Buck Converter Switching Frequency - What is considered acceptable for audio amplifiers?

I'm always in need of various bits and pieces like 6.3V DC or 12V DC regulators for vacuum tube heater supplies, or +/-15V DC supplies for op-amps.

I see a lot of DC buck converters available at very low prices, from surplus sellers, eBay, etc. Like this one:
https://www.mpja.com/07-31-23.asp?r=393065&s=18

It says it's based on a XL4015 regulator. I looked up the data sheet, and it looks capable, with up to 7A out with heatsinking and a super low 0.3V dropout voltage. But I remain skeptical and I have zero experience building with switching power supplies, which is why I'm asking about it here.

Also, the XL4015 data sheet says it runs at a fixed switching frequency of 180kHz (144kHz min, 216kHz max). I wonder how fearsome of a switching frequency spike that is. It's low enough that I'd be hesitant to inject a big spike at that frequency into whatever preamp, etc. I happen to be building. But $2.95 for a pre-made regulator PCB is very tempting, especially since it shouldn't need as big a heatsink as an LM317-based regulator of equivalent current capacity.

Is a switching frequency of 180kHz too low?
Are there other regulator chips that are similar to this XL4015 but work with switching frequencies of >300kHz?

Is using a switching buck converter like this doomed to fail if used for a 6.3V 1A vacuum tube heater supply down from something like a 12V DC raw supply?
Does this type of switching buck converter need to be used with only about 1.5V stepdown in voltage, so that it doesn't go into shutdown from inrush current at power on?

Finally, should I have posted this in the Valves/Tubes section? I posted it here because it is a power supply question.
 
The nominal frequency is usually ultrasonic, which is good, but the devil is in the details - some switching regulators drop their frequency down to a fraction of the nominal on light load (usually by going into burst mode, with spradic bursts of oscillation), and this can generate noise in the middle of the audio band - that type of regulator are best avoided (its a strategy for high efficiency at low load).

If the switching frequency is unstable enough it can leak energy at lower frequencies - for instance a bang-bang (hysteresis driven) switching regulator's frequency is very dependent on load. I learnt this from using this method for a custom stepper motor driver, the motor's windings emit a constant rushing sound due to the very dirty spectrum you get from this asynchronous method.

If you have several regulators with different switching frequencies their frequency difference might be in the audio band, so any non-linear element could mix these down to produce an audible frequency.

The higher the switching frequency the less likely it is to break through, especially as this allows smaller inductors which leak less EMI into the surrounding environment. And rail filtering is more effective when the switching noise is a long way away from audio frequencies. So I'd suggest go more for modern high frequency converters if you can - but be religious about following the exact layout and component selections given in the datasheet if making your own PCB.

As a data point I've used a cheap 70W 12V-in, high voltage out switching regulator in the past for a valve circuit, and it didn't seem noisy with some basic output filtering - it was in its own separate mini-chassis/heat sink and mounted apart from the input section.
 
  • Like
Reactions: 4 users
For that price...

The problem there is shipping. $6.95 minimum. I guess it would be worth it if I bought 4 or so and loaded up on other miscellaneous parts (although I can't think of any parts I actually need at the moment).

If the switching frequency is unstable enough it can leak energy at lower frequencies...

The XL4015 data sheet (https://datasheet.lcsc.com/szlcsc/1811081616_XLSEMI-XL4015E1_C51661.pdf) boasts "Fixed 180KHz Switching Frequency". Does that mean it's designed to be unwavering, stable at that frequency (+/- 20% or whatever) or is that spec only attainable with constant current draw? Actually, tube heaters are pretty much constant current once warmed up, aren't they?
 
... some switching regulators drop their frequency down to a fraction of the nominal on light load (usually by going into burst mode, with spradic bursts of oscillation), and this can generate noise in the middle of the audio band - that type of regulator are best avoided (its a strategy for high efficiency at low load).

A 'synchronous buck converter', that uses two MOSFETs can operate at the same switching frequency and uniform (predictable) output ripple and losses anywhere between no-load and full-load.

If the load is light enough (e.g. 1A), then an LDO may also be used at the converter output to clean up its ripple. However, this LDO must have a wide control bandwidth comparable to the switching frequency.
 
I have some experience with XL4015, its a buck reg and as such has appreciable output ripple at 180kHz. I get rid of that with a 2nd order LC filter on the output. The switching freq is fairly stable, it doesn't vary with load.

XLsemi has other chips which work at higher switching frequencies but the ones I'm aware of aren't buck type, they're boost-SEPIC and they run at 400kHz. However they're lower output power than the XL4015.
 
  • Thank You
Reactions: 1 user
I had plans to use the 4016 as a tracking pre regulator for something like a 108x

You can use a pnp plus some passives to control the linear post reg dropout voltage.

There’s an interesting app note relevant to the topic bij Jim Williams AN101 that gives some clues on how to get rid of the MHz switching residue.
 
  • Like
Reactions: 1 user