EEVblog® Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: gnif on October 18, 2013, 02:15:41 am

Title: How much ripple for a ATX PSU is too much?
Post by: gnif on October 18, 2013, 02:15:41 am
Hi All

My PC decided to start acting up the other day, random rebooting, etc... through a process of elimination I managed to identify the PSU as the fault. Now this is no cheap PSU, it is a Corsair HX1050 that feels like it is built well, and from what I can see past the fan, looks of decent quality. The device is still under warranty, so I am not going to open it, but I decided to power it up and stick it on my scope (which I am still learning to use).

I am seeing the switching on the output on all the voltages at around 20mv peek to peek every 14.5us, but on the 12V rail, I am seeing a 400mv spike. Now this PSU is on my bench and not under load, so obviously this is not a fair test. Is this normal or is this outside what would be expected for an ATX supply? or am I measuring this completely incorrectly because the PSU is not under any kind of static load?
Title: Re: How much ripple for a ATX PSU is too much?
Post by: kizzap on October 18, 2013, 03:09:33 am
Are you bandwidth limiting the scope at all?

According to wikipedia (http://en.wikipedia.org/wiki/ATX), the acceptable ranges of ripple voltages up to 20MHz specify that ripple should be less then 120mV, so your supply seems to be outside of that specification.

It might be worth seeing if it causes issues in another system?

-kizzap
Title: Re: How much ripple for a ATX PSU is too much?
Post by: jaxbird on October 18, 2013, 03:14:13 am
You can find the max allowed ripple in this design spec:

http://www.formfactors.org/developer/specs/ATX12V%20PSDG2.01.pdf (http://www.formfactors.org/developer/specs/ATX12V%20PSDG2.01.pdf)

Max 120mVpp (20MHz) for the 12V.

It would make most sense to measure it under load, at least 100W.

Title: Re: How much ripple for a ATX PSU is too much?
Post by: BravoV on October 18, 2013, 03:16:27 am
Make sure you turn on the 20Mhz BW limit before measuring the rail.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: Nerull on October 18, 2013, 03:36:58 am
Switching PSUs do not perform well without a load. Put a dummy load on it and see if it still does the same. My attempt to convert an ATX power supply into bench PSU had over 1V of peak to peak ripple before I stuck a power resistor across the 5V rail. Then it went flat.

Newer ATV12V PSUs need the 12V rail loaded instead, I believe.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: gnif on October 18, 2013, 03:54:25 am
Quote from: kizzap on October 18, 2013, 03:09:33 am
Are you bandwidth limiting the scope at all?

Yes, 20Mhz.

Quote
It would make most sense to measure it under load, at least 100W.

I don't have a static load I can put on it at the moment, and procuring one is a bit difficult since at the moment the roads into and out of where I live are blocked due to fires.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: kizzap on October 18, 2013, 04:23:35 am
Quote from: gnif on October 18, 2013, 03:54:25 am
I don't have a static load I can put on it at the moment, and procuring one is a bit difficult since at the moment the roads into and out of where I live are blocked due to fires.

Chuck as many HDDs that you can on, the older the better
Title: Re: How much ripple for a ATX PSU is too much?
Post by: Vgkid on October 18, 2013, 04:50:48 am
The unit uses good caps based on the review, but even new it had a higher ripple on the 12v rail. Maybe Phaedrus will chime in.
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245)
^^ Review^^
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245)
^^Under Load^^
Title: Re: How much ripple for a ATX PSU is too much?
Post by: gnif on October 18, 2013, 05:58:02 am
Quote from: Vgkid on October 18, 2013, 04:50:48 am
The unit uses good caps based on the review, but even new it had a higher ripple on the 12v rail. Maybe Phaedrus will chime in.
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245)
^^ Review^^
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245)
^^Under Load^^

Well, that is certainly not what I see on my 12V rail, will RMA the thing.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: wraper on October 18, 2013, 12:18:17 pm
Quote from: gnif on October 18, 2013, 05:58:02 am
Quote from: Vgkid on October 18, 2013, 04:50:48 am
The unit uses good caps based on the review, but even new it had a higher ripple on the 12v rail. Maybe Phaedrus will chime in.
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=245)
^^ Review^^
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245 (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story3&reid=245)
^^Under Load^^

Well, that is certainly not what I see on my 12V rail, will RMA the thing.

You forget ^^Under Load^^
How you can compare those pictures with what you get without load.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: Kjelt on October 18, 2013, 02:42:34 pm
You can use one or two of the high power halogen head light lamps (55W) of a car as a load.  ;)
They get pretty hot so mount them safely.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: dfmischler on October 19, 2013, 02:40:05 am
IIRC the ripple on an ATX supply is spec'ed right at the board (not the connectors) with a short ground lead.

EDN: Testing a power supply - Noise (http://www.edn.com/design/power-management/4411821/Testing-a-power-supply---Noise--Part-2--)
Title: Re: How much ripple for a ATX PSU is too much?
Post by: NiHaoMike on October 19, 2013, 05:40:29 am
A common problem is that consumer PSUs tend to have weak fans, causing the caps to fail early. An easy workaround (without voiding the warranty) is to externally supply better ventilation (such as pressuring the case), but the proper fix is to spend a little extra for a server grade PSU.
Title: Re: How much ripple for a ATX PSU is too much?
Post by: codeboy2k on October 19, 2013, 05:57:37 am
Quote from: gnif on October 18, 2013, 03:54:25 am
I don't have a static load I can put on it at the moment, and procuring one is a bit difficult since at the moment the roads into and out of where I live are blocked due to fires.

You mean you are trapped and encircled by fires??   :scared:

Don't you think you should be building a personal quadcopter, and getting yourself out of there?? The PSU can wait...