrary, with lower signal levels it drops significantly.

the noise in the audio band and the -50dB sample spur are most obvious and disturbing signals. They also ruin the linearity of course as they overlap low level signals

As said, when I have some time, I will listen again with some filters at the output...
 
  • Like
Reactions: 1 user
I promised some time go to post on my tests with chokes in the power supply. I just wrote a blog entry on this theme…
Link to blog post
Hmmm. Am I correct that you simply switched between different chokes with different inductance and DC resistance, but the C parts were always the same? This would mean that the performance of the CLC filter will be less than optimal for a given space/money/whatever constraint, and you're essentially comparing apples and oranges. I'd suggest to first define the constrains (money, space, weight, whatever...), then optimize the CLC filter to provide the best possible filtering given these constraints. The PSUD software can be quite useful for this kind of exercise.
 
It is not as bad as it reads from your comment. I used spice simulations for the optimization and the resonance frequency and overshoot, which are the most important, are far below the rectifying noise components. The AE and Lundahl are very much the same in terms of specs. The Hammond is much lower, but that was on purpose to see what the effect is when using apples in stead of oranges To use your words. The AE one was optimized with SPICE for the dddac psu and the Lundahl was similar, so the sonics difference for me was the amorphe core. Which is similar to other subjective listening sessions with chokes in tube amplifiers. Given the fact that the Hammond was so far away in terms of specs, but relatively close sound wise, was exacty what I was trying to do. Listen to real world chokes and see what it does. Conclusion stays. Any choke improves as long if you are not too far off. PSUD can indeed help here. If you follow my tests, it will be as good as it probably gets in terms of optimized result Without use and understanding of software simulations or paper calculations.
 
  • Like
Reactions: 1 user
Hello Doede,
I have a LL2771 with adjusted airgap . The two normally available are higher current rating so asking Lundahl to adjust it to one that can take 400mA with the two coils in parallel will give you considerable more mH ( 3500 mH) at no extra cost because they just give it a different airgap during " construction".
Of course this is a big boy but if we dont try we will not know.
Best would be to have a second test with choke input. It could well be that choke input could give a bigger difference than the differences between the 3 chokes in a CLC configuration. After all we wont have a shunt supply anymore so the load will vary.
I will be able to send it after arrival home around 6 March.
Greetings, Eduard
 

Attachments

  • 2771.pdf
    226.4 KB · Views: 29
Hi everyone.

I was wondering if these two not gates in series (circled in red) are only a buffer for the BCK signal going to the shift registers. Is this correct or do these gates have any other function?

1709472346642.png
 
Inspired by doede I took some bleeder resistors I have and connected them the ‘’doede way’’ after the Lundahls and Hammond choke, in my power sypply. The two resistors for the Lundahl chokes are rated 4 ohm and the one for the Hammond is 1 ohm. As the picture show, I have bigger trafos (80vac) than the stock ones from doede.
All the chokes are 1.5A, and 160mh (serial connection). The improvement was easy to hear. Tighter bass, in fact everything sound better.
 

Attachments

  • IMG_1499.jpeg
    IMG_1499.jpeg
    778.5 KB · Views: 88
  • Like
Reactions: 1 users
I have a 1794S (blue) mainboard and just stacked up two more decks from the last series. On one of the channels I can't adjust current. There is no voltage drop across the 100R resistor.
Solder joints are redone. Trimmer has been taken out and checked and is ok. Voltage across VB+ and the associated GND is 12,4V. Turning the trimmer doesn't change current, voltage across test point stays at around 35mV.
Voltage drop across the 7810 on the "good" channel is 1.85V, across the 7810 on the "bad channel" it is 2.12V.
Does anyone have a hint how to troubleshoot this further?
 
Last edited:
OK, so we have the cause, not enough voltage for the PCM1794 to work.
Could you please measure the following voltages:

at both BAD and GOOD side of the Dac Module:
Voltage across the resistors directly next to VB+

1710573026522.png


At the bad side:
Voltage input AND output vs GND of the 7810
Voltage input AND output vs GND of the LF33

Just ahead of things, as you are in Germany (?) if all fails, you can send it to me for check & repair...
 
ok, seems the error is in the embedded Tent Shunt or the DAC chip itself....

There is nothing you can do yourself here, as the Tent circuit is not public.
Also replacing PCM1794 is not easy if you do not have the right SMD soldering tools.

probably best is to send the stack (mainboard with dac-modules connected) to me. no chassis or PSUs needed....

best regards
doede
 
  • Like
Reactions: 1 user