PSU Noise on Output Voltage Lines

[Signal]

I've been chasing down an odd problem in which the PSU comes into play.

I starting looking at the output voltage lines on a scope and noticed some small spikes and even a harmonic with a Enermax Liberty 620W.

I replaced the PSU with a PPC 610EPS12V and don't see quite the same noise, but still see some suspected variances.

I realize that the general nature of a switching power supply that a small amount of noise on the output voltage is expected.

Also, with the PSU connected to the motherboard I assume that the motherboard itself might throw some noise onto the lines as well.

Also assume that the cap quality and overall design of how they are used also comes into play on smoothing out the noise.

But how much is considered "normal"?

Also, if the PSU is rated for a higher wattage output but the system draw is say 30-40% would the efficiency of the power supply affect the amount of noise and voltage stability on the lines?

[Signal]

Re: PSU Noise on Output Voltage Lines

Apologize for the poor wording above. Got distracted during my edit and can't change it now.

[Sparky]

Re: PSU Noise on Output Voltage Lines

How much noise are we talking about here? The ATX specs allow for up to 120mV of ripple noise. If you aren't seeing stuff beyond that then I wouldn't worry about it.

[Signal]

Re: PSU Noise on Output Voltage Lines

The general high frequency noise seems to be within reason (50mv) however I was seeing a harmonic where the hf noise varies into a wave form and hash spikes show up on the Liberty.

I don't see the harmonic on the PPC, but it still looks like hash spikes are there.
It's not a digital or LCD type of scope so it's tough to get a picture for an example.

[linuxguru]

Re: PSU Noise on Output Voltage Lines

On constant load and constant primary input voltage there should not be any harmonics in the output ripple (units with Active PFC may be an exception).

However, a PC motherboard is not a constant load - the processor keeps switching into different power-saving states, drive motors keep turning on and off, etc. The PSU's control loop will keep
adjusting the primary side duty-cycles, etc., and you will see changes in the ripple waveform on the secondary side. Depending on the time-constants in the control loop, the control loop may exhibit ringing (a quick burst of oscillation with overshoot/undershoot, generally considered undesirable) in its transient response to a load change - this may show up as harmonics in the output ripple.

It's generally inadvisable to modify the control loop - you could end up with further instability leading to oscillations, etc. However, you can try recapping the secondary caps with up to 50% larger low-ESR capacitors and see if the amplitude of the spikes decreases.

[starfury1]

Re: PSU Noise on Output Voltage Lines

The liberties have been known to "blow up"
(for want of the explanation of the exact nature of how they destroy them selfs and why)
under heavy load
there was a review at jonny guru's but I couldn't find it unfortunately
this may have given you a bit more insight as to what going on with the liberty.
(they are fairly in depth reviews on psu's)

from what I read I wouldn't push a Liberty pass 2 3rds of its rated max, at best to be safe
(there is a post here on popped caps in one driven fairly hard for about a year, oh I see you found it...search here there maybe a few more I think Gal posted about a blown one a while back)

Me, well linuxguru and a few others round here are far more up with how switchers work
so I'll leave it to them for the technical comment

I've been chasing down an odd problem in which the PSU comes into play.

If I may ask whats the "Odd problem" out of curiosity?

no help but just beware of Liberites...don't know if Enemax corrected what may have been the issues with this model

Cheers

BTW I'am using one but at best id say it coasts along in this box

[Galvanized]

Re: PSU Noise on Output Voltage Lines

I am not saying go out and buy one, but this is what jon over at jonnyguru.com uses to check ripple for the PSU reviews.
Seem to recall that it costs around $160.

http://usb-instruments.com/oscillo_stingray.html

[Tom41]

Re: PSU Noise on Output Voltage Lines

That's nothing - when measuring the output of an Apple iBook power supply cable, I saw a trace with 'spikes' moving along it. It was as if the trace was a conveyor belt, and the spikes were things on the conveyor belt moving slowly from one side to the other.

And to top it all off, when I measured my Mercury PSU in standby, I got a SPEECH waveform appearing as well as the voltage/ripple trace! I can only assume something was acting as a radio receiver and picking up a broadcast of mainly speech. Perhaps it had something to do with the vented cap in the PSU?

[999999999]

Re: PSU Noise on Output Voltage Lines

There is no harmonic that matters. Either the unit is staying within ATX specs (against which, motherboards must be designed to operate properly), or it isn't. The secondary concern is if it has a reasonable lifespan remaining or it would be better to pull it and repair (recap?) or replace. Certain signs like brand and model of caps or fan(s) may give hints about what, if any, changes are necessary for best expectation of longer remaining life.