microRendu & HQPlayer

[ted_b]

I'm not sure (you'll have to ask him yourself), but I don't think Miska's prime motive in creating the NAA was to improve SQ.

 

Mani.

 

I somewhat disagree. The NAA is a FIFO buffer machine (using networkaudiod.exe) and allows the HQP heavy lifting machine (possibly noisy due to its need to be heavy lifting when upsampling to things like DSD512 and DSD256 multichannel, using CUDA offloading, etc) to be isolated and dedicated. I think isolation and SQ is his primary goal; multi-zone is a nice byproduct (being that it isn't really multi, as no two zones can run simultaneously). Anyway....tomAto vs tomato.

[Miska]

I'm not sure (you'll have to ask him yourself), but I don't think Miska's prime motive in creating the NAA was to improve SQ. However, I think this was exactly John, Andrew, and Jesus's prime motive in creating the microRendu.

 

NAA's primary purpose is EMI/RFI isolation from the main computer that does heavy lifting. And secondary purpose is to make it possible to keep HQPlayer machine outside of listening room so that acoustically noisy fans and HDDs don't matter.

[manisandher]

I’ve completed my first set of listening tests. I’ve included links to some 24/192 files that I took from the output of the iFi Nano DSD, using a Tascam DA-3000. Listening back to these files, they give a fair reflection of the differences between the three configurations in my listening room. My comments are primarily focused on how closely these 24/192 files resemble the original 16/44.1 files. This should give a clue as to which configuration is the least lossy. (Of course, there will be some loss going through the ADC, but this will be identical for all three cases.)

 

1. HQPlayer -> microRendu (NAA) -> iFi Nano DSD

Here, HQPlayer was playing from my music server (i3-6100T CPU). No DSP was applied in HQPlayer and attenuation was set to 0dB, so the iFi should have received a bit-perfect 16/44.1 signal. The microRendu was connected to an Industrial Intona USB isolator (with the supplied USB adaptor) which in turn was connected to the Nano using true 90Ω USB cable. The microRendu was powered by an iFi 9V supply.

 

https://drive.google.com/open?id=0B0PU5LO5jVjfc3hVODlaQkR4OTA

https://drive.google.com/open?id=0B0PU5LO5jVjfSmpySXF6cjh1ZTQ

https://drive.google.com/open?id=0B0PU5LO5jVjfNHhDQVJPYkR0WlE

https://drive.google.com/open?id=0B0PU5LO5jVjfbWxKUDdRMnlpdTQ

 

Hmm… There’s clearly a loss of detail. The sound has become flattened and greyer. The sweetness from the original files isn’t there. Some ambience is missing.

 

2. HQPlayer -> Audio PC -> iFi Nano DSD

 

Here, HQPlayer was playing from my audio PC (specs in earlier post). No DSP was applied in HQPlayer and attenuation was set to 0dB, so the iFi should have received a bit-perfect 16/44.1 signal. The PC was in ‘Minimize OS’ mode (one of XXHighEnd’s features), which kills most of the OS processes. The audio PC was connected to an Industrial Intona USB isolator using true 90Ω USB cable which again was connected to the Nano using true 90Ω USB cable.

 

https://drive.google.com/open?id=0B0PU5LO5jVjfejZhS3VCdmZjdXM

https://drive.google.com/open?id=0B0PU5LO5jVjfVWNRT0ljektSNWc

https://drive.google.com/open?id=0B0PU5LO5jVjfcGE0V0hheEtReVE

https://drive.google.com/open?id=0B0PU5LO5jVjfbHROTTh4U29TX2M

 

OK, now closer to the original files. But there’s still some losses audible. Overall, the sound is slightly ‘deader’, flatter, greyer… but not to the same extent as before.

 

3. XXHighEnd -> Audio PC -> iFi Nano DSD

 

I wanted to include this because XXHighEnd has a trick up its sleeve that the other two configurations can’t do. It has a mode of playback called ‘Unattended mode’. What happens here is that all the music files are loaded into RAM before playback. So Ethernet activity during playback is virtually zero (it could be set to absolutely zero, but I need to connect to the audio PC via RDC). The rest of the configuration was as 2 above.

 

https://drive.google.com/open?id=0B0PU5LO5jVjfSWduXzdEalphSjQ

https://drive.google.com/open?id=0B0PU5LO5jVjfblNCbmxiRGFVWEE

https://drive.google.com/open?id=0B0PU5LO5jVjfV0ZjZEllaUY4bDA

https://drive.google.com/open?id=0B0PU5LO5jVjfdl8xNGNZODZDWWc

 

Another clear step up in SQ. Actually, larger than the step between 1 and 2. The ambience of the original files is there. There’s a clarity and sweetness that the other two lack. In a nutshell, simply the most musical sounding of the three configurations.

 

So these are my initial thoughts comparing a $700 piece of HW/SW to a $4k piece of HW/SW. For me, my audio PC is clearly better than the microRendu. However, we need to bear in mind that the microRendu was being powered by a $50 PSU – in the audio PC, the LPSU alone costs way more than the microRendu! I am definitely going to order an Uptone LPS-1 PSU as soon as they become available and see how that affects the microRendu’s performance. All the indications are that it should be a big step up. But it’s going to have to be quite a big step up to match, let alone exceed, my audio PC.

 

But I have to say, for a total cost of around $1k, this microRendu/iFi combo is pretty amazing!

 

More to come…

[bibo01]

Can you please post also the 4 original tracks?

I can measure the differences between different setups.

[Superdad]

Hi Mani:

 

I have an Intona and find that it detracts from SQ when placed after my MicroRendu. (Though it is effective at the output of a computer USB jack, then feeding a REGEN.)

You might want to try your test without it.

 

Best,

-Alex C.

[manisandher]

Happy to do this, but don't want to get into trouble for copyright violation. So what I'll do for now is just share one original track. But I need to do something else first...

 

I had the iFi's filter set to 'minimum phase'. If you're hoping to measure differences between different setups, your measurements will no doubt focus on the frequency domain. The 'minimum phase' filter purposely gives up accuracy in the frequency domain for accuracy in the temporal domain. From a SQ perspective, this adds a pleasant 'zing' and liveliness to the sound that's not really there in the original. For your measurements, I'll need to redo this one track with each of the three configurations, with the filter set to 'normal'. Let me do this, and I'll post the original track too.

 

Mani.

[firedog]

OK, Barrows still calls it the µRendu as seen below and Everything Matters according to him. Further, I thought he did that as most people do not know what µ means so saying micro is just easier.

 

I don't actually care what anyone calls it. I just was trying to point out that it isn't necessary to tell people to use the "micro" symbol, as that isn't the official name anymore. If people want to type a "u" for convenience, I'm fine with it.

 

BTW, I'm not a lawyer, but do have a bit of experience with names of products. Sonore and it's people should be consistent about what they call it. If not, their ability to have exclusive use of the name is more difficult to defend.

[bibo01]

Happy to do this, but don't want to get into trouble for copyright violation. So what I'll do for now is just share one original track. But I need to do something else first...

 

I had the iFi's filter set to 'minimum phase'. If you're hoping to measure differences between different setups, your measurements will no doubt focus on the frequency domain. The 'minimum phase' filter purposely gives up accuracy in the frequency domain for accuracy in the temporal domain. From a SQ perspective, this adds a pleasant 'zing' and liveliness to the sound that's not really there in the original. For your measurements, I'll need to redo this one track with each of the three configurations, with the filter set to 'normal'. Let me do this, and I'll post the original track too.

 

Mani.

 

It's frequency and time shifts.

No need to do measurements again with "normal" filter setting.

One original track is fine. Eventually, please tell me to which track corresponds in each of the three tests.

Thanks.

 

On the other hand, if you like to do me a favor, please do three captures in succession with exactly the same setting, so I can access how constant is your ADC gear. Sometimes even professional gear can have fluctuation of various dBs.

 

Please send me all info in PM, so we don't pollute this thread.

[manisandher]

I have an Intona and find that it detracts from SQ when placed after my MicroRendu. (Though it is effective at the output of a computer USB jack, then feeding a REGEN.)

You might want to try your test without it.

 

Hi Alex. That was actually my original intention - to go straight from the microRendu to the iFi. When I was describing my audio PC in an earlier post, I mentioned that I use a dedicated AC mains circuit for the hifi system, totally isolated from the house AC mains circuit (used for all the computers, etc). Here are the two configurations:

 

Hifi System AC mains circuit:

house meter -> ultra-low-impedance cabling -> dedicated CU -> 15kVA isolation transformer (neutral at secondary bonded to dedicated 'chemical' earth, measured at ~5Ω) -> PS Audio P10 regenerator

 

Computer(s) AC mains circuit:

house meter -> house earth -> house CU -> PurePower 2000 regenerator

 

The reason why I share this info is because it's really important to understand that in my case, I have to have full galvanic isolation between the computer (audio PC, microRendu, whatever) and whichever DAC is connected to the hifi system. When I was making the 24/192 recordings, I had the Tascam powered from the hifi system circuit, and the microRendu from the house circuit (as I do with music server, audio PC, etc).

 

Even though the microRendu has a galvanically isolated ethernet input, this isn't enough in my case - there will be a ground loop created due to the potential difference between the two circuits. Now, once I get hold of one of your LPS-1s, this simply won't be an issue any more! Hint, hint

 

But thanks for your thoughts. This is exactly what I'm looking for now - ways to get the performance of the microRendu to at least match (perhaps even exceed) my tricked out audio PC.

 

Mani.

[manisandher]

Eventually, please tell me to which track corresponds in each of the three tests.

 

Haha... So you're not intending to actually listen to any of this stuff ? I look forward to what your measurements bring up, but I think it's really interesting to actually hear what something like switching off ethernet activity does to the sound.

 

... please do three captures in succession with exactly the same setting, so I can access how constant is your ADC gear...

 

I'll try to give this a go. I think I should be able to switch between the three configurations quickly enough...

 

Please send me all info in PM, so we don't pollute this thread.

 

Will do.

 

Mani.

[bibo01]

...

I'll try to give this a go. I think I should be able to switch between the three configurations quickly enough...

 

...

Mani.

No, do not change any configuration! You choose any configuration you like (it's not important) and measure it three times - same song three times at same setting.

These three recordings are just to test fluctuation of your ADC gear.

[manisandher]

Got it. Will do...

 

Mani.

[manisandher]

Slightly more on topic, I've done some 24/192 captures of RoonReady vs. Roon+HQPlayer. I think the sonic differences should be pretty obvious.

 

Before I link the files, I want to share another experience first. I found an old 9V linear power supply sitting around, so decided to try this - I was happy to use it on the same AC circuit as the Tascam, so could try things without the Intona in place. But I very quickly put the Intona right back!:

 

 

[Edit: I should add that I'm not running the iFi from its own battery here - it is being powered by the microRendu, which probably explains the excessive noise without the Intona.]

 

So the two files below were both captured with the Intona in place. Sorry for the choice of 'music' - it only occurred to me later that maybe it wasn't the best choice - oh well.

 

1. RoonReady: https://drive.google.com/open?id=0B0PU5LO5jVjfY3VUVEhxU1BNSXc

 

2. Roon+HQPlayer: https://drive.google.com/open?id=0B0PU5LO5jVjfUmhqVjZkSTI3S2s

 

To my ears, Roon+HQPlayer sounds better. There's more detail. More ambience. And overall, it's just more tuneful.

 

Love to hear your thoughts...

 

Mani.

[manisandher]

Just to put a spanner in the works (who me?), I feel I have to disclose something that actually occurred to me a few days ago. I've been reluctant to do this because it just doesn't make any sense to me. But it's repeatable and real, and audibly captured in my 24/192s, I think.

 

HQPlayer -> microRendu NAA

 

... sounds different to...

 

Roon+HQPlayer -> microRendu NAA (linked to earlier)

 

Here's the 'straight' HQPlayer file:https://drive.google.com/open?id=0B0PU5LO5jVjfZUU1ZWc3V29rMEk

 

Compared to Roon+HQPlayer, it's a fuller sound. Richer. Less 'sparkle'. Seems to have less attack and more blunted transients. But perhaps more natural sounding?

 

Now this simply shouldn't happen. These files should sound identical. But to my ears, they don't. And as I said, this is repeatable and real - I've felt this for a few days now whenever I've compared HQPlayer to Roon+HQPlayer, i.e. it's not a strange quirk of the ADC.

 

If someone could explain to me WTF is going on, I'd really appreciate it.

 

Mani.

[kennyb123]

Compared to Roon+HQPlayer, it's a fuller sound. Richer. Less 'sparkle'. Seems to have less attack and more blunted transients. But perhaps more natural sounding?

 

Now this simply shouldn't happen. These files should sound identical. But to my ears, they don't. And as I said, this is repeatable and real - I've felt this for a few days now whenever I've compared HQPlayer to Roon+HQPlayer, i.e. it's not a strange quirk of the ADC.

 

Maybe the clue is on the Roon signal path screen. Roon claims they are outputting to HQP streaming. So maybe there is another layer.

 

If they were alternatively just passing the file path to HQP, and the HQP loads this as it would any file, they should sound the same. Hope Jussi weighs in.

 

 

[Distinctive]

I hope to live to see the day where HQP could run on a Cube-i (Sonic Orbiter) type device....

[Dev]

Just as another data point - I have posted in the uRendu thread a while back that the industrial Intona between the uRendu and DAC sounds much more pleasant in my system (which I wasn't expecting). If I take it out, the sound looses the bigger soundstage and dynamics a bit. I am using regular cat 6a with the uRendu on the ethernet side and Curious links on the usb end. The uRendu is powered by 9v battery supply with the charging disconnected when listening. My guess is that the effectiveness of Intona is also DAC dependent.

[kennyb123]

If I take it out, the sound looses the bigger soundstage and dynamics a bit. I am using regular cat 6a with the uRendu on the ethernet side and Curious links on the usb end. The uRendu is powered by 9v battery supply with the charging disconnected when listening. My guess is that the effectiveness of Intona is also DAC dependent.

 

I found that the Curious link was not a great match with the microRendu. A friend who tried the combo in his system agreed. He and I both prefer a much more expensive cable. But relative to this, the Curious rolled off the highs and reduced the soundstage. Maybe the Intona is counteracting this.

[Dev]

I found that the Curious link was not a great match with the microRendu. A friend who tried the combo in his system agreed. He and I both prefer a much more expensive cable. But relative to this, the Curious rolled off the highs and reduced the soundstage. Maybe the Intona is counteracting this.

 

I have tried several usb cables - Curious, Wireworld, Jcat, ppa and hard adapter. All with the same result. The Intona only made it better. So I don’t know if the effect of Intona is because of the type of USB cable used or not but I did find that with uRendu the differences in USB cables is very subtle at best in my setup. What cables are you using ?

[bibo01]

Just to put a spanner in the works (who me?), I feel I have to disclose something that actually occurred to me a few days ago. I've been reluctant to do this because it just doesn't make any sense to me. But it's repeatable and real, and audibly captured in my 24/192s, I think.

 

HQPlayer -> microRendu NAA

 

... sounds different to...

 

Roon+HQPlayer -> microRendu NAA (linked to earlier)

 

Here's the 'straight' HQPlayer file:https://drive.google.com/open?id=0B0PU5LO5jVjfZUU1ZWc3V29rMEk

 

Compared to Roon+HQPlayer, it's a fuller sound. Richer. Less 'sparkle'. Seems to have less attack and more blunted transients. But perhaps more natural sounding?

 

Now this simply shouldn't happen. These files should sound identical. But to my ears, they don't. And as I said, this is repeatable and real - I've felt this for a few days now whenever I've compared HQPlayer to Roon+HQPlayer, i.e. it's not a strange quirk of the ADC.

 

If someone could explain to me WTF is going on, I'd really appreciate it.

 

Mani.

Perhaps your server is injecting more noise on AC line when playing through Roon+HQP and it affects mRendu and/or iFi diffently than when playing through HQP straight.

As HQPlayer server are you using your i3 or i7?

[lmitche]

Perhaps your server is injecting more noise on AC line when playing through Roon+HQP and it affects mRendu and/or iFi diffently than when playing through HQP straight.

As HQPlayer server are you using your i3 or i7?

Where are the music files in this configuration? Where is Roon running? And Hqplayer?

[kennyb123]

I have tried several usb cables - Curious, Wireworld, Jcat, ppa and hard adapter. All with the same result. The Intona only made it better. So I don’t know if the effect of Intona is because of the type of USB cable used or not but I did find that with uRendu the differences in USB cables is very subtle at best in my setup. What cables are you using ?

 

Audience Au24 SE

 

I still have yet to try the hard adapter.

[Dev]

Audience Au24 SE

 

I still have yet to try the hard adapter.

 

 

What length did you compare the Curious and Audience cables ? I am using the Curious short regen link. Do you know if they make short usb cable ? Audience makes great stuff and I have their au24 power cables and the tssd power conditioner - both are just excellent. So I am not surprised :-)

[kennyb123]

What length did you compare the Curious and Audience cables ? I am using the Curious short regen link. Do you know if they make short usb cable ? Audience makes great stuff and I have their au24 power cables and the tssd power conditioner - both are just excellent. So I am not surprised :-)

 

I believe my Audience USB is 1 meter. I compared this to the short Curious REGEN link that you also have.

 

The shortest length The Cable Company lists for the Audience is 0.5m: https://www.thecableco.com/Product/AU-24-SE-USB-Standard

 

I purchased mine from them, but it was a show demo so I got a nice discount. It easily beat the well-regarded LightSpeed USB that I had previously.

[manisandher]

Maybe the clue is on the Roon signal path screen. Roon claims they are outputting to HQP streaming. So maybe there is another layer.

 

Miska?

 

Perhaps your server is injecting more noise on AC line when playing through Roon+HQP and it affects mRendu and/or iFi diffently than when playing through HQP straight.

 

Roon is always 'on' in the background, whether I use it as a player or not. This 'Roon+HQPlayer effect' happens irrespective of whether I plug the microRendu into the 'hifi' AC circuit or the house AC circuit. When it's plugged into the house circuit, it really is totally isolated from the rest of the hifi (and the Tascam) - it could inject as much noise as it wants into the house mains, it shouldn't affect things. And also, as a slight aside, I'm doing all this 24/192 capturing in the little 'studio' in my basement - it has a semi-Faraday cage built around it (my mobile phone signal goes to zero as soon as I enter). Here's an [old] pic:

 

 

Where are the music files in this configuration? Where is Roon running? And Hqplayer?

 

The music files sit on my Win10 music server (outside the 'studio'), which uses a low-power i3-6300T. Both Roon and HQPlayer run off this music server. As I said, Roon is always 'on'. (I'm really not prepared to move the location of Roon to try thngs out.)

 

Mani.