Valve DAC from Linear Audio volume 13

A musing about sigma-delta and multibit DAC'S (and then this present DAC): as we know, is harder to implement muitibit DAC's in entirely pure CMOS tech, see for example the efforts/tricks that for eg. Sanyo applied to some parts to made it work (eg: the LC7881 IC), and the other makers, like PCM63 and AD1865 having mixed techs (using bipolar tech), and at least, using laser trimmable R sections, for not having the issues/shortcomings found in DAC's like the LC7880 series (they are not very linear...), but this results in a expensive IC. And various other issues I know and some that I suppose. So, the sigma-delta seems to be very compelling to CMOS tech, to use it as a cheaper options than the expensive multi-bit ones, at the time that processing is more cheap, for implementing the sigma-delta modulator. Then just it was made popular with this tech.
Is interesting and at same time intriging to see that implemented with the oldest tech, the triode! At the time I readed the article, was very clear why this need. I can only imagine the mess about making a multi-bit pure valve DAC; the LC7881 issues for eg. reminds the reality....
If you think, valves are used for things like video AM modulation and QAM color modulation for TV's, in high frequency, so is again interesting to see that new application with old parts!

Great that Marcel choose to implement very clever/advanced sigma-delta modulation (for better simple valve demodulator performance) and digital filtering, so we can count with a refined option at all. I'm eager to see (listen...) the results!
 
Hi Marcel,

sorry I dropped off the radar after my inquiry about how to set RV3 and RV4. The cause was that I did not get any reportable conclusions from my investigations. All I can say is that to my ears there is a sweet spot or sweet ratio where the DAC sounds its best, alas, I did not find a repeatable way of reaching it. I blame the multiturn trimming, next time I'll use single turners.

But, I have a different matter today. My DSD-ValveDAC has started a habit of turning silent during playback. HQplayer and Roon keep on pushing the playback indication, counting time, but there is no sound anymore. Changing tracks does not do anything, reconnecting with HQP neither. Only switching off and on cures the silence.

I think I have sorted out that neither HQPlayer nor Roon are responsible, so my guess now is that somehow the mute function is engaged during playback. I will also ask JLsounds if they have an idea if there I2S board could be the culprit.

For now my question is - where can I check on the ValveDAC board it the mute transistors have been engaged to mute?

Cheers,
Achim
 
Hi Achim,

That's very unfortunate. For the DSD-only valve DAC, you could try measuring these things if they are safely accessable:

Voltage at pin 6 of U14. If the voltage is close to 5 V, the DAC is muting, if it is close to 0 V, it is not. If the voltage at pin 6 is 5 V when it should not be, please also measure the voltages at pins 3, 8 and 11, as that may help to narrow down where things go wrong.

If you don't see anything out of the ordinary at U14 pin 6 or if you can't access it, you could check the gates of Q1...Q6. The voltage at the gates of Q1...Q6 should be around -16 V in normal operation, around +0.5 V when muted. Mind you, these are fairly high-impedance nodes. When you measure the voltage to ground with a meter with 1 MΩ internal resistance, that -16 V becomes -7.6 V and the +0.5 V remains +0.5 V.

The voltage across C36 should be about -13.4 V during mute, -16 V in normal operation (about -12.4 V and -14.6 V with a meter with 1 MΩ internal resistance).

If for whatever reason the gates of Q1...Q6 are not properly pulled down, for example because R51 has come loose, the DAC can go into and out of mute at random depending on what leakage current wins. You probably hear a fade-out then.

The DAC can also randomly go into or out of mute when an input of U14 is loose, but muting would then happen rather abruptly, without fade-out.

The high voltage out of U14 should be within +/- 6 % from the theoretical value, but the other voltages depend on various tolerances, so don't worry about deviations up to +/- 20 %.

Best regards,
Marcel
 
Last edited:
1720981067232.png


Is it also ok to measure the output of U14-6 at the input of R45/R49? That should be easier, as I do not have test leads for the small U14 pins.
 
Ok, here are my results:

When playing ok I get 0,11V on the P12-mute, 4,6V at P12-DSDon.
When the dropout occurs (did not take long after first powering up cold), Mute goes to 4,6V. U14-6 at 4,6V as well.

But: I then pulled the Mute input at the Amanero connector.

When the dropout occured again (this time taking really long, 70 minutes or so), P12-mute stayed at 0V (of course). U14-6 at 0V as well.
In this state I then connected Mute again at the Amanero connector - this led to U14-6 at 4,6V.

So, are you with me here that this points to

a) everything OK on the DAC board
b) the hiccup occuring most probably on the I2S board or its communication with the PC/HQP

HQP steering a different DAC over ASIO does not run into dropping out, so it must be the I2S board, I think.
 
I think I remember that the JLsounds I2S board has a function to send digital silence as a DSD stream when in pause mode. Will see what Lyuben says.

It all points to the interruption happening there, not in the ValveDAC.

Phew, but could still be a can of worms... ;-)
 
The only commercial DAC I reviewed in my system against my DSD only ValveDAC was the Gustard R26. The Gustard was a bit better in resolution (because I run my VaveDAC DSD64 only), but did not have such a natural and relaxed presentation. Since I only run my ValveDAC in single ended output filter configuration, the range of my comparison is limited.
 
  • Like
Reactions: Oneminde
My self-muting problem seems resolved. After much head-scrubbing and looking at possible faults on the PC side, I accidentally spotted a new firmware version on the JLsounds webpage. After a reflash of the board, there has been no further unwanted muting.

Since there had been no other changes in the environment, I think it was not the firmware update but just the reflashing that solved the hiccup. Some bits in the EEPROM must have gone wonky. Blame the recent sunstorms...
 
My first time glancing over this thread so I have plenty of reading I still need to do to get there, but I have a tube rolling question I'd like to ask:

I don't have many E88CC tubes to hand (yes I could/will buy more) but I got to wondering why E88CC was chosen; presumably for its high transconductance? does tube mu & linearity matter in this part of the circuit? Perhaps with some major/minor (?) circuit modification other high(er) GM tubes could be used instead ? Has anyone tried other tube types? Perhaps 2mm banana plugs could be soldered on the pcb socket pins to allow for tubes with different pin arrangement and types of socket.

Thanks
 
Last edited:
  • Like
Reactions: rockies914