# PWM limitation for capacitor

Discussion in 'The Projects Forum' started by sriniketh, Jul 15, 2015.

1. ### sriniketh Thread Starter New Member

Jul 7, 2015
29
0
I am looking for capacitors with the following specifications .

voltage -1000 VDC
Peak current - 120 A
capacitance - 22µF and 150µF

I would like to use PWM signals to charge and discharge the capacitors using IGBT.

T-On of the pwm signal - 0.1µs
Time period of the signal - 1000µs

Is there any limitations to the duty cycle or the pulse frequency in selection of the capacitor ? Can i use any capacitors for this application?

2. ### ronv AAC Fanatic!

Nov 12, 2008
3,404
1,495
Let's check the numbers first.
Just rough estimates.....
Lets say you put 8 ohms in series with the cap to limit the current to 120 amps peak.
If you only charge it for .1 usec. at a time then rest for 1 second it will take a few minutess to fully charge it.
Is that what you have in mind??

3. ### AnalogKid Distinguished Member

Aug 1, 2013
4,709
1,301
Rest for 1 ms - 1000 us. Still, that's a 10,000:1 duty cycle. Gotta ask why.

ak

4. ### sriniketh Thread Starter New Member

Jul 7, 2015
29
0
We charge the capacitor with 800V for 1 second initially and we have two IGBT one for charging and one for discharging , the
charging time is (1000µs-300ns) by turning the IGBT1 and discharging time is 100ns using IGBT 2 and the total period is 1000µs.
It is also true that the capacitor is not fully charged to 800V before discharging

LTspice simulation shows the voltage is changing from 575V to 570 V in 100 ns and then reaches 575 before the next pulse .

Regards
Sriniketh

5. ### sriniketh Thread Starter New Member

Jul 7, 2015
29
0
I need such a short duty cycle to have a near square pulse from RC circuit

6. ### AnalogKid Distinguished Member

Aug 1, 2013
4,709
1,301
A short RC pulse only appears squarer because of a long scope timebase. And, of course, it isn't square; it's pointy. The shape of an RC exponential decay curve is a universal constant in electronics. 1 time constant = 65% decay, 3 time constants = 95% decay, etc. no matter what the actual times are (seconds, hours, whatever). If you want all of the pulse energy to be expended in a time period that is short when compared to other aspects of the timing cycle, that's fine. But *technically*, a 0.1 us pulse is no more or less square than a 0.1 month pulse.

ak

7. ### sriniketh Thread Starter New Member

Jul 7, 2015
29
0

yes indeed, capacitance charging and discharging follows exponential curves. the time constant of the circuit is close to 110 µs and we are seeing a period of 100ns with a voltage drop of 5v in LTspice simulation.

Is there any limitations to the selection of capacitors for this kind of application

8. ### AnalogKid Distinguished Member

Aug 1, 2013
4,709
1,301
For a 100 ns pulse I'd use a capacitor with low dielectric absorption to maintain pulse fidelity.

ak

9. ### ronv AAC Fanatic!

Nov 12, 2008
3,404
1,495
Zeros, So many zeros...
I think it takes 1250 amps to discharge 25 ufd. 5 volts in 100 ns. Not 125 amps.
Yes. In any case. see pulse capacitors.

Last edited: Jul 17, 2015
sriniketh likes this.