Self oscillating fun

[IVX]

I don't see the reason to have better 10% for such dielectric like X5R, X7R etc, due to noticeable thermal dependance of the capacitance. Try polyester (or polyprop) film caps instead of multilayer ceramic.

[Bender.ru]

Capacitor used in balanced line....

[IVX]

Originally posted by Bender.ru
Capacitor used in balanced line....

What??

[Bender.ru]

IVX, i also think so, but i'm not a designer

[Eva]

New magnetic snubber ready for testing in place of the old ETD29...

[Eva]

I have really learnt to hate film capacitors the hard way, actually the bigger they are the more I hate them. The medium and large sized ones become high-Q purely inductive above 500Khz or so due to geometry. They become pure junk when you need low capacitive impedance extending into the low RF range (no wonder they are the preferred ones by audiophiles, and the bigger the body and the longer the leads, the more they love them )

Conventional film capacitors have a great potential for spoiling any switching power circuit they come close to. This is particularly true for self-oscillating amplifiers which are the most sensitive ones to parasitistics and non-textbook waveforms. In self-oscillating post-filter feedback amplifiers, carrier frequency drop pattern and distortion are directly related to output capacitor inductance.

In these amplifiers, the self resonance of the output capacitor should happen at a frequency whose period is well below twice the propagation delay of the circuit, and it should be low Q, otherwise "carrier suckout" phenomena arises and a new oscillation mode will appear near output capacitor resonance when the output approaches the rails (BANG!!)

This is a DIY 1.35uF 250V non-inductive film. It works really good (I suppose that I don't have to explain where I'm using it in my amplifier):

[Eva]

Update:

The modulator is now working quite right, it's no longer producing multiple pulses or excessive frequency drop near the rails. PCBs are now full of patches

I have blown a pair of SPW20N60CFD while doing 1Khz 10% burst testing on 5.8 ohms load. Supply voltage was approx 260V and output voltage was +/-120V. This is 2500W peak but only 125W rms, it makes investigating circuit behaviour easier without burning a low of power. Short term peak MOSFET current including inductor ripple and diode recovery was probably 40A.

It may be time to use bigger MOSFET...

[luka]

Hi

Ok I can see how there was so much peak power, but how only 125Wrms?

[Eva]

It was 10% burst testing. It was 2 cycles of 1Khz at full power (2ms) and then 18ms of rest. I use a laptop to produce such a test signal. Dynamic circuit behaviour at high currents and voltages can be investigated in detail that way without burning 1KW or 2KW in the dummy load and without requiring a huge fuse and power supply. Fuse was just 2A. Note that even the IR2113 survived. No other parts blew except the two MOSFET.

[IVX]

Originally posted by Eva
I have really learnt to hate film capacitors the hard way, actually the bigger they are the more I hate them. The medium and large sized ones become high-Q purely inductive above 500Khz or so due to geometry. They become pure junk when you need low capacitive impedance extending into the low RF range (no wonder they are the preferred ones by audiophiles, and the bigger the body and the longer the leads, the more they love them )

Conventional film capacitors have a great potential for spoiling any switching power circuit they come close to. This is particularly true for self-oscillating amplifiers which are the most sensitive ones to parasitistics and non-textbook waveforms. In self-oscillating post-filter feedback amplifiers, carrier frequency drop pattern and distortion are directly related to output capacitor inductance.

Did you try SMD film caps? Pretty nice stuff, of course not so good as X7R e.t.c but better then conventional film caps yet, due to wide&flat case and legless.