Playing PCM to DSD Native with foobar2000

[ManniX]

If you want to see a comparison of how an implementation (Khadas Tone) of an ESS (9038Q2M) dac roughly performs here you are:

Thanks, very interesting!
I didn't thought about using MTA with DoP.
Unfortunately I don't have a decent USB ADC to use with MTA... only a UMC202HD.

 

[Sokel]

Thanks, very interesting!
I didn't thought about using MTA with DoP.
Unfortunately I don't have a decent USB ADC to use with MTA... only a UMC202HD.

Get one ASAP!
I use E-MU for a while since I already had it but I plan for a good one,it's eye-opening and absolutely necessary to set-up and test stuff.

 

[ManniX]

I'm open to any suggestion for a USB model that doesn't cost a fortune
Preferably not more than 100-150€.
I needed the scope for other reasons but this would be really just a fun gadget.

 

[Sokel]

I'm open to any suggestion for a USB model that doesn't cost a fortune
Preferably not more than 100-150€.
I needed the scope for other reasons but this would be really just a fun gadget.

I myself looking at E1DA Cosmos but I hear there's a new version coming.
We'll see...

 

[ManniX]

I myself looking at E1DA Cosmos but I hear there's a new version coming.

That's interesting but indeed there's a new version coming and scarce availability of the current one.
Probably cause there's a new one coming out...
But it's also only capable to measure balanced outputs.
I'd like to have something that can measure unbalanced outputs.

 

[Sokel]

That's interesting but indeed there's a new version coming and scarce availability of the current one.
Probably cause there's a new one coming out...
But it's also only capable to measure balanced outputs.
I'd like to have something that can measure unbalanced outputs.

I was under the impression that can do unbalanced too.Have to check that.
There's also an upcoming Topping interface this month that they claim it as lowest noise ADC ever.
Let's see if it will be usable for measurements.

 

[ManniX]

Get one ASAP!
I use E-MU for a while since I already had it but I plan for a good one,it's eye-opening and absolutely necessary to set-up and test stuff.

Got your advice and bought a Focusrite Scarlett 2i2 3rd Gen.
The Behringer was really utter trash... I bought it to use it for Room EQ but it's so bad it would have surely messed up also the mic readings.

Do you still have the Khadas Tone board? Can you try again with the latest MTA?
I'm surprised is so bad with DSD.
I wonder if there's something wrong with the board/drivers or maybe something was off with that version of MTA.

I have some weird issues testing DSD with MTA but could also be it's because of the awkward method I have to use.
Neither the Aune X8 or the Topping D50s are supporting DoP via USB.
For some reason I can't manage to use the integrated Realtek SPDIF Optical, neither with PCM or DoP.

My only way to test it is using the foobar2000 ASIO DSDTranscoder driver.
Maybe you can try that as well.
To use it you need to keep open foobar2000 in the background without anything playing or in pause.
There's a drawback as it doesn't work very well. DoP is engaged/disengaged in a flip/flop basis.
Hence it's impossible to use it with a multi step measurement; every each one I have to stop, clear and restart.
But it's fine for the MT32 test.
I have tried to replicate your for comparison.

That's where I had the useful but not so funny confirmation that my USB is incredibly noisy.
Usually also with strong harmonics at 4/8/16 kHz.
Was already suspecting it cause I could see random weird stuff with the scope.
I managed to find one USB port, the BIOS Flashback, which is slightly better but not that much.
Today I should get an Audioquest Jitterbug but I'm not so sure how much it can help.
I have ordered an ISOUSB211 eval board for full galvanic isolation but it will probably arrive only next week.

In summary up to DSD128 the quality is same as PCM 96 kHz; going to DSD256 there's a sensible drop.
Not sure at that level is audible but probably doesn't make sense to go DSD256.
Same goes for PCM, going higher with the rate will decrease all the figures.
But even with PCM going higher seems causing troubles with MTA, probably something wrong with the drivers.

The Aune X8 is doing a slightly better job handling the noise than the Topping D50s, seems.
But I have tested it before hunting for a better USB port.
Still not enough to justify the price difference in my opinion.

I've only tested briefly the JRC5532 op-amp with the Topping and it was worse than the stock 49720NA, especially in handling the harmonics.
On the Aune X8 the OPA2134 was slightly worse than the others but it's hard to say. Over many runs was always a tad worse.

This is the best reading I got with the D50s and 49720NA with PCM:

This is the best reading at DoP64:

This the best at DoP128:

At DoP256 there's a small decrease in general but the biggest problem is that the harmonics becomes untamable:

This the the JRC5532 at DoP64:

The Aune X8 with PCM and JRC5532 on the sub-optimal USB port was doing already a decent job, better than the D50s:

And it was doing well also with DoP64:

Not so good at DoP128:

But the 49720NA somehow managed to do better with the harmonics at DoP128 (maybe just cause it was tested 5 minutes later):

They all had the same figures at DoP256 like the 49720NA:

Just a tad worse the OPA2134:

 

[Sokel]

Ok,I tried some stuff,let's see:

That's DSD256 through foobar (and SACD processor)

That's DSD256 with MT's internal generator (a little better)

And that's PCM 352.8 Khz to see if high bitrates affect Khadas,but no.

That's Multitone32 DSD

...and Multitone 32 with PCM352.8


It's noise (probably something with my rig or the inability to play it at 0db,too hot for E-MU) that spoils DSD and I think it's also something that has to do with the format itself.
All of that are inaudible of course,but still...

 

[Sokel]

On another note,see how useful was to have the ability to diagnose problems for yourself?
For example we hear all the time that a good DAC takes care of a noisy port but as you saw the truth is far from that,interference by the bad USB happily passes through everything and appears at the end of the chain.

The biggest upgrade in a system is a decent ADC and a Multitone (or REW,etc) app to measure stuff,there's no better eye-opener!

 

[ManniX]

That's DSD256 through foobar (and SACD processor)

Are you sure it's going through foobar?
The output device should be the DSD Transcoder, not the Khadas.

And that's PCM 352.8 Khz to see if high bitrates affect Khadas,but no.

I'm not sure if it's the drivers or the USB issues in my case.
Have to try again when I can fix it.

That's Multitone32 DSD

This is were it's quite odd.
You should get results at least good as mine if not better.

I'm wondering if there's something wrong in how MTA is generating the DSD test tone.
I see the same oddities with mine.
The tones are not generated equal between PCM and DSD.

To make it comparable I had to hunt for a Low frequency limit for the test signal which gave me mostly comparable results between them.
Can you try to set as mine the low limit to 5 Hz instead of 20 Hz?
They seems pretty similar to mine as 20 Hz so maybe it works as well for you.

 

[ManniX]

On another note,see how useful was to have the ability to diagnose problems for yourself?

Absolutely
Indeed, I'm still hoping for a new Cosmos that can measure unbalanced...

For example we hear all the time that a good DAC takes care of a noisy port but as you saw the truth is far from that,interference by the bad USB happily passes through everything and appears at the end of the chain.

I knew this and it's a big problem especially on AMD. The USB is terrible when compared to Intel, unfortunately.

The biggest upgrade in a system is a decent ADC and a Multitone (or REW,etc) app to measure stuff,there's no better eye-opener!

Well, the biggest upgrade is when you manage to fix the stuff you uncover with it

 

[Sokel]

Are you sure it's going through foobar?
The output device should be the DSD Transcoder, not the Khadas.



I'm not sure if it's the drivers or the USB issues in my case.
Have to try again when I can fix it.



This is were it's quite odd.
You should get results at least good as mine if not better.

I'm wondering if there's something wrong in how MTA is generating the DSD test tone.
I see the same oddities with mine.
The tones are not generated equal between PCM and DSD.

To make it comparable I had to hunt for a Low frequency limit for the test signal which gave me mostly comparable results between them.
Can you try to set as mine the low limit to 5 Hz instead of 20 Hz?
They seems pretty similar to mine as 20 Hz so maybe it works as well for you.

If you see the test signal at foobar's measurement says "external".
I just played it from foobar and recorded it in Multitone,it's much easier (I don't use transcoder in foorba,I use ASIO+DSD,is much better and plays everything PCM and DSD up to 512 native changing everything automatically).

Here's a 5 to 20000Hz MT32 DSD:

 

[Sokel]

..and yep,it's level for me.
DSD don't like attenuation,increasing the level just 1db (after that E-MU clips) gets better:



I have to get my lazy a** to make some nice voltage dividers (too bad autoranger is no longer available).

 

[ManniX]

If you see the test signal at foobar's measurement says "external".

I just played it from foobar and recorded it in Multitone,it's much easier (I don't use transcoder in foorba,I use ASIO+DSD,is much better and plays everything PCM and DSD up to 512 native changing everything automatically).

I didn't noticed it...

I use the DSD Transcoder with the default ASIO driver cause the ASIO+DSD is unstable for me, at least with v2.0 x64.
Every now and then it hangs over the course of long listening sessions. I have to check it again, last time I tested was with the PO100 AK.

Have to try but somehow the Focusrite decided not to record anything so I can't test right now...

Here's a 5 to 20000Hz MT32 DSD:

You need to keep the frequency to test at 20 Hz, only change the test signal to 5 Hz.
If you see the generated test signal is odd and not going to -120 dBr as it should below 30 Hz.
You may need to find a better spot than mine, changing to 8 Hz the low freq of test signal generation makes a lot of difference and so on.

 

[Sokel]

I didn't noticed it...

I use the DSD Transcoder with the default ASIO driver cause the ASIO+DSD is unstable for me, at least with v2.0 x64.
Every now and then it hangs over the course of long listening sessions. I have to check it again, last time I tested was with the PO100 AK.

Have to try but somehow the Focusrite decided not to record anything so I can't test right now...



You need to keep the frequency to test at 20 Hz, only change the test signal to 5 Hz.
If you see the generated test signal is odd and not going to -120 dBr as it should below 30 Hz.
You may need to find a better spot than mine, changing to 8 Hz the low freq of test signal generation makes a lot of difference and so on.

Oh,now I see what you're talking about,we talked about this in MT's thread.
You just have to increase FFT size.

 

[ManniX]

Oh,now I see what you're talking about,we talked about this in MT's thread.
You just have to increase FFT size.

Oh damn, I didn't see it...
I tried at some point but didn't see any improvement.
Was mostly focusing on getting some comparable data.
Have to fix this issue with the Focusrite, I can't test anything now

 

[Sokel]

Oh damn, I didn't see it...
I tried at some point but didn't see any improvement.
Was mostly focusing on getting some comparable data.
Have to fix this issue with the Focusrite, I can't test anything now

(see the edited)

 

[ManniX]

(see the edited)

Not sure if something is wrong with this Focusrite...
Most of high rate PCM measurements ends up with this:

At some point during recording it gets crazy.
If I raise the gain with the know the noise floor gets much higher.
Not sure if I'm doing something wrong or not.

 

[Sokel]

Not sure if something is wrong with this Focusrite...
Most of high rate PCM measurements ends up with this:

View attachment 291294

At some point during recording it gets crazy.
If I raise the gain with the know the noise floor gets much higher.
Not sure if I'm doing something wrong or not.

It's not the interface,it's the PC.
Try to free as much memory and CPU usage as you can,close browsers,etc.

 

[ManniX]

It's not the interface,it's the PC.
Try to free as much memory and CPU usage as you can,close browsers,etc.

Uhm, could be
This is one of the most bloated Windows installation on the planet...
Let's see.