Jump to content
IGNORED

ESS Sabre and DSD Volume Control


Recommended Posts

I have read through multiple threads and have not found a definitive answer. So, I thought I would start a new thread, but just to get a specific answer. It isn't my intention to open another long thread that goes off topic with arguments over the merits of different formats, etc.

 

I just want to know once and for all how the ESS chips process DSD, specifically DSD volume control. I have read that in the ESS chips DSD is converted to PCM first. I have read that no, it is not converted to PCM, that it stays in Delta Sigma all the way, albeit multi-bit.

 

Either way, the ESS chip does volume control on DSD. Volume control is typically (and I assume most easily done) in the PCM domain. Volume control can be done in SDM too, though.

 

So which is it? Does the ESS do volume control for DSD in PCM, or does it do it in Delta Sigma?

Link to comment

First of all, I should say it's merely my speculation as ESS Tech company has published no technical documents that reveal any exact internal configuration details for a DSD play.

 

As you may have already known, ES9018 has a FIR OSF and a multi-level delta-sigma modulator (65 level output in default) and 64 1-bit switches.

No DSD play is available when you turn-off OSF.

 

Therefore, the volume control function is likely to be implemented with the FIR. This indicates a possible involvement of multi-level representation of sound intensity values after FIR.

However, it might not directly mean a usual PCM format.

Anyway, re-delta-sigma modulation to 65 levels must be essential.

 

Bunpei

Link to comment

I suspect that it is NOT pcm... I do think it is a multibit differential bitstream for volume control. As long as it is still an equivalent or greater sample rate differential signal, it is still "DSD" as far as I am concerned. What does concern me, though, is the state of the original 1 bit DSD signal after DSP and re-modulation.

 

 

 

First of all, I should say it's merely my speculation as ESS Tech company has published no technical documents that reveal any exact internal configuration details for a DSD play.

 

As you may have already known, ES9018 has a FIR OSF and a multi-level delta-sigma modulator (65 level output in default) and 64 1-bit switches.

No DSD play is available when you turn-off OSF.

 

Therefore, the volume control function is likely to be implemented with the FIR. This indicates a possible involvement of multi-level representation of sound intensity values after FIR.

However, it might not directly mean a usual PCM format.

Anyway, re-delta-sigma modulation to 65 levels must be essential.

 

Bunpei

Link to comment

http://www.esstech.com/PDF/sabrewp.pdf

 

look at section 3, this will give you an insight of what's going on.

Digital Sources: Optimised HP TouchSmart PC/CEC TL-1X CD Player/AMR DP-777 DAC/Theta Digital DS Pro Basic II (old)

Analogue Sources:Koetsu Jade Platinum MC Cartridge/Tri-Planar arm/Kuzma Stabi Reference turntable/AMR PH-77 Phono Stage

Amplifiers:The Gryphon Elektra Preamplifier/Convergent Audio Technology JL2 Signature Mk 2 Stereo Amplifier

Speakers:Kharma Grand Ceramique Midi[br]Cables:Nordost Valhalla (interconnect and speaker cables)/Shunyata Research power Snakes power cables

Portable: Sony PHA-1/PHA-2; Dragonfly 1.0/1.2; Meridian Explorer, Director; iFi nano iDSD, micro iDAC, micro iDSD; Geek Out; Hdta Serenade DSD

Link to comment

Here is John Siau's explanation of how DSD volume control is done in the Benchmark DAC2. The DAC2 uses the ES9018 DAC chip.

 

 

MW: How do you handle volume control in that final output stage? Do you convert to analog and then turn it up and down.

JS: We actually don’t. We do process that at the high sample rate and we have multiple 1-bit converters that are available to us. So the increase in word length that we get as a function of that volume control makes use of the redundant 1-bit converters that we have running in parallel.

MW: I see.

JS: So we’re not converting it…in a way you could look at that as if it’s PCM because there’s multiple 1-bit converters summed together in the analog domain. But that’s what you have to do to get volume control to work. The good thing is we don’t take it from 1-bit to multi-bit and back to 1-bit before we convert it to analog.

MW: Yep, as you were saying before.

JS: Instead of sending identical DSD signals to sixteen balanced 1-bit converters that are wired in parallel, we start sending different DSD signals to reduce the signal amplitude. All summing occurs in the analog domain. It is very cool!

 

Russ

Link to comment

I am trying to understand all of this with the mind of a layman. My limited understanding is this.. .(and I THINK that is what he is saying)

 

DSD is a differential signal. 1 is full voltage/volume. 0 is zero voltage/volume. If I have a stream of 1 0 1 0 1 0 1 0 1 0 1 0 the average of that is a 50 percent signal.

 

If I convert DSD to a multi bit system, (call it PCM, muti bit delta sigma, hybrid PCM/DSM, whatever you want) I can covert that 50 percent signal to say, 25 percent. 32 bit is full. 16 bit is half voltage. So a 16 bit pulse followed by 0, etc. etc... you get a 25 percent level signal.

 

I think the problem with it is, I think, is the same as with any digital volume control. You significantly reduce the resolution of signal while the noise level stays the same. Now, the audible effects? I don't know about that.

 

 

 

 

Here is John Siau's explanation of how DSD volume control is done in the Benchmark DAC2. The DAC2 uses the ES9018 DAC chip.

 

 

MW: How do you handle volume control in that final output stage? Do you convert to analog and then turn it up and down.

JS: We actually don’t. We do process that at the high sample rate and we have multiple 1-bit converters that are available to us. So the increase in word length that we get as a function of that volume control makes use of the redundant 1-bit converters that we have running in parallel.

MW: I see.

JS: So we’re not converting it…in a way you could look at that as if it’s PCM because there’s multiple 1-bit converters summed together in the analog domain. But that’s what you have to do to get volume control to work. The good thing is we don’t take it from 1-bit to multi-bit and back to 1-bit before we convert it to analog.

MW: Yep, as you were saying before.

JS: Instead of sending identical DSD signals to sixteen balanced 1-bit converters that are wired in parallel, we start sending different DSD signals to reduce the signal amplitude. All summing occurs in the analog domain. It is very cool!

 

Russ

Link to comment
If I convert DSD to a multi bit system, (call it PCM, muti bit delta sigma, hybrid PCM/DSM, whatever you want) I can covert that 50 percent signal to say, 25 percent. 32 bit is full. 16 bit is half voltage. So a 16 bit pulse followed by 0, etc. etc... you get a 25 percent level signal.

 

And that works only if you are fine each step of volume dropping level to exactly half. So with Sabre's 64 1-bit converters you would have 36 dB volume control range in 6 dB steps (6 steps). If you do it PCM way.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

Therein lies my misunderstanding, Miska. Thanks for chiming in, though. I respect your thoughts on DSD immensely.

 

Now, I understand what you are saying if the volume control is done in they DS converter. But per my example done in DSP before the converter, if you had each pulse rendered in 32 bits, you could use any number of levels. You could use 30, 29, 28 and everything in between.... and you would still be left with a differential signal.

 

It seems to me this should work. But I don't know. Futhermore, I think the only people that know what is going on in the ESS chip have signed an NDA and aren't talking!

Link to comment
But isn't my idea a good one though? Come on, why cant someone build a 32 bit DSM?? A 4 million plus level modulator would do the trick well!!

 

4 billion... It would need just 4 billion 1-bit converters and have about four times the number of transistors the latest Core i7 CPU has... :)

24-bit would be 16.8 million.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
What is your final understanding after discussions and comments here?

 

Digital volume SUCKS! Why be cheap? Get a real preamp.

My Dedicated 2CH System Gallery

 

Custom C.A.P.S. Reference Music Server with UpTone Audio JS-2 External Linear Power Supply > Bel Canto REFLink Asynchronous USB Converter > AT&T ST Optical Glass Fiber > Bel Canto DAC3.7 DAC > Pass Labs XP-20 Preamp > Pass Labs XA160.5 Class A Mono Blocks > Martin Logan Summit X Speakers

 

Powered By Balanced Power Technologies - UpTone Audio JS-2 Linear Power Supply - CyberPower Sinewave UPS

Link to comment

I hate to agree, but I must- emphasis on real. To date I have always preferred my Buffalo DAC with my TVC. preamp

Digital volume SUCKS! Why be cheap? Get a real preamp.

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment

If noise floor was it, a simple differential input op amp is all that would be required for the ultimate sound. Say what you will, but the op amp version of my DAC was my least preferred.

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment
If noise floor was it, a simple differential input op amp is all that would be required for the ultimate sound. Say what you will, but the op amp version of my DAC was my least preferred.

 

I have not yet found analog volume control that would have lower noise, better accuracy, channel balance and distortion... No pre-amp is best pre-amp.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
What is your final understanding after discussions and comments here?

 

 

Actually, not from comments here. From a trusted source. ESS won't tell anyone, not even designers that work in close collaboration, what is really going on inside the chip. They keep it very guarded.

 

There are a few things we can glean from looking at the output of the ESS with instrumentation, along with a few assumptions we can reasonably make based on what we do know about the ESS chip.

 

And that is as follows.

 

 

1) The internal processing uses an ASRC which in turn incorporates a digital filter

2) The digital filter is claimed by ESS to work at 32 Bits

3) The digital Volume control is claimed by ESS to work at 32 Bit Precision

4) The digital filter and ASRC can be bypassed, if you do this DSD no longer works

 

Number 4 is very telling. The ESS does sample rate conversion on DSD. If you turn off the ASRC, bye bye DSD. This combined with their precision 32 bit volume control, and the digital FIR filter,

 

Strongly suggests DSD is converted to a common internal format, which is some form of PCM.

 

You can't do ASRC, digital filtering, and volume control in Delta Sigma (miska's software aside. I am pretty confident that the ESS chip is not doing volume control ala Miska).

 

The output on the 'scope confirms the above conclusion. As a matter of fact, if you want to see a similar scope trace, you can convert DSD to PCM in Jriver and set the filter to permissive.

 

So in conclusion, I am pretty confident that the ESS chip converts DSD to PCM. All of this above comes from an excellent source, that knows as much about it as anyone.

 

In the end, it is all in the listening. I am sure that DSD still sounds most excellent on the ESS. Conversion to 32bit 352.8 khz (a guess) PCM isn't the worst thing in the world.

 

 

P.S.

 

If you want to know more about DSD, go read the QA article on Audiostream with Thorsten Loesch.

Link to comment

Hi, Andrew!

 

I appreciated your very comprehensive explanation very much!

I agree with you on most of the points.

However, just one point: Except the recent version of ES9018-K2M, ESS chip of Sabre32 architecture does not allow us disabling or bypassing its ASRC at all.

 

By the way, a user "dusfor99" of this forum must be one of main designers of ESS chips though he has posted nothing about them on this forum.

Link to comment

In theory I completely agree, in practice it varies IME. I have sources that are not digital and need other inputs as well. Even though my pre has bypassing, I still prefer the sound using the transformers. I must like the distortion.

I have not yet found analog volume control that would have lower noise, better accuracy, channel balance and distortion... No pre-amp is best pre-amp.

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment
In theory I completely agree, in practice it varies IME. I have sources that are not digital and need other inputs as well. Even though my pre has bypassing, I still prefer the sound using the transformers. I must like the distortion.

 

I think actually good way to handle vinyl would be to use DSD ADC and then apply RIAA EQ in realtime in DSD domain and use DSD DAC to play it back... ;)

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
If you want to know more about DSD, go read the QA article on Audiostream with Thorsten Loesch.

 

There's bunch of errors there, especially in diagrams. For example the PCM DAC diagrams forget the path:

[PCM] -> [8x oversampling digital filter] -> [sample-and-hold/linear oversampling] -> [delta-sigma modulator] -> [delta-sigma converter]

 

It also forgot to mention that CS4398 has Direct DSD mode capable of both DSD64 and DSD128, while the referred Wolfson chip is capable of only DSD64.

 

And no, CS4398 when using DSD processor (for digital volume control) doesn't do decimation, DSD is kept at original rate in that case too. So for 5.6 MHz DSD, volume is processed at 5.6 MHz sampling rate. And AFAIK, same thing for Sabre. (Wolfson does do decimation to 8x PCM for digital volume)

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

AFAIK, we didn't get a clear answer to the original question. Does the ESS chip convert the DSD to PCM in order to do Volume control?

 

If the answer is yes, how does a company like Mytek specifically say their DSD 192 DAC does native playback of DSD 64 and 128?

Main listening (small home office):

Main setup: Surge protector +>Isol-8 Mini sub Axis Power Strip/Isolation>QuietPC Low Noise Server>Roon (Audiolense DRC)>Stack Audio Link II>Kii Control>Kii Three (on their own electric circuit) >GIK Room Treatments.

Secondary Path: Server with Audiolense RC>RPi4 or analog>Cayin iDAC6 MKII (tube mode) (XLR)>Kii Three .

Bedroom: SBTouch to Cambridge Soundworks Desktop Setup.
Living Room/Kitchen: Ropieee (RPi3b+ with touchscreen) + Schiit Modi3E to a pair of Morel Hogtalare. 

All absolute statements about audio are false :)

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...