Pink Faun 2.16

[flkin]

Ah, good to see a Pink Faun thread at last. I recently picked up a 2.16x. Arch Linux OS with USB output. Mine has 4 OCXO clocks.

 

Jord did say he can install a windows OS if required but I opted for his recommended Arch Linux OS which was described to be "as stable as a 16 legged elephant”. Haha

[flkin]

Does anyone here use his HQplayer embedded? If so what settings are you using?

[flkin]

9 hours ago, FredM said:

Congrats, thats a mighty server configuration!

 

Thanks, it's still breaking in but sounds quite good already. The streamer is playing with 3 clocks now. The forth one will come when the new LAN board is complete and hopefully with the new low phase noise OCXO (-130 DB@10hz) installed.

 

1 hour ago, lpost said:

 

I plan to try it out but I will be feeding a DirectStream which takes all inputs to 20x DSD so I'm not sure there is much value in spending CPU cycles on the server when the FPGA is going to do it by design.  

 

Did you get HQPlayer trial or licensed?  I haven't asked Jord but the website says a licensed, embedded version is included but via email he mentioned a trial version, perhaps its changed.

 

My embedded HQPlayer is licenced and was included without me asking for it. 

 

Ive been using HQPlayer on my Mac for years and somehow I preferred the sound of redbook feeling the upsampled version was too thin and sharp. Not analogue enough for my liking. 

 

But with the Pink Faun, I've found that upsampling is another experience altogether. Pretty amazing details, presence and holographic soundstage. 

 

20 times DSD? Wow, thats some heavy computing. 

 

[flkin]

Sounds like you have a very nice system there!

 

Yes PF's new clock should be great especially on board component. Looking forward to trying it out once it's launched. I understand it's going to be soon.

 

The 2.16 was already sounding great when I first set it up but has improved quite a bit in the past week of constant playback. I think the clocks need at least a month to bed in properly. It's a great month for me, like going to concerts every night ?

 

 

[flkin]

Love those speakers. Had a chance to hear a pair during a show a month ago with a Chord front end and they were very real sounding. Reminds me of the Opera Diva singer from 5th element. ?

 

I have to mention that during this burning in period, I found the PF benefited when the USB signal was processed through a reclocked SOtM tX-USBultra (by a Mutec Ref10 clock) and iFi Purifier prior to the DAC. Clarity and presence was improved.

 

I hope that over time I can retire these additional components and play from the PF directly to the DAC.

[flkin]

Jord shared this with me about the new clocks. The key would be they are on the streamer boards at the required frequency rather than external to the streamer as a reference clock and having to be converted to a correct frequency through multiple connections. 

Phase noise 26 MHz.pdf

[flkin]

Just saying that external ref clocks (like with Mutec or Cybershaft reference clocks) have even lower phase noise. My Cybershaft OP14 works at -114@1Hz. Unfortunately for them to operate they have to undergo multiple connections and frequency conversions as compared to PF's approach with on board clocks at the required frequencies. It wasn't about the clock swaps. 

 

On that matter, swapping add on card clocks would be a simple matter of mailing in the old card but for the motherboard or processor clock that would probably require returning the streamer in my case. I had Jord move the latter 2 clocks to the bottom of the motherboard to shorten the signal path of the clock.

[flkin]

41 minutes ago, ddetaey said:

 

I am using HQplayer embedded in the combination f the Pink Faun Streamer 2.16 connected via USB to the T+A 8 DSD dac.

I upsample everything to DSD128 (via DoP) as the Amanero USB interface does not (yet !?) support native DSD higher than DSD64.

Benefit of this mimitation is that I can use the most demandign filter settings.

It all sounds fantastic, but still would love to be able to apply native DSD512.

I have listened to native DSD512 at the Pinkfaun office on their DAC 2.32, and it sounded great as well  !

it does not make sense trying to compare, because everything else in the chain is so different - room, amplification, speakers,  ....

 

Attached are my current HQe settings

 

Dirk

 

Hi Dirk

 

Thanks for sharing your settings. I was trying to get the same poly sinc xtr ls upsampled 44.1 to DSD512 but wasn't able to do playing natively to my LIO DAC2.0. Managed it with the -2s settings from 44.1 to 512/48 but I think the filter doesn't sound as good as the non -2s setting.

 

Thanks for the idea to try it DoP rather than native. Will give that a go.

[flkin]

Ok I see. Its that your HQPlayer settings use the full xtr filters upsampled to DSD512. I can't get this setting to work with 44.1 files and the only difference is the DoP setting.

 

My unit contains the AMD Rizen 8-core chip. Is that the same as your unit?

 

My head is spinning with the huge number of possible combinations available. And not all filter combinations work with different music file resolutions. The only DSD512 stable settings I can find is with the -2s/ASDM options. Although not my preferred filter, it sounds pretty good anyway.

[flkin]

Just want to confirm that it is indeed easier to use the harder filters under DoP setting. But I'm finding I still prefer the Native setting for my Vinnie Rossi LIO DAC2.0.

 

Another question for Pink Faun streamer owners, do you leave the machine on 24/7 or switch it to standby (one pink led on).

 

Jord informs when in standby mode the motherboard and processor clocks are still on but not the card clocks for synchronisation issues. 

[flkin]

Deleted 

[flkin]

It may be a little more complicated than that as the clock boards are separately powered and regulated. Not sure if ordered without the clock, the power supply is pre-wired especially given their  complicated star-earthing system they have in place?

 

Anyway, a quick email to Jord will clear this up if interested.

 

On a side note,

 

the motherboard and CPU clocks can also be installed under the motherboard (as an option) to shorten the signal path. The benefit of clocks are said to be highest when installed closest to the point they are required. 

[flkin]

Yeah, I recognise those photos, it's my unit. It shows the 2 clocks installed on one board under the motherboard. It has separate power regulation despite being on the same board I'm told.

 

It's not clear how this shortens the signal path though, I wonder how the connection is made to the MB and processor? I guess it's through the holes and into the bottom of the MB. This is one highly optimised and dense design of a machine!

[flkin]

Yeah, it think it's around 400euro or so more per clock.

 

A better clock makes for quite large perceived difference in my experience to date. 

 

Each time I switch to a better clock, what I thought of as being near SOTA sound would be bumped up to another level. There seems no end to this improvement!

 

Or spending ?

[flkin]

3 hours ago, jjkale said:

 

Do any of y'all have an opinion on what sounds best for redbook output:  native, 24/192 or DSD512?

Regards!

 

 

As my 2.16 is breaking in, and it does seem to take quite some time for this to happen, the sound is becoming better playing in NOS mode. No upsampling at all, just Roon playing at original resolutions. 

 

It was fun experimenting with HQplayer and Roon upsampling and certain settings (polysinclp or polysincxtr mp/lp or Sinc) sound very nice but all that's changing as I approach full burn in. Now I find 44.1 as is it amazingly good - focused and full of detail, yet organic and real.

 

That said, I output as USB into a tX-USBultra reclocked by a Mutec Ref10. Managed to retire my IsoRegen and iPurifier in the chain but not the tX-USBultra. That still seems to add an extra sense of reality to the music.

[flkin]

Yes, I find that many who use the 2.16 uses I2S but at this time my Vinnie Rossi DAC2 doesn't have this input option. From the chatter, it seems the I2S may be a better input technique so I'm curious about this. However this will require a specialised board made to order as the I2S wiring standards are not fixed and therefore is device specific. 

 

Mind you the Pink Faun OCXO USB output is no slouch! It's by far the best my system has ever sounded and not by a small margin. And this is before optimisation of power cables, USB cables etc. All to come once the break-in process is complete.

 

@austinpop , it's still work in progress at this time as the Pink Faun streamer is still breaking in - Jord did inform that it takes as long as 500 hours before it settles in. It's quite a change in sound from the first day it arrived 3 weeks ago and I've been running it 24/7 since with the occasional thermal rest in between. Immediately, the sMS-200ultra wasn't comparable so that went. This was connected via the streamer's standard LAN port. My OCXO LAN card wasn't ready yet as Pink Faun was building a new card so I plan to revisit this comparison at a later date.

 

During the first week, the IsoRegen could be seen to dull the music despite Stammhein power cleaning boards and PoE DC wires so I stopped using it. Then in the second week, the iPurifier started to make the music hard so that had to go. This past week, all upsampling had to be removed as it wasn't necessary anymore. Straight 44.1 (for redbook files) sounded most organic and real without further need for "sharpening" and "focus" using upsampling techniques.

 

Presently the tX-USBultra alone drops the sound quite a bit. It's only when the Mutec Ref10 is added (with the Habst clock cable) that brings the sound a touch higher than without the tX. This may change when the break-in completes or when the newer Pink Faun clocks are out. Jord informs that those are custom built for them and has substantially lower phase noise than the current clock - by a whopping 20db! ??

[flkin]

My unit was shipped in a wooden crate with an external shock sensor. Now I know why.

[flkin]

Does the i2S board come with clock options?

[flkin]

Actually it's probably closer to 20db difference.

 

The existing clock is a Connor Winfield OH200 series clock which has a phase noise at 10Hz at around -110db/Hz according to their spec sheet.

 

I posted the Pink Faun prototype clock data sometime back with readings of -130db/Hz giving a difference of around 20db. 

 

Hot off the press, here is the official clock performance of the new Pink Faun clocks. -132.5db/Hz at 10Hz ??

 

This will make a big difference for sure!

Pink Faun OCXO phase noise.pdf

[flkin]

Found the Mutec Ref10 phase noise figures for 10Hz. It's -142dBc/Hz so that yet another level lower. This is a special reference clock and is designed for external use unlike the Pink Faun clock installed on board closer to where they are needed. 

[flkin]

Ah, that's not the case in my experience. Every dB seems to make a difference where phase noise is concerned especially at the 1 Hz mark. 

 

I have a Cybershaft OP14 clock which has a phase noise of  -114dB at 1Hz. The Mutec Ref10 is -116dB at 1Hz  and is nearly double the price for good reason.

 

A change of 20dB at 10Hz is effectively a few orders of magnitude change. It's going to be a whole new streamer. ?

[flkin]

I wasn't talking literally which I believe is 10 times (only an order of magnitude) for a 20db change, I was trying to say figuratively that the effective perceived sound improvement would be massive. Logs and the such was another lifetime for me too. ?

 

I should clarify that earlier I mentioned the Mutec Ref10 has a -142 dBc/Hz at 10Hz but these new Pink Faun clocks will have -132.5 dBc/Hz but at 20Hz. It is apparently easier to obtain low phase noise at lower frequencies. 

 

What makes me so excited about these clocks is that they will be used without having to convert the frequency from 10MHz to the final required MHz required by the electronics while having the clock signal pass through multiple connections and cables to get there. 

 

Jord says the frequencies required in the 2.16 server is 20MHz  for the USB bridge, 24.576 for the I2S bridge / SPDIF bridge and 25MHz for the system and processor clocks. So the 20MHz clock chart I attached earlier would refer to the USB bridge but is probably indicative of the low noise their other clocks will eventually achieve.

[flkin]

Ok I see the confusion.

 

The Ref10 clock outputs a clock signal at 10Hz for other devices to use. This signal has a phase noise curve in dBc which changes depending on the Hz of measurement. The usual measurement for comparisons between clocks is at the 10Hz or 1Hz noise on the curve of the clock.

 

When I said Pink Faun's clock was based on a 20Hz clock that's the frequency of its output signal and not referring to the frequency of noise measurement on the chart.

 

It's easier to get a low phase noise with a  signal output frequency at 10hz so you can't directly compare the phase noise of a 10hz clock with a 20hz clock. 

[flkin]

4 hours ago, m3lraaHnevetS said:

Try 44.1 (or anything else) coming from Roon through HQP with PCM filter settings at none, none, none, 44.1, 192

this only works when the first file you play is indeed a 44.1 file, after that you can also play hires files natively using these settings.

To me this sounds better than any upsampling and also much better than just Roon.

 

I read about this sometime back but haven't been able to get it to work.

 

Using the setting you @m3lraaHnevetSrecommend above, 44.1 files play fine but DSD files simply hang.

 

Would I need to set the HQPlayer Output mode to PCM or Auto? Or is there another trick I missed?

 

Thanks!

[flkin]

16 hours ago, m3lraaHnevetS said:

You are right, DSD files do hang with these settings, i had not noticed this since i have very few DSD files. But still, to me PCM (not just 44.1 but also higher res) does sound amazing with these settings.

 

I have to do more comparisons for Roon 44.1 and HQPlayer 44.1 (in my system comparison isn't easy) but I agree with your findings that HQP 44.1 is better, but in my system only subtly. 

 

I've discovered that the hanging issue is solved if I set :

 

-Output mode: Auto

-Direct SDM ticked 

-Options: Auto rate family (not sure if this is important)

-PCM settings : none, none, none, 44.1, 192

 

then HQPlayer 44.1 plays as 44.1, hi res PCM plays at their original hi res settings and DSD files play as per HQP's SDM settings.