# Spanning high voltage between capacitors in series

first of all thanks for your time.

I've studied a lot before to make this post but theory and real world rarely make a good marriage.

I have 20kVAC of spike coming from an ignition coil (the one used for making the sparks in a car engine).

The spark is made through an oscillator and so I can modulate the frequency, the amplitude and the duty cycle.

At the output of the coil I put an HV diode bridge to make it DC and, after the bridge, I have a long series of electrolytic capacitors (50 units, SAMWUA 400WV HE 105°C (M), rated at 400VDC/82uF each) put in series.

I made a simulation with PSPICE and the software says that, at each spark, every capacitor is fully charged at each pulse (once put in series the capacitance of each cap is only 1,64uF).

Before to apply the 20kVDC to my capacitor's bank in the real world I would know if I can do it safely or I will risk a dangerous explosion or to damage the bank (I will make particular attention to the polarization, of course).

The second question is: is it true that - in the real world -, at the terminals of each capacitor, I'll have 400Vdc?

What about the first and the last capacitor of the series? Do they have to withstand 20kV or simply, like the others in the series, they will carry only 400V?

• How about a circuit diagram? I'd be worried about over-stressing the caps due to tolerance differences and some will inevitably receive more voltage than their theoretical share. I'd also be worried about the diodes you are using - there will be large impulses of current to charge the caps and this sounds like disaster for HV diodes that I've used. Generally best to run electrolytics at no more than 75% rated voltage too. – Andy aka Apr 12 '14 at 18:11
• @Andy Aka: a circuit diagram for what? To represent a series of capacitors, at the ends of which is applied a pair of 20kVDC terminals? The experimental circuit that I'm building is only a test to verify - physically - a principle; this test has to be done with these exact components as a proof for a theory I have in mind. The number of 50 units has been used to represent ideally the overall idea...the real number of the capacitors that will be involved is, of course, greater to allow a big tolerance: 20kV are a lot! There were also planned the bleeding resistors... – Devesh Apr 13 '14 at 21:22
• Please yourself dude my comment still stands as relevant but you can do whatever you wish. It's a free world. – Andy aka Apr 13 '14 at 21:26
• @Andy aka:It wasn't my intention to offend you, Andy! I have a big esteem of you: I read a lot of your interventions and answers on this forum and I loved all of them. Sometimes it is difficult, from one language to another, to express the real feeling which stands behind the words...I'm from Italy and, perhaps, my ability to manage the english language is not so appreciable, so I apologize for the misspelled sense of my sentences. In the end, i thank you so much for your precious advice that will be put in great consideration. – Devesh Apr 14 '14 at 12:56

You should use balancing resistors across the capacitors, assuming it's a simple series connection of electrolytic capacitors.

The appropriate resistor to use across each capacitor is

R = $\frac {n V_M - V_b}{0.0015C V_b}$

R is in M ohms

n is the number of capacitors

C in uF is the capacitance of each capacitor

$V_M$ is the voltage allowed on each capacitor

$V_b$ is the total voltage across the string

If we allow 400V across each cap and total voltage will not exceed 15kV then we have a balancing resistor value of 555K ohms.

Each one needs a power rating suitable for the highest possible voltage, so

$P_R = \frac {V_M^2}{R}$ = 0.29W for this example

The balancing resistors will suck 0.54mA in this example, which is 8.1W total, a significant power loss, so you may not be able to get full voltage from your coil.

• thanks in advance for the precious info, I didn't know the formula to calculate the exact value for a balancing resistors! Can it be applied for the bigger capacitors (ultracapacitors rated @ 3,000F/2.7V)? Just a question: did you had a direct physical/real experience about the power loss you stated above? 8.1W of power loss are tremendous higher than the power which can be stored in the capacitors themselves... – Devesh Apr 13 '14 at 21:36
• Such a long string of electrolytic caps is rare, but the equation makes sense to me. A typical leakage specification is 0.02CV for an e-cap, so they're suggesting that the leakages are matched to within +/-7.5% of maximum. – Spehro Pefhany Apr 13 '14 at 21:41