Coil whine of (meanwell) SMPS and also graphics cards.

[rock soderstrom]

I am currently researching the meanwell offer for suitable (dual) SMPS for a project and have read quite a few reviews on Amazon that report disturbing coil whine. This also seems to be related to the load, less load seems to go hand in hand with more whine.

What are your experiences?

Additional question: My Geforce graphics card has also developed an annoying coil whine over time. Can anything be done against it?

 

[mkiijam]

We used to solve coil whine on CRT TV's with dipping them in varnish or from spray can.

 

[JohnRoberts]

This may just be coincidental but about 20 years ago I inherited a design to consult upon (1RU DSP). It had a clever two voltage output DC to DC switching power supply. Instead of the typical one output voltage single inductor design, it used a switching transformer to generate two different regulated output voltages. For whatever reason it had an annoying hum/buzz, I tried many tweaks to the circuit but nothing got it quiet. I searched for an alternate source for the HF switching transformer with zero luck. I redesigned the PS (that I inherited) to be a single output switcher, with a 3 terminal regulator to generate the second lower voltage. It was less efficient but dead silent.

The quiet single inductor was potted, the buzzy transformer not and made too much noise to use in a premium rack unit.

JR

 

[GeorgeToledo]

I’d probably take it out and see how the thermal paste has held up. If it looks like it is degrading I would put some new paste on it.

There are probably other things that could be done but that’s my first thought.

 

[k brown]

Just wondering if thermally conductive pads might provide more damping than paste.

 

[rock soderstrom]

Thanks for the tips guys, I'll try that out on my graphics card when I get back from vacation.

As for the Meanwell power supplies, I was surprised after reading the reviews on Amazon that some of them have the problem right out of the box, whether single or dual solution.

That's why I was surprised because some people here seem to be using this successfully in studio equipment and it would certainly be considered particularly annoying here.

But there don't seem to be any problems with that, at least nobody is talking about it.

In the Amazon reviews, some people found the Meanwell coil whine too annoying even when used (behind the interior paneling) as a room lighting PSU...

I'll buy some Meanwell modules when I get back and see if it's really a problem, I hope not. My recent experiment with SMPS in audio gear has been very promising, I'd like to expand on that.

 

[k brown]

"Meanwell Coil Whine" - great name for a band.

 

[rock soderstrom]

"Meanwell Coil Whine" - great name for a band.

Yes for a US-Japanese-German avant-garde band with 6 guitars, no bass and two drummers. But the main instrument is a Coil Whinophone!

 

[mrclunk]

We used to solve coil whine on CRT TV's with dipping them in varnish or from spray can.

Me too, yacht varnish, warm part up in oven before dipping.

 

[rock soderstrom]

We used to solve coil whine on CRT TV's with dipping them in varnish or from spray can.

This sounds like a good plan for larger coils, perhaps it can also be applied to the rather small SMT(?) coils of a graphics card. I'll have to take a closer look at which components make the sound.

 

[jp8]

Also, keep an eye out for noisy MLCCs.

https://www.ti.com/lit/pdf/ssztb09

 

[gyraf]

I'd consider it mainly a historical problem: With the maturing of the technology, most recent-design smps psu's depend on switching frequencies far beyond hearing limit, as higher frequencies comes out more efficient for amount-of-material-used

Usually only annoying if you have two very loud similar ones interfering..

/Jakob E.

 

[jp8]

Not so under light load conditions, where the converter might switch to intermittent bursts. This is also reported in the first post by @rock soderstrom

 

[JohnRoberts]

I'd consider it mainly a historical problem: With the maturing of the technology, most recent-design smps psu's depend on switching frequencies far beyond hearing limit, as higher frequencies comes out more efficient for amount-of-material-used

The transformer switching supply that was making annoying buzzes was operating at 100kHz or so. I didn't try to varnish it quiet, but as I recall it was not sealed, so that might have helped.

Usually only annoying if you have two very loud similar ones interfering..

/Jakob E.

Speaking of beat frequency, I recall one odd case while working at Peavey. Two cheap computer monitors were free running with their raster frequency close but not exactly at the same frequency. They generated a 1Hz beat frequency that was driving the engineer crazy trying to locate it. The computer monitors were interfering with each other through a paneled wood office wall.

As I recall the engineer was designing a powered mixer and couldn't find the 1 Hz noise source corrupting his mixer noise floor. The noise mysteriously stopped when either one of the two monitors in adjacent offices was turned off.

JR

 

[Cqwet Dbdfte]

SMPS noise can be a real annoyance, for those with good hearing. I can't hear much beyond 12KHz, but beat frequency interference tones would be a problem. Maybe a reason why SMPS control chips and some modules have a sync pin. A litte hard to implement on adjacent equipment. Old gapped ferrite HV transformer in TV receivers were famous for this noise.
Some SMPS' employ a spread spectrum function, and you can get clock multiplier chips with varying degrees of random spreading useful reducing this problem.
Meanwell makes a good piece of kit, I never heard any noise myself, in hundreds of applications, but well smaller modules when in overload protection. It depends on how heavily loaded the inductor or forward transformer is, and as suggested, how well potted it is.
Maybe Meanwell needs to hear about this.

 

[Matt Syson]

Commercial 'industrial ' linear power supplies used for 'audio' had noise specifications of 1millivolt or LESS measured DC to 500KHz so if you are replacing these your completed supply should be at least this good which requires good filtering since most switchers quote 150 Millivolts or thereabouts. Since it is now possible to record high frequencies it is even more important to do a thorough job.