Develop ultra capacitor power supply and LiFePO4 battery power supply

Try the experiment I mentioned - powerful processors tend to have lower latency due to increased cache. Noise floor and ripple are only some of the factors affecting sound.

Anyhow, these findings are proven by many other now. I still don’t understand why a 40 core processor sounds better... but it does.

I’m looking forward to LifePO mkIII as well, just waiting for my answers to come back.

Yeap.....the answer is in the listening right :D
 
Well... I hate to contradict part of this but I've been running Roon core on different platform upsampling everything to DSD 256 and some Pi as bridge. In this case, the Pi is just passing already rendered data to the DAC and using Roon's low latency protocol to communicate with the core. I've tried 12 cores processors, 48 cores dual Xeon server and a few others on the Roon Core side running different Linux distros, just for the DSD upsampling. From the 12 cores i7 Extreme CPU to the 48 cores dual Xeon, no difference in sound whatsoever... Just much faster upsampling to DSD. On the Bridge side, I've tried NUC and Pi and again, when using my linear PSU, there was no difference there. Only swapping out Linux distros I found some differences.

It's probably more the hardware design, layout, possibly the type of components, the OS, PSU, etc... More a combination of everything that in the end affects the chain, I don't think the CPU is the only factor here. As for Euphony, it is a good player but I can't justify its super high yearly price tag for the end result compared to Roon or Moode/Volumio. I gave it a full month test and while it is good, it just sound different, not better... Question of taste in the end, but I could not justify a 900$ early price tag for my home setup, not worth anywhere close to that in my book...

Anyways, this is all question of personal taste and experiences and no one is forced to agree with me, just my two cents...

That said, Ian's LifePO4 will definitely rock the house!

Do
 
Well... I hate to contradict part of this but I've been running Roon core on different platform upsampling everything to DSD 256 and some Pi as bridge. In this case, the Pi is just passing already rendered data to the DAC and using Roon's low latency protocol to communicate with the core. I've tried 12 cores processors, 48 cores dual Xeon server and a few others on the Roon Core side running different Linux distros, just for the DSD upsampling. From the 12 cores i7 Extreme CPU to the 48 cores dual Xeon, no difference in sound whatsoever... Just much faster upsampling to DSD. On the Bridge side, I've tried NUC and Pi and again, when using my linear PSU, there was no difference there. Only swapping out Linux distros I found some differences.

It's probably more the hardware design, layout, possibly the type of components, the OS, PSU, etc... More a combination of everything that in the end affects the chain, I don't think the CPU is the only factor here. As for Euphony, it is a good player but I can't justify its super high yearly price tag for the end result compared to Roon or Moode/Volumio. I gave it a full month test and while it is good, it just sound different, not better... Question of taste in the end, but I could not justify a 900$ early price tag for my home setup, not worth anywhere close to that in my book...

Anyways, this is all question of personal taste and experiences and no one is forced to agree with me, just my two cents...

That said, Ian's LifePO4 will definitely rock the house!

Do

Exactly my thinking here.....
 
stretchneck,
Thank you for your very complete post. I will read these threads with great interest. I think the continued evolution of the art led by the DIY community is advancing the SOTA for digital. Certainly, if I get forced to listen to my older gear, the difference is pretty significant.
 
Well... I hate to contradict part of this but I've been running Roon core on different platform upsampling everything to DSD 256 and some Pi as bridge. In this case, the Pi is just passing already rendered data to the DAC and using Roon's low latency protocol to communicate with the core. I've tried 12 cores processors, 48 cores dual Xeon server and a few others on the Roon Core side running different Linux distros, just for the DSD upsampling. From the 12 cores i7 Extreme CPU to the 48 cores dual Xeon, no difference in sound whatsoever... Just much faster upsampling to DSD. On the Bridge side, I've tried NUC and Pi and again, when using my linear PSU, there was no difference there. Only swapping out Linux distros I found some differences.

It's probably more the hardware design, layout, possibly the type of components, the OS, PSU, etc... More a combination of everything that in the end affects the chain, I don't think the CPU is the only factor here. As for Euphony, it is a good player but I can't justify its super high yearly price tag for the end result compared to Roon or Moode/Volumio. I gave it a full month test and while it is good, it just sound different, not better... Question of taste in the end, but I could not justify a 900$ early price tag for my home setup, not worth anywhere close to that in my book...

Anyways, this is all question of personal taste and experiences and no one is forced to agree with me, just my two cents...

That said, Ian's LifePO4 will definitely rock the house!

Do

Definately listening preferences and system synergy play a part... for example I couldn't get along with the sound of Roon running on a NUC with ROCK installed. Some people may love it.

Try Gentoo Player (it's free) running HQ Player embeded, or as a Roon server... it's marvelous!

Of course I prefer one box solutions - so this is how my comments should be taken. There is every reason to believe that server/renderer can also be very good and maybe technically better for some depending upon what you want to do.
 
Will the 'UcConditioner 5V' be suitable for direct power supply of DAC chip eg PCM1704?
PCM1704 requires two rails +5V and -5V DC. A pair of PCM1704 (for stereo) consume +5V: 10mA and -5V: 60mA. Will the voltage from 'UcConditioner 5V' fit within the required tolerance between 4.75 and 5.25V?
Will the charger power supply quality affect the sound quality or it completely irrelevant?
 
I have been studying this for a while now, and higher powered CPU's make a significant difference (but you need a great power supply otherwise the sound can become harsh). ....

Having said all of this, the theory behind LifePO4 is very sound - very significant uptick in sound quality for a Raspberry Pi I'm sure and it's massively cost effective. ...

State of the Art has moved on a great deal since these earlier experiments / practices you quote.

Talking about best sound I became frustrated by harshness of computer/NUC players and came across this. Check out if you haven't seen. Very cheap simple, low powered chinese SD Card player outputting I2S. I have one powered by Ian's LifePO4 modded with Ultracaps. A bit hair shirt but simply the best SQ, without any hardness and a joy to listen to.

https://www.tirnahifi.org/forum/viewtopic.php?f=15&t=4541
 

Attachments

  • SD CARD LifePO4 Ultracaps mod 1.jpg
    SD CARD LifePO4 Ultracaps mod 1.jpg
    837.4 KB · Views: 532
  • SD CARD LifePO4 Ultracaps mod 2.jpg
    SD CARD LifePO4 Ultracaps mod 2.jpg
    843.6 KB · Views: 532
stretchneck said:
I mentioned - powerful processors tend to have lower latency due to increased cache. Noise floor and ripple are only some of the factors affecting sound.
Hardware is one part of the problem, software plays a crucial part, if not more than hardware. Modern hardware is fast more than enough to process audio streams. Software, in particular, an operating system's kernel, plays a significant part. Make sure any kernel you use is optimised for sound processing.

I see a problem in one particular post. A poster wrote they are planning to use a NUC which, as far as I know, is intended for gaming. However, gaming focuses hugely on graphic processing which can be an issue. You need your kernel to focus on sound reproduction, not graphics.

A reply to my last paragraph just above, may be, GPUs are used instead of CPUs in the case of graphic processing. That is obvious, but the main CPU is what runs execution threads which in turn delegate work to a GPU.
 
Hardware is one part of the problem, software plays a crucial part, if not more than hardware. Modern hardware is fast more than enough to process audio streams. Software, in particular, an operating system's kernel, plays a significant part. Make sure any kernel you use is optimised for sound processing.

I see a problem in one particular post. A poster wrote they are planning to use a NUC which, as far as I know, is intended for gaming. However, gaming focuses hugely on graphic processing which can be an issue. You need your kernel to focus on sound reproduction, not graphics.

A reply to my last paragraph just above, may be, GPUs are used instead of CPUs in the case of graphic processing. That is obvious, but the main CPU is what runs execution threads which in turn delegate work to a GPU.

I disable the embedded GPU in bios, some people do use Nvidia CUDA for offloading HQPlayer upsampling - but I don’t use this.

I have given links earlier - check them out. There is not a problem with my approach, it is based on what several other members at AudiophileStyle have undertaken. The approach is proven by evidence.
 
Upgrad my system with UcMateConditioner, UcHybrid and UcAdapter

I upgraded my system today, the power supply section is now:

1. LifePO4 MKII
2. UcAdapter PCB
3. UcMateConcitioner (5V)
4. UcHybrid (3.3V 4x)

The upgrading job was pretty easy and straight froward. Both MKII and MKIII are good to go. MKIII will have some dedicated connectors. The integration was seamless. When I turned it on, everything was OK. The blue conditioning LED started lit in a couple of second. No smoke, no fire:D.

I have some new discovery and still feel the improvements.

The small board behind my PRi DAC stack is my HDMI receiver board. I use it to receive the true DSD signal the I hacked from my Oppo bluray/SACD player through ReceiverPi.


UcUpgradeKitIntegration
by Ian, on Flickr


UcUpgradeKitIntegration1
by Ian, on Flickr

Ian
 
Last edited:
I upgraded my system today, the power supply section is now:

1. LifePO4 MKII
2. UcAdapter PCB
3. UcMateConcitioner (5V)
4. UcHybrid (3.3V 4x)

The upgrading job was pretty easy and straight froward. Both MKII and MKIII are good to go. MKIII will have some dedicated connectors. The integration was seamless. When I turned it on, everything was OK. The blue conditioning LED started lit in a couple of second. No smoke, no fire:D.

I have some new discovery and still feel the improvements.

The small board behind my PRi DAC stack is my HDMI receiver board. I use it to receive the true DSD signal the I hacked from my Oppo bluray/SACD player through ReceiverPi.



Ian

new LEGO toys!:D
 
I have been studying this for a while now, and higher powered CPU's make a significant difference (but you need a great power supply otherwise the sound can become harsh). There is a huge thread on Audiophile style.... I can't possibly go through everything here but their member AustinPop posted the following recently:

A novel way to massively improve the SQ of computer audio streaming - Page 649 - Music Servers - Audiophile Style

I think this is a good summary of the direction of travel, you can see that higher powered computers are ranked to sound better. My own experiences are the same (e.g. Allo USB Signature didn't sound anywhere near as rich and powerful in sound as my NUC, the USB Sig sounds comparatively flat and lifeless in comparision... both where powered by good quality linear power supplies).

But all of the same rules apply, low latency, low OS overhead, storage type, networks clocks, reclocked USB. Operating system makes a big difference as well. If you have a gaming PC (something with a high powered CPU) then try booting it with a USB stick which has a copy of Euphony or Gentoo Player on it... you might be amazing by how good it sounds even though the noise floor and lots of other things might not be optimised.

I'll post more on my build once I know the answer to putting LifePO4 in series and also the height of the product... anyone able to let me know?

Another good thread, Nenon is at the cutting edge: Building a DIY Music Server - Music Servers - Audiophile Style

You would think that a 40 core dual processor setup would be absolutely ridiculous... but it apparently sounds amazing. Powering this beast is a real issue, so I'll only ever go down the single processor route. Some people have given up with DIY and reached deep into their pockets and have spent $20,000 on the SGM Extreme. Apparently nothing comes close in sound quality, and I can well believe it.

Having said all of this, the theory behind LifePO4 is very sound - very significant uptick in sound quality for a Raspberry Pi I'm sure and it's massively cost effective. I am guessing that LifePO4 and a USBSig, with my favourite software Gentoo Player would easly sound better than many expensive commercial products and would completely annihilate the sound you'd get playing iTunes from a standard PC or Macbook. I'm really keen to try it with my NUC.

State of the Art has moved on a great deal since these earlier experiments / practices you quote.

Hi,

In reply to your request concerning height of the Mk2 board; I assume same as Mk3, height from bottom of board to top of battery holder is 30mm. There are up to 2.5 mm high SMD components under the board.

Martin.
 
@ Stretchneck,

I too have been following Nenon's sever build. Powering a Z390 with high power CPU by batteries, depending on OS, may require many batteries or frequent recharging. As an example. I have used one board at 13V to power a Chord M Scaler. With attached UCs, I can achieve in excess of 2 hours between charges. The M Scaler power consumption is rated at 16W. Nenon powers the CPU using a 60W supply. Using two boards and utilising the last two 3.3V batteries of each board in series for another 13V, you would get 1 hour 10 minutes per charge. Nenon said he averaged 30W so if using Euphony, no upsampling, charge intervals may be extended. The 24 pin ATX and peripherals still also need a similar amount of power. The case I use to accomodate two boards, the charging power supply and one full set of UCs has dimensions 186 x 225 x 141mm. It should start to become apparent that this is going to be neither a cheap nor convenient way to power such a server. I'm currently investigating MP Audio's offerings.

Martin.