Setting up a HQ Player (w/ convolution) 2.2 system

[semente]

I am considering the move to a 2.2 system.

My current setup consists of HQ Player, NAA, DAC, integrated and stereo speaker pair.

 

I would like to know if it is possible to high-pass the main speakers using filters made with REW.

 

Do I need a digital preamplifier with sub outs?

A 4-channel DAC (which cheap ones)?

Another NAA (latency issues)?

[blue2]

I have thought about this too. If you use HQP to upsample and do the convolution then you'll need a 3 or 4 channel DAC. I don't think there's a technical reason that HQP couldn't output to your stereo DAC and also the line out on your computer (for the subs) but it's not supported. Maybe @Miska can clarify this point?

If you go with a digital preamp it needs to handle 2-> 4 channel and convolution. HQP might still be useful in this case if you need to convert DSD to  PCM or want to upsample PCM.

Also note that if you're a fan of DSD you lose that by going with convolution whereas it's retained using analogue filters on your subs, but the mains won't have high pass.

My thinking at the moment is along the lines of something like ATC active speakers, without convolution. Although DRC and speakers such as D&D 8c are alluring 🙂

[Miska]

24 minutes ago, blue2 said:

I have thought about this too. If you use HQP to upsample and do the convolution then you'll need a 3 or 4 channel DAC. I don't think there's a technical reason that HQP couldn't output to your stereo DAC and also the line out on your computer (for the subs) but it's not supported. Maybe @Miska can clarify this point?

 

They are not clocked from the same source. And most OS don't support launching multiple unrelated devices in sync.

 

There are multichannel DACs on the market, so that's not an issue.

 

27 minutes ago, blue2 said:

Also note that if you're a fan of DSD you lose that by going with convolution

 

Why would you?

 

[semente]

4 minutes ago, Miska said:

There are multichannel DACs on the market, so that's not an issue.

 

So the best (only?) option is to get a multi-channel DAC?

[Miska]

1 minute ago, semente said:

So the best (only?) option is to get a multi-channel DAC?

 

Yes, if you'd like to run up to DSD256 then the options are pretty much exaSound and Merging...

 

[semente]

7 minutes ago, Miska said:

 

Yes, if you'd like to run up to DSD256 then the options are pretty much exaSound and Merging...

 

 

Thanks.

All very expensive options. I'm better off getting speakers with more LF extension.

 

[DuckToller]

How about the OktoResearch DAC 8 pro  (up to DSD128) at 1200 Euros ?

https://www.oktoresearch.com/dac8pro.htm
Sorry, linking did not work due to an unspecified error of the site design ....

[semente]

43 minutes ago, DuckToller said:

How about the OktoResearch DAC 8 pro  (up to DSD128) at 1200 Euros ?

https://www.oktoresearch.com/dac8pro.htm
Sorry, linking did not work due to an unspecified error of the site design ....

 

I thought it would be more expensive than that, its price is actually quite close to that of the RME DAC.

[blue2]

22 hours ago, Miska said:

 

(A) They are not clocked from the same source. And most OS don't support launching multiple unrelated devices in sync.

 

(B) There are multichannel DACs on the market, so that's not an issue.

 

 

(C) Why would you?

 

(A) that's a pity. In a multichannel system this would be a showstopper, but for subs the frequency range is probably < 200Hz, even < 80Hz. Would synchronisation really be noticeable on subs?

(B) It is for me as I like my DAC (T+A DAC8 DSD). I suppose that would have to be part of the compromise if I went down the convolution path i.e. DRC and/or digital filtered 2.2 system

(C) I've read many times that DSD can't be mixed/processed except by conversion to PCM despite some mixing software claims. As an aside Sonoma & Pyramix claim to process native DSD: do you know if they internally use PCM?

Anyway in a hifi context I'd want to play DSD source files and you're saying HQPlayer can do convolution outputting as DSD so that makes it indispensable in this use case - thanks for the correction. I believe Roon claim to have the capability too.

Can HQP apply convolution filters to DSD natively? I guess it's a bit academic if HQP can do DSD in DSD out with convolution and there is no loss of SQ, but as you know especially on this site folk are quite pernickety, and I'm curious.

Thanks

[CJH]

If you want to play native DSD and keep it as DSD, then perhaps analog electronic filters after your DAC is a good solution. Not much different than adding an analog preamp between DAC and amp.

CJH

[Miska]

16 hours ago, blue2 said:

Can HQP apply convolution filters to DSD natively? I guess it's a bit academic if HQP can do DSD in DSD out with convolution and there is no loss of SQ, but as you know especially on this site folk are quite pernickety, and I'm curious.

 

Let's say you play DSD64 upsampled to DSD256 with convolution. The path looks like this:

DSD64 -> convolution at 2.8 MHz sampling rate -> 4x SDM conversion to 11.3 MHz sampling rate -> DSD256

 

And for 44.1/24 PCM upsampled to 352.8/24 with convolution the path looks like this:

44.1/24 -> convolution at 44.1 kHz sampling rate -> 8x PCM conversion to 352.8 kHz sampling rate -> 352.8/24

 

HQPlayer has two separate DSP engines, one for PCM and one for SDM. Improvements in sound quality are similar both DSD and PCM.

 

16 hours ago, blue2 said:

(A) that's a pity. In a multichannel system this would be a showstopper, but for subs the frequency range is probably < 200Hz, even < 80Hz. Would synchronisation really be noticeable on subs?

 

I rather not enter such gray areas...

 

[bogi]

3 hours ago, Miska said:

And for 44.1/24 PCM upsampled to 352.8/24 with convolution the path looks like this:

44.1/24 -> convolution at 44.1 kHz sampling rate -> 8x PCM conversion to 352.8 kHz sampling rate -> 352.8/24

I am curious, what's the path please, when you listen to 44.1/16 content, want to upsample to DSD128 and your IR WAVs are 192/24 ?

[Miska]

6 hours ago, bogi said:

I am curious, what's the path please, when you listen to 44.1/16 content, want to upsample to DSD128 and your IR WAVs are 192/24 ?

 

44.1/16 -> convolution at 44.1 kHz sampling rate -> 128x PCM-to-SDM converstion to 11.3 MHz sampling rate -> DSD128

 

[bogi]

Thanks. So when the the IR WAV sample rate does not match the source content rate it is always first resampled to the source sample rate to perform convolution at that sample rate.

 

And in the DSD case

12 hours ago, Miska said:

DSD64 -> convolution at 2.8 MHz sampling rate -> 4x SDM conversion to 11.3 MHz sampling rate -> DSD256

are the IR WAVs at first upsampled to DSD64 to perform convolution at 2.8 MHz?

[Miska]

27 minutes ago, bogi said:

Thanks. So when the the IR WAV sample rate does not match the source content rate it is always first resampled to the source sample rate to perform convolution at that sample rate.

 

Yes, although "resampled" is wrong term, it is scaled using algorithm designed for that particular purpose.

 

28 minutes ago, bogi said:

are the IR WAVs at first upsampled to DSD64 to perform convolution at 2.8 MHz?

 

Yes, although here too "upsampled" is wrong term. So they are scaled and converted to suitable form, and optionally with the "HF Expansion" that extends the frequency response flat beyond Nyquist frequency of the filter sampling rate.

 

 

Overall, the filters always go through bunch of adaptation steps to make them suitable for actual processing.

 

[jamesg11]

On 10/11/2020 at 11:27 PM, CJH said:

If you want to play native DSD and keep it as DSD, then perhaps analog electronic filters after your DAC is a good solution. Not much different than adding an analog preamp between DAC and amp.

CJH

This is where I’m at, though not yet sorted a computer build to handle 256 with best modulators

ie a TI Japan “high-speed precision FET active crossover” added into analogue passive pre, distributing to Naquadria Lucien, Elac mains & to Perreaux 9000B, Naquadria 12” long-throw 2xsubs. 2 channel.

Now also need DSP filter for HQP convolution.

[semente]

19 hours ago, Miska said:

 

44.1/16 -> convolution at 44.1 kHz sampling rate -> 128x PCM-to-SDM converstion to 11.3 MHz sampling rate -> DSD128

 

 

I'm using a UD-501 at DSD128. Should I export the convolution WAV filter at 44.1kHz or at 384kHz?

 

Most of my content is 44.1 but I also have 48kHz, 88.2 and 96kHz.

[Miska]

32 minutes ago, semente said:

I'm using a UD-501 at DSD128. Should I export the convolution WAV filter at 44.1kHz or at 384kHz?

 

Most of my content is 44.1 but I also have 48kHz, 88.2 and 96kHz.

 

If the tool can create a proper filter at higher rate, it is better to create it for a rate same or higher than highest rate content you have. So if possible at least 192k or 352.8k. 352.8k should cover all PCM content there is available (since I haven't seen any 384k) and also bandwidth wise cover all possible content frequencies.

 

[semente]

50 minutes ago, Miska said:

 

If the tool can create a proper filter at higher rate, it is better to create it for a rate same or higher than highest rate content you have. So if possible at least 192k or 352.8k. 352.8k should cover all PCM content there is available (since I haven't seen any 384k) and also bandwidth wise cover all possible content frequencies.

 

Thanks. Will replace 384k with 352.8k

[jamesg11]

10 hours ago, Miska said:

 

If the tool can create a proper filter at higher rate, it is better to create it for a rate same or higher than highest rate content you have. So if possible at least 192k or 352.8k. 352.8k should cover all PCM content there is available (since I haven't seen any 384k) and also bandwidth wise cover all possible content frequencies.

 

/& if highest rate content is dsf256?

[Miska]

15 minutes ago, jamesg11 said:

/& if highest rate content is dsf256?

 

The same, 352.8k filter has frequency response spanning from 0 to 176.4 kHz which should be enough for any musical content.