Is the flatness of DC output of a typical ATX power supply unit worse than a power adapter?

Question

I converted an ATX power supply unit (Seasonic, 330W) into a standalone power adopter (by adding load, etc.), to which I connect multiple things. When I plugin a computer and a monitor each to a 12VDC line, they work fine, but when I also plugin an audio speaker to another 12VDC line, it gives a noise that is at an unusually bad level. If I, instead, just plugin the speaker to an ordinary AC-DC switching power adopter, then, the noise dissapears. I was wondering if the flat level of the DC output of an ATX PSU is that bad. I had though that switching adaptors should be worse than a power transformer (which is used in ATX PSU) with respect to flatness, but the reality seems to be the opposite. Am I wrong?

Answer 1

The ATX power supply is a switching power supply. It's important to be sure of the difference - a switching power supply switches the voltage at a much higher frequency (e.g. 65kHz) so smaller transformers can be used (transformers work more easily at higher frequencies) The downside is high current switching can easily cause EMI problems if not dealt with properly, and often in cheap power supplies (e.g. ATX) the filtering is very basic and may not always meet regulations.

A "standard" supply uses the 50Hz or 60Hz mains frequency to change the voltage, so the transformer for the same amount of power would be much larger. The supply size will likely be over twice as big for the same rated output. However with low frequency and (usually) linear regulation, these supplies are usually quieter. Are you absolutely sure your "AC/DC" supply is a switching supply?

It is relatively easy to make a quiet supply using either design though, so you may find a well made SMPS (can include linear regulation after initial switch regulation) is quieter than a cheap (possibly unregulated) standard supply.

I think your noise may well be present all the time, and causing issues with the sensitive (active I assume :-) ) analogue speaker circuitry. It may be (as mentioned previously by Madmanguruman) from the PC/monitor plugged causing problems with regulation, or could be a grounding problem of some sort. Depending on the frequency (be good to know roughly how high/low it is) it would be easier to guess at what is causing it. An scope on the lines would tell the full story.

Answer 2

We have a few ATX units converted in the office, we didn't try with an big external load but loads we use are around 3W. When we build the things we looked at the output with a scope in AC coupled there was noise (more than bench supply) however it was at an acceptable level. Perhaps only 30-40% more then bench supply for a given voltage output.

Sounds like speaker is not well designed as some high frequency noise creeping back into the output. In this case, I guess supply noise may be the issue, but the speaker should be able to cope with it, it sounds like it is not well designed.

Answer 3

ATX supplies specify their ripple to be 1% of the rail voltage. What I suspect is happening is that the other loads are disturbing the rail and the speaker is picking up the disturbances. Even though there may be multiple connections to 12V, I suspect that there's only one feedback loop for all the feeds, so the noise coming from the other loads is enough for the speaker to transduce.

Try the speaker on the ATX supply without the monitor and/or computer and see if there's an effect.

Answer 4

I assume you are refering to noise on the audio output of active speakers. The power supply itself could also directly produce audible noise, which is a different problem.

The noise could be caused by a ground loop.
In this case, the difference of the two power supplies could be that the output is connected (or not connected) to ground in a different way.