[tomb]

  USB can be anything from 4.5v to 5.2v with a nominal of ~5v.  In reality it fluctuates depending on supply load, for example plug in a USB memory key and copy some data to it while monitoring the voltage and it will jump all over the place.
 
One thing i have found the PupDAC susceptible to is dropouts due to spikes in the output ground. I can fairly consistently make it drop off by power up my valve amp connected to it, or even turning on my cheap Chinese eBay soldering station can cause it to drop off, though interestingly the USB connection to my UPS drops off too when i use it.
 
It doesn't seem to make any difference if i power it from the computers USB output, Doodlbug or a completely overkill power supply / Isolator box i previously built so it has to be small voltage spikes coming back into the RCA output grounds that is causing it, or a voltage differential between the input and output grounds.
 
Whatever amp you are connecting to the PupDAC, make sure its powered from the same power point as the computer, other than that i haven't really got any solutions to avoid it.

I've experienced some of this, but seems like with the DoodleBug it comes back online without a manual disconnect-reconnect.  It sort of reminds me of why we went from BUF634's with headphone amps in the early days and went to discrete parts, instead.  The OPA2836 may provide ample protection against frying the DAC - a serious problem with the Starving Student and Alien/BantamDACs and Gamma 1's.  It's one reason I'm a whole lot more careful with any source when connecting/disconnecting to a powered-up amp (I simply do NOT do it anymore.).  Yet, what we're describing might be attributed to less output protection circuitry in the opamp.
 
Of course, whenever comparisons were made - the less protection circuitry, the better something sounded.  So, I would rather accept the hassle of it knocking offline sometimes with the understanding that it will sound better  The BUF634 has a lot of protection circuitry on the output, but it always sounded kind of "tame."

  I don't get that impression with the OPA2836 (maybe a little with the OPA2835).  I'm not saying for sure that's what is happening with the Pup, but the DoodleBug seems to put it all in the context of either a trivial hassle at worst, or at best - it improves everything, especially sound quality.

 

[mcandmar]

That is interesting, i have found when its offline it emits static out of one channel so that would imply it is the output opamp that shuts down. If you unplug the DAC and wait for the capacitors to discharge it goes away, reconnect it again and away you go.  I've actually gotten used to powering up the amp before the DAC's power supply so its not an issue for me.
 
I certainly wouldn't want to add anything that would alter the sound as i love this thing just the way it is, only bit of audio hardware i own that i haven't felt compeled to modify.

 

[COOKIESnPIE]

Post deleted by user.
 
 

 

[mcandmar]

The noise i was referring to is only when the DAC gets knocked offline.  Are you saying you get a noise when the amp is connected and running?   

 

[COOKIESnPIE]

Post deleted by user.
 
 

 

[tomb]

Even my "official specs" are pretty poor compared to a PupDAC connected to a DoodleBug.  Still, your graphs are much worse than any I've seen on a PupDAC.  Unfortunately, it only confirms what we already know - something is wrong with your PupDAC.
 
You did not build from a Beezar kit.  We haven't found anything amiss, necessarily, on your build's pics.  So, maybe there's an issue with the parts themselves.  We talked about the resistors and apparently you confirmed that they are of good quality.  However, what about the rest of the parts?  What kind of oscillator did you purchase?  Is there anything different about the onboard regulators you chose?  Could the ferrites be rated too low in amperage and introducing resistance into the circuit where they shouldn't?  Is there a particular reason you chose to use the OPA2835 instead of the OPA2836?  The difference in the two chips is extreme, to say the least.  That said, the OPA2836 would probably be less stable in this situation, being a much faster chip in MHz, slew rate, and current impulse.
 
Anyway - look at those parts.  The charge pump might be a source of trouble, too.

 

[mcandmar]

  Oops - I thought you got the noise even if the DAC wasn't knocked offline. I've made several posts about the noise (it's from the DAC and occurs regardless of amp or computer), and the fact that I have high DC offset. Start here at this post and work your way down.

 
Sorry missed that.  I agree with Tomb its most likely something to do with the output opamp, since the DAC doesn't use any DC blocking caps in the signal path it either the chip or the half a dozen resistors in the signal path.  Personally i don't like those resistors, i certainly wouldn't use them in a signal path so for the sake of a few $'s bin them and fit some decent low noise vishay's as per the BOM sheet.  I would also try measuring for DC around resistors 5,7,16, and 18 to verify the DC is being generated from the opamp or further upstream in the main PCM1794.

 

[COOKIESnPIE]

Post deleted by user.
 
 

 

[tomb]

  I'll look at replacing U6 and U7 when I'm shipping in parts for my next project, and maybe pickup some mil grade Vishays as well. My oscillator is not the one stated in the BOM, as that (like the OPA2836) was sold out at Digikey at the time. I used the subsitute recommended by tomb (ASV-12.000) instead. The ferrites and regulators are the official ones on the BOM.

How would you measure the DC around the resistors?

Stick one probe of your DMM in Ground, then use the other probe to touch the resistor leads of each resistor and note the measurement.  It would probably help if you measured the "output" lead in each case, since that would represent the voltage after the resistor.
 
FYI, if the DC measurements are way off prior to the opamp, something else is wrong.  If the DC is equal, same resistor on left vs. right channel, then the opamp is the culprit.  At least in terms of offset ...

 

[COOKIESnPIE]

Post deleted by user.
 
 

 

[mcandmar]

That didn't look bad so i measured mine out of interest and the figures are very close, though mine is silent.
 
R5 130mv/122mv
R7 130mv/122mv
R16 129mv/121mv
R18 129mv/121mv
 
Right -1mv
Left -5mv
 
P.S. I started expiermenting with RMAA, how did you guys get the noise floor so low?  I tried two mic inputs my machine has, onboard Realtek, and a Sound Blaster X-Fi and both are ~100db when i have the mic level set for ~ -1db level?

 

[COOKIESnPIE]

Post deleted by user.
 
 

 

[mcandmar]

/face palm moment. Good point sir, don't know what i was thinking there...doh!
 
I'm wondering if repeating the same tests but measuring for AC would tell us anything, i.e. is the noise measurable, and is it measurable before the opamp. I think the DC offset is a red herring and not related to the noise issue you are having as its not much higher than my readings.

 

[COOKIESnPIE]

Post deleted by user.

 

[mcandmar]

Now were getting somewhere...
 

 

 
@COOKIESnPIE I'm all out of ideas