[xiamen]

Hi, maybe some one has an idea using HQPlayer and NAA.
does’ anyone have good experience up sampling DSD64 (SACD) to DSD512.
I can’t find any combination that sounds better then bypassing DSD directly to the DAC ( es9038Pro ).
Sending DSD directly to the DAC has better Transients, and deeper soundstage.
Up sampling PCM to DSD has a big Sonic impact . Everything is improved with PCM.

My Chain:
PC ( Rayzen 7 3800x ) ---USB---> iFi Micro ---USB---> Singxer SU-6 ---HDMI---> DAC ( L.K.S Audio MH-DA004 Dual ES9038pro ) ----> Accuphase E-470

In theory (according to https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/) and I quote "This is also why advanced player software, such as JRiver, offers Double-Rate DSD output. Using player software that upsamples DSD64 to DSD128 or DSD256 significantly improves performance by putting the digital artifacts octaves above audibility allowing more advanced noise-shaping algorithms and less severe digital filters. Note these extremely high sampling frequencies are why ultra accurate clocking is more important in DSD vs. PCM recordings.". What filter are you using. With your CPU, you may give ASDM7EC a try to upsample to say 128/256K (128K should definitely work for your CPU), using poly-xtr-mp as your filter assuming you are on HQPlayer 4. Benefit of upsampling to 128K instead of 256K is you don't need a top notch GPU for cuda offload with the poly-xtr-mp filter. But if native DSD64 sounds better to you, then why not stick with it.

 

[Xoverman]

In theory (according to https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/) and I quote "This is also why advanced player software, such as JRiver, offers Double-Rate DSD output. Using player software that upsamples DSD64 to DSD128 or DSD256 significantly improves performance by putting the digital artifacts octaves above audibility allowing more advanced noise-shaping algorithms and less severe digital filters. Note these extremely high sampling frequencies are why ultra accurate clocking is more important in DSD vs. PCM recordings.". What filter are you using. With your CPU, you may give ASDM7EC a try to upsample to say 128/256K (128K should definitely work for your CPU), using poly-xtr-mp as your filter assuming you are on HQPlayer 4. Benefit of upsampling to 128K instead of 256K is you don't need a top notch GPU for cuda offload with the poly-xtr-mp filter. But if native DSD64 sounds better to you, then why not stick with it.

Thanks for the reply.
That's exactly what I experimented when yousing HQPLAYER. HQPlayer is probably the best program around to do the job.
Upsampling PCM to 256K with poly-xtr-mp / ASDM7EC is no problem. But I actually prefer poly-xtr-mp / SDM5 / 512KHz . For me it has better transient response. Upsampling PCM with HQPLAYER has a big sonic benefit on the LKS.
That's why I just can't believe that the es9038pro dos a better job upsample dsd64 then HQPLAYER.

 

[piaseczek]

Hi,
I wonder, will I ommit the first stage of upsampling in the es9038 with external upsampling (8x) to 352.8kHz?

 

[Lennym]

In theory (according to https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/) and I quote "This is also why advanced player software, such as JRiver, offers Double-Rate DSD output. Using player software that upsamples DSD64 to DSD128 or DSD256 significantly improves performance by putting the digital artifacts octaves above audibility allowing more advanced noise-shaping algorithms and less severe digital filters.

JRiver (I have MC 23.) does not seem to permit native format DSD upsampling using the Amanero-WASAPI.

It does permit upsampling after conversion to DoP. For native format upsampling you need to use ASIO.

 

[MartinWT]

My new streaming front end is now complete and sounding magnificent. Asus Tinker Board S streamer running Volumio with Qobuz and Spotify subscriptions. Mutec MC-3+ USB reclocker. LKS MH-DA004 modified with Coherent femtoclock. AudioQuest Carbon USB and AES cables. Coherent QP-1 supercap power supply running Asus and Mutec.

 

[b0bb]

Hello it's been quite long,
Bobb, can you please report the sonic differences you experienced in the last 2 months between the SU-1 vs SU-6 using i2s ?

I use a SU-1 myself. I olso experimented a little with some mod's.
1. Installing Sorbathan feet, and setting it on a concrete stone. ( sonic improvement )
2. Filling the SU-1 with quartzsand. ( big sonic improvement )
3. Remove the sand and installed ss7805 and SS1117-3.3 from sparkoslabs.com ( same sonic improvement as SU-1 with sand )
4. Fillid the SU-1 with quartzsand again . ( big sonic improvement over just the ss7805 and SS1117-3.3 )

I didn't try the RF sheets jet, that's the next experiment.

SU-6 is an incremental improvement over SU-1. My SU-6 is stock.

The main reason I use it is improved I2S compatability with the LKS-004, no more thumps, loud clicks, bangs and howling when starting/stopping tracks and changing sample rates

Sound quality wise it brings smoothness without sacrificing ability to resolve the soundstage, with the SU-1 there is a tradeoff between the two aspects.

This is most evident on complex tracks for example the opening of Dies Irae from Mozart's Requiem in D-Minor.

The SU-1 is not able to clearly separate out the orchestral and choral sections during the opening when orchestra and the choir go with the full might.

On the SU-6 orchestra and chorus is clearly rendered and within the choir, the different sections can be clearly identified (soprano, alto, tenor, bass)
SU-1 renders the the chorus as a single mass of sound, I could barely make out the latin lyrics


Tried this on 2 versions of the track with HQPlayer DSD512 output polysinc-xtr-mp resampling filter and ASDM7 512+ DS modulator

On the SU-1 the increased dynamic range of the DSD128 version was not at all evident, it was quite clear on the SU-6

Amadeus Soundtrack, Neville Marriner conducting the Academy of St Martin in the Fields PCM 44.1k


Choir of St Jacob Stockholm, Stefan Skold Conducting DSD128

 

[b0bb]

Hi, maybe some one has an idea using HQPlayer and NAA.
does’ anyone have good experience up sampling DSD64 (SACD) to DSD512.
I can’t find any combination that sounds better then bypassing DSD directly to the DAC ( es9038Pro ).
Sending DSD directly to the DAC has better Transients, and deeper soundstage.
Up sampling PCM to DSD has a big Sonic impact . Everything is improved with PCM.

My Chain:
PC ( Rayzen 7 3800x ) ---USB---> iFi Micro ---USB---> Singxer SU-6 ---HDMI---> DAC ( L.K.S Audio MH-DA004 Dual ES9038pro ) ----> Accuphase E-470

The stock 004 performance of the DSD256 and higher has a lot of room to improve, this is one of the reasons I swapped out the XO.

The filter you choose in HQP greatly affects the performance, if you like good transient handling use the -mp minimum phase filters and stick to the polysinc family.
On the longer polysinc filters, a big NVIDIA GPU helps, HQP is adaptive it scales the work to the available computing resources.
DSD to DSD upsampling takes about 4X the computational power compared to PCM->DSD and this is where GPUs have a role to play.

If straight DSD64 sounds better, you might have a bottleneck somewhere in your playback chain.

 

[xiamen]

The stock 004 performance of the DSD256 and higher has a lot of room to improve, this is one of the reasons I swapped out the XO.

The filter you choose in HQP greatly affects the performance, if you like good transient handling use the -mp minimum phase filters and stick to the polysinc family.
On the longer polysinc filters, a big NVIDIA GPU helps, HQP is adaptive it scales the work to the available computing resources.
DSD to DSD upsampling takes about 4X the computational power compared to PCM->DSD and this is where GPUs have a role to play.

If straight DSD64 sounds better, you might have a bottleneck somewhere in your playback chain.

If you are using the standard clock on the LKS, that may be the bottleneck. Accurate clocking is important for re-sampling at high frequencies. The other suspect will be the LKS power supply. Variation in voltage will cause jitters amplified with the re sampling. That's probably why many people here are doing mods on those tow components.

 

[xiamen]

JRiver (I have MC 23.) does not seem to permit native format DSD upsampling using the Amanero-WASAPI.

It does permit upsampling after conversion to DoP. For native format upsampling you need to use ASIO.

Jriver has a upgrade discount now on for the upcoming jriver 26. For me, one of the big enhancement is with the newer versions is a 64-bit version of jriver which optimize system resources assuming your OS is also 64-bit.

 

[MartinWT]

Accurate clocking is important for re-sampling at high frequencies. The other suspect will be the LKS power supply. Variation in voltage will cause jitters amplified with the re sampling. That's probably why many people here are doing mods on those tow components.

Exactly that. I have a Coherent femtoclock installed, Mad Scientist Donuts and RFI shielding around the two toroid transformers and run it from a regenerator for very stable power. The level of insight, even from 16/44 material, is extraordinary.

 

[Xoverman]

The stock 004 performance of the DSD256 and higher has a lot of room to improve, this is one of the reasons I swapped out the XO.

The filter you choose in HQP greatly affects the performance, if you like good transient handling use the -mp minimum phase filters and stick to the polysinc family.
On the longer polysinc filters,
a big NVIDIA GPU helps, HQP is adaptive it scales the work to the available computing resources.
DSD to DSD upsampling takes about 4X the computational power compared to PCM->DSD and this is where GPUs have a role to play.

If straight DSD64 sounds better, you might have a bottleneck somewhere in your playback chain.

Thanks for the explanation Bobb. Actually my 004 is quite intensive modded as your may remember.
Oscillator, capacitors, copper wall, and grounding upgrades.

Danger!!! This Mod is simple but also very dangerous if you don't know that you are doing!!!
My last mod is pretty easy to do but has a bigger impact then it thought it would have. I disconnected ground from the mains plug,
and screwed it to the chassis ground of the amplifier. ( very very big sonic impact! Deeper and better controlled bass, better harmonics )

My Stereo:

My DIY X-over

 

[Xoverman]

@BobB and the rest

Let me be more pressies about my system :

PC ( Rayzen 7 3800x + RTX 2080 HQPLAYER ) --- NAA ---> PC ( Rayzen 5 2400ge ) ---USB---> iFi Micro ---USB---> Singxer SU-1 ---HDMI---> DAC ( L.K.S Audio MH-DA004 Dual ES9038pro ) ---RCA---> Accuphase E-470

Bobb can you please share your exact dsd upsampling settings with us. Maybe that will help. Thanks in advance

 

[b0bb]

Thanks for the explanation Bobb. Actually my 004 is quite intensive modded as your may remember.
Oscillator, capacitors, copper wall, and grounding upgrades.

The T-Shaped absorber over the XO will cause a few problems. (Is that the Fairrite?)
1) Part of the input stage JFET is covered, it can disrupt the electric field around the JFET affecting its operation
2) I/V is no longer fully balanced as only one half of the I/V is affected.
3) One channel has the in-phase (+) affected and the other channel has the inverted-phase (-) affected
4) Similarly the CCHD-575 and CCHD-950 performance is altered if an absorber sheet is placed over them

The symptom is degraded high sampling rate performance especially DSD.

The absorber over the XO appears to smooth the sound quality but in reality it takes a lot away from the top end and smears the transients.
To take the edge off the top end audio, add absorbers (both 3M + Fairrite) over the HDMI connector and transciever chip

 

[b0bb]

@BobB and the rest
Bobb can you please share your exact dsd upsampling settings with us. Maybe that will help. Thanks in advance

My HQP settings: poly-sinc-xtr-mp, ASDM7 512+ upsample PCM and DSD to DSD512

You appear to have ground borne noise problems, try isolating the PC with HQP from the rest of the playback chain, I use fiberoptic ethernet.

I have the iFi micro as well and found it does more harm than good when connected to the SU-1 playing high rate DSD.
I got better performance by absorbing the noise inside the SU-1 and removing the iFi from the chain.

 

[Stefday2]

hi all,
I would like know arround tihis budget what are the alternative of the MH-DA004 ? or if it's still the best choice
i curently have the DA003 (stock) since 2014 and i'm looking for a serious updgrade, i recently tested AKM based dac without succes (Bryston BD3 and RME ADI dac 2)
regards !

(i'm french and if you would like have more information about my current systeme it's here :
https://www.homecinema-fr.com/forum/viewtopic.php?p=180081038#p180081038
)