Kapanadze Cousin - DALLY FREE ENERGY

[d3x0r]

I got 2 mosfets and a cap(rophone)

Pulse generator Simulator

Generates 4.69MW Pulses in nano seconds.
Be better if I had a DSRD model.




Mind you I was working with Dally voltages... so there was a seperate 12v or 5V input voltage from the source... and a 12V drive voltage already from the lower inverter...


The capacitance of the gates needs to be figured in that you need more in the source than the target or you get 0.5 of the voltage in both...  and if it's too small, it won't pass quick enough to be a good sharp pulse.  10-100 times the capacitance seems to work pretty well.


The RC pulse width constance needs to account for gate capcitance as a part of the figure...  Eventually you will be purely the gate capacitance...

There's also things you might not have seen in the options for Falstad simulator... Like this simulation is set to step at 250ps (picosecond)

[TinselKoala]

High current to gates will be good for sharp turn on gates; but in a controlled dosage, so it's a high current (but tiny time).  Espcially if the mosfet is already stressed by a high voltage, partial on's are the worse.  The driver would be tuned for the mosfet so it has a current source (capacitor) available to quickly dump to the mosfet;  It's also got good switching to be able to pull the voltage back off the mosfet (good dump to ground).  In this circuit, the pull to ground is 50% of the duty cycle (by the other mosfet)... so the ground side isn't as critical...


But since they measure gate capacitance in (1 actually more like 30-50) pF, even at 1mA it fills in 1ns (or 30-50).


In this case almost any current is quick enough to turn on.

Have you looked at the actual gate or input capacitances given in mosfet data sheets? I think your estimate is a "bit" low for most of the mosfets we are familiar with.

And "almost any" is very different from "none", isn't it?

The point remains: a finite current supplying charge to the gate is required to turn a mosfet on, and a finite current moving charge away from the gate is needed to turn the mosfet off. Further, when AC signals are applied, the mosfet's gate capacitance can allow currents of large magnitude to flow through the mosfet from gate to source or gate to drain or vice-versa.

[d3x0r]

Have you looked at the actual gate or input capacitances given in mosfet data sheets? I think your estimate is a "bit" low for most of the mosfets we are familiar with.

And "almost any" is very different from "none", isn't it?

The point remains: a finite current supplying charge to the gate is required to turn a mosfet on, and a finite current moving charge away from the gate is needed to turn the mosfet off. Further, when AC signals are applied, the mosfet's gate capacitance can allow currents of large magnitude to flow through the mosfet from gate to source or gate to drain or vice-versa.

http://www.irf.com/product-info/datasheets/data/irfi4020h-117p.pdf
Ya this is higher power 1240 just good response times...  But this is the power one with 200V 9A (but it's pulsed so it's okay)
so 1.2nf which is 1us instead of 40ns... but still sufficient for 100kHz or less... (which is 10us)


so okay maybe reduce the gate voltage divider to 5Kohm instead... so you get 2mA. (2.8 actually).. and effectively double that to the gate.


http://www.nxp.com/documents/data_sheet/PHE13009.pdf
Ya and what I ended up using was transistors for the front two.


But then again the point really is, if you look at your part; and pick appropriate parts for what you're doing it's not so hard.  And this royer is definately not something that requires a canned IC part just to drive a stupid mosfet


The resistances keep any surges from happening, and the diode to the collector is only a connect to ground when the other is open.


If one really cared about it... add some 10nf caps in front of the resistor... though that won't help much because the diode will have cleared that too... maybe a cap between there and another resistor so it's at a higher voltage... well then again... it's in the dead of the storm, so you have quite a bit of time to recover before opening the switch... even if you were stalled to say 1us; if you allow turn on at 25% of the LC tank swing (since when it comes on it's already at a low point, and going lower)... which is actually the other 50% of the cycle to have to turn on ....


so that's 1Mhz... and just reducing the resistances will increased alowable frequency and keep voltages within tolerance.


EDIT
Digikey page with 379 parts for 20-50V with 1-40pf gate capacitance.

411 digkey parts 50-1100V with gate capitance from 1-40pf

[Khwartz]

Is it _still_ necessary to point out that "Watts" are not a measure of energy, and mere multiplication of peak power means essentially nothing?

Energy is conserved. Peak power can be multiplied as much as you like, but you still aren't showing any greater energy coming out than going in.


.../...

Hi!
That's why could be Verpies' Watts-Volts converter should be used any time as he said it checks all the need values both on input and output for only few bugs.
Cheers

[d3x0r]

Hi!
That's why could be Verpies' Watts-Volts converter should be used any time as he said it checks all the need values both on input and output for only few bugs.
Cheers

Still think it would be nice if it had a sliding window average that was the integral of power.  Just because I can get a GW pulse in nano seconds... doesn't mean I'm not just using watts over longer time periods