SMPS in audio

[Rutherberg]

Hi,

More and more, SMPS are used in audio designs with success. What are the requirements and the key aspects for an SMPS to suits audio needs ?

- Which topology is the more adapted ?
- What is the minimum switching frequency ?
- As I can understand, switching frequency can vary with load in some topologies. I guess it's best to have a fixed switching frequency ?
- Aside from output power, is there different requirements between preamp and power amp SMPS ?
- Is it possible to do a better filtering of a noisy SMPS to adapt it to audio ? I'm interested in small SMPS modules (1 to 5W) like XPpower, Recom, MeanWell,... to power a preamp. I guess it is too noisy for audio.

Thanks in advance !

[Richard Crowley]

Your questions are rather broad and generic to be answered with anything specific.

As with most things, there are "audio applications" that are sensitive to power supply noise, and others that are less sensitive.
And there are SMPS that are noisy, and others that are quiet enough for most audio applications.
A switching frequency well above 20K would be very desirable for obvious reasons.
If the output ripple/noise is properly filtered, it should not matter whether the switching frequency is fixed or variable.
And since SMPS operate on super-sonic frequencies (vs 50-60 Hz power mains freq) it is much easier to filter HF ripple.

[Rutherberg]

More specifically, I would like to use SMPS in musical instrument amplification. Some circuits are high gain and very sensitive to PSU noise.
The advantage here is that the bandwith is more narrow than in HiFi (70 to 5-6kHz).
I've always had bad experiences with wall wart and generic smps, it is very noisy. Not hum but hiss and squeals. I tried to tame it with passive components, without success. I know it is possible, but didn't find it yet. BTW any hint would be appreciated.
That's why I would like to understand what are the differences between generic SMPS and the ones used in audio.

Thanks !

[Richard Crowley]

You can't really blame "hiss" on the power supply (of whatever type).  That comes down to circuit design and component selection.

And it is not clear whether "squeals" really has anything to do with the type of power supply, either.  Might it be more likely to be a power rail bypass/decoupling issue?

I would certainly handle circuit design/layout as a separate and independent issue from what kind of power source you use.  If the circuit has hiss and or squeals with a proper bench supply (or batteries) then it isn't really the power supply that is the source of the problem(s).

[Rutherberg]

Well, I know I'm in the beginners section...

But if I ask this, it's because I've tested several circuits with a "proper bench supply" and even SMPS designed for audio, and it works perfectly well, without noise.
I only get noise when I use several generic SMPS to power the circuit.
So I ask it again : what is the difference between noisy generic SMPS and silent SMPS that works well with audio, in this case a high gain preamp environment very sensitive to noise ? Switching frequency ? Filtering ? Topology ? All of this ? Is there a prefered topology for lower switching noise ?

Or am I misformulating the question ?

PS : by squeal, I mean high pitched ripple noise.

[basinstreetdesign]

I take it that when you say "generic" SMPS you are talking about one NOT designed specifically for audio use.  These supplies are probably designed for profit more than for performance.  So the designers probably put in little output filtering or abnormal operating mode suppression.

There is no such thing as a "silent" SMPS.  I have used a boost converter to create the HV for a couple of tube-based guitar products and the noise generated by the switching can be pernicious and infectious.  The noise finds its way into other circuits by conduction through the SMPS output and radiation from any/all inductors and conductors that handle the high voltage.

Any suggestions/arguments that the noise is easily filtered out because the frequencies involved are above human hearing are fallacious.  Even if the supply is working at a nominal 100KHz, its regulation will use frequencies that extend all the way down through the audible range, all the way to a few tens of Hz.  The feedback paths in the regulator will be modulated with a significant amount of random or not-so-random noise so as to also modulate, not only the output voltage but also the timing of high voltage/current switched signals with audible noise frequencies.  These high voltage signals will be broadcast through the air to every conductor within the unit, thus carrying the audible modulation there as well.  High frequency noise as well as the audio-frequency modulation noise conducted via the SMPS output will not only be directly injected into any circuit that it supplies but also take advantage of that power network as an antenna system to further infect everything else in range.

So, what to do?  The solutions are the usual.

Stop the noise at the generator:
1 - try to keep control of the sources by quelling unwanted PS operating modes such as undampened control loops, sub-harmonic oscillation, etc.
2 - try to keep radiated magnetic fields low with toroidal inductors.
3 - shield the entire (or at least the noisiest part of the) power supply with a shield can, ground planes, etc.
4 - filter outputs with R-C or L-C networks at the point where the output leaves the above shield can.

Kill the noise at the loads:
5 - apply more filtering on supply rails at the point of use using R-C's, L-C's, ferrite beads.  This can be applied on a module, channel or individual load basis.

If there is ANY high pitched audible noise directly to be heard from a purchased SMPS then its not for you.  That noise is either from a switcher which is operating from a too-low frequency or it is operating in a bad sub-harmonic mode.  That noise will infect any amplifier that is not designed to withstand it.

The most paranoid measures must be applied for high-gain amplifiers.  Put them in their own shield boxes, on separate PCBs, more filtering, whatever you can do to isolate it from inevitable noise pickup.

That's all I can say.

[Audioguru]

The current used by a preamp is almost nothing so why use a SMPS for it? Switched mode is used for a high current power amplifier so that much of the power supply current is not wasted by making heat.
A wall wart cannot cause a preamp to make a squealing output, maybe it is acoustical feedback from the speaker to the microphone?

[BrianHG]

My problems with SMPSP have always gone down to 2 things:
1.  Conduction to the mains AC, where isolation transformers would fix this problem, I'm left with I might as well have just used a transformer in the first place.
2.  When amplifying a coil based MIC or Pickup, the coils in these devices also pickup the emi from the high frequency switching in the switch-mode supply and thanks to harmonics, even if the switch-mode supply was really high frequency, it still created noise at many other lower audible frequencies.