Cockroft Walton multiplier redesign

Things at the limits.

Cockroft Walton multiplier redesign

Postby JonathanH13 » Sun Sep 18, 2011 7:13 am

I am currently modifying my Cockroft Walton multiplier. It has too many stages (9) and lots of apparent voltage, but not much under load. So I'm reconfiguring it using the parts that I have, for more current and a lower voltage.

The small caps are 3300 picoFarad at 10KV, the larger caps at the bottom of the stack are 4800pF at 12KV; each diode is 500mA at 15KV (high speed). The new design will incorporate 2 sets of 3 caps in parallel, in series with each other, for each stage (see images).

Using these values with a 40KHz supply:

Edrop = I /(f*C) * (2/3*n^3 + n^2/2- n/6)
Edrop = 1/(40000*4.95e-9) * (18+4.5-.5)
= 111111 volt drop per amp or 111 volt drop per mA (ignoring diode losses)

Before I go ahead and build this, I would like some suggestions / criticism on the design.

A couple of years ago someone from this forum made the following suggestions:

1) place the capacitor columns in a triangle
2) include protection resistors in-between the capacitors
3) place equipotential rings at each stage
4) put the large caps on the outer columns for more current
5) use a large metal sphere or torus on the output

I have implemented suggestion (1), capacitor columns in a triangle, into the design.

Since I have a large ballast resistor on my fusor (50Kohm 1000W) and the diodes are over-rated, I intend to not implement (2), the protection resistors, because they will complicate an otherwise robust construction.

Would like to implement 3), but I don't know at which points in the circuit the equipotential rings connect.

Would like to implement 4), but don't have sufficient large caps, so they are going on the first stage for ripple.

Will implement 5), a metal sphere on the output, when everything else is done.

The brass plates in the image will be waterjet cut, and their corners rounded off.

Cheers,
Jon

CW multiplier.jpg
Existing CW multiplier stack


equipot.jpg
Proposed configuration



Doug Coulter on 2011-16-08 09:36 (from fusor forum)

This looks real nice. You connect the rings at the DC output of each stage, and can use one at the top as well - they want to be fairly large in diameter (cross section) so they aren't corona points themselves.
I wouldn't use resistors in series myself. You could make a case for having them in series with diodes, but they'd have to be able to stand off full supply volts without arcing across them -- that's hard to achieve. And of course, they add even more series impedance. To the extent you want some (I believe you do) -- get it by using smaller caps on the AC part of the multiplier.

If you have more than a single cap size, you'd put the larger ones on the bottom. But! This loses the equal voltage division in a quick discharge (arc) and causes diodes to have to carry potentially nasty peak currents in that case. If you find yourself killing diodes in that case, this would be the first thing to look at. Perfect prevention of any arcs can be a real challenge.

You can "get away with" a fairly "squishy" supply for a fusor, and it can even be desirable as a sort of built-in current limit that protects the driver stuff in the case of a too-heavy load. That is, as long as you can hit your target voltage under the desired load. It might even be desirable to have a fairly high output impedance so you don't need as much fancy protection in the electronics. You will surely let in too much gas at some point and create a situation where the thing will act as a 600 volt (or so) zener diode, or what amounts to a short across the supply. Voltage drop across the series AC caps can be a big advantage in this case as it doesn't make heat anywhere, and limits peak driver current.
User avatar
JonathanH13
 
Posts: 53
Joined: Tue Aug 10, 2010 3:20 pm

Re: Cockroft Walton multiplier redesign

Postby JonathanH13 » Sun Sep 18, 2011 8:21 am

A quick update on CW stack redesign progress:

Only one major change to this design - I recently received an off-forum suggestion from an ex-FusorForum member suggesting that some of the large capacitors could be redeployed:

"The first capacitors on each side arm, but only the first, hold back half the voltage of the others,
so 2 in series wastes 2 caps and halves the capacitance. These are critical as all of your power has to go through these..."

If this is correct then half of the large caps from the stage1 side-arms can be moved higher up the stack. The new design looks like this:

model.jpg


I had some brass plates water-jet cut, which is always a pleasure to watch. That machine cuts through 50mm stainless steel like butter :)

cut2.JPG


Brass disks.JPG


A spice simulation to get the details right before the rebuild:

New CW config.jpg


New CW simulation.jpg


I am also thinking about building a new H-bridge inverter circuit using five FETs in parallel for each leg of the H. We can discuss the pros and cons of that in another thread...
User avatar
JonathanH13
 
Posts: 53
Joined: Tue Aug 10, 2010 3:20 pm

Re: Cockroft Walton multiplier redesign

Postby Doug Coulter » Sun Sep 18, 2011 12:13 pm

Be sure no arc can take place "inside" your ballast to the fusor, else you will kill more than just the top two AC side diodes in discharging the AC sides of the stack in an arc. Gets expensive quick.
Those are high quality caps that can make thousands of amps peak in a discharge.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Cockroft Walton multiplier redesign

Postby chrismb » Sun Sep 18, 2011 12:53 pm

Jon,

Are you sure you want to rebuild your stack? I don't think you've yet got the best out of it, maybe you could test with an improved driver first?
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Cockroft Walton multiplier redesign

Postby Doug Coulter » Sun Sep 18, 2011 1:42 pm

He was having corona issues, arcs, and not enough output under load, all related to the stack, so he's making it nicer. The driver seems to work fine (or that's what he said when he was here).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Cockroft Walton multiplier redesign

Postby chrismb » Sun Sep 18, 2011 1:58 pm

Oh, OK. I hadn't read there'd been arcing. (Sorry if I missed that, I'm a bit 'under the weather' these days.)
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Cockroft Walton multiplier redesign

Postby JonathanH13 » Tue Oct 04, 2011 8:01 am

Yes, actually I am rebuilding the stack and the driver. The stack is the priority though, and is nearing completion.

Some of the capacitors have small cracks and chips in them, which is a bit of a concern, but I will power up slowly and see how it goes.

I have tripled up the diodes, and will only ever power up this stack slowly through a variac, which I hope means that it will never see the huge current spikes present during initial charge up.

(I mean to ensure that the ballast resistor does not arc over, as Doug suggests)

J
Attachments
CW stack.jpg
User avatar
JonathanH13
 
Posts: 53
Joined: Tue Aug 10, 2010 3:20 pm

Re: Cockroft Walton multiplier redesign

Postby Doug Coulter » Tue Oct 04, 2011 9:07 am

This definitely wins the prize for looking good! It seems that a lot of the challenge in all this is just getting all the basic subsystems really solid and stable first, so you can "play" later and try new things with confidence that you can get meaningful results. When you have this big chain of things that have to all work well at the same time it's much harder.

I know Jon knows this, but a little example -- I'd make a bet with 70% chances on market trading, in a New York minute. But if we have two "ifs" at 70% "in series" our probability of a win goes to 49% - not so good at all. Once you have a few things, the probability of each working has to be pretty high to have much time when they all work at once! Even the high voltage supply system has several components prone to failure - the stack, the driver, the ballast, and the feed-through. Now add all the stuff in the vacuum system, the instrumentation, and so on - it's not trivial to have it all reliable at the same time.

But hey, if this was easy, anyone could do it, and success wouldn't be so rewarding. :D
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Cockroft Walton multiplier redesign

Postby JonathanH13 » Wed Oct 05, 2011 3:30 am

Yes, absolutely - it is not enough that it 'just works' or 'sometimes works', it has to work reliably every time and preferably have some spare capacity. I feel that in order to make progress we need fusion 'at the flick of switch' (close to what Doug has achieved), systems that we can rely on to reduce some of that ambiguity. I have started on the new H-bridge, it is going to use 4 sets of 5 parallel FETs to drive this stack, but we can explore the pros and cons of this on a new thread...
User avatar
JonathanH13
 
Posts: 53
Joined: Tue Aug 10, 2010 3:20 pm

Re: Cockroft Walton multiplier redesign

Postby Doug Coulter » Wed Oct 05, 2011 9:02 am

That thing is going to look really nifty with the corona rings on it, I'd bet. And yes, if a bunch of things are changing under your feet, how do you measure the effects of a new, deliberate change?
This seems a poster child for the value of just doing it right from the get-go.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to High Voltage, High Power

Who is online

Users browsing this forum: No registered users and 5 guests

cron