DACs need to be warmed up for a while. Fact or Fiction?

Not for all equipment. I use a HEKA brand amplifier and ADC/DAC in my lab for work. It can accurately measure current down to 3 picoamps at a 100kHz sample rate and needs zero warm up time.

 

Perhaps we should have a discussion about how long the warm-up period is and not whether it exists. As @Cspirou has noted (and is undeniably true) the bulk resistance of a material changes with temperature. Some high power capacity ceramics have as high as 1000ppm! Now, with better resistors, this change is very low (single digit ppm for metal foil, wirewound).

The gain drift for a PCM1704 (an R2R chip of yore) is stated at 25ppm of the full-scale range per degree C. The chip in the Yggdrasil is stated at 0.04ppm per degree C. It seems highly unlikely that the DAC chip itself is the source of such a significant change. Perhaps the output stage or the jitter stability of the clock source.

Regardless, I'd challenge us to think about the reality of how long it takes electronics to warm up. I'd put a reasonable amount of money on the presumption that 90% of the warmup happens within the first 20 minutes with less than a handful of degree change over any remaining amount of time. It seems highly unlikely to me that there is too much difference over a much (order of magnitude or more) longer time period.

 

I'd professionally question your setup's accuracy at any warm up time, much more so wrt to temp change.
Perhaps you meant you have resolution down to picoamps?
There are ofcourse ways to shorten warm up effects to neglible time,

 

You sir have made me open the manual to check out my own BS.

It advises a warmup of at least a half hour before starting my experiment. I've never actually timed my startups but I do know I turn on the amp in the morning way before I start my experiments so I don't have to worry about redoing everything. Phew.

Still, it is nowhere near the 6-48 hour warm up times I have heard of for the Yggdrasil which makes me wonder what is going on with it.

 

Just a hunch but it maybe that the 5791 packages themselves are experiencing some stress from PCB flexing due to thermal growth which settles in after the recommended times of 6-48 hours. Its hard to tell from what I can see on the interwebs there seem to be no relief cuts on the Yggdrasil board for this. I think the 5791EVM board has a cut right across the board to prevent this.

Other possible explanations in Analog's white paper about precision DACs.
http://www.analog.com/library/analogdialogue/archives/44-04/AD5791.pdf

 

Would be great to get comments from the professionals like @schiit or @baldr.

Why do R-2R dacs need to warm up? Does it go faster after equipment is burned in?

e.g. If a dac takes 48 hours to warm up and reach its potential when brand new, does it only take 4 hours after being used for hundreds of hours?

Can anyone speak to the technical aspects of what's going on?

 

I did PM Jason to ask him that very question, but he is probably way too busy to read my PM. I was thinking of emailing Schiit and asking, but I have bothered Nick too much already.

 

Read:

http://forums.stevehoffman.tv/threads/schiit-bifrost-dac-my-break-in-diary.285693/

 

This all looks anecdotal. It's an interesting link, but it's bereft of technical detail or actual data and knowledge. Unless there's a post buried in that thread that actually explains what's going on, it's just another thread talking about burn-in like this one without anyone qualified to speak to what's happening, if anything.

 

thank you Original Ken - Does it make a difference if after burn-in is achieved if you turn on your DAC (in my case the Bifrost Uber) for a while before you start your everyday listening sessions? Or can you just turn it on when you're ready to listen and it will sound the same without warming up (after burn-in period)

 

Please forgive my ignorance, but does this "allow warming up period before use" procedure is actually written (or enforced) in the user manual of most DACs out there? Or perhaps in troubleshooting section even, like:

• Q: "Just received my unit brand new and it sounds bad, really bad!"
  
• A: "Leave the unit on for 5 days straight then we'll talk again, drama queen. Jeez, kids these days."

 

Not to diss the Schiit team but I doubt mission critical applications have the fortune of 48 hr cal times.

On the Yggdrasil I'm not seeing any thermal relief cuts on the large PCB board which would prevent board flex from affecting the 5791 chips. If you read the paper above AD recommends that the 5791 be mounted on as small as possible boards or relief cuts on 2 or 3 sides of the package if separate boards are not possible.

 

I asked Kingwa about the Audio-GD Master 11 (4x PCM1704UK chips) warm-up time, and he said it needs at least 20 minutes to sound great.

That's a far cry from what the rules of thumb I've read would indicate (e.g 2x number of bits = hours for warm-up).

I'm not sure if that means (A) 20 bits * 2 = 40 hours or (B) 20 bits * 4 chips * 2 = 160 hours.

Maybe @purrin can clear up the 2x Number of Bits rule for Multibit/R-2R DAC warm-up time.