|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
85.147.82.48
In Reply to: RE: cPlay - the open source high-end audio player using ASIO posted by cics on May 05, 2008 at 12:31:58
Hi all,Summer in The Netherlands is at it's end, so it's time too pick up my indoor hobby again. Which means its time for: Power Supply improvements: part 2.
As already stated by other inmates (Gstew, Reylands, Bertel, Theob, carcass93, Rayban, Sondale, fmak and many others) improving the power supply is a very worthwhile and very rewarding exercise.
Optimizations of the Operation System, Kernel and the BIOS all together do have a nice effect on sound quality, but I was really amazed what BIG IMPACT (!) a linear PSU on the P4-pin had on sound quality in my setup.Note: my setup is a fully optimized PC as per Cics hard- and software recommendations. So my findings are based on a setup like this: see: http://www.cicsmemoryplayer.com. With only 2 exceptions: I do not under-volt the FSB nor RAM . I over-volt both with + 0,2 V.
Improving the power supply to the P4 pin with a linear PSU turned out to be surprisingly easy (and also relatively cheap), Click on picture to see photo's in Picasa webalbum:
cMP2 project: Optimizing power with linear PSU to P4 power connector'
So I now would like to try optimizing the power to the P24 connector.Just before summer (in june) more knowledgeable inmates who already tried to improve the power on the P24 (Bertel, Gstew and others) gave me a lot of suggestions and possible directions.
-> Bertel pointed out the importance of choosing good quality low-ESR smoothing caps but also and the use of low resistance wires (!).
"to keep power supply as clean and quick as possible to keep the rise time between 0 and 1 ("no power" and "power", changing ever so quickly) at a minimum! IMHO this can best be done with caps sized the "right" way, plus cables from the caps to the board which are short, have a lot of internal conductors (that brings speed, e.g. I use AWG 6 which is 133*AWG27) and a very low internal resistance (<.5Ohm/1000ft, just to be safe - that's in the region of an average PCB's traces)".-> Gstew suggested use a tweaked Pico PSU (extra smoothing caps) on the P24 pin.
So that's what I did (wires and picoPSU) last week, with unexpected and very surprising results !!
* using LOW RESISTANCE internal power wires to P4 and P24.
Although Bertel already had pointed out to me the importance of using low resistance wires, I still wired everything together with standard computer wires (18AWG wire). Why? because its easy and I'm lazy. Standard computer 12 v power wires are constructed of 18 AWG and can simply be bought everywhere and they have already all the right plugs and connectors attached and in place. Regular wire to distribute power inside pc's is: 18AWG = 6,5 ohm/1000ft. Bertel suggested to use 6AWG which is: 0,4 ohm/1000ft. http://www.interfacebus.com/Copper_Wire_AWG_SIze.htmlUntill last Sunday. A suddenly cancelled social appointment left me with nothing much to do on a rainy Sunday afternoon. With electronic stores being closed on Sunday in The Netherlands, I had no other wire at hand than the type of wire that is used to wire the electricity in housing (with solid 2 mm copper conductor).
I braided 2 solid copper wires together until I had 1 meter of braided solid copper cable. With still 1 cm standard 18AWG computer wire left at the P4 plug I soldered the braided solid core copper wire to the P4 plug (a P4 plug has 2 black 18AWG wires + 2 yellow 18AWG wires)
I saw no other way than too solder the 2 mm solid copper wire to the P4 plug than trough leaving 1 cm of the original 18AWG standard wire attached to the P4 plug. The best solution would be to solder the low resistance wire directly to the metal-connectors inside the plastic plug. But I don't manage / don't know how, to get the metal-connectors out of the plastic plug.Click on picture too see photo's of the solderingproces of low resistance multi-stranded wire to the P4 plug in the picasa webalbum.
cMP2 project: low resistance wire between linear PSU's and P4 / P24 power connector
I also braided 1 meter of 2 times 2 mm solid copper wire to also connect the picoPSU with this braided wire to the linear PSU that is feeding the picoPSU.Working with this braided 1 meter 2 x 2 mm solid coper wire is a hassle. The braided wire is not flexible and one has to bend the braided solid copper wire in place all the way from the linear PSU, through the pc case, to the P4 and P24 connectors. Of course it's much easier to use multi-stranded wire (6AWG or similar).
I didn't expect much of this construction because there still was a bottle neck of 1 cm 18 AWG standard computer wire attached to the P4 plug. So I expected the whole wrestle with the stiff braided cable would probably not be worth the trouble.
But WHOW! I couldn't be more wrong! It made big and massive impact!!
Bass, mids, treble everything changed dramatically. (I'm not writing this for effect purposes) Especially drum kicks, plugged strings, or any type of 'explosive sounds', where very much more heard and present. A drum kick, now sounds like a real drum kick !!!
The changes where that big, that I first thought, that there was something wrong. That I only added some kind of reverb like distortion. But after playing more and more music and tracks I started to realize that this was no distortion at all. This is probably how the music was recorded and mastered.
* Optimizing power to the 12Volt rail of the P24 with linear PSU + picoPSU.
- Chosing the right picoPSU model.
It turned out that not al picoPSU are made by the same concept. Some picoPSU models leave the 12 V line untouched and pas it straight on too the MoBo, where other picoPSU models don't leave the 12 V line untouched and regulate the 12 V line.So shit happens and I first bought the wrong picoPSU model. A PW-200-M w/P4-ATX. This model regulates the 12 V line which results in almost no sound quality improvement. So it's not worth the trouble to change a 'smoothing capped' ANTEC ATX PSU for an PW-200-M w/P4-ATX.
By now I have ordered an other picoPSU model and installed it. The model picoPSU-150-XT leaves the 12 v line untouched and passes the 12 V line straight on too the MoBo. Now there is a very nice improvement in sound quality !
However a very nice improvement, it's not as much as with a linear PSU on the P4 pin.
In comparison: a picoPSU-150-XT on the P24 gives about 1/3 of the improvement of a linear PSU on the P4 pin. (So improving the 5 Volt and the 3,3 volt power also needs to be done. Especialy the 5 v line. But thats for PSU improvements: part 3, because I haven't made up my mind yet on how to do it the easy way). But although only 1/3, it's still a worthwhile and recommended improvement !So chosing the right picoPSU model (which leaves the 12v line untouched) is important! Installing a picoPSU model that regulates the 12V line, it's not worth the trouble IMHO.
cMP2 project: Optimizing power with linear PSU + picoPSU-150-XT to 12V rail on P24 power connector'
* First question on internal power wires.
Are there any other inmates who swapped out the standard PC power wire type (18 AWG) and replaced these internal pc-power-wires for low resistance wires like Bertel suggests?
(AWG 6 wire or lower resistance)
Or, are there any inmates willing to experiment with low resistance type of wire between their linear PSU and the P4 pin and/or their picoPSU ? And report back there findings ?* Second question on internal power wires.
If this would be analog audio, I could easily understand what is happening and understand the effect. But what is happening here ? Is this really the effect Bertel pointed at? "keep the rise time between 0 and 1 ("no power" and "power", changing ever so quickly) at a minimum!"Any comments, suggestions, feedback is very much appreciated. Or even better: please try this at home and report findings back here at the forum !
Power supply improvements part 3, will on the P24 5 volt and 3,5 volt lines but I haven't made up my mind yet on how do to this the easy way. Probably there is no easy way.
Mark
Note:
I don't encourage anyone too use solid core 2 mm copper wires. It's a hassle. I just had nothing else lying around at that time. But I do encourage to try multi-stranded 6AWG wire (or any other low resistance wire) between the linear PSU and the P4 / P24 connecter and report back the findings to this forum.
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 11/02/09 11/02/09 11/02/09 11/03/09 11/03/09 11/08/09 11/09/09 11/11/09 11/11/09 11/11/09 11/11/09 11/11/09 11/11/09 11/15/09 11/23/09 01/18/11Follow Ups:
I think I'm almost ready to take this same journey. All this information is very exciting to me. Do you have any pictures you can share?
Also, I'm interested to know, do you have just one linear PSU powering the P4 and the pico or are you using two separate linear PSUs? Also, which brand/model linear PSU are you using? Or did you build it yourself?
Thanks for sharing!
Hi Edward,
Since you asked I made some picture to share via Google Picasa webalbums this week, for your and other inmates convenience .
See added links to Picasa web photo albums in original post.
fully AOB optimized cMP2 PC -> Lynx AES16 -> XLR AES/EBU -> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Thanks for the pix! So then I take it you are going from the PSU straight to the P4 & P24? And not using any "smoothing" capacitors in between? Have you found that with the linear PSU, the caps are not needed?
Also, I'm intrigued about using lower resistance (higher gauge) wiring. I'm confused though. Correct me if I misunderstood, but did you not say you braided two 12 gauge solid core wires together? In the photo it appears to be a multi-stranded wire. I'm not an expert on these matters nor an electrician, but I am a fan of the concept of solid core cables. So I'm wondering if it would be sufficient to use 10 gauge solid core copper (and try to go directly to the P4 connector without using the 18 gauge stranded wires).
Thanks again for your contributions and information.
Hi Edward,Yes from the linear PSU the wires go straight to the P4 / straight to the picoPSU on the P24.
I’m also not an expert nor trained / skilled in electronics, so we both should be careful not to become: two blind people, who are leading each others way. :-)
But my understanding is that smoothing caps are already employed inside a linear PSU after the rectifier bridge and or Voltage regulator. I don’t know if extra caps outside the linear PSU betters things extra.I’m not trained /skilled in audio electronics and/or digital electronics. So I can’t say that I’m a fan of solid core cables or stranded wire. I just don’t know which one is better for a given situation/application. I just wanted to try Bertels suggestion of using wires with a much lower resistance than the standard used 18AWG wire. At that time (Sunday afternoon) I only had solid (1,5 mm core) copper wire lying around as used in the electricity system in walls in housing. I braided 2 strands of this solid core copper wire together, to get 1 metre cable.
Since the 1 meter cable of 2 braided solid copper wires is really stiff and a hassle to work with, this week I bought some 2,5 mm (outside diameter including isolation) multi-stranded copper wire and made again: a P4 extension cable of 4 times 2,5mm multi stranded copper wire. Soldering of these 4 multi-stranded 2,5 mm copper wires to the four 1cm 18AWG wires at the P4 plug, is what you see on the pixs.
Both wires types (solid copper and multi-stranded) sound the same as used for P4 extension cable / power cable to the picoPSU at the P24. But 2,5 mm multi-stranded copper wire is much more easy to work with.I haven’t received any feedback from other inmates yet on sound quality when replacing 18AWG wire for a wire type with a much lower resistance as Bertel suggests (6AWG or similar).
I hope other inmates do want to try this and report their findings to the forum. Because – also very much to my surprise ! - it had a very, very big impact on sound quality in my setup.As for your question: “So I'm wondering if it would be sufficient to use 10 gauge solid core copper (and try to go directly to the P4 connector without using the 18 gauge stranded wires)”, I can’t answer your question as I don’t know that.
fully AOB optimized cMP2 PC -> Lynx AES16 -> XLR AES/EBU -> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 11/09/09
Hi Edward,I bought a simple Voltcraft 12V/2A inear PSU (40,- euro’s) at www.conrad.nl (part number 510172). I have not enough knowledge nor skills to build my own.
Voltcraft is their home-brand and is produced in Chechoslowakia. But one can buy this type of simple linear PSU’s everywhere. For what is worth: the technical specification shows a max ripple of 2,5 mVolts. But I don’t know if there are any other distortions present on the 12Volt output.As Reylands already suggested it really very easy and simple to connect a linear PSU to the P4 socket if one uses a P4 extension cable. http://www.atxpowersupplies.com/p4-extension-cable-adapter.php . Just cut off one connector and attach that side of the extension cable too the linear PSU and put the other end into the P4 connector on the MoBo. That’s all. This will take you no longer than max 15 minutes. And there you go !
If one P4 extension cable is to short (they are about 14 inch / 36 cm which usually is to short), just use 2 or 3 and connect them together to get the desired length.
But using these handy ready made P4 extension-cables comes at a sonic price as I discovered this weekend. They are constructed with AWG 18 wire. And as inmate Bertel already suggested to me, using low resistance wire (for instance AWG 6 or so) might be far better. And Bertel was o so right !!
If one uses low resistance wire as Bertel suggests, than there’s again a big sound quality improvement.I think many inmates did the same thing like I did, and wired everything together with standard AWG18 wire. So they are not getting the maximum sound quality improvement possible. That’s why I decided to post my findings and encourage other inmates to try the same and report back there findings on sound quality.
Right now I have 2 linear PSU’s outside my pc-case to power the cMP2 PC and 1 standard ‘dirty’ ANTEC Earthwatts ATX PSU mounted inside the PC case. The ANTEC now takes care of the job that originally was indented for the Granite PSU’s.
I don’t like the Granite solution to feed the USB-port, HHD and DVD-drive. When using the Granites one has to bring 230 AC wires inside the PC-case to power these Granites. I don’t like that in terms of noise, hum, RF-polution, etc.
On the other hand I don’t worry very much about ‘horror stories’ on how ‘noisy’ it supposably is inside a PC case. A PC entirely runs on 12/5/3,3 volt DC, so I don’t worry at all about ‘RF-noise’ inside my PC-case. I just don’t like 230 AC wires and unshielded 230 AC -> 12V DC converters inside my case.Having 2 linear PSU’s was not a deliberated choice, it just went this way because I first optimized the power on the P4 pin. I found this to be so rewarding that I decided to also give the P24 a try. So I had to buy second linear PSU 12V/5Amp to feed a picoPSU on the P24 (Using a picoPSU model that leaves the 12V line untouched !)
But when using 1 linear PSU it will have to be a big one because it will have to provide 7 or 8 amp total. May be there is also some benefit from using 2 separate linear PSU’s. They cannot influence each other, ect.
If you buy a linear PSU be sure to buy one with low ripple specs and passive cooling (without a noisy cooling fan).
Furthermore I do suggest that you first carefully read the ‘Advanced’ section chapter 12.b on http://cplay.sourceforge.net/ and their subsequent links to other inmates here on the forum. Some of those inmates are far more knowledgeable than I am. I just read all their recommendations and suggestions and than try something in practice and report back my findings to the forum.
If you don’t mind an extra PSU box beside you PC-case, than a linear PSU on the P4 is a must do. When using P4 extension-cables as suggested by Reylands, you are done within 15 minutes.
It even gets much (!) better, when you don’t use the ready made P4 extension cables (with AWG 18 wire), but when you use low resistance wire (something like AWG 6 wire as Bertel suggests) .Succes and happy listening.
fully AOB optimized cMP2 PC -> Lynx AES16 -> XLR AES/EBU -> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 11/03/09
I have made a 12v linear psu to power up the 12v p4 line on my cMP. The unit is about 2 feet away from the load (p4 pin). I have used 10,000uF caps for the filter capacitor & 1000uF for the output capacitor after the 7812 regulator. Since the wire from the output cap to the p4 pin is roughly about 2 feet long (and twisted), do i need to place a small cap close to the load????
Vellerman's PS1503SB (for ~$90) looks very impressive with RMS ripple specifications of ≤1mV and ≤3mA!
According to manual: for stable voltage supply you need to set current output to Max (3A), turn on and adjust voltage to 12V. Anybody tried this?
PS - I've added a link to your excellent post for Hybrid P24 power supply (Advanced section).
Hi Cics,* using one linear PSU (instead of two)
When I first red your suggestion: “how about 1 linear PSU of 12/3A ? ”, I thought: well a 3Amp linear PSU is not enough. Other inmates on this forum reported that the current draw on the 5 volt P24 line alone was between 4 and 6Amps. If one would ad this up to the also reported:
- 0,5 Amps for the 3,3 Volt
- 0,5 Amps for the 12 Volt
And for the P4:
- 0,4 / 0,5 amp during boot (0,3 – 0,4 while playing music)
That would make a total of:
- P24: 6A + 0,5A + 0,5A
- P4: 0,5A
Total = 7,5 A.
Say 8 Amps to be on the save side.But it’s seems that 4-6 Amps trough the 5 Volt rail on the P24 pin (as other inmates have reported) isn’t correct.
Since I now use 2 linear PSU’s, of which one is feeding the picoPSU at the P24 pin, it’s has become very easy to measure the current flow through the 12 Volt line that is powering the picoPSU. So that’s what I did this afternoon, after reading your suggestions too use only one 3 Amp linear PSU. My measurements gave the following readings:
The current flow on the 12 Volt DC line towards the picoPSU is:
- while booting up: between 1,5 and 1,7 with peaks to 2,0 amps
- while starting XP: between: 1,75 and 1,85 with peaks to 2,0 amps
- when Xp is at rest: 1,75 Amps
- when starting a program: 1,8 – 1.85 Amps
- While playing music: 1,76 – 1,79 AmpsSo when using only one linear PSU, the unit would have to deliver:
- P24: max 2 amp
- P4: 0,5 amp
makes a total of: 2,5 Amp.
So your suggestion of a linear PSU at 12V/3 Amp would indeed be enough.
I indeed would like only one linear PSU for space and energy consumption reasons. But also because only than both the P24 and the P4 will get exactly the same voltage. And there will also be no a voltage difference between the 2 black minus poles of each PSU and the ground.
On the 12V/2Amp linear PSU the voltage between the black minus pole and the grnd always was: 0 Volt.
But on the 12V/5Amp linear PSU the voltage between the black minus pole and the grnd always was: -0,003 / -,021 volt. No matter what I did, I couldn’t bring it also back to also: 0 volt.
If this where analog audio, one wouldn’t want this difference between two PSU’s. I don’t know if this also matters when powering digital data processing, but this didn’t give me ‘ease of mind’.I now connect both lines feeding the P4 and feeding the P24 to one linear PSU as you suggested. The 12V/5Amp one. It works fine. The safety-circuit hasn’t shut it down. Not when switching on the PC (run-in currents), booting up or when playing music. On first ear, I doesn't make any differences in SQ, But I will do some carefull listening on that in near future.
* low resistance wire and linear PSU’s with better specs.
Unstill now I haven’t had any feedback from inmates who also changed from 18AWG wire to a wire with much less resistance and how this improves sound quality in their setup.
That’s a pitty, because the sound quality in my setup improved really, very, very much !!
And also the MoBo liked it very much :-) I now can cold boot without any failure at only 130 mhz FSB speed. And warm boot at even 125 mhz FSB speed !If low resistance wires have such a (real big !) impact on sound quality (in my setup) than I probably should have paid much more attention to linear PSU specs, such as very low Voltage ripple and very low Amp ripple, when I ordered one. I more or less bought the first low cost linear PSU, with reasonable low Voltage ripple, I came across.
I think the Velleman linear PSU you suggest, looks like a good buy. It’s only € 57.50 euro and has much better specs than the 12v/2A one that I bought for € 40,-. And it’s looks better too! Thankx for your suggestion. Although I won’t buy another linear PSU in the coming months. All funds will be directed to my upcoming 1 month holiday in New Zealand and Australia in February 2010.
I hope other inmates will soon report their findings on linear PSU’s with better specs and swapping their internal 18AWG power wires for low resistance ones.
fully AOB optimized cMP2 PC -> Lynx AES16 -> XLR AES/EBU -> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 11/08/09 11/09/09
Thanks again for the info. And it brings up another question. On your measurements of 2 amp (P24) and 0,5 amp (P4), I assume this is based on underclocking and undervolting? Does it make any difference if the CPU is run at full speed with default volt?
For example, if I get a PSU at 3A, but then experiment with lowering FSB and voltage further and then go too low to the point it will not boot - then I will have to reset the CMOS/BIOS which will boot at defaults. Will 3 amps still be enough at default speeds?
Hi Edward,Good point. your assumption is correct. 3 Amps is not enough with the BIOS set at: ‘Load optimized Defaults’ :-(
I connected the multi-meter again and loaded BIOS settings: ‘optimized Defaults’. Then I repeated the booting process several times. The highest peak I saw on the multi-meter while the BIOS was starting the PC was a short peak of 3,44 Amps. One of these booting sessions I recorded on camera, which can be seen here:
During this boot up session you can see a short peak of 3,33 Amps in the 18 / 19 second (is fase where the BIOS is starting the PC). Another short peak of 3,44 Amps can be seen later on when (a normal full version of) XP starts up.So a linear PSU of 3 Amps is only okay if it handles / tolerates short peaks of 3,5 Amps.
Too be on the safe side a Linear PSU which has to feed both the P4 and P24 should be cappable of supplying: 4 Amps.I also was curious so what the maximum Amp need of the system was. So I played a HD-movie (1920 x 1080) in a normal XP version with Windows Media Player Classic Home Theatre Version (1 Gb RAM installed). The PC than needs 3,75 / 3,85 amps for playing a high-definition movie
I also played the same HD movie in Windows 7 with Windows Media Player 11. In Win7 the PC needs 3,90 / 4.00 for playing a High Definition movie.So a linear PSU capable of providing 4 Amps is enough for the BIOS to start the system with setting 'Optimized Defaults'. It's also (just) enough for the system to play a high definition movie.
* PC system:
-> Gigabyte GA-G31M-ES2L, 1 RAM HyperX, E7300 Intel Processor.
* PSU’s:
-> the linear PSU 12V/5A powers: P4 = processor. P24 = MoBo, Lynx AES16 digital audio interface and integrated VGA on MoBO (1920x1080).
-> Antec 430 ATX PSU powers: USB, HDD and DVD.
fully AOB optimized cMP2 PC -> Lynx AES16 -> XLR AES/EBU -> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 11/12/09
nt
Will 3 amps still be enough at default speeds?
I posted some measurements with the Gigabyte G21 board back in April. They were buried deep in one of the cMP^2 threads but the key bit was:When playing flac-format music data without upsampling with the CPU clock at 140MHz, VID at 0.75000, PSU at precisely 12v, EIST off, 512 MB Kingston "ValueRAM" and a "Full Monty" cMP2 setup, the motherboard and CPU typically draw about 1630 ma.If you reset to defaults, the current draw will get quite close to 3 amps. Running flat out 24/7 with a 3-amp unit is cutting things a bit fine but, given you're going to be resetting fairly quickly, my guess is it'll be just fine.
HTH
Dave
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: