Pink Faun 2.16

[lpost]

Where do I tell HQPe about the URL?  The benefit of routing through Roon is the control.  I can easily change from local files, Tidal and to Spotify 'internet radio' all within Roon.  

 

If I feed direct to HQPe, I expect I will have to contend with 'Roon lost control...'

 

For others that may want to set this up, here's the path:

 

Windows (or other supported machine on network running) Spotify client > Stream What You Hear (SWYH) > http stream fed to Roon Internet Radio station > Roon feeds HQPe > DAC.

 

Spotify Connect from any other client can control selection and play, I just have to select the Roon Internet Radio station.

 

I've tried the uPNP route but haven't found a combination of server/renderer/control point that all want to work together.  Plus it would be another interface outside of Roon or HQPe that I would have to interact with...making it more cumbersome than my current setup via Roon Internet Radio.

 

Now onto setting up Soundiiz to automatically move my 'discoveries' in Spotify to Tidal playlists.

 

I also need a scheduled task to ensure Spotify is stopped/paused and not consuming bandwidth endlessly.  Minor as it's just 320k but still a waste to forgot and leave it streaming.

 

Thank you Jussi for your assistance and patience!

 

 

[flkin]

The Pink Faun 2.16x will be Munich Hi End Audio Show 2019. Partnering with Aries Cerat equipment. 

 

I've always wanted to hear the Kassandra II DAC with the PF. If any one goes to the show please comment!

 

https://www.monoandstereo.com/2019/04/aries-cerat-at-munich-high-end-audio.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed:+monoandstereo/HOym+(MONO+AND+STEREO+Ultra+High-End+Audio+Magazine)&m=1

 

 

[Pink Faun]

Kassandra II DAC and streamer 2.16x in DUAL configuration ready to rock Munich 

 

 

[Miska]

I this running HQPlayer or something else? Anyway, I'll come and listen!

 

[flkin]

The all black faceplate looks nice 👍

[Pink Faun]

10 hours ago, Miska said:

I this running HQPlayer or something else? Anyway, I'll come and listen!

 

The server runs ROON core, the renderer can do ROON bridge or HQPe. We have to decide when the full system is setup in the demo room. See you there!

[flkin]

My 4 Pink Faun extreme clocks have arrived! 😄👍

 

Actually they arrived a week ago but I was in the middle of some tests and wasn’t able to install them into my Pink Faun 2.16x streamer at that time. But I have since started with the first one and it’s been in my system for a week now.

 

I started with the 20Mhz USB clock first as it was the easiest one to replace on the USB board. Three plastic clips hold the clock mini-board to the main board which are replaceable. A quick squeeze of the flare tips and the replacement was completed. 

 

 

A reboot and all worked as expected again. The new clock is physically different from the old one. It’s thicker by half and has Pink Faun markings on it. From what I can see, the mini-board is otherwise the same as the old board. The picture below shows the older Connor Winfield OH200 clock. The specs show this older 20MHz clock to operate with a phase noise of around -110dbC/Hz at an offset of 10Hz. The new Pink Faun extreme clock operates at a quieter -132dbC/Hz - a whopping 22dbC less than the old clock! 

 

 

Whether the reduced clock phase noise makes a significant difference is the question now. In the past I had compared the Pink Faun USB (with the older CW mini-board) directly to my DAC vs. the same via a tx-USBultra with the Mutec 10MHz reference clock attached. I covered this in my earlier review of the streamer and I found no appreciable improvement in sound. 

 

Why would this be as the Mutec Ref clock has an amazingly low phase noise of -142dbC/Hz? It’s due to the way reference clocks are attached to HiFi system. The reference signal has to pass through a ppl converter to generate the required frequency prior to being connected to the USB board. Along the way the signal is processed from 10MHz to 20MHz passes through at least 2 sets of cables, noise from a few power supplies injected and a multitude of connectors. This would seem to degrade the clock contribution to the level roughly matching a directly connected USB clock with a phase noise of -110dbC/Hz.

 

And so now after burning in the clock for a week, is there a difference in sound quality? Isn’t the original 20MHz clock at  -110dbC/Hz enough? Does the new extreme clock matter?

 

Matter it does, and not subtly.

 

I was concerned that the listening test would be spoilt by expectation bias. The clock switching was a one way action (it isn’t but I didn’t want to fiddle too much) so I invited 2 other friends over for their listening opinions before and after. A clock really needs at least 24 hours to warm up and stabilise to sound it’s best so a direct swap from warm clock to a cold one isn’t really a direct comparison but was what it was.

 

Even cold, the extreme clock immediately brought a new level of naturalness to the music. It felt like there was less effort to enjoy the music. Initially it felt quieter and then we realised it was from the speed of dynamic sounds being controlled faster. Music felt more alive and exciting and with the increased speed, new details and emotions revealed themselves. My friends felt the mid bass tightened and become more bouncy as well.

 

There was an improvement in 3D spatial accuracy too. Especially obvious in complex music with many performers, the separation of the different musical threads from each performer was clearer and easier to follow. Live music is now more of a joy to listen to, the clarity of the stage, audience engagement and subtle echos recreate the live experience so well. I especially enjoyed the following No Woman No Cry (reprise) track no.20 on the Sound Of Blomberger Songfestival 2007-2011 album available in Tidal. With the new extreme clock, it’s something to behold, especially the drum solo at the end, amazing transient speeds.

 

 

People who have played with clocks before will recognised much of what I have said. The benefits of a better clocked sound is an all round better sound with few negatives. Again, mostly true in my case and the reduction of a further 22dbC/Hz does makes quite a difference. Where will it stop? How far lower does phase noise go before it makes no appreciable difference? I don’t know the answer for this but a directly connected -132dbC/Hz, 20MHz USB clock makes a clear difference to me. There are other streamers that say clocks don’t matter - like the new SGM Extreme server with 2 CPUs and external power supply. I’m finding it difficult to accept this claim as so far to date, every better I’ve clock used brings about a whole new level of refinement to the music.

 

I did find however that the sound character of my system changed slightly. Not for the worse, just different. It used to be a darker side of neutral but after a week, the new extreme clock sound comes across as brighter side of neutral. Perhaps it’s still settling in and requires more time to break in, or perhaps I have to tune the system again with cables. As it is now, it’s a new level for my Pink Faun 2.16x streamer just when I was thinking how it can’t possibly be improved. Bravo Jord, thanks for the upgrade!

 

Next up, the motherboard and CPU extreme clock upgrade. This is a little more tricky as it involves some soldering so I’m planning to do this next week and will post my impressions here after.

[lpost]

What DAC are you using?  It seems the DAC would need to use the clock from USB output board for the phase noise to matter at all.  The DAC I own does not use any clocking from anywhere but it's own right before the conversion to analog.

 

 

[flkin]

I’m using the Vinnie Rossi DAC2. It has a femto level clock on board but not OCXO. I have no idea why clocking upstream from the dac matters at all. This has been discussed at length by some with the conclusion that there must be something at play not understood yet. Still re-clocking works and the benefits are easily heard.

 

 

[auricgoldfinger]

10 hours ago, flkin said:

Initially it felt quieter and then we realised it was from the speed of dynamic sounds being controlled faster. Music felt more alive and exciting and with the increased speed, new details and emotions revealed themselves.

 

This has been my experience using a 19v Farad power supply with my AL NUC endpoint.

[austinpop]

Had a great chat with Jord from Pink Faun on Thursday, about server design, clocks, and power supplies. Their show system was pretty extreme (in a good way), with the wacky looking (but excellent sounding) Aries Cerat chain!

 

[flkin]

Rajiv, are those 6 power amps stacked up high in the pic? Talk about tri-amping with mono blocks! Wow

[austinpop]

I didn’t really dig into that. It does look like tri-amping...

[Pink Faun]

On ‎5‎/‎12‎/‎2019 at 1:50 AM, austinpop said:

Had a great chat with Jord from Pink Faun on Thursday, about server design, clocks, and power supplies. Their show system was pretty extreme (in a good way), with the wacky looking (but excellent sounding) Aries Cerat chain!

 

Thank you, it was a pleasure meeting you.

[flkin]

An update about the latest AudioLinux kernel  4.19.37 rt19 1 rt bfq and the 235 menu for Pink Faun 2.16.

 

The CPU Governor under Expert mode doesn't work properly or doesn't display it if it does work. Setting exclusive cores also is slightly buggy. I'm was able to isolate 2 cores sometimes!

 

However from the BIOS, under OC Tweaker, one can set CPU frequency and voltage change to manual and specify a value. The frequency fixed here still doesn't display as fixed under AL Menu Status as fixed. So there's something amiss at this stage. 

 

However during my experimentation I discovered a very nice setting sound wise and is now my preferred setting.

 

In addition to my usual settings for AL -  Extreme priority, Normal Boot mode, Ramroot, I am finding that with the addition of selecting to using only 1 CPU core and setting in Bios the CPU frequency to 2500MHz and voltage to 1.000v the sound becomes a lot more analogue.

 

Initially I though it was less dynamic but on longer listening it was simply more emotional and less hard. Details, focus as good and scale as large as before but smoother. I sat through 3 Beethoven complete symphonies at a single sitting and felt zero fatigue. Switched back and forth many times with the same effect.

 

I'm told that changing the frequency/voltage will break some of the AL scripts and isn't recommended - not until PF/AL makes some fixes to the current software version I suppose. But what I am hearing is very nice so I'm staying with it. Perhaps only for those willing to experiment...

 

 

[flkin]

Just realised that I didn't update the record here with my motherboard and CPU clock switches from normal to extreme as promised earlier. The new clocks are larger than the old ones and proprietary to Pink Faun apparently. Shown below. The board and the other components seem to be much the same as before.

 

 

Installation was at the bottom of the motherboard accessed through a bottom removable panel of the streamer. I requested for this option so the clocks could be attached as close to the motherboard and CPU as possible 

 

 

From my listening notes, I noted that immediately there was more air and space around the music. Deeper soundstage and clearer echos. Somehow the tempo improved and became more infectious - toe tapping. Interestingly the changes are not more of the same as when the USB clock was upgraded. It's more complimentary, with the new clocks working nicely together with the USB one.

 

The sound was however brighter again in the same way I experienced with the initial USB clock change. Bass tighten up to the extent of changing the speaker character. My Wilsons are usually bassy but with the additional 2 clocks has becoming noticeably tighter and bouncier.

 

After 2 weeks or so and with an update of the AudioLinux kernel / AL Menu (described a few posts above), the sound character settled down. Treble balance back down to normal and bass a touch looser again. But the original feeling of soundstage space, depth and lightness remains. How much this has to do with the OS changes and how much the clocks I cant say but the overall effect is a relaxed, enveloping detail sound. Very happy with where I'm at 😁 

[flkin]

On 6/19/2019 at 7:21 PM, flkin said:

An update about the latest AudioLinux kernel  4.19.37 rt19 1 rt bfq and the 235 menu for Pink Faun 2.16.

 

The CPU Governor under Expert mode doesn't work properly or doesn't display it if it does work. Setting exclusive cores also is slightly buggy. I'm was able to isolate 2 cores sometimes!

 

However from the BIOS, under OC Tweaker, one can set CPU frequency and voltage change to manual and specify a value. The frequency fixed here still doesn't display as fixed under AL Menu Status as fixed. So there's something amiss at this stage. 

 

However during my experimentation I discovered a very nice setting sound wise and is now my preferred setting.

 

In addition to my usual settings for AL -  Extreme priority, Normal Boot mode, Ramroot, I am finding that with the addition of selecting to using only 1 CPU core and setting in Bios the CPU frequency to 2500MHz and voltage to 1.000v the sound becomes a lot more analogue.

 

Initially I though it was less dynamic but on longer listening it was simply more emotional and less hard. Details, focus as good and scale as large as before but smoother. I sat through 3 Beethoven complete symphonies at a single sitting and felt zero fatigue. Switched back and forth many times with the same effect.

 

I'm told that changing the frequency/voltage will break some of the AL scripts and isn't recommended - not until PF/AL makes some fixes to the current software version I suppose. But what I am hearing is very nice so I'm staying with it. Perhaps only for those willing to experiment...

 

 

 

Update on my Pink Faun 2.16x settings for best sound after further experiments in the Bios.

 

Audiolinux settings:

- CPU isolating 1-core

- Normal Boot mode

- Extreme for priority 

- Ramroot

- Roon no upsampling, no HQPe 

- Set ethernet speed 100Mb/s

- Expert exclude modules:

echo 1000 > /proc/sys/vm/stat_interval

echo 0 > /sys/bus/workqueue/devices/writeback/numa

echo 8 > /sys/bus/workqueue/devices/writeback/cpumask *

echo -1 > /proc/sys/kernel/sched_rt_runtime_us

echo 0 >/sys/devices/system/machinecheck/machinecheck3/check_interval

echo 0 > /proc/sys/kernel/watchdog

echo 0 > /proc/sys/kernel/nmi_watchdog

echo 1500 > /proc/sys/vm/dirty_writeback_centisecs

 

Not sure if the above exclusions actually work (doesn't seem to be any harm so far) but this one does:

 

echo 8 > /sys/bus/workqueue/devices/writeback/cpumask

 

* For cpumask, the 8 depends on what which CPU cores are isolated. In this case it’s CPU3 (core 4). Calculate this figure from http://store.tomandmiu.com/cal-cam.php Non hyper-thread, enter the core number and get the affinity number. If the mask number is incorrect, sound becomes more detailed and sharper. Not nice at all.

 

Bios settings:

- CPU frequency 2800Mhz

- CPU voltage at 0.9volts

- RAM voltage at 1.1volts

 

Note : Before experimenting with the Bios, good idea to make a backup at the bottom of the "tweaked" tab. If the streamer hangs (mine did during trials of settings not described here) there is a reset button behind the 2.16x which resets the Bios to original settings. Then you can return to your last working Bios settings from the restore option.

 

Adding a reclocked switch helps my system further. The treble is smoother.

[FredM]

On 7/12/2019 at 4:11 PM, flkin said:

Adding a reclocked switch helps my system further. The treble is smoother.

 

Hi flkin, can you share which switch you use? I wonder if you also have considered an internal (clocked) network card? (powered by the 2.16 or with a external psu, I.e. JCAT NET).

[flkin]

Hi Fred, prior to the 2.16x one box solution I was using the SOtM spaghetti system of which I still own. Part of that system had a standard D-Link switch that was modified by SOtM to add a clock input and linear regulators internally. I then used a Cybershaft OP14 clock through a tx-USBultra to generate the clock signal for the switch. Then I passed the ethernet signal through this D-Link switch before entering my 2.16x. 

 

Its a a little more tricky as I have to power the D-Link and tx-USBultra and so used my existing 2 LPS1 supplies - themselves requiring supplies to operate. And with LPS1 devices you have to ground the negative to drain leakage currents. In total a lot of wires, but something I’m used to from my spaghetti days 😄!

 

Yes I agree that an internal clocked card would be a neater and better solution. I ordered a PF LAN card with OCXO clock a long while ago but, due to intel firmware issues, the card isn’t ready yet. When it does  eventually arrive it will be interesting to see if an external clocked switch still makes a difference.

[Jambo]

I have ordered my pf 2.16x in this colour. Contact Jord for other colour options and prices. A "pink" Pink Faun. Very nice.

[FredM]

5 hours ago, flkin said:

Hi Fred, prior to the 2.16x one box solution I was using the SOtM spaghetti system of which I still own. Part of that system had a standard D-Link switch that was modified by SOtM to add a clock input and linear regulators internally. I then used a Cybershaft OP14 clock through a tx-USBultra to generate the clock signal for the switch. Then I passed the ethernet signal through this D-Link switch before entering my 2.16x. 

 

Its a a little more tricky as I have to power the D-Link and tx-USBultra and so used my existing 2 LPS1 supplies - themselves requiring supplies to operate. And with LPS1 devices you have to ground the negative to drain leakage currents. In total a lot of wires, but something I’m used to from my spaghetti days 😄!

 

Yes I agree that an internal clocked card would be a neater and better solution. I ordered a PF LAN card with OCXO clock a long while ago but, due to intel firmware issues, the card isn’t ready yet. When it does  eventually arrive it will be interesting to see if an external clocked switch still makes a difference.

 

Haha, you must have had a deja vu when your setup increased with 5 boxes at once 🙂. Good to see a Pink Faun Network card is in the pipeline, nice! It would sure declutter things and very likely sound better than the D-Link/OP14/PSU’s/..  

[flkin]

1 hour ago, Jambo said:

I have ordered my pf 2.16x in this colour. Contact Jord for other colour options and prices. A "pink" Pink Faun. Very nice.

 

Brilliant colour! Nice!

 

[lpost]

On 7/12/2019 at 7:11 AM, flkin said:

Bios settings:

- CPU frequency 2800Mhz

- CPU voltage at 0.9volts

- RAM voltage at 1.1volts

 

 

 

My system became unstable at these low (lowest possible) voltages, at least for CPU.  I've returned to default voltages for both and haven't noticed any SQ degradation.  The lower CPU frequency is definitely better sounding to me.  I'm back to using HQPe though, it just sounds more natural to me than Roon alone.

 

I worked with Piero a bit regarding the pstate issues and AMD Ryzen.  He's on holiday but will work on it upon his return, though I'm perfectly happy setting the values in bios and be done with it.

[k27R]

My 2.16 is on its way to me.  I had Jord install a JCAT NET card in my build since his pink faun lan card isn't available yet.  

[lpost]

Congrats on the 2.16.  It's a fantastic machine.  Mine is just over a year old.  You're getting in at a great time in Piero's AudioLinux development.  It's not always been so slick and well done.

 

I put in a, long since retired from work, PCIe fiber nic in 6 months ago for a modest improvement.  Drivers were already on the system so it was no fuss whatsoever.