Jump to content
IGNORED

Does Native PCM exist or has existed ?


Recommended Posts

I'm shopping for a DAC (second hand T+A DSD8 and Holo XXX??? on the short list ; feedbacks welcome) and wonder if different routes for PCM and DSD makes sense. I understand the logic behind upsampling everything to SDM with sigma delta DACs and that defeats the argument that most SACD have been PCM processed at one stage or another, that native pure DSD is a rare beast. But how about native PCM ? I have been looking for ADC as well and seems to me that all are Sigma Delta as well, at least nowadays. So, is there any logic in shying away from SD at DAC stage (that is using the R2R PCM route of a Holo or Denafrips) if most probably the PCM is of SD origin, having had its Analog to Digital Conversion made by SDM?

 

 

HQ Player 4 Mac Mini M1

Link to comment

Pure 'native' PCM ADCs do exist, at least as chips. But I'm not aware of current production ADC boxes which use such chips. Therefore almost all PCM has indeed been through an S-D modulator (not very likely to be 1bit though) at some stage in its life.

 

I'd not pay as much attention to whether a DAC is S-D or R2R as to the particular implementation details.

Link to comment
  • 3 weeks later...
3 hours ago, Mike48 said:

it's implementation quality -- engineering quality in a broad sense -- that makes a DAC superior. It is not a particular chip or technology that is the key, though manufacturers like to advertise those.

 

OK, here's one (manufacturer); he has some comments as well. :-)

 

For a pure PCM DAC (like ours) it sure is the most crucial what "chips" one uses. Thus, when it would be something like the Holo, this is a full (as in "huge") discrete design and each corner of the circuitry requires enormous attention not to do it wrongly (think mostly noise). There's always the impedance sh*t following you, with for key requirement the low output impedance, like a couple of Ohms would be that. With a discrete R2R DAC this is virtually impossible (but t can be done). Voltage vs Current (R2R) ladders ... there's really a LOT to design and think of. That is, if you really want to do it in a best fashion.

 

The discrete way is always very noisy, thus ... it really is not the best way to design a ladder = PCM = (mostly) R2R DAC. If such a DAC can do a resolution of 16 bits (no matter how many physical bits it has), you may prize yourself lucky (as the designer).

But wasn't the Holo tuning the ladders (in real time or otherwise) ? So that will achieve more resolution (if the noise can be held out of the way). You could call this engineering quality, but if you'd ask me this is technology in explicit sense.

 

This brings us to chips ...

Okay, which chips ?

The only 24 bit chips which ever existed are the PCM1704U-K (-K for the better quality version). They are end of life for quite some time now and the only ones still available, well, are in my possession. So ... currently it would not even be possible to design a "PCM" DAC around a 24 bit chip (unless it is me doing it - LOL).

 

The conclusion could be that for a PCM DAC the ladder technology used (be that chip or discrete) is the most crucial of everything.

Next the engineering around it is as crucial, but that counts for all DAC's. However, one could argue that for PCM DACs this is more difficult (for the discrete ladder DACs for sure, and for the chip version still so because of the "to voltage" conversion).

There's the aspect of the output stage which is quite crucial to the sound (a lot of engineering to be done there), but there is even more engineering going on pre-D/A stage, thinking of the reconstruction (digital filtering). For a PCM DAC this is more crucial than for an SDM DAC because what you feed the PCM DAC will be 1:1 for the conversions in the filtering area (16 bits are transferred to 24 bits but the code is and remains PCM). Thus with a PCM DAC you will be hearing much more of what the filter is doing, hence it is there where the sound really is "created". And these days we do that in the playback software ...

 

Those knowing about the Phasure NOS1 DAC know that this is very explicitly NOS/Filterless (as how this was hot, say 20 years ago), BUT that the filtering is (per my requirement) to be done in software. Although more NOS DACs still exist (though not 24 bit - 20 Yes) their manufacturers do not tell you to apply filtering in software. The NOS1 (in today's NOS1a/G3 incarnation) is the only one explicitly wanting that, and their owners know that there's only one filter(-type) sounding the very best - that's what's in the XXHighEnd software, made for it. IOW, any software player can be used, but they all sound way "less" because they don't use the proprietary filtering I made.

... And this is how the "engineering" as such, is very very broad and it even extends to software.

 

Quote

It is not a particular chip or technology that is the key, though manufacturers like to advertise those.

 

So I just did. 🤪

 

On 4/15/2021 at 9:37 AM, Jean Paul D said:

second hand T+A DSD8 and Holo XXX??? on the short list

 

Personally I would take the Holo because it is a much more technically interesting product. Don't ask me how it sounds.

 

Peter

Lush^3-e      Lush^2      Blaxius^2.5      Ethernet^3     HDMI^2     XLR^2

XXHighEnd (developer)

Phasure NOS1 24/768 Async USB DAC (manufacturer)

Phasure Mach III Audio PC with Linear PSU (manufacturer)

Orelino & Orelo MKII Speakers (designer/supplier)

Link to comment
3 hours ago, PeterSt said:

The only 24 bit chips which ever existed are the PCM1704U-K (-K for the better quality version). They are end of life for quite some time now and the only ones still available, well, are in my possession. So ... currently it would not even be possible to design a "PCM" DAC around a 24 bit chip (unless it is me doing it - LOL).

What about the PCM1754, PCM1789, PCM1792, PCM1798 and many others?  Or maybe the statement refers to non delta-sigma type chips specifically? 

Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade.  Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones.

Link to comment
3 hours ago, Confused said:

Or maybe the statement refers to non delta-sigma type chips specifically? 

 

Sure it does. 🙂

This is inherent to the "PCM" question. Thus, a CD is encoded in PCM, and it can be 1:1 converted by the "PCM" chip which does not oversample internally (for example, the PCM1792 does that and with that can also do DSD (as a kind of hybrid).

 

Of course each SDM chip will "do" the PCM encoded files just the same, but the internal (SD) modulator is always active (same in the PCM1792).

Some times it is not completely clear what a chip can do, as its functionality is/should be written in the datasheet of it, and certain register settings should depict its possibilities. But this is not simple to try out and usually one just gives up on the particular function, just in case it does not work (mind you, it always requires PCBs and the parts to be sourced and if something was overlooked or uncertain it is all down the drain ($ and especially time).

 

I recall from the 1792 that theoretically it should be able to work in native PCM mode, but it can't shut off its internal filtering (or whatever was "wrong" with it). This makes it again "worthless" for someone who wants the filtering outboard (in advance of it - in-software).

 

Designing a DAC is quite exciting. For example, the PCM1704 has always been specified as being able to do 24/96. Still the datasheet gave hints to otherwise, and I just gave it a try at 24/768 and it worked. Since then the 1704 can officially do 768 (many people still don't know, but alas, the chip is EOL anyway).

Anyway the fun for me : in the beginning of 2011 our NOS1 could do 24/768 as the very first (that I know of). In 2010 it could do 24/384 which was also a world's first, with mind you, a Juli@ soundcard of 24/192 as the base.

Lush^3-e      Lush^2      Blaxius^2.5      Ethernet^3     HDMI^2     XLR^2

XXHighEnd (developer)

Phasure NOS1 24/768 Async USB DAC (manufacturer)

Phasure Mach III Audio PC with Linear PSU (manufacturer)

Orelino & Orelo MKII Speakers (designer/supplier)

Link to comment
32 minutes ago, PeterSt said:

Designing a DAC is quite exciting. For example, the PCM1704 has always been specified as being able to do 24/96. Still the datasheet gave hints to otherwise, and I just gave it a try at 24/768 and it worked. Since then the 1704 can officially do 768 (many people still don't know, but alas, the chip is EOL anyway).

 

The 'headline' sample rate for the chip is 96kHz but it was always assumed it would be used together with an 8X OS digital filter. The spec table shows only one rate for all those detailed measurements, which is 768kHz (96k*8).

Link to comment
On 4/15/2021 at 10:50 AM, opus101 said:

 Therefore almost all PCM has indeed been through an S-D modulator (not very likely to be 1bit though) at some stage in its life.

 

Quite a lot of gear use TI's PCM1804, PCM4202 or PCM4204 chips which are all pure 1-bit converters. This includes number of RME ADC's and cards like HDSPe AIO.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
9 hours ago, PeterSt said:

I recall from the 1792 that theoretically it should be able to work in native PCM mode, but it can't shut off its internal filtering (or whatever was "wrong" with it). This makes it again "worthless" for someone who wants the filtering outboard (in advance of it - in-software).

 

All these "Advanced Segment" TI chips can operate to max 6-bit in native PCM mode. Rest 19-bits (MSB + 18 LSBs) are through very rude simple 3rd order 5-level delta-sigma modulator. But they can operate in native DSD mode by reconfiguring the conversion section into analog filter for DSD with four different arrangements. Their modulator produces quite a set of spurious tones and noise floor modulation, so you certainly get best performance out of TI/BB chips by running them in DSD mode at DSD256.

 

See the "Theory of operation" section in the datasheet.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...