I rarely post praises of commercial gear as a prefer to build my own, knowing that each component is used in the correct topology is important to me. I am no shill as you can see from my 2000 posts on head-fi.

I have to say that this 19MK3 is an outstanding piece of gear. They have done three things extremely well:

1. PMD100 - I'll say it till my death it is the pinnacle of the digital golden age. One day these will be worth $1000 a chip, mark my word.

2. PCM1704K - Erno Borbely told me personally that this is the finest DAC chip ever made, some would argue the AD8162(Jocko) or the PCM63K, but regardless it is a much better design than sigma-delta chips.

3. Zero feedback analog stage - all these $1000 DAC with C-moys on the output make me sick, this is the way to go.

Results is that this DAC can convey the ambience of the recording as well as any source I have heard, I can listen to good recordings that that the guitar would sound distant and dull on lesser DAC's but with the 19MK3 the instruments have life!

I hope we see a trend toward R2R DAC's in this price range with discrete zero feedback output stages.

I have to admit that I am at once intrigued by and skeptical of the audio-gd stuff. But I agree that its nice to see the good chipses in use.

To what other DACs would you compare the 19mk3?

I'm also finding the various models and naming conventions to be extremely confusing on the audio-gd website.

I'm a cynic, but I'd have to guess they were using knockoffs of the chips in question

Does Audio-gd use non-fused power supply for their DACs?

I'm glad you brought up the zero-feedback -- that is intriguing, and I don't think it's been mentioned before.

Colin (cetoole's) stage is zero feedback; jocko's (and all the variants I've seen including the rbroer one I have used) are also zero feedback. I like the sound based on what I've used heard, particularly compared to opamp stages (which by definition use a lot of feedback).

It would be interesting to know if the chips are legit. How would one check?

Ugh, I never considered the possibility of knock-offs. Unfortunately, you're right grawk, that's a valid concern.

PMD100s haven't been made for what? A decade? And now someone is introducing a product built using them? I'm not sure what the projected volume of these things is, but 100? 200? I haven't seen any sellers touting having that many of them. Legit? Perhaps...

Yeah, but that doesn't mean they don't have a legitimate stash of them. Not everything is on eBay.

It should be fairly easy to test, no? Does it decode HDCD?

Oh, I'm not suggesting it's something other than a PMD100 and 1704, just that they're knockoffs of the chips.

Which would make them something other than ersatz PMD100 and PCM1704, no?

PMD100 knockoffs are certainly easy to generate since you just need to copy the code over to another chip. Of course China also is most likely where most of the old stock is located. PCM1704s I would guess are pretty difficult to knockoff.

PMD100 knockoffs are certainly easy to generate since you just need to copy the code over to another chip.

And presumably, that'd be just as good since the code is really the thing.

PCM1704s I would guess are pretty difficult to knockoff.

Pfft. You just need a steady hand with the laser trimmer. And really, really good eyesight.

PMD100 knockoffs are certainly easy to generate since you just need to copy the code over to another chip. Of course China also is most likely where most of the old stock is located. PCM1704s I would guess are pretty difficult to knockoff.

I'm not sure. The PMD200 is a Blackfin, but I suspect the PMD100 may be hardwired silicon.

PCM1704s I would guess are pretty difficult to knockoff.

Also: this. So yeah, either they're (real/genuine/non-cloned) PCM1704s, or they're not. I'm not sure they could be "knocked off" (pretty sure that isn't the correct way to put that in the past tense, but...whatever).

They could also be "non authorized", meaning the tooling was stolen or what have you.

IIRC, some people have tried marking regular PCM1704 chips as the K grade ones.

What they are doing is pulling and re-using old chips from equipment others throw away. The cost to tool up and reverse engineer a PMD100 or PCM1704 would be exponentially higher than finding junkyard chips. Unless he is selling a million units a years there would be no payback.

And yes I veryified that the PMD100 decodes HDCD. Also swapped the chip with a PMD100 from a 1994 Audio Alchemy and they sounded the same.

In a decade DAC's like this with a good digital filter and an R2R DAC are going to be like gold, just like old tube equipment is now. Sigma-Delta chips are flat out inferior.

As far as stamping a "K" on the (non K) PCM1704's, I wouldn't put that past anybody, even TI !

I'm glad you brought up the zero-feedback -- that is intriguing, and I don't think it's been mentioned before.

The Adcom you heard sunday had no NFB in the analog stage.

They could also be "non authorized", meaning the tooling was stolen or what have you.

Or possibly rejects, parts which work, but didnt meet spec.

I'm not sure. The PMD200 is a Blackfin, but I suspect the PMD100 may be hardwired silicon.

Almost certain the PMD100 is an ASIC.

I think you mentioned that yours didn't; I didn't know it about the Audio GD.

The Adcom (modified by Colin) rocked. It was right up there with my Modwright modded Pioneer.

What they are doing is pulling and re-using old chips from equipment others throw away.

:( all those poor dacs

dacs have feelings too!

Naa, Kingwa doesn't bullshit about parts being real. If anything, it pisses him off as much as it does us. He asked me once about where to buy some or other part overseas as he completely didn't trust places in China.

I did briefly try a DAC19MK3 belonging to a Japanese customer who was having issues (with his computer as it turns out, not the DAC) and it was significantly better than the Compass in detail. Other than the Ref 1, I didn't have, nor could I access another other DAC to compare it to, but I can say it sounds like its specs suggest (PCM1704UK, PMD100) -- smooth, but not over-smooth like the Assemblage DAC3. Somewhere in a thread on the other site someone posted the RMAA from a Japanese site of it, which was flat with the usual roll-off at 20k from the filter. I didn't notice anything particularly specific using the different inputs, but I wasn't trying to. I reckon it would be great value if it were balanced and not just SE. Would be interesting to compare it to the usual suspects between $500-1000, such as the Stello, Benchmark and PS Audio and see how it fares in, say, a $1k Stax rig with vintage Lambdas. IMO and all that.

Currawong, how is the coloration with the DAC19MK3?

The Pico is very nice but after spending time with it for 3 weeks, I suppose I could use something more neutral.

Currawong, how is the coloration with the DAC19MK3?

The Pico is very nice but after spending time with it for 3 weeks, I suppose I could use something more neutral.

It's voiced neutral, and is smooth and natural sounding. It doesn't have the colouration of the Compass, for example. Still the same overall sound signature though, which to people who are used to artificially bright gear, sounds "dark".