SMPS Noise

[microsim444]

Heloo.

I bulit a SMPS, ( the A-AND-T-LABS) www.a-and-t-labs.com

but when connecting the supply to the load a noise can be heard form the output TRANSFORMER ETD-49. (ssshhhshshh)
remove the load sound dissappears!

that noise can be heared clearly at the amplifier when connected.

Transformer winding.

Primary 28 Turns --> 1mm
Secondary 14,CT,14 Turns --> 1.6mm

I am not that professional into transformers, so i dont know why i hear this nois, and does this nois makes my driving MOSFETS gets hot quickly?


Any help will be more than appreciated.


Thanks all

[Geek]

Hmmm, that "shhhhshhhh" noise is parasitic oscillation instability in the power supply.

Does it do this for all loads, or are you powering something weiredly reactive?

Cheers!

[Eva]

This is either noise from the error amplifier and modulator or mild instability at low loads. It should not cause problems, though.

Note that the design of this power supply does not appear to be particularly good.

[Geek]

Originally posted by Eva
This is either noise from the error amplifier and modulator...

If it is, that can be cured by slowing down the time constant a wee.

Microsim, is that thing TL494 based?

Cheers!

[infinia]

Hi
Like Eva says. These PS outputs need light loads for stability/ and or duty cycle control. Should not be a problem if connected to a class AB amp with quiescent current. Otherwise use a couple of dummy loads... 10 Watts or so until your amp is connected and biased.
edit > read the manual on testing 4.2 http://www.a-and-t-labs.com/K6_Sw_Am...nual_wfigs.pdf

[microsim444]

hi.



The SMPS is running with the SG3525
putting any load at the output will not solve the problem, even 10K resistor.

Mesure the wave at the gates with no loads, wave looks nice,
put the load , the wave gets very ugly and jumping, distorted.


I spent houres trying to find the problem, i failed.

is the gate drive transformer a problem?

the mosfets gets very hot at 33Ohm load per rail + / -

[ChocoHolic]

Originally posted by microsim444
hi.


....

Mesure the wave at the gates with no loads, wave looks nice,
put the load , the wave gets very ugly and jumping, distorted.
....

This is indicating parasitic resonances.
Also the lot of RC-dampers in the schematic are looking like the R&D guys where struggling with undesired resonances. IMHO this topic is the most difficult part of designing power switch mode circuits.
How to fight it? Layout, component selection, dampers.
Not funny if a professionally sold design shows massive issues with that. As a first uncritical modification you could try to place a RC(L) snubber directly at the MosFets across the rail.
This RCL has to be adjusted to the frequency of the undesired resonance. If not adjusted properly it will be worthless.
The same sort of snubbers (but with small caps: May be 100pF + 10 Ohms + parasitic L) you can apply from drain to source of each MosFet. BUT take care - at these voltage levels you may get quite high losses in these drain-source snubbers. Which switching frequency do you have? Schematic is indicating values which lead to 80kHz. If you have 80kHz, then 100pF will already give you 0.8W losses in each 10 Ohms resistor.
Also sometimes it already helps to place fast & low inductive ceramic caps accross the rail, but you should choose a LOSSY ceramic type (large tan delta in the data sheet).
All other mods which come to my mind, would directly go into critical circuit and layout parts...

Good luck!

[ChocoHolic]

80kHz and driving 4 of these heavy MosFets directly from the SG3525 ...75 Ohms gate resistors? ...
Hard to imagine that the gate signals look Ok, even in no load condition.

[ChocoHolic]

....the gate drive transformer can be an issue, but not necesarily... Depending of its leakage inductance and losses it will form a more or less critical resonance with the gate capacitances. But I think this effect should already be visible without load.

[ChocoHolic]

I am referring to this schematic:
http://www.a-and-t-labs.com/K6_Sw_Am...atic/k6pwr.pdf