Voltage Multiplier Design: Larger capacitors on first stage

[FonziDaytona]

Hi all,

So I decided to work on a 7 stage full wave CW for my fusor PSU upgrade and have a question about design. I’ve seen one instance where a commercial multiplier used larger capacity capacitors for the first stage…I think that this would assist with charging the following stages, but don’t understand how to quantify this benefit. Is it worth doing/messing up symmetry (I have severe OCD when it comes to layout:-)) over? Or is it more of a product specific design?

I’m asking because I found that I have just enough 150nF (1600v) capacitors to do this with if it offers some nice benefits. The rest of the multiplier will use 100nF (1600v) capacitors. The 150nF are sized a little larger which will cause some uneven layout…

First stage capacitance would work out to be 75nF and the remaining stages would all be 25nF. If there’s considerable benefit, I’d be able to ignore the asymmetric layout, but if there’s little benefit, I’d rather not. 

Thanks!
Matt

[davekni]

Is it worth doing/messing up symmetry (I have severe OCD when it comes to layout:-)) over? Or is it more of a product specific design?

Run a simulation to see.  Used LTSpice to simulate mine.  For minimum voltage drop for a given input frequency and output current, capacitance should be high at the bottom and low at the top and middle in the middle.  That's what I did for my low-current CW multiplier (to use in place of a van der graaf generator).
Most HV-forum members use CW multipliers to generate sparks/arcs.  For that use, uniform capacitance has a distinct advantage (except that the first input-coupling capacitor should be 2x value).  With non-uniform capacitance, an output spark discharges top (low-value) capacitors first, forcing more current through diodes.  With uniform capacitance (and 2x initial capacitor), all capacitors discharge together.
BTW, the first 2x capacitor sees only 1/2 voltage of other capacitors.  Thus 2x capacitance is the same stored charge.

[FonziDaytona]

I did a little more reading and it would seem that there isn’t a lot of benefit from having just the first stage larger…Looks like I’d need to make the first couple of stages larger, which I don’t have enough capacitors on hand for. I’m only using two capacitors in series for the first stage anyway, so getting close to what I’d get by using these larger, and more cumbersome, capacitors.

I may look to see if I can get more of these larger capacitors before I give up on it…

[Uspring]

The choice of best cap values depend on what you are aiming for.
For the least voltage drop under load, the best cap values for a 7 stage CW have the ratios 1:2:3:4:5:6:7, counting from the top cap down to the bottom.
For the least ripple, the ratios should be 1:sqrt(2):sqrt(3):sqrt(4):sqrt(5):sqrt(6):sqrt(7).

[klugesmith]

Dave pointed out:
>>Most HV-forum members use CW multipliers to generate sparks/arcs.

If you are making a fusor or something,
please do not abuse your power supply by making useless arcs and sparks.
A high power resistive load is more work, but more meaningful and informative.

Be different if you were making a big Marx generator,
where you can feel the impulse from each great big spark.

[FonziDaytona]

I’ll be wanting to have the least voltage drop during maximum current draw. 

No worries, I won’t be letting this short/arc :-) I have a ballast resistor setup to limit current in the event of an arc over (external). This (7 stage full wave) will be my third multiplier, replacing my previous 6 stage full wave (something wonky going on inside the capacitors themselves) and my original 4 stage half wave.

Looks like I’ll probably be better off just using the same capacitors throughout since I already have them and larger sizes of the same type are going to cost a bundle.

-Matt

[Uspring]

You don't need to follow the 1:2:... rule strictly to reduce voltage drop. The idea is to put additional caps there, where they have the biggest effect. If you have spare caps, double the first stage, then the second and then the third. After this, triple the first stage and so on until you run out of caps. I don't know exactly the order in which to add caps. A simulation would help there.

[FonziDaytona]

I could always parallel my 6 150nF capacitors with 6 100nF capacitors (equivalent to 1 300nF capacitor in series with 1 200nF capacitor) to make up the first stage. This would make for a 120nF first stage and all remaining stages 25nF.

-Matt

[Uspring]

I just noticed your remark:

First stage capacitance would work out to be 75nF and the remaining stages would all be 25nF.

Each stage contributes to the voltage drop. Making a single stage cap larger and larger is a matter of diminishing returns. Since your first stage cap is already much larger than the others, it is better in this case to add capacitance to the second stage, or better even if possible, to spread the available extra caps over the second and third stage.