HQ Player

[Miska]

23 minutes ago, Kalpesh said:

NOTHING works but dff... I think I had not played a dsf file for a while ; could you grant me access to a 4.19 version or any before a component update that could explain ?

 

There have been no changes in the DSF code since June this year, when I added one more workaround to deal with corrupted DSF file headers:

"Fix DSF reading problem where header claims to have metadata while there's actually none"

Before that the previous change is from June 2019:

"Create more header fixups for corrupted DSF files"

 

These have been added when people encounter malformed DSF files. But without these changes, HQPlayer was just refusing to play those files altogether.

 

So I would say this is something else, maybe some setting has changes since? Can you send me one of such files, as small as possible, for testing.

 

[Rovo]

1 hour ago, Miska said:

 

What kind of NAA hardware is this? You could try putting Intona USB isolator between the NAA and DAC. If it's a regular PC motherboard, it will have a bunch of switching regulators / DC-DC converters anyway. So you kind of have LPS feeding SMPS.

 

If you use the low noise Intona USB3 version, do not use any aux power, it is not needed and may insert unexpected noise path. 7054 should be already fine.

 

P.S. Also good to check that you are not using 100 Mbps side of the EtherRegen.

 

 

The NAA endpoint is equipped with a JCAT USB XE card with an external power supply Plixir BDC. The T+A DAC 8 DSD does not need a 5V power line from the endpoint. A jumper is removed from the JCAT USB XE card to cut off this 5V to the DAC. This resulted in a step up in sound quality . The motherboard is a Supermicro X11SCL-F with the graphics turned off (jumper), also a step up in sound quality. 

 

I only use the A-side of the EtherRegen. 

[Miska]

33 minutes ago, Rovo said:

The NAA endpoint is equipped with a JCAT USB XE card with an external power supply Plixir BDC.

 

But that doesn't provide galvanic isolation... So it is not a substitute for the Intona. With the Intona isolator you wouldn't need the JCAT card either. I would say you get better objective results with the Intona.

 

33 minutes ago, Rovo said:

The T+A DAC 8 DSD does not need a 5V power line from the endpoint. A jumper is removed from the JCAT USB XE card to cut off this 5V to the DAC.

 

Gigabyte motherboards with DAC-UP USB ports also provide this option from BIOS.

 

33 minutes ago, Rovo said:

The motherboard is a Supermicro X11SCL-F with the graphics turned off (jumper), also a step up in sound quality. 

 

I use UP Gateway. For some months I used Ferrum Hypsos PSU for it, feeding the Holo Spring 3 DAC. But since it didn't make noticeable difference in this respect I moved Hypsos to power Ferrum Oor. Spring 2 is getting fed by my HP Z4 G4 Xeon W2245 workstation (with RTX2080 Founders Edition GPU).

 

For some time, I used OnLogic CL100 as a NAA for the DAC8 DSD. But nowadays the CL100 is a NAA for SMSL D300 (with the rather rare Rohm DAC chip). And Fitlet2 is a NAA for TEAC NT503. If you use optical networking, Fitlet2/Fitlet3 is good choice because it can take SFP module and is very low power device otherwise.

 

T+A DAC 200 is getting fed by the i5-7600T machine through Intona 7054. Same machine is also feeding Accuphase DAC-60 through same Intona model. T+A HA 200 is getting fed by my Ryzen 7 5800X + Radeon RX 6700XT workstation through the USB3 Low Noise Intona.

 

Another USB3 Low Noise Intona is left for my measurement rig that I use to check with/without with various DACs. For measurements, primary source is the Xeon workstation.

 

[dericchan1]

@Miska so I have bitten the bullet and upgraded from a NEO to the IFI IDSD PRO (non signature version), should my HQplayer settings change accordingly or DSD256, 7ECV2 is likely still the "sweet spot"?

 

Cheers

 

Deric

[Kalpesh]

6 hours ago, Miska said:

 

There have been no changes in the DSF code since June this year, when I added one more workaround to deal with corrupted DSF file headers:

"Fix DSF reading problem where header claims to have metadata while there's actually none"

Before that the previous change is from June 2019:

"Create more header fixups for corrupted DSF files"

 

These have been added when people encounter malformed DSF files. But without these changes, HQPlayer was just refusing to play those files altogether.

 

So I would say this is something else, maybe some setting has changes since? Can you send me one of such files, as small as possible, for testing.

 

I Wetransferred you such a file. THe MIni M1 is 99% dedicated to HQP and there has been no change but Monterey updates (no Ventura) except 425 NAA image for the Up NUC. Will try 430 later (image from November 7 ?!)

[Miska]

1 hour ago, dericchan1 said:

@Miska so I have bitten the bullet and upgraded from a NEO to the IFI IDSD PRO (non signature version), should my HQplayer settings change accordingly or DSD256, 7ECV2 is likely still the "sweet spot"?

 

I would say likely DSD256 is still the sweet spot, although the difference is not huge for NEO. DSD512 trades some audio band performance for lower out-of-band noise. One could consider this system-dependent.

 

[Kalpesh]

NAA 430 IMAGE does not solve the issue.

I don't expect much from your analysis of the file I wetransferred you since I played dsf and dff files for ages via HQP without issue

 

Please grant me access or transfer me a pre june version so that I can test

 

It's a huge issue for me

[Miska]

11 minutes ago, Kalpesh said:

NAA 430 IMAGE does not solve the issue.

I don't expect much from your analysis of the file I wetransferred you since I played dsf and dff files for ages via HQP without issue

 

Let's continue this in email. Please just send me email.

 

The file is playing fine for me on my Mac Mini M1 with latest HQPlayer.

 

[shoubhikbhatti]

@Miska I am using the UPBoard as NAA. I am using the latest NAA image (naa-430-x64ramfs.7z) . When I turn on the NAA machine, I get this very loud static noise. On connecting with my HQPlayerDesktop, the white noise amplifies a lot in PCM mode. If I use DSD mode, the noise disappears but it appears between two songs. I am using Roon, but I have also seen the same behavior with HQPlayer Desktop in standalone mode. The DAC is Delta Sigma Dac. The white noise is extremely overbearing. I am using MacOS (I have the same behavior on both Intel and M1).  Here is my HQPlayer desktop setting.

 

[Kalpesh]

2 hours ago, Miska said:

 

Let's continue this in email. Please just send me email.

 

The file is playing fine for me on my Mac Mini M1 with latest HQPlayer.

 

We resolved the issue by email actually but I want to publicly laud @Miska's amazing support

[Miska]

2 hours ago, shoubhikbhatti said:

@Miska I am using the UPBoard as NAA. I am using the latest NAA image (naa-430-x64ramfs.7z) . When I turn on the NAA machine, I get this very loud static noise. On connecting with my HQPlayerDesktop, the white noise amplifies a lot in PCM mode. If I use DSD mode, the noise disappears but it appears between two songs. I am using Roon, but I have also seen the same behavior with HQPlayer Desktop in standalone mode. The DAC is Delta Sigma Dac. The white noise is extremely overbearing. I am using MacOS (I have the same behavior on both Intel and M1).  Here is my HQPlayer desktop setting.

 

The Amanero Combo384 does that when it doesn't keep up with the data and has lost clock sync. It also depends a lot on the particular firmware version flashed on the Amanero. Is this a DSD-only DAC or does it also support PCM?

 

Does it change if you change the Buffer time setting to "Default", 10 ms or 100 ms?

 

Reason for the reappearance between DSD outputs is that the interface may momentarily switch to PCM mode between track changes. If the track format doesn't change and there's natural change from track to another, this shouldn't happen. But if you use next/previous or jump between different tracks, it may happen. If it incorrectly switches to PCM mode while getting DSD data, you certainly get a lot of loud white noise.

 

[shoubhikbhatti]

1 hour ago, Miska said:

Is this a DSD-only DAC or does it also support PCM?

It does support PCM. Because if I hook up the dac directly to the laptop and use Roon without HQPlayer then it works perfectly.

 

1 hour ago, Miska said:

Does it change if you change the Buffer time setting to "Default", 10 ms or 100 ms?

I tried all the setting here, but nothing seems to be working.

 

1 hour ago, Miska said:

The Amanero Combo384 does that when it doesn't keep up with the data and has lost clock sync.

Does this happen even if nothing is playing? Because as soon as I turn on the NAA device, i am getting the static noise.

 

 

Is there a pass through mode in HQPlayer, that just passes the input signal to the output without fixing the output frequency and bits?

[Miska]

Just now, shoubhikbhatti said:

It does support PCM. Because if I hook up the dac directly to the laptop and use Roon without HQPlayer then it works perfectly.

 

With the same output format you are using with HQPlayer and the same NAA hardware, under Linux?

 

Just now, shoubhikbhatti said:

Does this happen even if nothing is playing? Because as soon as I turn on the NAA device, i am getting the static noise.

 

Does it work with HQPlayer correctly when connected directly to your HQPlayer server, without NAA?

 

 

[shoubhikbhatti]

5 minutes ago, Miska said:

Does it work with HQPlayer correctly when connected directly to your HQPlayer server, without NAA?

Just tried, even with HQPlayer directly hooked up at 384K on Mac i am getting the white noise. I tried the same thing on Holo May as well, there also the same thing.

[shoubhikbhatti]

7 minutes ago, Miska said:

With the same output format you are using with HQPlayer and the same NAA hardware, under Linux?

When directly playing through Roon, it is the native PCM format from Qobuz. All my music is coming from Qobuz.

[AudioDoctor]

On 12/12/2022 at 2:40 PM, Miska said:

 

If it works for you, it is great! Just want to warn that it may have unexpected side effects, or may not work for someone else. I have many times spent a lot of time trying to solve problems on similar setups - when something doesn't work as expected. For example unreliable streaming to NAA (occasional drop-outs) due to missing multimedia QoS service.

 

 

Yeah, my i5-7600T does DSD256 with ASDM7ECv2 with under 35W and without GPU. With HQPlayer Embedded on Linux. But lowest power consumption to do DSD256 with these settings goes to Apple's M1 CPU.

 

And i5-11600 doing DSD512 with ASDM7ECv2 under 65W without GPU. HQPlayer Embedded on Linux too.

 

I have also i9-10900K machine, with RTX2080Ti GPU, doing 4 or 8 channels DSD256 to Merging Hapi. It runs totally stock Windows 11. Inputs are through RME HDSPe AIO card.

 

 

Miska, I use those same settings in my totally passively cooled monsterlabo case with a 5900x and a 3080 gpu and it LOAFS along on ubuntu. It does use more power though and is totally silent. Unlike some I don't want to run hardware to the ragged edge, I prefer to keep it running for a long time to maximize my investment.

[AudioDoctor]

On 12/12/2022 at 2:40 PM, Miska said:

 

If it works for you, it is great! Just want to warn that it may have unexpected side effects, or may not work for someone else. I have many times spent a lot of time trying to solve problems on similar setups - when something doesn't work as expected. For example unreliable streaming to NAA (occasional drop-outs) due to missing multimedia QoS service.

 

 

Yeah, my i5-7600T does DSD256 with ASDM7ECv2 with under 35W and without GPU. With HQPlayer Embedded on Linux. But lowest power consumption to do DSD256 with these settings goes to Apple's M1 CPU.

 

And i5-11600 doing DSD512 with ASDM7ECv2 under 65W without GPU. HQPlayer Embedded on Linux too.

 

I have also i9-10900K machine, with RTX2080Ti GPU, doing 4 or 8 channels DSD256 to Merging Hapi. It runs totally stock Windows 11. Inputs are through RME HDSPe AIO card.

 

 

Miska, I use those same settings in my totally passively cooled monsterlabo case with a 5900x and a 3080 gpu and it LOAFS along on ubuntu. It does use more power though and is totally silent. Unlike some I don't want to run hardware to the ragged edge, I prefer to keep it running for a long time to maximize my investment.

[Miska]

10 hours ago, shoubhikbhatti said:

Just tried, even with HQPlayer directly hooked up at 384K on Mac i am getting the white noise. I tried the same thing on Holo May as well, there also the same thing.

 

Can you post your exact settings with Holo May connected directly to your Mac?

 

Also make sure that you don't have Roon running at the same time.

 

[Miska]

10 hours ago, shoubhikbhatti said:

When directly playing through Roon, it is the native PCM format from Qobuz. All my music is coming from Qobuz.

 

Qobuz also with standalone HQPlayer, without Roon running? With DAC directly connected to your Mac?

 

[Miska]

10 hours ago, AudioDoctor said:

Miska, I use those same settings in my totally passively cooled monsterlabo case with a 5900x and a 3080 gpu and it LOAFS along on ubuntu. It does use more power though and is totally silent. Unlike some I don't want to run hardware to the ragged edge, I prefer to keep it running for a long time to maximize my investment.

 

The part you quoted was about running HQPlayer on modified Windows Server-edition, which is not officially supported environment for HQPlayer. Ubuntu is a different thing. Not about passive cooling... I understand passive cooling from the point that it is totally silent acoustically, if/when you keep the computer in the listening space. I don't agree though that it would make a difference in sound quality.

 

I have only one passive cooled HQPlayer server, rest are fan cooled. Mac Mini M1 is also fan cooled, but very quiet one. I don't have liquid cooled servers yet, but when I have money for 7000X-series Ryzen, it will be liquid cooled, in the new Fractal Design North case. My next hardware build will be low power 13th Gen Intel in the new small Fractal Design Ridge case since it is cheaper. (Electricity prices have increased here 5x due to the war in Ukraine, and it is winter now 🥶)

 

These will be just for testing and optimizing HQPlayer further for these new chip generations.

 

I run the hardware as much to the edge as necessary, and for example when building HQPlayer OS images, my workstation is all cores 100% load for several hours in row. And to make a release, it often takes many builds and some testing between. Same workstation is also used for HQPlayer development (Desktop and Embedded) which is also constant 100% build load and then running HQPlayer with what ever heavy settings. I consider such a normal workstation use the machine is designed to take. I've been using HP workstations since the beginning. These come with 3 year next business day on-site warranty, which I've been extending to 5 years. So far the five years have always run out before there's a single issue. In fact, the machines have become obsolete before a single issue has appeared. I have many workstations and servers in the storage that are in perfect operating condition, but just became obsolete.

 

If you think about data centers like the ones run by Google, they likely want their servers run close to 100% utilization to maximize ROI. And they also likely want minimum number of faults, since it costs to fix and replace hardware, and with thousands of servers you will notice quickly if something is prone to break. 

 

 

From hardware longevity point of view, the only part that is actually sensitive to temperature are electrolytic capacitors. So from this perspective what matters are the PSU internal temperatures and temperature inside the case. This is easy to manage with fan cooling. Like usually my machines have two or three fans blowing air in from the front and two fans blowing air out from the back. In addition the PSU has it's own fan controlled by the PSU temperatures.

 

Usually all broken machines I've got have got blown electrolytic capacitor(s). These have been usually internet-facing servers in the machine room where at summer time ambient temperatures could get to 40C (not anymore, fixed by heat air pump). These servers have never been running under high CPU load. So this is why I think the inside case temperature readings are more important than CPU core temperatures.

 

This is also reason why I power down all warm running audio equipment when not in use, such amps, DACs etc. Especially hotter running ones like Holo Audio etc.

 

Well, that was a lengthy write up. 😅

 

[The Computer Audiophile]

13 minutes ago, Miska said:

I have only one passive cooled HQPlayer server, rest are fan cooled. Mac Mini M1 is also fan cooled, but very quiet one. I don't have liquid cooled servers yet, but when I have money for 7000X-series Ryzen, it will be liquid cooled, in the new Fractal Design North case. My next hardware build will be low power 13th Gen Intel in the new small Fractal Design Ridge case since it is cheaper.

 

That North case is really nice! I hadn't seen it yet. Thanks for the link. 

 

 

13 minutes ago, Miska said:

(Electricity prices have increased here 5x due to the war in Ukraine, and it is winter now 🥶)

 

This made me think of the ultimate in Winter passive cooling. Keep the server outside :~)

 

 

Cheers from a snowy Minneapolis. Outside my house this morning:

 

[Jud]

1 hour ago, Miska said:

This is also reason why I power down all warm running audio equipment when not in use, such amps, DACs etc. Especially hotter running ones like Holo Audio etc.

 

Well that opened up a can of worms! (Don't know if something like this expression is common outside the US.)

 

Pretty much any DAC? Especially those with more large discrete parts? Any notion what the tradeoff might be between heat cycling (warming up when turning on then cooling down when turned off) and constant warm temperature (always on)?

[Reg19]

@Miska I have a question on the HQPlayer volume control quality.

 

Power Conditioner: Audioquest Niagara 5000 (using all AQ Hurricane power cables)

NAA: SoTM SMS-200 Trifecta

DAC: Holo May

Preamp: VAC Renaissance Mk V

Power Amp: Hegel H30

Speakers: Focal Maestro Utopia III

 

Generally, I keep the volume level on the preamp relatively high and control the music volume level using the HQPlayer volume (through Roon). Currently, I have adjusted the preamp volume level so that at ideal listening level, the HQPlayer volume is at -35dB.

 

The question is: ideally, at what volume level should I keep HQPlayer at (I can adjust my preamp volume so that my normal listening level is at that HQP volume level)?

 

I have a Intel NUC (running Roon) connected via Ethernet to network. I have a M1 Mac Studio running HQPlayer connected via Ethernet to network. I generally upsample to either 768KHz using sinc L / LNS15 or (if source is DSD) DSD256 using ASDM7ECv2 / poly-sinc-ext2.

 

Thanks.

[AudioDoctor]

5 hours ago, The Computer Audiophile said:

This made me think of the ultimate in Winter passive cooling. Keep the server outside :~)

 

A Pass Labs amp is good for keeping the toes warm as well...  ;-)

[AudioDoctor]

5 hours ago, Miska said:

The part you quoted was about running HQPlayer on modified Windows Server-edition, which is not officially supported environment for HQPlayer. Ubuntu is a different thing. Not about passive cooling... I understand passive cooling from the point that it is totally silent acoustically, if/when you keep the computer in the listening space. I don't agree though that it would make a difference in sound quality.

 

I was only mentioning it for the sake of highlighting how little stress I am putting on my components in the hope that they will last a long time being passively cooled. All PCM gets upsampled to DSD256 with ASDM7ECv2 N1 rates are poly sinc gauss long and Nx rates are poly sinc gauss hires lp, and files that are already DSD just get played as is.

 

I know some here think it makes a difference for sound quality to be passively or non passively cooled. My only concern is that this is in the listening area. I have used a fully water cooled PC before and heard no discernible sound quality differences.