Non Oversampling DACs versus High Rez Hype

[jkeny]

I would love to see what Bruce thinks about it and compare it to the Mitech! Bruuuuuuuuce!!!

Bruce kindly received the DAC32 a while ago (I had Paul Candy send it to him) & he has just PMed me his impressions. I would like to thank Bruce for his time in doing this, much obliged, Bruce! I'm sure he will post something in due course about his impressions.

 

[Bruce B]

Would first like to thank John and Paul for the opportunity to audition this wonderful DAC. I've always been skeptical of most of the lower priced units because I've never really heard one that I could live with. We've had DAC shootouts at our PNWAS meetings and I've really been underwhelmed at what I've heard so far. I've invested over $100k trying to chase the ultimate in A to D and D to A technology over the past year. I now have my reference converters that I use on clients material and couldn't be happier.
When I was approched and asked if I'd like to try out a new DAC, at first I put it off. I did some research and was also trying out new native software in the studio and needed a DAC to try it out. So I told John to send it to me.
The software I'm using is Magix Sequoia. It's a native digital audio workstation that is capable of sample rates up to 384kHz. I also used it on the server software "Emotion" by Merging Technologies, the same company that makes the Pyramix workstations we use. I also used the DAC with JPlay by itself. Unfortunately I could not get JPlay to work with anything greater than 192. I'd just get noise/artifact with 352.8 and 384 files.
When I got the DAC I opened the package and saw a nondescript black box with RCAs and a USB port on one side and a switch on the other. I connected the DAC using a new JPS Labs USB cable and SC3 interconnects.
The first cut I listened to was the Oscar Peterson "We Get Requests" that we mastered for FIM. The DAC was very coherent from top to bottom. Nothing sticks out to me in a bad way. The hardest part of a DAC is to recreate the soft parts. I can sense the size of the venue and reverb tails go out all the way without falling off into the black.
The bass is nice and tight. Imaging on upright basses are spot on. They have good definition and body. Nothing bloated or unnatural. Snare hits are fast. I can hear the body of the snare for 1-2sec. It doesn't fall off like other converters. Piano is strong throughout the whole range. Voices are very natural and brass doesn't make my ears bleed.. especially muted trumpet. The only thing I found lacking are tiny cymbal hits. We mastered the Sheffield Track and Drum album for FIM as well and the quiet cymbal hits aren't quite there. I can hear the initial transient, then it falls off too quickly into the noise. If I turned the volume up, I could hear the decay though. I really had to nit pick to find anything wrong with this converter. Remember, right behind me was the best of the best, so I knew what it had to do to impress me.
All in all.. I'd say it's the best inexpensive converter I've heard yet. I'd put it up against anything under $5k. We'll be getting the Mytek 192/DSD DAC in here again and would love to directly compare the 2 head to head. From what I remember, the Mytek has its work cut out for it! Great job John!!

 

[jkeny]

Bruce, thank you for your impressions & your time. Glad you enjoyed the DAC. I too would be interested in a head-to-head against the Mytek. When do you think you will be getting the Mytek?

 

[flez007]

Sounds encouraging to say the least! Thanks for the write-up Bruce! ...glad to see that gear at this price range are that promising.

 

[jkeny]

Sounds encouraging to say the least! Thanks for the write-up Bruce! ...glad to see that gear at this price range are that promising.

I hope mine punches way above it's weight as Bruce has stated But if you know another 32/384 DAC at around this price that sounds as good I would like to know as I'm always interested in learning/improving & hence pushing the boundaries!

 

[Bruce B]

When do you think you will be getting the Mytek?

Michal just asked me over the weekend, so hopefully it will be either this or next week.

 

[jkeny]

Michal just asked me over the weekend, so hopefully it will be either this or next week.

Great, so hold onto it until you get hold of the Mytek & have a comparison, if you are interested & have the time to do so!

In the meantime you might be like to try comparing Jplay with Foobar or other playback software with bit-perfect output & see if you can hear a difference. Could be an interesting area for discussion in another thread, maybe ?

 

[jkeny]

BTW, something occurred to me which is on the topic of the thread:
- the digital filter I use on this DAC chip (PCM5102) is the minimum phase filter which is much closer to the sound of the non-oversampling DACs than the typical digital filters found in high-res DAC chips. This type of filter was originally proposed by Peter Craven as a steep filter with only post-ringing or post-echo. It was used & well reviewed in products like Meridian 808i.2, Linn Klimax and Ayre C-5xeMP.

There are actually two filters that can be chosen on the chip - they call them: a FIR normal filter & a low latency IIR filter (although it doesn't make sense that this is a IIR filter). In my testing of both filters I found that the FIR filter sounded a lot muddier than the low latency filter (minimum phase) which seemed to snap the sound stage into place. In a later experiment I used Audacity to overlay an attenuated pre-echo of the signal on a piece of music to evaluate the audibility of pre-echo & it proved to be very audible at 60 samples in advance & fairly easily audible at 30 samples. I introduced these test files here in another thread

Anyway, I thought I would write this as it has a bearing on the thread title & may account for a lot of the sonic advantages that non-OS seems to have over High-res DACs. To me the pre-echo introduces a time-smear around instruments & voices which leads to a less distinct sound & it has resulting effects on the sound stage definition & apparent dynamics of the sound.

 

[LL21]

In my testing of both filters I found that the FIR filter sounded a lot muddier than the low latency filter (minimum phase) which seemed to snap the sound stage into place. In a later experiment I used Audacity to overlay an attenuated pre-echo of the signal on a piece of music to evaluate the audibility of pre-echo & it proved to be very audible at 60 samples in advance & fairly easily audible at 30 samples. I introduced these test files here in another thread

Anyway, I thought I would write this as it has a bearing on the thread title & may account for a lot of the sonic advantages that non-OS seems to have over High-res DACs. To me the pre-echo introduces a time-smear around instruments & voices which leads to a less distinct sound & it has resulting effects on the sound stage definition & apparent dynamics of the sound.

Thanks for that...interesting. I am no techie...i have read that designers of NOS DACs refer to the lower phase aberrations of NOS designs (or perhaps the higher potential for phase aberrations due to noise shaping and digital filters required in high-res dacs) as one of the reasons they prefer NOS DAC designs. Any thoughts on this from the techie professionals of WBF? I think NOS DACs may have some inherent design advantages...but i am also told NOS designs based on the older chips tend to be noisier than later chip designs. I am wondering if someone can do a NOS DAC design and somehow focus on getting it quieter/lowering noise floor. Any thoughts are welcome!

 

[opus111]

Thanks for that...interesting. I am no techie...i have read that designers of NOS DACs refer to the lower phase aberrations of NOS designs (or perhaps the higher potential for phase aberrations due to noise shaping and digital filters required in high-res dacs) as one of the reasons they prefer NOS DAC designs.

NOS tends to have amplitude aberrations rather than phase - the top end 'droop' reaches 3.2dB @ 20kHz. There are various means of correcting for this - some analog filtering is done on, e.g. the AMR product. However the implementations I've seen generally use inductors which have a tendency to pick up EM trash. The shape of the droop curve doesn't lend itself to passive RC solutions without some compromises - I'm using RC in one of my prototypes but it only really works up to 17kHz.

OS designers tend to believe that the lowest phase (and amplitude) aberration is best, and hence use equiripple, linear phase digital filters almost exclusively. Time domain response duly suffers from the almost obsessive freq domain focus. The BB chip jkeny mentions is the first one I've seen which gives a minimum phase option. Perhaps other semiconductor houses will catch on here.

I think NOS DACs may have some inherent design advantages...but i am also told NOS designs based on the older chips tend to be noisier than later chip designs. I am wondering if someone can do a NOS DAC design and somehow focus on getting it quieter/lowering noise floor.

You're right - the development of multibit (sometimes called 'R2R') chips reached its zenith with the PCM1704 and nothing since then has come out to handle hi-res. So the S-D chips have been sporting much better noise figures - however they do suffer from noise modulation which isn't mentioned in the datasheets (ESS is the exception). I'm working on multibit designs for lower noise though its very early days - take a look at my blog if you're curious.

 

[LL21]

NOS tends to have amplitude aberrations rather than phase - the top end 'droop' reaches 3.2dB @ 20kHz. There are various means of correcting for this - some analog filtering is done on, e.g. the AMR product. However the implementations I've seen generally use inductors which have a tendency to pick up EM trash. The shape of the droop curve doesn't lend itself to passive RC solutions without some compromises - I'm using RC in one of my prototypes but it only really works up to 17kHz.

OS designers tend to believe that the lowest phase (and amplitude) aberration is best, and hence use equiripple, linear phase digital filters almost exclusively. Time domain response duly suffers from the almost obsessive freq domain focus. The BB chip jkeny mentions is the first one I've seen which gives a minimum phase option. Perhaps other semiconductor houses will catch on here.



You're right - the development of multibit (sometimes called 'R2R') chips reached its zenith with the PCM1704 and nothing since then has come out to handle hi-res. So the S-D chips have been sporting much better noise figures - however they do suffer from noise modulation which isn't mentioned in the datasheets (ESS is the exception). I'm working on multibit designs for lower noise though its very early days - take a look at my blog if you're curious.

Hi Opus, Thank you. From a designers standpoint, do you have any thoughts on the Zanden digital or STahl-Tek Vekian Opus digital? Ultimately, it is how the unit sounds that matters (to me)...but i enjoy learning from much more technical people whether the unit is built well, designed well, etc. Thank you!

 

[opus111]

From a designers standpoint, do you have any thoughts on the Zanden digital or STahl-Tek Vekian Opus digital?

I wasn't familiar with the STahl-Tek but its webpage presses all the boilerplate marketing buttons. So we have Extensive...exceptional...custom...enhanced... redesigned in the features list for example.

Then moving on the technical stuff we have the ticklingly underwhelming claim of 0.02dB flatness from 0Hz to 100Hz. Perhaps that's a typo and they really intended to claim (the physically impossible with the given sample rates) 100kHz upper lilmit ? THD doesn't have a level at which its measured, perhaps they're claiming <0.005% THD at all levels? If so its very hard to verify at lower levels when noise dominates. At 54lbs, its a bit of a brute isn't it?

Not at all confidence inspiring from the technical point of view, but perhaps there's a Chinese wall between marketing and engineering?

 

[LL21]

Thanks, Opus111!!

It is a heavy piece of equipment, and the earlier generation i heard was remarkable...truly remarkable in how it took a recording from Proj Johnson (Reference Recordings) i know very well...and allowed me to hear 35 strings and a harpsichord as 5 separate units playing contrapuntal music to each other...i honestly did not realize this is what they were doing on my Zanden...because the Zanden cannot keep the dynamics and resolution going when the dynamics and number of instruments starts to build.

I suspect its higher noise floor and general abilities falter when things get super complex. The flow is still there better than any DAC i have heard, but the individual bits were not there in as specific and easy-to-read a way as with the Stahl-Tek Vekian...and the Opus is 2 generations later.

As for their technicals...i am completely ignorant. I would like to listen to the Vekian again myself in a direct shoot-out with the Zanden...but there are a number of strong positive endorsements from people who've heard it including some well regarded manufacturers. Ralph, the owner of Atmasphere tube amps, posted that he presented at an audio show with Stahl-Tek...and i recall he wrote he kept running back and forth to see if the turntable was playing...that says something to me.

 

[opus111]

It is a heavy piece of equipment, and the earlier generation i heard was remarkable...truly remarkable in how it took a recording from Proj Johnson (Reference Recordings) i know very well...and allowed me to hear 35 strings and a harpsichord as 5 separate units playing contrapuntal music to each other...i honestly did not realize this is what they were doing on my Zanden...because the Zanden cannot keep the dynamics and resolution going when the dynamics and number of instruments starts to build.

I don't suffer this kind of confusion with my (very simple, circuitwise) NOS DACs, so I did a bit of research into what's going on inside the Zanden. The designer, a Mr Yamada has a patent on his filter stage (US Patent 6,721,427) and the patent document itself is remarkably informative (for a patent) about the circuit he's using. Its a succession of LC tuned 'traps' to cancel out the image frequencies from the D/A zero-order hold function. Its entirely possible that the large number of inductors are contributing to this confusion effect. I'm solving the image frequency problem myself by different means, so its very interesting to explore other solutions that have been tried. I will try a simulation using the circuit supplied in the patent to see how it fares.

I also checked out a Stereophile review of the Zanden - its very positive subjectively but the measurements do suck rather. The frequency response rolls off prematurely at both ends (-1dB @ 12kHz and 130Hz) which is something that could be fixed up. The distortion for LF signals is also a total joke - even at -30dBFS they found above 1%. To me that says they used a bad transformer(maybe too cheap, nickel cores are fairly expensive). Perhaps its that transformer which is the cause of the confusing sound?

<edit> Having read through the subsequent articles on the Zanden I see that the poor LF response was due to a manufacturing error on the sample JA tested.

 

[LL21]

I don't suffer this kind of confusion with my (very simple, circuitwise) NOS DACs, so I did a bit of research into what's going on inside the Zanden. The designer, a Mr Yamada has a patent on his filter stage (US Patent 6,721,427) and the patent document itself is remarkably informative (for a patent) about the circuit he's using. Its a succession of LC tuned 'traps' to cancel out the image frequencies from the D/A zero-order hold function. Its entirely possible that the large number of inductors are contributing to this confusion effect. I'm solving the image frequency problem myself by different means, so its very interesting to explore other solutions that have been tried. I will try a simulation using the circuit supplied in the patent to see how it fares.

I also checked out a Stereophile review of the Zanden - its very positive subjectively but the measurements do suck rather. The frequency response rolls off prematurely at both ends (-1dB @ 12kHz and 130Hz) which is something that could be fixed up. The distortion for LF signals is also a total joke - even at -30dBFS they found above 1%. To me that says they used a bad transformer(maybe too cheap, nickel cores are fairly expensive). Perhaps its that transformer which is the cause of the confusing sound?

<edit> Having read through the subsequent articles on the Zanden I see that the poor LF response was due to a manufacturing error on the sample JA tested.

Wow, Opus111! You are thorough! thank you...yes, there was a 2nd unit sent to Stereophile for testing because the 1st unit was had a manufacturing error. In any event, Zanden have been very positive in my discussions with them about my comparison of Stahl-Tek to Zanden, and i am told everything i have emailed them has been discussed with Mr. Yamada himself. They have encouraged me to send the unit back for their latest upgrades which apparently will include updated circuits, improved shielding and better cables which connect the power supplies to the main unit.

I look forward to any further observations you make about the Zanden design...you seem highly knowledgeable and also extremely helpful. Thank you!

As many here can confirm, my search for digital has been ongoing for over a year...upgrade Zanden vs hi-res...

so far, i have heard Stahl-Tek, Emm CDSA, Esoteric X-01SE, Metronome Kalista Ref/C2A, ARC CD7, Meridian 808.2i, Wadia s7i, DCS Scarlatti, Krell 505. I have been encouraged to listen to a few others still...but so far only the Zanden (TDA1541A NOS) displays the effortless, ease/flow of delivery that has [as strangely commented by Roy Gregory, Marc Michelson, Linnman and Srajaen Ebaen] stopped me caring about SACD, hi-res, DVD-A...to which i have compared this red-book only player. As Marc Michelson wrote, 'how did that happen?'...I just listen to music. Even after all these years, 2 of them still advise it remains their favorite digital source. That i am aware, only Linnman has heard the Stahl-Tek. He owns Orpheus Heritage and Zanden...i am not sure he kept the Stahl-Tek though he positively reviewed it.

Opus111...thanks again for taking the time to write, read, research and write again! Your positive endorsement of TDA1541A is helping me consider the Zanden upgrade offer more seriously. Look forward to reading more of your posts. I did check out your blog and will do more reading this weekend...do not always understand the technical speak, but interesting nonetheless!

 

[opus111]

Wow, Opus111! You are thorough!

That's the positive way to look at it. Others of a more cynical disposition have called me pedantic

In any event, Zanden have been very positive in my discussions with them about my comparison of Stahl-Tek to Zanden, and i am told everything i have emailed them has been discussed with Mr. Yamada himself. They have encouraged me to send the unit back for their latest upgrades which apparently will include updated circuits, improved shielding and better cables which connect the power supplies to the main unit.

Being the inveterate skeptic that I am - ask them to ask Mr Yamada this - 'What have you learned since your original design which has led to these upgrades?'. In other words, do your best to separate the marketing spin from the real progress - if there were really mistakes made in the original then let's hear what they were. From reading the response of Mr Pheils in the Stereophile article, he does to me sound rather prone to the former. Let me quote just the final sentence from his hagiography of Mr Yamada " Only someone like Mr. Yamada, who understands the strengths and weaknesses of both playback systems, could design a digital audio system like the 5000S DAC and 2000P transport: a system that combines the best aspects of digital and analog." Notice him playing the guru card there?

That i am aware, only Linnman has heard the Stahl-Tek. He owns Orpheus Heritage and Zanden...i am not sure he kept the Stahl-Tek though he positively reviewed it.

Positive reviews are two a penny, no reviewer enjoys publishing bad news and hence risking losing advertising revenue. So I always go on whether the reviewer kept the unit or returned it. Actions speak loudest.

I did check out your blog and will do more reading this weekend...do not always understand the technical speak, but interesting nonetheless!

Comments and questions most welcome on the blog - you'd have to sign up to diyaudio I think but that's not really a hardship.

 

[LL21]

Being the inveterate skeptic that I am - ask them to ask Mr Yamada this - 'What have you learned since your original design which has led to these upgrades?'. In other words, do your best to separate the marketing spin from the real progress - if there were really mistakes made in the original then let's hear what they were. [/FONT][/COLOR]

They said the power cables now use the Safety Wave Fabric used inside to shield against RF/EM. There are new gold plated circuit boards, all new wiring, and significantly more shielding inside the DAC unit using the same Safety Wave Fabric which apparently provides equivalent shielding to 30mm of aluminum. They also change the i2s cable between the Transport and DAC. They said the result is the same voice but with far quieter, blacker background and much lower noise floor.

Positive reviews are two a penny, no reviewer enjoys publishing bad news and hence risking losing advertising revenue. So I always go on whether the reviewer kept the unit or returned it. Actions speak loudest.
[/FONT][/COLOR]

The reviewer has owned the Zanden for at least 5 straight years as his reference and eventually bought the Orpheus Heritage as well. I am not aware if he bought the Stahl-Tek or not.

Comments and questions most welcome on the blog - you'd have to sign up to diyaudio I think but that's not really a hardship.

I will do that! . Thanks again.

 

[docvale]

Lloyd,

Have you ever tried the Playback Designs MPS-5?

 

[LL21]

Lloyd,

Have you ever tried the Playback Designs MPS-5?

That is one of the 2 that i've been recommended to hear...but not had the opportunity.

 

[mep]

Are NOS DACs the digital equivalent of SET amps?