I.T.E AMS47-0501000FV 5v 1A audible oscillation

[Per Hansson]

Hi, looking for decent 5v PSU's for my fiber converter I came across this Amigo I.T.E AMS47-0501000FV 5v 1A PSU.
It's full range 100 > 240VAC input and has no external optocouplers, so quite nice.
It has a LinkSwitch-CV LNK624PG switching controller which integrates all, even feedback:


It had really REALLY bad audible oscillation going on though, measuring the original Su'scon SK 400v 4.7uF primaries they had 6.5 and 9.5 in ESR respectively.
Replacing them with UCC KMG caps of equal rating made no difference though.
I had already checked the secondaries and they measured good, they where Su'scon MC 330uF & 680uF 10v caps.
But to make sure I replaced them with Rubycon ZLH 220uF 16v and Panasonic FR 1000uF caps respectively, still no difference.
Felling bummed out I tried adding SMD caps on the unpopulated position C9 right on the output, but this also made no difference.
And finally I populated the empty L3 position with an inductor but this too made no difference.

So I started poking around and realized that if I touch my multimeter probe to the drain of the LNK624 the oscillation became much worse.
Even if it was just the bare multimeter probe not connected to the multimeter.
So I figured I'll try replacing diode D5 with an ultrafast, even though the attached documentation advises against it.
There was no difference though.
Next I tried poking the feedback pin of the LNK624PG instead, and that made the oscillation much better, almost gone.
So I figured that probably resistors R4/R4A or R3 would be out of spec, but desoldering them and measuring I found they where perfectly in spec!
So I ghetto modded potentiometers in their place so I could adjust them but this also made no difference.
Well of course except for the fact that it allows to calibrate the 5v output.

At this stage I've decided to give up on it but if anyone has some ideas about what I could try I'm all ears.
Also it might help someone Googling in the future
The PCB is marked AMS47-2 REV: 01

[Th3_uN1Qu3]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

Try putting a 10n or 100n capacitor from pin 1 (FB) to negative of primary filter cap. This will make the feedback loop very slow, but if your load does not exhibit rapid variations it doesn't really matter.

[budm]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

If you look at the spec sheet page 13, there is RC network R5/C5 for the FB pin, are they present on the board?
I wonder if the PCB designer of the board follow the PCB guides as shown on the spec sheet.

[stj]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

ferite beads on the right wires would probably fix it.

[Per Hansson]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

Th3_uN1Qu3 and budm, seems you have the same idea
I saw the example in Figure 13 (not page 13) before.
Actually I have spent some time with the datasheet because it's very well made and I'm really learning from it.
Otherwise I would have thrown out this shitty POS much sooner! :P
I will see if I can scavenge some components together tomorrow and see how it does.
It's odd though because it must have worked like this at one point in it's life.
What could cause it to degrade, might the Schottky Rectifier D7 be damaged or the transformer itself to cause it?
I guess the RC network for the FB pin was left out to begin with since it increases ripple...

stj: I did add an inductor at empty position L3, it was just a wire with ferrite bead that I stole from the original shitty PSU for my fiber converter.
There is no possibility to add one to the Feedback pin, not without cutting the trace and introducing even more noise by making it longer.

[stj]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

not an inductor, a bead you thread onto component legs to reduce high frequency crap.

it's common to see them on the gate leg of fast fets for example.

[Per Hansson]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

Th3_uN1Qu3 and budm: I added the RC network per Figure 13 in the PDF.
It did not fix the issue though, the RC network itself worked and the response was slower; I could hear that the oscillation was different.
Just as you would expect if you look at the scope graphs before/after in the PDF.
The drain voltage increased significantly too, it was 450v with a load connected before, and after it became 550v.
It's still within margin but damn it feels crazy with such high voltages in such a small device with traces so close to each other!

I also tried replacing the ceramic capacitor because I saw in a Fairchild PDF that could cause audible oscillations too, but no difference there either.
For good measure I also added Y1 caps in the two empty positions but of course that also had no effect.
And I've doubled the capacitance from 1uF to 2uF on the bypass (BP) pin on for the controller thinking it might be starved for current but no difference there either.

It's worth mentioning that on the output Schottky rectifier I see ~7VAC with my UNI-T UT61E multimeter.
But my Fluke 28-II is not able to read anything here, I'm assuming the frequency is too high for it but a little disappointing to be honest but I digress.
Measured across the first filtering cap the ripple is already below 50mV and on the final output cap it's below 10mV.
So wherever the oscillation is occurring must be on the primary.
I think I'll just have to invest in a high-tech instrument to find it: a straw


stj:

stj: I did add an inductor at empty position L3, it was just a wire with ferrite bead that I stole from the original shitty PSU for my fiber converter.
There is no possibility to add one to the Feedback pin, not without cutting the trace and introducing even more noise by making it longer.

[budm]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

BTW, does it get quieter when it is fully load?

[Per Hansson]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

No the other way around, it's pretty silent at idle and really noisy under load.

[retiredcaps]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

It's worth mentioning that on the output Schottky rectifier I see ~7VAC with my UNI-T UT61E multimeter.
But my Fluke 28-II is not able to read anything here, I'm assuming the frequency is too high for it but a little disappointing to be honest but I digress.

Didn't see this comment, but until I read into this post

https://www.badcaps.net/forum/showpo...2&postcount=12

but the specs for ACV on the Uni-T 61E is 45Hz to 10kHz. The Fluke 28 II specs for ACV is 40Hz to 30kHz.

[Agent24]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

Loose windings\core on the transformer?

[rievax_60]

Re: I.T.E AMS47-0501000FV 5v 1A audible oscillation

Consider replacing the LNK624PG if you find nothing else at fault. I have been replacing many of the SO version in F&P washer motor controllers. They usually start off by repeatedly shutting down when loaded by the lid lock solenoid, then exploding.

I last ordered the LNK626, hoping that it will be reliable.