Review and Measurements of Soekris dac1421 Multibit DAC

[andreasmaaan]

I think on this forum we have come to the realisation that any differences in DAC performance are minimal unless poor design/implementation or tailoring is apparent.

It is sad that individuals get so hung-up on this essentially high performance gear over more flawed items in the reproducing chain.

I wish we could move on from these discussions of trivialities.

I guess I just wish we could be told what measurement would reveal the difference, make the measurement, and come to a conclusion, before moving on...

 

[Wombat]

I guess I just wish we could be told what measurement would reveal the difference, make the measurement, and come to a conclusion, before moving on...

I believe that has been made clear many-a-time. Threshold of hearing, subjective bias and lack of evidence to the contrary.

 

[andreasmaaan]

I believe that has been made clear many-a-time.

What's the magic measurement then?

(I'm also ready to move on of course - I'd just like to be told by a designer/proponent of R2R DACs what the evidence is supposed to be so that it can be tested and dis/proven)

 

[Wombat]

What's the magic measurement then?

(I'm also ready to move on of course - I'd just like to be told by a designer/proponent of R2R DACs what the evidence is supposed to be so that it can be tested and dis/proven)

I think such absence from proponents is enough to give little credence to their claims.

 

[soekris]

With current methods and equipment you can only measure a limited set of basic characteristics, then you need to switch to a much more sensitive method, the human hearing using music.... Unfortunate that require good ears, patience and time, so some prefer to stay with what measurement equipment can measure.

When I design I first get the basic characteristics right, then afterwards its a job for the human hearing to finish the design.... Just like what most high end audio equipment manufacturers do.

But that's just my subjective opinion.

 

[andreasmaaan]

With current methods and equipment you can only measure a limited set of basic characteristics, then you need to switch to a much more sensitive method, the human hearing using music.... Unfortunate that require good ears, patience and time, so some prefer to stay with what measurement equipment can measure.

Do you have a theory as to what it is that your ears are able to detect that equipment cannot?

 

[March Audio]

DonH56 explain it better than I could.... I also believe that simpler designs wins, and a R-2R DAC is simple, parallel bits in, audio out, no processing, that's it. While a DS DAC have a lot of processing to get something usefull out of those 1-5 bits.... Of course a great DS design can beat a crappy R-2R design, but a great R-2R DAC will beat a great DS DAC in sound quality, and my customers feedback confirm that.

I am one of your customers and sorry but I can't confirm that. When accurately level matched and blind tested it sounds fine, but not fundamentally better than a good DS dac.

Unfortunately there is no real evidence to justify your words there.

 

[soekris]

I am one of your customers and sorry but I can't confirm that.

Unfortunately there is no real evidence to justify your words there.

That's ok, I have plenty of other customers that have sent me very happy emails, love getting those to confirm the hard work is not wasted.

 

[DonH56]

Thanks Don,

<elided>

Oh BTW there appear to be amipro converters, found this in a quick search

http://file-convert.com/a_ami.htm

Forgot to say thanks for the converter. I actually bought that a year or two ago, but it failed miserably with my files, which were mostly presentations or papers with lots of graphics. It totally hosed the graphics when it converted them at all, and of course the graphics were what I really wanted to preserve. Oh well. I did find instructions to build a system using the older Word files, and even a site that has a way to load and install AmiPro on a Win8/10 machine, but the process looked involved and too likely to mess up other things (e.g. it would change the system fonts, among other things). The really frustrating part is I cannot find the old hardcopies and backup Mathcad/Matlab files to recreate the data -- I have a feeling they are on floppy disks in a box in the basement. Might as well be in Andromeda.

 

[DonH56]

Your contributions are always hugely informative @DonH56, thanks.

This last point of yours surprises me a bit though. The better-measuring R2R DACs like @soekris' seem to have distortion in the order of 0.005%, which is very unlikely to be audible.

The cheaper/worse R2R DACs (e.g. Schiit's cheaper models) have much higher (probably audible) distortion, but it doesn't seem to be lower order particularly.

And then the better DS DACs these days have distortion levels well below 0.001%, often more like 0.0001%.

Finally, for all good DACs, we seem to have noise levels well below -100dB.

Is there really any possibility that, when looking at decent-measuring DACs, distortion and noise could be the source of audible differences?

Knowing what we know about thresholds of audibility, it seems extremely unlikely to me.

Thank you!

Should have noted I was thinking about DACs in general and not Soekris' DACs which seem to have both exceptional performance and have taken great pains to compensate the typical problems of conventional DACs.

I strongly suspect that the DACs themselves are audibly the same unless something is wrong. But, many other things can go wrong, such as poor grounding adding noise and/or jitter (including lack of isolation/buffering from the incoming data power/ground lines), poor output filter and buffer amp design (leading to images above the audio band that could be mixed/modulated down to cause other problems in the audio band), poor power supply design, poor interface design (e.g. USB or S/PDIF or TOSLINK transceivers), etc. I have insufficient basis from which to draw any sort of conclusion in general; sometimes it's hard to keep from musing out loud even during a technical'ish post. Some of it gets into masking effects and psychoacoustics that is definitely not my area. Amir and others have done a much better job and describing why audible differences might exist; I tend to (try to) stick with the technical issues I actually know something about. I often fail, however...

There are a wealth of other tests that could be performed that might highlight differences among DACs. But that takes a lot of time and most of us don't have Amir's equipment to play with. Heck, I still want a simple 'scope, but think I may sink money into better surround speakers instead.

 

[March Audio]

That's ok, I have plenty of other customers that have sent me very happy emails, love getting those to confirm the hard work is not wasted.

Oh it's not wasted, I would happily use your dac, the sound is good and competitive with the market. I just don't hear this alleged magic that is supposed to accompany multivitamin DACs. There seems little evidence to support it.

Haha auto correct is fab, that should say multi bit

 

[restorer-john]

I just don't hear this alleged magic that is supposed to accompany multivitamin DACs. There seems little evidence to support it.

I equally wonder whether anyone hears DS D/A converters as being better.

At the time DS hit the market, multivitamin (I like that one) D/A converters had reached their peak performance for 16/44. The early DS converters did have a 'softer' sound (read: noisy). I think they were a step backwards, but obviously they have improved immensely in the 28 years we've had them.

It was the wholesale commercial adoption of D/S and the complete extinction of the more expensive laser trimmed ladder networks on multibits, that created the opportunities for 'diehards' who held onto their SOTA multibit machines to extol their virtues to anyone who would listen. Discrete R2R D/As are a result of this IMO.

The massive cost savings of DS won out. You only need 1/3 to 1/4 of the regulated power supplies to run DS D/As and they are not remotely as critical. The D/A chips cost a tiny fraction of the expensive big 28pin Burr Browns of yore. Manual (up to 4 bits) MSB THD trimming was unnecessary.

I love my older D/A converters and one box players. They simply perform better in every respect on 16/44 than any external transport/combo that can be dreamed up.

 

[gvl]

I equally wonder whether anyone hears DS D/A converters as being better.

At the time DS hit the market, multivitamin (I like that one) D/A converters had reached their peak performance for 16/44. The early DS converters did have a 'softer' sound (read: noisy). I think they were a step backwards, but obviously they have improved immensely in the 28 years we've had them.

It was the wholesale commercial adoption of D/S and the complete extinction of the more expensive laser trimmed ladder networks on multibits, that created the opportunities for 'diehards' who held onto their SOTA multibit machines to extol their virtues to anyone who would listen. Discrete R2R D/As are a result of this IMO.

The massive cost savings of DS won out. You only need 1/3 to 1/4 of the regulated power supplies to run DS D/As and they are not remotely as critical. The D/A chips cost a tiny fraction of the expensive big 28pin Burr Browns of yore. Manual (up to 4 bits) MSB THD trimming was unnecessary.

I love my older D/A converters and one box players. They simply perform better in every respect on 16/44 than any external transport/combo that can be dreamed up.

I have a Parasound D/AC-1000 with a pair of 20-bit PCM63P-Js inside. I recently re-purposed it for my desktop near-field system. Listening to Jimi Hendrix's Live at The Fillmore East, for a change, and it is hair-raising. I know, you can't measure that.

 

[restorer-john]

I have a Parasound D/AC-1000 with a pair of 20-bit PCM63P-Js inside. I recently re-purposed it for my desktop near-field system. Listening to Jimi Hendrix's Live at The Fillmore East, for a change, and it is hair-raising. I know, you can't measure that.

It is a beautifully constructed D/A converter, three transformers, Nippon Precision 8x OS digital filter chip and an absolute classic.

It is a keeper for sure.

One thing I would do however, is remove the top panel and inspect the two Rubycon Black Gate capacitors on the +/- supplies for the opamps. They are often glued down with the notorious Sony Bond Glue (or similar) and it corrodes the component legs around it. And inspect the Nichicon PSU caps for the same stuff.

 

[restorer-john]

I found a picture on the internet for you of the offending glue and marked it up:

 

[gvl]

Thanks, will do, I had it for some time now but didn't have a good reason to peek inside. I've been contemplating to swap out the SPDIF receiver and the DF with a kit that brings 24-bit support, but frankly never felt I really needed it.

 

[Newk Yuler]

The Soekris DACs have isolation transformers on the electrical S/PDIF input connections. Many manufacturers place them in their source components' output connections. Is it not true that running two isolation transformers is unnecessary and possibly degrading or otherwise somehow detrimental? Amir, could this have had an affect on your 1421 tests?

To put it bluntly, is this a design benefit or something Soekris might have better left out? It's all subject to reclocking anyway.

 

[DonH56]

Many S/PDIF and AES circuits I have seen use transformers. It usually does not hurt to have on both sides; helps provide galvanic isolation for both components, but if you go balanced make sure the shield is truly grounded at one end (only, usually). That said I think I have seen them more often on the receiver than the transmitter but not something I pay much attention to since it is a digital signal.

 

[amirm]

Amir, could this have had an affect on your 1421 tests?

No, there would be no difference.

 

[SIY]

Perhaps the designers of these older units were able to creatively balance the inherent nonlinearities of these older processors to add more apparent depth and detail, but which is somehow an artifact. That may very well be. Also perhaps there is an amusical character to the filter types inherent to DS designs that is not present in many or most R2R designs.

Or perhaps sighted listening with no level matching could be the issue.