|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
217.186.111.228
In Reply to: RE: cMP - the open source high-end Memory Player posted by cics on December 30, 2007 at 05:42:01
dear forum,
can anyone explain which parts of a motherboard uses which voltage(es) ?
PCI uses 3.3 V for example for the soundcard and certain other processors/chips do presumably use all individual voltages.
we are in process of putting together a linear PSU and would like to know which certain areas of the motherboard are consuming which part of the P24 ATX connection.
in michailov´s PSU project there is no direct advice of which "ground" connection is part of which power circuit.
or in other words : given we have three linear supplys (3,3V,5V &12V) do they all use the same "ground" ?
how to assemble the P24 ATX connector correctly -that´s the big question.
please help, if you can.
kind regards
Follow Ups:
The no-longer-active and lamented JackWong96 seemed to have the best grasp of how power is used on motherboards. He describes some of that in this post:
http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=75904
But noone has yet performed and published a comprehensive survey of all of the voltage consumers on the recommended motherboard(s) AFAIK, sorry.
I do suspect that 3.3v is only being used to power some of the less-critical circuits in the MoBo, but I am not sure if you'll get a good boot without that voltage. Worth a try... Mihaylov, Bibo01, or others, have you tried booting one of these boards with the 3.3v not active? JackWong96's experiments with removing and/or un-powering various peripheral chips did strongly suggest that the board will not be upset if at least some of the non-core chips were not active, so you might get away with this. I do know that with the GA-G31M-S2L mobo, I am only supplying about .3ma on the 3.3v line, so it is lightly used at best. JackWong96's experiments suggested that a number of the uses were non-critical chips (like the DVI to HDMI converters) and removing them and similar chips nearly halved the power consumption on that line.
Then on grounds, as Mihaylov has indicated, all of the grounds connect together at the ATX-24 socket on the motherboard (at least on the Gigabyte boards we are using... I haven't looked at every board out there!). That is also true for each of the multiples of the same voltage rail... for example, all the 5v connections on the ATX-20/24 come together at the socket on the mobo.
One technique I've seen recommended is to pair each voltage line with a corresponding ground line to minimize the radiated noise and magnetic fields. On my supply, I used twisted pairs or braided triples or quads for this. I used a 20-pin plug (see attached picture for a good diagram)
I paired up the following:
- For +5v, I paired up the grounds on pins 16 & 17 with the +5v on pins 19 & 20 into a 4-cross braid. I also paired up the grounds on pins 5 & 15 with the +5v on pins 4 & 6 for a 2nd 4-cross braid.
- For +12v, I paired up the ground on pin 7 with the +12v line on pin 10 as a twisted pair.
- For +3.3v, I combined the +3.3v lines 1, 2, & 11 into one wire, then twisted that with the ground on pin 13.
- For the +5v PWR_OK & +5VSB on pins 8 & 9 respectively, I used the ground on pin 3 with them to make a 3-wire braid.
Of course, feel free to make your own choices for pairing wires up if you decide to do this.
Bibo01, thanks very much for the Molex part numbers. I used a scavenged ATX-20 connector for my 1st supply, but I'd like to do it with a virgin connector for my next. You've made that easier!
BTW, to confirm something I said in my monster Juli@ mod thread awhile back, that soundcard does not use the 3.3v from the motherboard. This doesn't appear to be useful for anyone involved in this sub-thread now, but might be useful to someone else making their own linear supply another time.
Best of luck to you on this. I think you'll find it very rewarding!
Greg in Mississippi
P.S. I've been corresponding separately with another inmate who has posted here very little. He worked up a linear supply for his cMP and it had easily adjustable voltages on the 5v and 12v lines. He played with them and found that lowering them by something in the range of .2v -.4v produced SQ benefits. I know he's been otherwise-occupied and I'll let him tell his story when he can, but based on his experiments and similar accounts others posted here in the past, I tried it and found it worthwhle too.
Everything matters!
Hi Greg!
I have checked up. The system board GA-H55M-UD2H does not start without +3.3V.
Tell me please why you attach foil slices on capacitors and SATA cable?
Edits: 10/30/10
Mihaylov,
Thanks for checking. Dang, it would have been a nice simplification to the linear supplies.
As for the foil-ish things, those are damping materials.
I'm very big on vibration control.
I can't say that I heard any diff from the material on the SATA cable.
And didn't try those caps with and without.
But other cap, case, and heatsink dampings have been very audible. Additional damping on my amps (on the clips that attach the devices to the the heatsinks) was what made B38 preferable to B37 in my system.
The foil-looking stuff is EAR material available from Michael Percy. I also use Dynamat. I generally prefer Dynamat on cases & the EAR mtl on components.
On my linear supply, all the caps, transformers, and rectifiers are damped. Also the structural materials were chosen for vibration control... bamboo mounting structures with brass hardware (no steel hardware anywhere in supply). Bamboo bases were also damped with combo of Dynamat Xtreme & regular.
My Zalman case is also extensively damped, as are my other supplies and the heatsinks on the motherboard. Also my amps, my volume control, and speaker crossovers & structure.
Much more to try, but this has been a fruitful excursion over time.
Must listen to the effect when doing... when I damped the caps in the linear supply, first used Dynamat Xtreme... great clarity and extended & strong bass, but tunefulness was impared. Replaced with EAR damping, very slightly less clarity & bass, but very tuneful by comparison. And I hadn't expected any changes with that damping as the caps are already mounted to the bamboo bases with a resiliant double-stick tape.
Later!
Greg in Mississippi
P.S. If you haven't tried these, I can strongly recommend these two tweaks to your linear supplies:
1. Add another supply for providing the control voltages instead of taking them off the ATX-20/24 +5v.
2. Seperate the AC input to the P4 & peripheral supplies and take them to a separate AC circuit in your listening room as per cics recommendations.
Everything matters!
Thank you Greg! About two tweaks. I see second tweak and I think that is very interesting to try. But I don't understand first tweak. What is "control voltages" and why it is necessary to use a separate power supply unit for them?
Mihaylov,
Very welcome.
On that 1st tweak, the control voltages are the +5v for the +5VSB & the PWR_ON signaling.
When I first setup my linear supply, I saw that the motherboard didn't appear to need those voltages (along with the -12v) after it had booted, so I supplied them with a battery and turned them off after boot.
When I was troubleshooting why my system would not stay running, I decided to try to make those voltages permanent. I put in small AC transformers along with diodes, filter caps, and regulators for both the +5v control voltages and the -12v. Doing this fixed the issue with it shutting off randomly... and after we realized that the -12v was not needed for anything but the GA-G31M-ES2L board, I pulled that supply out.
Later, I thought it might sound better to supply the +5VSB & PWR_ON from the main +5v ATX-20/24 supply due to a simpler setup and grounding. When I wired that up, I lost some clarity and individualization of sounds. So I've gone back to the separate supply for _5VSB & PWR_ON.
The difference, while not huge, was very definite & positive.
BTW, I didn't notice any difference between supplying the +5v control voltages from the small AC supply or a battery, but the battery was only a 9v battery, so a better battery might be a different matter.
This leaves me with 5 separate AC tranformers in my linear ATX-20/24 & P4 supplies! Gotta lotta iron!
Later!
Greg in Mississippi
P.S. On the motherboard, the control voltages are separate from the main +5v ATX-20/24 line. When I lowered my +5v down to +4.6v or so, it didn't cross over to the +5VSB & PWR_ON.
Everything matters!
Thanks! I see.
I am building an ATX linear supply following Mihaylov's project.Although I am sure Mihaylov can add to me, I believe that the P24 supplies all share the same ground.
One word of advice -
The latest Gigabyte MB reccomendation does not need -12V for startup. However, knowing that you employ a LynxTWO-B card like me, this card requires:
+5V 600mA
+12V 500mA
-12V 150mATherefore, no need of +3.3V for it, but all 4 linear supplies are required, including -12V. I am also evalueting the possibility of double consecutive stabilization on +12V circuit as well as +5V.
Re. P24 connector, it is possible to make such connector from scratch but you require the right tools to crimp. So, I went to my local computer repair shop that throws away dead ATX PSUs every other day and cannibalized one for P24 and P4 with cables. For device cables+SATA conenctors I had one left (actually a few) from my last PSU which had detachable cables.
Eventually, if needed, the P24 connector is MOLEX 39-01-2240; the P4 connector is MOLEX 39-01-2040. You can search for those numbers at, for ex., RS Online.
Any connector explanation can be found here: http://pinouts.ru/pin_Power.shtml
Edits: 10/29/10
"I believe that the P24 supplies all share the same ground" - it's true!
Pay attention on the page 6 from http://www.cicsmemoryplayer.com/index.php?n=CMP.ApdxBAdvanced?action=downloadman&upname=B_Linear_PSU.pdf:
"The terminal block (position 9 in the fig.1) is the center of the PSU’s «ground star». The ground buses from all three circuit boards (are soldered by a thick wire to filtering capacitors of rectifiers – see the position 8 in the fig.1) converge here and diverge to the power consumers."
Bibo, whence you have known that +3.3V is not required for power supply of Lynx Two?
I have known that +3.3V is not required for power supply of Lynx Two directly from Lynx support.
Also, even if one might use analog side of the card only, all three supplies are required.
http://www.lynxstudio.com/forum/forum_posts.asp?TID=3421
Edits: 10/29/10
Thanks!
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: