Accuphase DC-91: multi multi-bit

That 1995 DAC would sell for over $17,000 today. I hardly think it would even be in the same league as a stock Gustard Pro ($869 delivered....which, btw, just beat a DAC section in the $21,000 Esoteric K01X). The Gustard can play up to DSD512, all pcm, has a digital volume control and sounds good out of the box....with mods it is outrageous. The Aries Cerat would kill that thing. The Gustard has 16 DACs per channel (eight for the positive phase and eight for the negative phase (each of the two ESS DAC chips has 8 balanced DACs in it......all used in parallel). The signal to noise in the Gustard is way, way better than parallel PCM 63s. A nice boat anchor.....he he. Who wants yesterdays DAC....who wants yesterdays girl....nobody in the world......please sing along.

By the way. I made a DAC with 4 parallel PCM 63s back then and it sounded pretty darn good for its day with HDCD digital filter, UtraAnalog input receiver and discrete fet output IV converter. It had 10 power transformers (5 for digital and 5 for analog), shunt regulators, Marigo dots, shielded DAC chips, hardwired digital cable, damping, etc. etc....the whole nine yards. However, even modded players soon beat it. The beat goes on. I have some pictures if anyone is interested.....it is wild beyond belief....just another pile of parts now. Found some pics.....one below....I have close ups too. Actually, I lied, I forgot about the transformer for the muting.....that makes 11 transformers. Here it is: Mike's Max Stax.......it started out as a Stax Talent DAC and got transformed (slightly) by the tweak crazy me and Mike:

morricab said:

(...) FWIW, the measurements of the DC-91, according to Stereophile, are extremely good...as good as pretty much anything before or after that Stereophile measured.
(...)

Click to expand...

At that time the Stereophile measurements were only a few and very basic, analyzers did not have the resolution or the capabilities of the current measuring systems. At that time the challenging measurement that people loved because it resulted in a nice picture was the -90 dB 1KHz - something where the PCM63 would be great due to its dual R2R structure! But there is a lot more to consider in a DAC.

IMHO modern DACs, such as the Vivaldi have much better measurements.

morricab said:

Actually I think the DC-91 is still a very good measuring machine. It has low noise, good linearity nice waveforms. Plus it has the BBPCM63...a whole bunch of them. However, as we have all learned along the way, hopefully, the measurements are by no means a direct translation to sound quality.

I hear a quality to BBPCM63 based DACs that I don't hear in any of the DS based ones I have owned/heard. I haven't heard a DCS stack yet that I found musical. If you think its great that's fine but throwing numbers into the game says nothing about sound quality.

Personally, I have not heard either the DC-81L, which is one of the earliest implementations of a discrete 20bit R2R on the market, or the DC-91, which was Accuphase's ultimate MMB machine. If you have heard either or both please comment on the sound. Maybe they are not good because their analog stages are typically "over" engineered. I doubt that they do not sound good because of their digital implementations, which are still contemporarily relevant, IMO.

Click to expand...

I only referred the measurements because in order to comment measurements I needed measurements ... As you know I think measurements are an important think, but not to rank high end equipment.

I am always afraid of the unobtainium in high-end, independently of the reason. Perhaps it limits my views, but dreaming about products I never listened, or will never be able to own is not in my audiophile activity. And yes, I owned a PCM63k CD player - the Sony ESDX7, even modified it and commented on it before in WBF. I suppressed all the analog stage and replaced it with an LC filter and 50 ohm precision resistor. IMHO it sounded better than the original. As far as I remember I replaced it with one of the Ben Duncan Audio Synthesis DACs using the UltraAnalog chip.

R2R chips have characteristic technical problems - mainly in the differential and integral linearity and even being non-monotonic. But as written in the dstasheet : The key specification for the PCM63P is total harmonic distortion plus noise (THD+N). All else and their influence in sound quality is forgotten in audio products datasheets.

IMHO, R2R chips have a characteristic sound - if it aligns with our preference, it is great.

posting just to ensure i get to keep a watch on this thread...interesting.

I have spent 35 plus years tweaking products and making high end thangs including DACs.....like the one shown. What I know, is that I know nothing. The game is infinite. Every single thing you do to something changes the sound. Yes, all DAC chips have an inherent sound......everything does. However, the idea that one DAC technology, no matter how it is implemented is the "only" way to go is just plain silly. I can make a Gustard Pro with its parallel Delta Sigma chips sound pretty much any way I want. I really don't think these new ESS DAC chips have much sound on their own....but we will find out......eventually. Because every thing matters. For instance, the ESS DACs are one of the only DAC chips that uses its own master clock....right on the a DAC pin. So, the sonic signature of the clock/clock power suppy is what you will hear.....very directly. The "Crystek Femto Clock" ($25) is the one most people use (Mytek, Resonessence, Exasound, etc.). Sorry, but this Crystek part sounds forward in the treble and not that refined.....I know because I just replaced the Crystek clock in the modded Gustard with the Pulsar Clock ($400, my cost). You mean to say the sound of the ESS DAC is not what I thought? I am hearing jitter? Yup, that is what I am saying. The clocking, power supplies, the capacitors, the damping, the output stages, etc, etc. etc. all make the sound. One not done perfectly right thing somewhere along the chain and that is what you will hear. Why do you think I made such a crazy thing like the Max Stax.....it is because I wanted it all. However, today I would do practically nothing that I did back then (20 some years ago). That is how much we have learned. The output stage and the large transformers are still current.....but everything else including the wire and solder and jacks and regulators, resistors, diodes and DACs, etc. etc. would not be up to the latest sonic marvelousness. The game is ever changing.

It took MSB years of tweaking of discrete R2R DACs to make the MSB Select II. They started with discrete R2Rs back around 2000. Will my fully modded Gustard Pro (approx. $2200 including all mods) sound as good as the $110,000 Select with the dual power supplies and 33 Femto clock? No way. Will it be embarrassed?..... not likely. But, we shall see...er hear...

Here is another thing to ponder. With the advent of HQplayer and people raving about upsamping all PCM to DSD512 then the whole idea of what a DAC is goes out the window. Now you just need something to do DSD at that super speed. T+A and Lampizator have DACs that do DSD without any DAC chips. The ESS DAC is suppose to sound its best with DSD. So, using a high powered Windows computer and running HQplayer into a DAC running at DSD512 may be better than just about any other way to listen to PCM. You see folks, the times they are a changing. The Gustard will do DSD512 through its I2S interface and soon there will be a box that can send it. Fun times ahead. Many who have listened to 512 upsampled music through a great DAC are no longer listening to their turntables......yup...is that good. Of course, there are super turntables, etc.....but at what cost?

As I said, I know nothing.....but I know what I hear....and what I think is that if you spend a lot on digital right now....those thing will be boat anchors soon.....at least with the MSB Select II you have updates for life thrown in.....at $90,000 plus they should give you a Mercedes as well!!!

morricab said:

Perhaps you have the DAC chip wrong in that Sony?

http://www.dutchaudioclassics.nl/the_complete_d_a_dac_converter_list/

Looking on this list for any Sony that used a PCM63 I found exactly 0 that had that chip. Some with other BB (PCM56, PCM60, PCM64) but not PCM63. If the one you had is on that list then I would love to see it. I could not find it in a google search. Nonetheless, Sony is notorious for making an overly complicated and not terribly good sounding output stage. I remember Allen Wright showing me his output stage upgrade board that he offered from Vacuumstate for the Sony SCD-1 and SCD-777ES. He showed me that they must have had a fire sale on opamps because Sony was using just a ton of them in the SCD-777. The Vacuumstate board was all discrete transistors (yeah I was surprised too given that it was from Vacuumstate!). He said the digital part was basically SOTA but that the analog section was ****. So yes, you can ruin a good digital section with bad analog...this is well known.

IF it was a PCM63 in your Sony, not so sure based on my searching, then you cannot just to a resistor IV conversion with no gain of any kind afterwards...you would not be able to drive a preamp properly like that...it needs amplification of some sort. The Monarchy uses a passive IV conversion followed by a SRPP stage using a single 6922/6DJ8 (ECC88 or E88CC) per channel for gain and load matching.

Click to expand...

If you say so, perhaps it was another BB chip (PCM58 - 18 bits?) or not this Sony, it was long ago, but I still remember having to cut all the pins of the PCM63J and taking them one by one to replace the chip with the glorious K grade.

As far as I remember using the 50 ohm resistor, the output reduced to around 100 mV RMS - enough for high gain preamplfiers, having more than 30 dB gain, that were usually operated around 8-9 o'clock position with normal CD players!

LL21 said:

posting just to ensure i get to keep a watch on this thread...interesting.

Click to expand...

See 'thread tools' at the top? You can subscribe more effficiently using that feature

microstrip said:

R2R chips have characteristic technical problems - mainly in the differential and integral linearity and even being non-monotonic.

Click to expand...

True enough but is there any evidence that those specs matter most towards SQ in an audio converter? In my experience what matters in an audio converter is dynamic performance and DAC datasheets don't talk about this much, if at all. Yet its a fairly consistent finding that the subjective dynamics decrease the faster a DAC is asked to update, all other aspects being controlled. Which could well be why NOS is so popular.

All else and their influence in sound quality is forgotten in audio products datasheets.

Click to expand...

Ain't that the truth?

opus112 said:

See 'thread tools' at the top? You can subscribe more effficiently using that feature

View attachment 30798

Click to expand...

thanks! very nice.

opus112 said:

...In my experience what matters in an audio converter is dynamic performance and DAC datasheets don't talk about this much, if at all. Yet its a fairly consistent finding that the subjective dynamics decrease the faster a DAC is asked to update, all other aspects being controlled. Which could well be why NOS is so popular...

Click to expand...

Hi Opus...its been a while since your posts on digital which i still remember (though not enough to be able to recite them back being a non-techie! )...i am curious if you have observed any new digital that intrigues you and that you feel has significantly pushed digital playback forward? Probably since your posts a couple of years ago, there has been the Vivaldi 2.0, the Aries Cerat, Audio Note DAC 5th Element (also AD1865 chips), and no doubt several others.

LL21 said:

Hi Opus...its been a while since your posts on digital which i still remember (though not enough to be able to recite them back being a non-techie! )

Click to expand...

Hi Lloyd, I often go back to see how many views that thread has attracted, we were up to about 65k last time I checked. So its getting an audience which is gratifying

...i am curious if you have observed any new digital that intrigues you and that you feel has significantly pushed digital playback forward? Probably since your posts a couple of years ago, there has been the Vivaldi 2.0, the Aries Cerat, Audio Note DAC 5th Element (also AD1865 chips), and no doubt several others.

Click to expand...

Interesting you should ask this, it just so happens I have made what seems to me a fairly important discovery in recent months. Not in existing high-end products but I rather suspect it could be influential in the more affordable end, perhaps eventually trickling up. You'll note that the market at the top is fairly well divided between proprietary D-S (exemplified by dCS) discrete R2R (like MSB) and then various incarnations of multibit chips. Recently though Mola-Mola has introduced another approach - proprietary PWM. There are off-the-shelf incarnations of PWM but they're primarily aimed at FDA : 'full digital amplification'. Its these chips that I have found recently show a lot of promise when re-purposed as DACs, so much so that I've put my multibit DAC designs into suspended animation while I explore the possibilities they offer.

Here's one device on Taobao which will give you an idea - I have modified this basic design in a couple of ways firstly to drive headphones direct, then to make a PWM DAC.

https://item.taobao.com/item.htm?spm=a230r.1.14.57.HThuTt&id=19260362926&ns=1&abbucket=18#detail

How this approach wins out over traditional multibit DACs is in the area of dynamics. The operating voltage of the D/A stage can be much higher than a normal multibit DAC chip - which when coupled to the output via a step-down transformer translates into jaw-dropping dynamics, provided your amp is up to the task!

opus112 said:

Hi Lloyd, I often go back to see how many views that thread has attracted, we were up to about 65k last time I checked. So its getting an audience which is gratifying



Interesting you should ask this, it just so happens I have made what seems to me a fairly important discovery in recent months. Not in existing high-end products but I rather suspect it could be influential in the more affordable end, perhaps eventually trickling up. You'll note that the market at the top is fairly well divided between proprietary D-S (exemplified by dCS) discrete R2R (like MSB) and then various incarnations of multibit chips. Recently though Mola-Mola has introduced another approach - proprietary PWM. There are off-the-shelf incarnations of PWM but they're primarily aimed at FDA : 'full digital amplification'. Its these chips that I have found recently show a lot of promise when re-purposed as DACs, so much so that I've put my multibit DAC designs into suspended animation while I explore the possibilities they offer.

Here's one device on Taobao which will give you an idea - I have modified this basic design in a couple of ways firstly to drive headphones direct, then to make a PWM DAC.

https://item.taobao.com/item.htm?spm=a230r.1.14.57.HThuTt&id=19260362926&ns=1&abbucket=18#detail

How this approach wins out over traditional multibit DACs is in the area of dynamics. The operating voltage of the D/A stage can be much higher than a normal multibit DAC chip - which when coupled to the output via a step-down transformer translates into jaw-dropping dynamics, provided your amp is up to the task!

Click to expand...

Interesting...is any commercial producer using this new approach?

I haven't seen any commercial vendors adopting the approach of using one of these FDA chips as a DAC, no.

There are at least three manufacturers of silicon I'm aware of which can be re-purposed to create such a DAC - TI, STM and Intersil.

opus112 said:

I haven't seen any commercial vendors adopting the approach of using one of these FDA chips as a DAC, no.

There are at least three manufacturers of silicon I'm aware of which can be re-purposed to create such a DAC - TI, STM and Intersil.

Click to expand...

Very interesting. Setting this aside for a second then and coming back to my earlier question, are you aware of any recent COMMERCIALLY AVAILABLE DACs which intrigue you?

To my way of thinking the Mola-Mola looks the most interesting because its using PWM, along with a transversal filter on the back-end.