ESS IMD hump with AKM DAC? (specifically AK4499EXEQ in Topping E70 Velvet)

[KSTR]

Do you think there are any tests I could do on the E70 Velvet to look deeper into what's going on?

Not much, I fear, without measurement equipment. Note that the ripple problem also exists with ADCs and that's why RME have added a compensation in the ADI-2 Pro (AKM Version) where the ripple (and general roll-off) is compensated for, for the 384kHz sample rate.

A test you could do is playing shaped impulses and try to listen if sounds different with different filters and on different DACs. I've attached such a test pulse.

 

[mike7877]

Not much, I fear, without measurement equipment. Note that the ripple problem also exists with ADCs and that's why RME have added a compensation in the ADI-2 Pro (AKM Version) where the ripple (and general roll-off) is compensated for, for the 384kHz sample rate.

A test you could do is playing shaped impulses and try to listen if sounds different with different filters and on different DACs. I've attached such a test pulse.

Thanks. I'll try it out when I get home

 

[mike7877]

OK, understood.
Anyway, the issue is there with almost any DAC chip, to different amounts of course.
Now we know the reason why datasheets seldom show close-up filter responses :
View attachment 355960
ESS allows for custom FIR filters so maybe the E70 ESS version is using that...

From 9018 datasheet I was flipping through for unrelated

 

[Sokel]

Is it possible to, say... on PC, resample PCM to DSD? Like, have the PCM upsampled and interpolated and line of best fit and then analyzed and output in DSD?

edit: forgot my reply to you: Good point/good to know about the hump essentially disappearing when DSD is he source material

You can do it using foobar with SACD and DSD processor (they are in the same file when you download SACD plug-in) plug-ins,easily.

 

[mike7877]

You can do it using foobar with SACD and DSD processor (they are in the same file when you download SACD plug-in) plug-ins,easily.

Nice. Is it something you do on occasion or with certain DACs?

I had a random thought: although I don't understand all the ins/outs of DSD vs. PCM, it seems easier to end up at an analog waveform representing the original with it. And cheaper.
With enough processing power to do a full resolution PCM > DSD conversion probably available now with [even] an older ARM chip like Snapdragon 8 - with room to spare. Maybe this will be the direction ESS/AKM take their DACs in the future? If I was at the helm at AKM I'd at least be experimenting with it for chips intended for high tier, non battery operated things (things that could handle 1.5-2.5 watts more dissipation and draw). The resulting stream wouldn't be any less accurate than the 256x or similar oversampling they do with their DACs currently, and transforming data to analog is even better handled by AKM's current 4499EX when DSD is the format (allegedly lol). Do you think anything like this is likely to happen? Or just incremental improvement in low-bit (4, 5, 6, 7, 8) DACs with sporadic improvements to delta sigma modulators.

Anecdotally, in the PS Audio review I posted earlier in this thread of the E70 Velvet, the reviewer (not Paul McGowan if you know who he is lol - great guy) was especially impressed with its DSD playback ability. If I recall correctly, he said for notable DSD sound quality improvement over the E70 Velvet, you'd need to move up to ~7-8x its price. That's quite the endorsement!... the last time I made an endorsement like that was regarding ROI on room treatment for sound quality improvement. And using a passive preamp if line level to your power amp's gain level is high enough to reach your preferred max listening level. Generally, with audio, you get what you pay for (if you have an eye for snake oil and dishonesty and aren't prone to blind brand loyalty and do due dilligence), but I still wouldn't buy anything at all from anyone at all if the seller's returns and exchange protocol was "no refunds, exchange for same device only"

 

[Sokel]

Nice. Is it something you do on occasion or with certain DACs?

I had a random thought: although I don't understand all the ins/outs of DSD vs. PCM, it seems easier to end up at an analog waveform representing the original with it. And cheaper.
With enough processing power to do a full resolution PCM > DSD conversion probably available now with [even] an older ARM chip like Snapdragon 8 - with room to spare. Maybe this will be the direction ESS/AKM take their DACs in the future? If I was at the helm at AKM I'd at least be experimenting with it for chips intended for high tier, non battery operated things (things that could handle 1.5-2.5 watts more dissipation and draw). The resulting stream wouldn't be any less accurate than the 256x or similar oversampling they do with their DACs currently, and transforming data to analog is even better handled by AKM's current 4499EX when DSD is the format (allegedly lol). Do you think anything like this is likely to happen? Or just incremental improvement in low-bit (4, 5, 6, 7, 8) DACs with sporadic improvements to delta sigma modulators.

Anecdotally, in the PS Audio review I posted earlier in this thread of the E70 Velvet, the reviewer (not Paul McGowan if you know who he is lol - great guy) was especially impressed with its DSD playback ability. If I recall correctly, he said for notable DSD sound quality improvement over the E70 Velvet, you'd need to move up to ~7-8x its price. That's quite the endorsement!... the last time I made an endorsement like that was regarding ROI on room treatment for sound quality improvement. And using a passive preamp if line level to your power amp's gain level is high enough to reach your preferred max listening level. Generally, with audio, you get what you pay for (if you have an eye for snake oil and dishonesty and aren't prone to blind brand loyalty and do due dilligence), but I still wouldn't buy anything at all from anyone at all if the seller's returns and exchange protocol was "no refunds, exchange for same device only"

I test every DAC that falls to my hands to see,mostly out of curiosity.
It seems that DSD is handled differently,that's only by observation,I don't have the knowledge to go deep.
But it's enough to know that the format has nothing to do probably,it's how each device handles it that matters.

Why am I interested?Cause I listen to classical and there lots of really nice stuff on DSD.
So...

 

[PlasticDoc]

Cause I listen to classical

You may already know them, but I was surprised by the passion and love for classical music in these two clips: 10 Essential 20th Century Symphonies and 10 Ridiculously Underplayed Symphonies.
Please excuse me for going off-topic.

 

[Sokel]

You may already know them, but I was surprised by the passion and love for classical music in these two clips: 10 Essential 20th Century Symphonies and 10 Ridiculously Underplayed Symphonies.
Please excuse me for going off-topic.

They're almost classical themselves,love them!

 

[pma]

A test you could do is playing shaped impulses and try to listen if sounds different with different filters and on different DACs. I've attached such a test pulse.

Klaus, would you kindly elaborate what kind of changes should we hear?

 

[Music1969]

It seems that DSD is handled differently

Yes you bypass all ESS chip filtering when you feed it DSD256/512.

Just the ESS modulator is not bypassed.

So the IMD hump might be related to ESS filtering... maybe related to @KSTR 's comment about the filter plot he showed

 

[Sokel]

Yes you bypass all ESS chip filtering when you feed it DSD256/512.

Just the ESS modulator is not bypassed.

So the IMD hump might be related to ESS filtering... maybe related to @KSTR 's comment about the filter plot he showed

The trend follows DSD64 and DSD128 too:



PCM,where the hump is in all its glory



DSD64,greatly reduced



..and DSD128 even more

(going higher reduces it even more)

 

[mike7877]

I test every DAC that falls to my hands to see,mostly out of curiosity.
It seems that DSD is handled differently,that's only by observation,I don't have the knowledge to go deep.
But it's enough to know that the format has nothing to do probably,it's how each device handles it that matters.

Why am I interested?Cause I listen to classical and there lots of really nice stuff on DSD.
So...

I listen to a bit of classical. I think my favourite is the Telarc Tchaikovsky compilation with the real cannons. It was recorded in 1979, and when they recorded the cannons they used the real machinery (with blanks obv). I read on the insert it blew out five floors of glass windows from a building over 500ft away lol. Where do you get your DSD files?

 

[MaxBuck]

Is any of this stuff potentially audible? Asking for a friend.

 

[Rednaxela]

ESS called, they want their hump back.

 

[Sokel]

I listen to a bit of classical. I think my favourite is the Telarc Tchaikovsky compilation with the real cannons. It was recorded in 1979, and when they recorded the cannons they used the real machinery (with blanks obv). I read on the insert it blew out five floors of glass windows from a building over 500ft away lol. Where do you get your DSD files?

The usual suspects,NativeDSD,SA-CD net.,etc.
Some even have direct (as they say) DSD recordings but the music is usually not very interesting,is mostly "look how well I can record".
Small ensembles though,nothing grand and exciting.

 

[KSTR]

Klaus, would you kindly elaborate what kind of changes should we hear?

View attachment 356085 View attachment 356089

Pavel, the idea is if a DAC filter produces relatively strong pre- and post- echo pulses it might be audible with such a test pulse directly. Since the filters often have a length of at least several dozen samples those extra pulses are like 1..2ms away from the main pulse, maybe much more. That might prevent full masking.
It's just an idea, ideally one would first do experiments with artificially added pre- and post-pulses

 

[KSTR]

Is any of this stuff potentially audible? Asking for a friend.

A sense of graininess perhaps, less "see-through" properties with regard to soundstage and decaying reverb tails, more long-term listening fatigue than with better DACs, that subtle kind of things... for an experienced listener maybe.
The unstable and hashy nature of the distortion is what's potentially problematic.
So it might be audible to some but it's certainly not a deal breaker, not rendering a product "unlistenable".

 

[KSTR]

Yes you bypass all ESS chip filtering when you feed it DSD256/512.

Just the ESS modulator is not bypassed.

So the IMD hump might be related to ESS filtering... maybe related to @KSTR 's comment about the filter plot he showed

Well, from what I've seem to have found the root problem is the modulator scheme in the final 5/6-bit DAC stage in PCM modes. In DSD modes this modulating scheme is replaced with something else only ESS knows to feed that final DAC stage. Maybe it's just a resampled 5/6-bit PCM capture of the original 1-bit DSD stream. Think of a digital oscilloscope when you are viewing the direct DSD signal...

 

[Music1969]

Well, from what I've seem to have found the root problem is the modulator scheme in the final 5/6-bit DAC stage in PCM modes. In DSD modes this modulating scheme is replaced with something else only ESS knows to feed that final DAC stage. Maybe it's just a resampled 5/6-bit PCM capture of the original 1-bit DSD stream. Think of a digital oscilloscope when you are viewing the direct DSD signal...

Well ESS sum 1-bit DACs. Maybe something related here.

@Sokel can you followup your PCM96k vs DSD256 with one more comparison

If you upsample to PCM353kHz you will by-pass ESS first stage oversampling - does that help IMD hump ?

 

[KSTR]

Well ESS sum 1-bit DACs. Maybe something related here.

That's just their nomenclature for the final 6 or 7-bit thermometer-coded final DAC.

If you upsample to PCM353kHz you will by-pass ESS first stage oversampling - does that help IMD hump

In my experiments the ESS hump phenomen was completely unaffected by sample rate... that was for an asynchronous implementation of ES9038q2m, running on a fixed 100MHz clock for the final DAC. With a synchronous setup of the chip it might be different but I doubt that.