• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

PCM vs DSD

Your DSD128 plot with -60dB ESS volume control looks close enough to Archimago's DSD128 DNR plot which did not use the ESS volume control, but Archimago's DSD256 plots in all tests (THD/IMD/DNR) are pretty pathetic.

Anyway, according to your tests, using the ESS volume control makes DSD measure like PCM.
I think Archimago must obtain some decent DSD256 test signal or something else is at play.
At my rig DSD256 is not much different than DSD128:

256.PNG


Almost the same (about 2db worst) DR but decent overall.

Edit,the above was controlled by DACs side,by the analyzer's side is a little worst but by no means close to Archimago's plots:

an.PNG
 
Last edited:
For the DAC that he measured - yes.

Cannot generalise....
Similar results applies to Oppo UDP205 and Topping D90SE too that Archimago tested too, and...
With ESS, with DSD input you get 1-bit DSD converted to 32-bits for digital volume control.

But the original sample rate is maintained, so you bypass all ESS oversampling DSP.

So it stays at the DSD MHz original sample rate.
Regardless of how the volume control actually works, the results show that using the ESS volume control on DSD input caused distortion when compared to a test signal generated at for example -20 or -30dB.


I think Archimago must obtain some decent DSD256 test signal or something else is at play.
At my rig DSD256 is not much different than DSD128:

View attachment 314363

Almost the same (about 2db worst) DR but decent overall.
The DSD256 distortion does not happen on other DACs Archimago tested, for example, RME ADI-2 series. So it is not related to the input DSD256 signal at all.
 
Similar results applies to Oppo UDP205 and Topping D90SE too that Archimago tested too, and...

Regardless of how the volume control actually works, the results show that using the ESS volume control on DSD input caused distortion when compared to a test signal generated at for example -20 or -30dB.



The DSD256 distortion does not happen on other DACs Archimago tested, for example, RME ADI-2 series. So it is not related to the input DSD256 signal at all.
It seems all ESS Dac's with "preamp mode" didn't bypass the PCM conversion. May be in direct as well?
 
Ok,that was a tricky one:
I took Archimago's DSD256 signal,upsampled it to DSD512 and that's what I get despite all the strange connection that had to be made:

upsample.PNG


Decent somehow (considering I had to reduce DACs volume as anything above -2db it too hot for my ADC)

Edit:and that's at -60db from DACs side:


-60DSD512.PNG
 
Last edited:
Even if there is conversion, if the conversion quality is decent there should be no obvious distortion, so I am a bit disappointed.
You're right,all test here (big disclaimer for that "here" ) show that the main reason is VC overall.
(specific one gets a hit too despite that at the peak of it's hump but that's another story as it doesn't happen with DSD)
 
Even if there is conversion, if the conversion quality is decent there should be no obvious distortion, so I am a bit disappointed.
The conversation itself is not the problem, later how it's handled in the following stages seem to be the problem (Hump I/V conversation HF glitches....)
 
To complete the thread a little since we now have the ability thanks to @pkane who added perfect native DSD test signal to his Multitone,here's a couple of native DSD measurements.
(I try to keep the same size at pics so they can be comparable and easy to sequence with all the previous ones in the thread )


SINAD DSD128.PNG

1Khz SINAD DSD128 native

100Hz DSD128.PNG

100Hz DSD128 native

SMPTE.PNG

IMD DSD128 native

MT32 DSD128.PNG

Multitone32 DSD128 native.

Not much of a difference than DoP I think but still nice and clean without the added "energy" but at the expense of couple of dbs' elevated noise compared with PCM.
 
Last edited:
Not much of a difference than DoP I think but still nice and clean without the added "energy" but at the expense of couple of dbs' elevated noise compared with PCM.
With a few of the DACs that I used for testing (SMSL DO100, RME ADI-2 Pro FS) I found no measurable difference between DoP and Native DSD. Multitone makes it easy to switch between these by just changing one setting. It was also quite nice to be able to output 512x48k DSD test signal and capture it using 768k ADC :)

index.php
 
With a few of the DACs that I used for testing (SMSL DO100, RME ADI-2 Pro FS) I found no measurable difference between DoP and Native DSD. Multitone makes it easy to switch between these by just changing one setting. It was also quite nice to be able to output 512x48k DSD test signal and capture it using 768k ADC :)

index.php
Such DSD tests (but not only tests,even single test signals) were simply non-existent before you implement (or better say invent them from ground up) them in Multitone Analyzer.
It;s not only unique but one and only!
There's no way to thank you enough for this!
 
With a few of the DACs that I used for testing (SMSL DO100, RME ADI-2 Pro FS) I found no measurable difference between DoP and Native DSD. Multitone makes it easy to switch between these by just changing one setting. It was also quite nice to be able to output 512x48k DSD test signal and capture it using 768k ADC :)

index.php

And just for DSD lovers (and haters :) ) here's a native DSD 64x44.1k test set up like the 512x44.1k capture above:

Capture2.PNG


And a comparison of 64x to 512x:

Spectrum.png
 
Last edited:
It was also quite nice to be able to output 512x48k DSD test signal and capture it using 768k ADC :)
What DAC is this because the DSD noise should not be rising at 40kHz for DSD512

Should be more like 200kHZ
 
What DAC is this because the DSD noise should not be rising at 40kHz for DSD512

Should be more like 200kHZ

By the way, I know you use HQPlayer, so here's a comparison with the same DAC. HQPlayer vs. Multitone. Same signal upsampled to 512x44.1k by both, HQP is red, MTA is blue. The noise floor appears pretty identical despite completely different implementation. H3 and higher order harmonics seem a bit higher with HQP, not sure why:
hqp.PNG
 
By the way, I know you use HQPlayer, so here's a comparison with the same DAC. HQPlayer vs. Multitone. Same signal upsampled to 512x44.1k by both, HQP is red, MTA is blue. The noise floor appears pretty identical despite completely different implementation. H3 and higher order harmonics seem a bit higher with HQP, not sure why:
View attachment 334104
This doesnt look like DSD512 is being converted to analgue

The conversion might be running at a lower rate

For example there are AK4499EX DACs out there supporting DSD512 input

But it is well known this chip actually runs at DSD128 and DSD256
 
This doesnt look like DSD512 is being converted to analgue

The conversion might be running at a lower rate

For example there are AK4499EX DACs out there supporting DSD512 input

But it is well known this chip actually runs at DSD128 and DSD256

You might be right. There are certainly some funky things in the DSD implementation of DO100. Easy to see when measuring with MTA :)
 
You might be right. There are certainly some funky things in the DSD implementation of DO100. Easy to see when measuring with MTA :)
That's how the same comparison looks at Khadas but captured at only 192Khz so we can't see higher than 96Khz :

DSD64 nf.PNG
DSD64

DSD512 nf.PNG
DSD512

both.PNG
Both


Looks smoother or is just the scale?
 
Now that I had some time I thought it would be nice to see how this thing handles the two formats distortion-wise only vs level:

Rig is the same as with all the rest of the measurements,the only difference are the two formats.
Chart shows both THD+N vs level (white trace ) and THD only (orange trace) vs level to become more apparent.



THD vs Level PCM.PNG
THD vs Level PCM

THD vs Level DSD.PNG
THD vs level DSD

It was interesting watching the spectrum chart as the measurement proceeded,PCM had no stable pattern vs level,at some H2 was dominant at some H3,etc.
The big lump around -35 to -20db consisted almost entirely of H3,H5,H7 at the same amounts at some time with PCM signal! (imagine all 3 harmonics at about the same level)
On the other hand you can see the non-existent amount of distortion of DSD at the same levels.

That's broken in my book.

For reference,fun and for the WTF? of it,that's how the loopback of the 20 yo,heavily abused interface used to measure the above looks like:


THD vs level E-MU PCM.PNG
THD vs Level E-MU's loopback.

Again,I don't think it's the format,it's the way a device handles it,in my case Tone 1 does a really poor job at PCM despite it's really good review for the price (or not the price as it eventually ended up in a finished DAC with way higher price if I remember well)

Feel free to add the benefit of about -5db better THD+N shown in Amir's review as my ancient gear is a joke compared to the AP.
 
Back
Top Bottom