Negative discharge effect

[Dave45]

Most circuits out there, if you study them are either boost converter's, buck converter's or a variation thereof,
Look at the topology, how their switched, from which rail, how the pulse is hitting the coil, is the output inverted.

[TinselKoala]

@ayeaye:a capacacitor is frequency-sensitive. Different frequency means different reactance, different reactance means different amount of charge transfered per period via Miller-Capacitance.
See here: http://www.sengpielaudio.com/calculator-XLC.htm
Capacitive reactance X_C = 1 / (2 · π · f · C)
It´s not worth to persue IMHO, exept what Utkin proposes. I have to say however that I could not replicate his results with the negative charged capacitor. Energy in form of negative voltage was always less than existed before in the cap as positive voltage.



Kator01

The voltage on a capacitor in a ringing circuit reverses polarity with every half-cycle of ring.  There is no such thing as "negative" or "positive" voltage as such, because voltage polarity and magnitude are always relative to some reference level. If a ringing cap starts out with a positive peak voltage, the next half cycle will reverse the polarity of the voltage on the cap as it discharges to zero and then _recharges_ to the negative peak voltage, which, sure enough, will be less than the original positive peak because of losses in the circuit. Let it ring long enough without resupplying energy and you get that beautiful exponential decay ringdown, voltage reversing polarity on every half-cycle. So if you arrange your snubbing circuitry properly you can "shut off" the discharging ringing capacitor with either polarity of charge and at whatever level, below the initial charge energy, you like.

Consider two 12volt batteries connected together in series. Measure across the stack. What is the voltage and polarity at the most positive terminal? It is 24 volts (nominally) positive WRT the most negative terminal. Now... what is the voltage and polarity at the "center" terminal where the two batteries are connected together?

SO... I think the voltage on the cap in the device under test will be sensitive to the frequency for at least two reasons: First, the capacitive and inductive reactances depend on the frequency as you note, and also the mosfet is going to be turned on and off at different points in the "ring cycle" by the signal and thus will trap different amounts of energy on the cap, with polarity that could be either way depending on the frequency of drive and its relationship to the resonant frequency of the tank circuit. I think. I haven't done the experiment yet but I may fire up the kit later on today if I have a chance.

[TinselKoala]

The 1n5399 diode is a HV rectifier diode, rated 1 kV and 1.5 A. But it is _slow_, having a typical reverse recovery time of 2 microseconds.

UF4007 is similar in voltage, a little lower in current handling but is quite a bit faster at 75 nanoseconds max.

[TinselKoala]

I see.

But why the effect appeared on different frequency when i repeated the experiment? While everything remained the same, the mosfet, the coil, the diodes, the capacitors. What changed, the mosfet, the coil's core, or the capacitors?

The mosfet or the capacitor or both could be changing. Mosfets can fail progressively especially if they are subjected to lots of avalanching, and the equivalent series resistance of the electrolytic caps can also change if they are subjected to short HV spikes that could exceed their dielectric rating, even as the cap still works as a capacitor.
I doubt if the core of the coil could be changing under your experimental conditions, but the mosfet and caps can definitely change. But I don't think that component changes are needed to account for your results. I'll know more after my scopes warm up.

I just pulled a handful of IRF630s out of an old monitor chassis, by strange coincidence...

[TinselKoala]

I will, i had no idea that there is any kind of diode inside.

All ordinary mosfets have a "body diode" that is the result of the manufacturing process. This is often omitted from the mosfet symbol but it is there. It is a reverse-biased Zener diode with anode at Source and cathode at Drain. If your diode-check function on your DMM has enough range you can check this diode in your mosfets yourself.

http://en.wikipedia.org/wiki/Power_MOSFET#Body_diode
http://hephaestusaudio.com/media/2008/11/mosfet-body-diode.pdf

http://www.youtube.com/watch?v=RBJGOOTEwfU