[ohcrapgorillas]

It blows me away that there are so many like Rob Watts of Chord who think delta-sigma are better than R2R DACs. I can't stand delta-sigma DACs. I listened to a DAVE/M-Scaler at canjam and wasn't impressed. I have a Spring 3, and the only other DAC I have had that I really liked was my old Bifrost 2. R2R DACs seem to me to be better. I'm not sure what the draw of delta-sigma is. Maybe ease of manufacture.

Jussi seems to not be too interested with going all out to help build up knowledge of his product. Lachan from PassionForSound tried to reach out to him when he did his review of HQPlayer, and Jussi apparently didn't want to be interviewed for some reason.

I agree wholeheartedly about R2R DACs sounding better to me, but this hobby is subjective and D/S DACs do have tons of detail. Not to mention, good D/S DACs are a lot cheaper than a great R2R DAC.

Anyway, if you are using R2R DACs, why bother with DSD oversampling in HQPlayer? Just do PCM oversampling at the maximum sampling rate accepted by your R2R DAC and don't worry about needing a powerful PC. PCM processing is vastly less intensive than DSD and can be done with something as simple as a RPi, so your laptop or PC will have no problems at all. Either your DSD signal is being converted to PCM in your R2R DAC, so DSD is an unnecessary intermediate step, or it's using a different 1-bit DAC (that's what my Gustard R26 does) which bypasses the ladder that you paid a premium for.

EDIT: also, if you're going to be doing PCM oversampling into a NOS R2R DAC, make sure that you've set the DAC bits correctly for dithering. For example, since the Bifrost 2 is linear down to 16 bits, the DAC bits need to be set to 16.

 

[Jake2]

[

It's just a sliding scale, there isn't really a "this CPU is enough and this one isn't" barrier. I'm not sure where the line is for which a particular setup can or could not run x512 ASDM7EC-Super Sinc-L for example.

I would love to know where that line is for this specific modulator and filter combo at x512.

Background is after recently upgrading to HQP5 and getting more familiar with filter & modulator combos over the last few weeks I really like what I'm hearing at x256 with that combo with my Mac Mini M1 16GB, albeit it's a bridge just too far with occassional dropouts. Below which my current pref is, IIRC, Sinc-Mx/7ECv3, but not quite on the same level.

I can see from folks' sigs on this thread what hardware would more than get me there, some quite pricey rigs, I'm still not sure what the minimum spec and spend to do so would be.

Interesting reading how very close the latest gen Apple silicon gets to this, if not quite re Sinc-L at x512 it seems.

Yes I know, I know, I need to (re)visit the mega AS thread(s), and probably just cut to the chase and ask Jussi...

 

[Jake2]

I agree wholeheartedly about R2R DACs sounding better to me, but this hobby is subjective and D/S DACs do have tons of detail. Not to mention, good D/S DACs are a lot cheaper than a great R2R DAC.

Anyway, if you are using R2R DACs, why bother with DSD oversampling in HQPlayer? Just do PCM oversampling at the maximum sampling rate accepted by your R2R DAC and don't worry about needing a powerful PC. PCM processing is vastly less intensive than DSD and can be done with something as simple as a RPi, so your laptop or PC will have no problems at all. Either your DSD signal is being converted to PCM in your R2R DAC, so DSD is an unnecessary intermediate step, or it's using a different 1-bit DAC (that's what my Gustard R26 does) which bypasses the ladder that you paid a premium for.

EDIT: also, if you're going to be doing PCM oversampling into a NOS R2R DAC, make sure that you've set the DAC bits correctly for dithering. For example, since the Bifrost 2 is linear down to 16 bits, the DAC bits need to be set to 16.

I agree re PCM sounding damn good with R2R dacs and far less demanding and expensive to compute. PCM768 + Sinc-Mx-MG/a (can't quite make up my mind) + LNS15 sounds superb on my R26 and has been my default in recent days. It's just DSD on the above settings at x256 is a step up in resolution, refinement and dynamic range to my ears. DSD512 will be even more so from the tantalising glimpses I can manage with my M1. But boy it'll be expensive to get there.

 

[bogi]

I'm not sure what the draw of delta-sigma is. Maybe ease of manufacture.

Price/performance ratio and total majority of low cost applications everywhere. Standalone DACs are not used by many people. The amount of low cost D/S applications for ordinary people with no high end interest is many many magnitudes higher.

For $500 you don't get better R2R DAC than D/S DAC (some Cayin users could oppose) and for $1000 it is also questionable.

HQPlayer also utilizes delta sigma technology. Just on better level than is reachable in performance compromised low cost delta sigma chips. So it is not about D/S to be "bad", just there is very little amount of other than DAC chip based implementations.

 

[genefruit]

Thanks. Yes, I also have found sinc-long sounding better than sinc-L, but haven't used it as it's been too heavy with DSD1024. However with PCM I can do both at max rate.

Today I've been playing with bit depths and also TPDF vs NS5/LNS15. NS5 is still nice, but LNS15 goes too far, making things artificially soft somehow. NS5 as well, but to much lesser extent. Overall noise shapers make the sound more deep in some sense, like I'm going "there". But with TPDF it's the other way around: music is here. Especially in transients TPDF has more that "raw" realism. It's like NS5 makes me drift and TPDF makes me awake. I really like sinc-long vs TPDF combination, especially with acoustic music. Really musical and superb fit with acoustic music. I usually don't listen to country, but I've been lost in some random music for 2h.

P.S. I've done some further comparison with bit depth. 23 was still clearly preferred over lower ones, but I also started to prefer 24 over 23. Especially with originally 24 bit content there is clear difference in liveliness IMO. Not sure if truncating rounds things into mellowness or if 24 bit has some distortion that I like, but 24 "just works" for me, feeling very real, natural, kind of "live".

This day was a big surprise as well.

Tinkering with PCM and 24bit today. Perhaps it's simply the sound of something different but it's hitting some aspects I'm enjoying, especially staging.

 

[Superunknown2]

I agree re PCM sounding damn good with R2R dacs and far less demanding and expensive to compute. PCM768 + Sinc-Mx-MG/a (can't quite make up my mind) + LNS15 sounds superb on my R26 and has been my default in recent days. It's just DSD on the above settings at x256 is a step up in resolution, refinement and dynamic range to my ears. DSD512 will be even more so from the tantalising glimpses I can manage with my M1. But boy it'll be expensive to get there.

What is ultimately the difference between what DSD does versus PCM on HQPlayer? Why does DSD need so much more computing power?

 

[bogi]

The Wandla has some HQPlayer functions built in, but it is a delta-sigma DAC and so it can't do NOS.

Wandla appears on Jussi's list since he designed digital filters used in Wandla FPGA. You are true Wandla is ESS based D/S therefore it does not appear on the AS list.
Denafrips models on the AS list are capable of direct DSD. They don't appear in the PCM part of the list.

 

[GoldenSound]

With a delta sigma DAC you'll still get the benefit of the HQPlayer oversampling filters, you just can't take full advantage of the high performance noise shapers since the DACs will be running their own noise shapers/modulators on top anyway

 

[genefruit]

What is ultimately the difference between what DSD does versus PCM on HQPlayer? Why does DSD need so much more computing power?

math - the file size created by a DSD file is significantly larger than PCM and to create it in real time requires moar power

 

[bogi]

What is ultimately the difference between what DSD does versus PCM on HQPlayer? Why does DSD need so much more computing power?

That would be for very long post. Short answer: For the same output quality as D/S chips are capable of it would require equally low power. I partially answered few posts above.

 

[bogi]

With a delta sigma DAC you'll still get the benefit of the HQPlayer oversampling filters, you just can't take full advantage of the high performance noise shapers since the DACs will be running their own noise shapers/modulators on top anyway

That's totally incorrect.

Many delta sigma DAC chips provide direct DSD path possibility and many DAC manufacturers utilize that possibility with D/S chips. Just look at the DSD part of the AS list. DAC chips used in those devices are ALL of delta sigma design. You don't find ESS based DACs on that list but not only ESS chips are delta sigma design ...

Also discrete DSD designs like Holo, T+A, including Jussi's DSC1 and DSC2 ARE delta sigma design, since they work ONLY with delta sigma modulated input signal (what DSD actually is).

Look how Jussi described his DSC1:
"DSC1 is an Open Hardware, "discrete" delta-sigma D/A-converter specifically designed for 256x and higher sampling rates and optimized for seventh order modulators used in Signalyst HQPlayer."
DSC1 does not contain any delta sigma modulator. But it requires delta sigma modulated signal on its input. Like all direct DSD capable DACs. DSD signal is result of delta sigma modulation and no other than delta sigma based D/A section can accept such signal.
DSC1 and DSC2 open hardware DACs were inspirations for DAC manufacturers like Holo. They share the same design principles. That's all delta sigma. Just other way than is implemented in few dollar costing and hardware compromised chips.

 

[GoldenSound]

That's totally incorrect.

Many delta sigma DAC chips provide direct DSD path possibility and many DAC manufacturers utilize that possibility with D/S chips. Just look at the DSD part of the AS list. DAC chips used in those devices are ALL of delta sigma design. You don't find ESS based DACs on that list but not only ESS chips are delta sigma design ...

Also discrete DSD designs like Holo, T+A, including Jussi's DSC1 and DSC2 ARE delta sigma design, since they work ONLY with delta sigma modulated input signal (what DSD actually is).

Look how Jussi described his DSC1:
"DSC1 is an Open Hardware, "discrete" delta-sigma D/A-converter specifically designed for 256x and higher sampling rates and optimized for seventh order modulators used in Signalyst HQPlayer."
DSC1 does not contain any delta sigma modulator. But it requires delta sigma modulated signal on its input. Like all direct DSD capable DACs. DSD signal is result of delta sigma modulation and no other than delta sigma based D/A section can accept such signal.

I was referring to PCM playback. Delta sigma DACs will run their own noise shapers/modulators by necessity which limits the possible performance improvement from using a more aggressive noise shaper (and depending on the one you pick could actually end up with more noise overall).

DSD is a different question and dependent on whether the DAC itself has a native DSD path or not. R2R DACs can't convert DSD but many R2R DACs such as Holo etc have a separate DSD converter inside. And many delta-sigma DACs can't handle DSD natively and will convert to PCM internally whereas others can convert DSD natively. Just need to check for the specific DAC you're looking at

 

[Superunknown2]

With a delta sigma DAC you'll still get the benefit of the HQPlayer oversampling filters, you just can't take full advantage of the high performance noise shapers since the DACs will be running their own noise shapers/modulators on top anyway

What does a "noise shaper" and a "modulator" do on HQPlayer?

I have seen some people here say that with a Holo Spring 3, there is no need to use DSD in HQPlayer. Would you agree with that?

 

[bogi]

I was referring to PCM playback.

OK, that was not clear from your post.

When using HQPlayer PCM output with delta sigma DAC, it's not only on chip D/S modulator which cannot be skipped. It is also the 2nd oversampling stage of these chips which cannot be skipped. Delta sigma modulators operating frequency appears in MHz range, older models often 5.6 MHz, newer usually 11.3 or 22.6 MHz. So when one upsamples in HQPlayer for example to 705.6k, the rest of oversampling up to for example 11.3 MHz (that's 16x oversampling ratio) is still performed in DAC chip. That 2nd oversampling stage is much more compromised implementation than the 1st stage which can be skipped by HQPlayer upsampling. The 2nd oversampling stage is source of audio band images at output of delta sigma DACs. I's happening because DAC chips don't have enough resources to perform digital filtering at such high sample rates. Jussi demonstrated that on many measurements.

 

[GoldenSound]

What does a "noise shaper" and a "modulator" do on HQPlayer?

I have seen some people here say that with a Holo Spring 3, there is no need to use DSD in HQPlayer. Would you agree with that?

Technically a modulator and noise shaper are sort of the same thing, though modulator specifically implies modulating to a lower bit depth.

Basically, when you convert a higher bit depth piece of info to a lower bit depth piece of info, you encounter a 'quantization error'. Simple example of this:

- You have a device with a maximum output of 3V
- You need to output exactly 2V
- This can be accurately represented by a 2-bit system, because two bits can be '11' (100%/3V), '10' (66%/2V), '01' (33%/1V) or '00' (0%/0V). So you just choose '10' and you're good.
- If you now wanted to get this output with a 1-bit system, you can't do it accurately. A 1-bit system can only be '1' (100%/3V) or '0' (0%/0V). And so picking the closest value of '1' (3V), you now have a quantization error of 1V, the difference between the value you intended and the actual value.

This can be shown visually too. Here for example is an image on the far left, represented with 8-bit accuracy, meaning there are 255 different possible shades from black/white that can be used to represent each pixel.

If you now wanted to represent each pixel with just 1-bit, meaning ONLY pure black or pure white, you can do so by just throwing away the bottom 7 bits and keeping only the top bit '1' (black) or '0' (white). But you've thrown away all the finer detail information about the particular shade and this leaves a fairly inaccurate result due to what is called truncation distortion.

We can address this by dithering, which is the process of intentionally adding some random noise to prevent any strong signal-correlated truncation distortion errors. This can get us closer to a more perceptually ideal result, but it's still not perfect.

We can take things a step further by applying dithering in a more advanced way, called noise-shaping. This means that rather than noise being added in a more random fashion and evenly distributed, a feedback loop system is employed to correct remaining noise/quantization error and provide more dynamic range within the area we want or the area of perceptual relevance, and shape the quantization error out to areas where it's not as much of an issue.

In audio, this means that the noise shaper is designed in a way which provides higher dynamic range in the audible band itself, and shapes the quantization errors out to higher frequencies, where they can just be filtered out with an analog filter.

This image shows a 1khz signal, represented by 768khz 8-bit PCM. No dithering was applied, and so we can see unwanted extra content, truncation distortion.

Now if we apply some standard TDPF dither, this adds a small amount of random noise, but in doing so removes the truncation distortion and actually results in better effective dynamic range:

We can then try with a noise shaper, HQPlayer's LNS15 noise shaper in this instance, which prioritises dynamic range in the audible band, and shapes the noise out to higher frequencies and gives us even better dynamic range up to about 100khz, and the higher noise above 100khz can just be filtered out by the DACs analog filter:

Noise shaping can be applied when upsampling just in PCM as shown above, because the new samples generated as part of the upsampling are only as accurate as the bit-depth of the output allows. They can't perfectly represent the ideal real value, so quantization error is still a concern even though you're not actually reducing bit depth.

But when doing upsampling + modulation, such as converting 16 bit PCM info to higher rate, 1-bit DSD info, you have enormous quantization error, and thus there's a lot of work for the modulator to do to try to get a high dynamic range output in the area you want.

SUPER TLDR:
Even if something can only be 'on' at 100% or 'off' at 0%, you can still get an output of say 80% by turning it off and on a hundred thousand times rapidly, with it being 'on' about 80% of the time, and then filtering out the excess high-frequency noise.