Home Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Linear Power Supplies for ATX & P4

I wanted to start a separate sub-thread pulling together the info on this upgrade path. I'd like to try and prevent having this info hidden across a number of other sub-threads with seemingly un-related topics.

My goal is to have a single place to go and get the info on building and/or assembly a linear supply for your computer. As we have seen in the past, power supply improvements translate directly into sound quality improvements... whether chosing a 'good' standard computer supply, installing additional filtering on that supply, building a 'hybrid' supply with a linear supplying a PicoPSU, or now fully linear supplies.

There have been some very interesting developments regarding fully linear supplies over the last 6 months. The first I'm aware of was Gene's linear-supplied computer using off-the-shelf linear power supplies. He also demistified the ATX startup sequence, at least for some boards. That thread is here:

http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=65198


Then there was Mihalov's truly herculean linear supply described in this thread:

http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=72756


Then Jackwong96's battery-powered linear supply that he described here:

http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=74498


Inspired by Gene's, I started pulling together the parts to build a linear ATX/P4 supply back in December. I liked his use of a 9v battery to supply the -12v needed for startup, but not needed afterwards if one is not using a Juli@'s analog stage powered from the motherboard.

My first tries were unsuccessful. I initially tried it with a GA-EG45M-UD2H board and never got a successful startup... although I was lucky and never damaged the board.

Not wanting to risk one of my small supply of the 'magic' GA-G31M-S2L boards, I next tried an ES2L version... and varying the sequencing of how I powered up the -12v(-9v), the main supplies, and signaling POWER_OK, I did finally get successful bootups... sometimes.

It was not until Jack posted details of his supply and then I went back and re-read Gene's that I found the startup sequence that works ALMOST all the time for either an S2L or ES2L board:

1st - Apply power to the -12v (-9v) rails AND to the 5vSB (By flipping a switch that connects one 9v battery to the -12v and another 9v battery to a regulator chip for the 5v control voltages... 5vSB is energized at this time)

2nd - Apply power to the 12v P4 and the ATX 3.3v, 5v, 12v (By plugging in the main supply AC)

3rd - Press power button

4th - Apply 5v to the ATX PWR_OK (By flipping a switch that applies 5v to the POWER_OK line)

Note that this is done with just two switches on the power supply, no relays or anything like that.

After the board starts powering-up, I flip the two switches (-12v(-9v)/5vSB and POWER_OK) off to conserve battery life on the 9v's. This does not effect running and so far, I hear no sonic difference with these switches on or off.

Also note that on the S2L and ES2L boards, if going into the Bios for modifications, if you flip the switch that controls the POWER_OK off just before you exit the Bios, then after a second or two, flip it back on, the board will reboot ok. Otherwise, you have to power down the main supplies and do the power up sequence again.

This setup still does not work with the GA-G31M-UD2H board, with it's more sophisticated startup sequence. This may also be true for other more-recent Gigabyte motherboards... I am just guessing here, but I suspect the reason that Jack gets a good startup on these with his battery supply is due to using the PicoPSU to intiate and control the startup. This needs more investigation.

In any case, I've attached a few pictures of my supply. I'll give more details as I get it debugged... I'm currently testing it in my system and finding some significant sound quality gains, BUT I'm having issues with it overheating and a regulator shutting down... either the 5v or the ATX 12v (which was very close to the 5v on the heatsink). I've just restructured it a bit with the existing heatsink to try and keep the chips a bit cooler. If this doesn't work I can move the ATX 12v and 3.3v to their own heatsinks while sticking with the same physical package, which is designed to fit inside of my Zalman case. I have several other avenues to pursue before relenting and using a larger heatsink on the ATX 5v and P4 12v lines, the current-hogs. But when I do that, I'll likely need to move it outside of the Zalman, something I'm trying to avoid doing for now.

In any case, I'll continue adding info on my supply and pull some of the details of the other's into this thread as time goes on.

And if you have questions on these or want to share your experiences with fully linear computer supplies, please feel free to post here!

Greg in Mississippi






Everything matters!



Edits: 05/31/10

This post is made possible by the generous support of people like you and our sponsors:
  Analog Engineering Associates  


Follow Ups Full Thread
Follow Ups

FAQ

Post a Message!

Forgot Password?
Moniker (Username):
Password (Optional):
  Remember my Moniker & Password  (What's this?)    Eat Me
E-Mail (Optional):
Subject:
Message:   (Posts are subject to Content Rules)
Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:  Automagically notify you when someone responds.