ATX12v EPS to 5v Possible? \ VOGONS

ATX12v EPS to 5v Possible?

Topic actions

Post a reply

First post, by ptr1ck

Posted on 2022-02-23, 04:05

ptr1ck Offline

Rank Member

Rank: Member
Posts: 288
Joined: 2022-02-18, 01:03

I have been thinking of how to use a modern style power supply to properly power old equipment. I had a light bulb moment the other day when reading some old threads, specifically this post: Re: New solutions to the ATX -5V problem

My idea is to use a common 24 to 20 pin adapter, an EPS connector (or maybe PCIE) and a 12v to 5v 40a step down converter. Terminate the 5v at the 24 pin connector. 12v in from a modern power supply's CPU or PCIE connector will feed the step down converter. The output of the converter connects to the 5v pins on the 20 pin ATX connector to motherboard.

Thoughts:
12v power from modern supplies is abundant and under used on an old system.
The EPS connector is useless on most anything pre-P4 and Athlon XP (some XPs).
Would this be too inefficient or noisy?
By freeing up the rail shared by 3.3v/5v now by allowing it to be 3.3v only, would that allow more wattage to be dedicated to 3.3v or just prevent possible brown-outs?

It would be about $50 in parts to build, but would also be modular in that it's not a permanent modification to the motherboard or power supply. If an electrical engineer type person would chime in, it would be greatly appreciated.

"ITXBOX" SFF-Win11
KT133A-NV28-V2 SLI-DOS/WinME

Reply 1 of 5, by pancakepuppy

Posted on 2022-02-23, 06:28

pancakepuppy Offline

Rank Newbie

Rank: Newbie
Posts: 91
Joined: 2019-12-26, 06:30
Location: NW USA

I designed and built something similar a while ago (producing both 5V and 3.3V from 12V) using 30A fully integrated synchronous regulator ICs. Definitely helped level out the voltages versus plugging the group regulated ATX12V supply directly into the Athlon XP board, but I probably should have went with a 2 phase controller and discrete FET setup for the 5V to beef it up to 40A+. I don't have a real clear idea of what 3.3V current load is like on a 5V-heavy AGP system so the 3.3V half of my board was probably unnecessary, that's something I need to measure in my system before whipping up a new design. Maybe even add -5V generation with a fun inverting switcher circuit like a Ćuk converter, LM2611A or something. Many fun possibilities for power circuit design.

Reply 2 of 5, by darry

Posted on 2022-02-23, 08:20

darry Offline

Rank l33t++

Rank: l33t++
Posts: 5805
Joined: 2014-01-20, 06:27
Location: Canada

I am no expert on this, but if the plan is to use something like in the below link, care should taken about ripple. ATX max allowed ripple is ~~120mv~~
50mv on the 5v line, whereas the below is rated at 200mv .

https://www.amazon.ca/Cllena-Converter-Regula … r/dp/B07VGG5YQ3

Last edited by darry on 2022-02-23, 22:03. Edited 1 time in total.

Reply 3 of 5, by ptr1ck

Posted on 2022-02-23, 16:41

ptr1ck Offline

Rank Member

Rank: Member
Posts: 288
Joined: 2022-02-18, 01:03

darry wrote on 2022-02-23, 08:20:

I am no expert on this, but if the plan is to use something like in the below link, care should taken about ripple. ATX max allowed ripple is 120mv on the 5v line, whereas the below is rated at 200mv .

https://www.amazon.ca/Cllena-Converter-Regula … r/dp/B07VGG5YQ3

That's an excellent point. I have seen a few 50a units with ratings at <120mv. Would a higher amp rated converter allow for lower ripple if it's not running at capacity?

"ITXBOX" SFF-Win11
KT133A-NV28-V2 SLI-DOS/WinME

Reply 4 of 5, by darry

Posted on 2022-02-23, 22:05

darry Offline

Rank l33t++

Rank: l33t++
Posts: 5805
Joined: 2014-01-20, 06:27
Location: Canada

ptr1ck wrote on 2022-02-23, 16:41:

darry wrote on 2022-02-23, 08:20:

I am no expert on this, but if the plan is to use something like in the below link, care should taken about ripple. ATX max allowed ripple is 120mv on the 5v line, whereas the below is rated at 200mv .

https://www.amazon.ca/Cllena-Converter-Regula … r/dp/B07VGG5YQ3

That's an excellent point. I have seen a few 50a units with ratings at <120mv. Would a higher amp rated converter allow for lower ripple if it's not running at capacity?

Sorry, I was thinking of max ripple on the 12v rail when I wrote 120mv . 5v and 3.3v rails are each allowed up to 50mv ripple .

I corrected my previous post accordingly.

Reply 5 of 5, by ptr1ck

Posted on 2022-02-24, 04:11

ptr1ck Offline

Rank Member

Rank: Member
Posts: 288
Joined: 2022-02-18, 01:03

darry wrote on 2022-02-23, 22:05:

ptr1ck wrote on 2022-02-23, 16:41:

darry wrote on 2022-02-23, 08:20:

I am no expert on this, but if the plan is to use something like in the below link, care should taken about ripple. ATX max allowed ripple is 120mv on the 5v line, whereas the below is rated at 200mv .

https://www.amazon.ca/Cllena-Converter-Regula … r/dp/B07VGG5YQ3

That's an excellent point. I have seen a few 50a units with ratings at <120mv. Would a higher amp rated converter allow for lower ripple if it's not running at capacity?

Sorry, I was thinking of max ripple on the 12v rail when I wrote 120mv . 5v and 3.3v rails are each allowed up to 50mv ripple .

I corrected my previous post accordingly.

OK wow. I'm surprised most old power supplies would have been that tightly regulated, even when new.

"ITXBOX" SFF-Win11
KT133A-NV28-V2 SLI-DOS/WinME

Go to top of page Go to top of page

Back to General Old Hardware

Main menu

Common searches