Polywell Assembly Overhaul

26 08 2012

It’s clear from the inconclusive results of the symmetry test that our experimental apparatus isn’t durable enough. It needs to for a long period of time over the course of many trials, without breaking. Every time it breaks, we need to open up the chamber and fix it. In doing so, we inevitably make some slight change to the alignment of the components, the material makeup of the assembly, etc. Each of these changes introduces unknowns. This makes it difficult to compare the results of one set of trials to another, and thus difficult to accumulate the date we need to actually demonstrate something conclusively. To make matters worse, we aren’t getting deep vacuums because the much of the plastic and rubber components of the assembly are out-gassing.

It’s as though instead of running one experiment a thousand times, we run a hundred slightly different experiments ten times each.

In a word, the three main problems with the old polywell assembly are alignment, structural integrity, and vacuum compatibility. I designed a new assembly which should remedy these problems. instead of multiple separate components, it will be one solid piece of 3D printed ceramic which include a core, acclerator anode, hot cathode, and langmuir probe, all bolted to the 8” conflat flange:

The two feet (bottom left) will be bolted to the conflat flange.

Close-up of the hot cathode holder:

The cathode holder is actually two pieces which sandwich two 2o mm lengths of 10-gauge solid copper wire between them into the inside grooves (A). The grooves on the outside of the cathode holder accept zip ties which will hold the copper wires firmly in place (B). Once the wires are secured, the whole thing is put into place against the left column, and another zip tie is slipped through the hole (C) on the extreme left of the cathode holder and looped around the column. This connection will be strong  enough to prevent the cathode holder from moving during setup/normal operation.  A light bulb filament, serving as the cathode itself, is soldered to the ends of the copper wires.

Close-up of the anode holder:

The anode, a copper cylinder,  is put into the crescent moon shaped space (A), and a zip tie goes around it and through the hole (B), and secures it to the assembly.

Langmuir probe holder:

The langmuir probe fits int the groove in the cylinder (A), and attached with a zip tie or perhaps teflon tape. It extends into the center of the core, indicated by the blue sphere (B). The cylinder is oriented and positioned such that the langmuir probe will extend into the center of the core.

Here’s the new langmuir probe:

The new langmuir probe is a strand of wire inside a very thin ceramic tube. The 9-volt battery on the right is for scale.


The coils fit into the cavities in the core, and then covers go over them. The covers will be secured with zip ties or perhaps hose clamps.

Other than the zip ties and wires, all of this will be made of ceramic. The zip ties will be made of tefzel, a strong, heat tolerant, and highly vacuum compatible material similar to teflon. All wire insulation will be teflon.  This assembly will be heat resistant, electrically insulating, and much more rugged than previous designs. Ideally, we will be able to put it together, put it into the chamber, pump down to much deeper vacuum, and do hundreds of trails without anything breaking. moving, or changing shape.

In order to work at all, this design has to be compatible with the vaccum chamber and conflats that we already had. If one dimension is even slightly off, then the whole thing fails. To prevent that, I first took measnurements of the chamber and flange, and maodeled exact copies of them in OpenSCAD, and built this assembly inside the chamber :

Here you can see the conflat flange (left) and the chamber. Notice that the blue sphere, which indicate the center of the polywell is not centered in the chamber. By offsetting the core slightly, I was able to get more clearance between the walls of the chamber and the core, which in turn allowed for a larger coil radius.

Another pic of the whole thing:

While I have uploaded these models to shapeways, they will probebly not be the ones we actually have printed. This is more of a first draft.

The source code is here

Domenick Bauer

Inconclusive Symmetry Test

14 08 2012

The symmetry test was an experiment we ran to determine if our potential well was symmetrical across its x-axis. It featured three langmuir probes, one in the center, on at the extreme left, and one at the extreme right. more details are in this post.

While there were a few interesting results few interesting results, on the whole the experiment was inconclusive, and totaled our electron gun assembly.

The first problem we encountered was the vacuum level. We barely got into the 10^-4 Torr range, and when we turned on the electron beam, the pressure went up into the 10^-3 Torr range. High pressures like these don’t render the experiment impossible, but they certainly don’t help. Ideally, the only particles in the chamber would be electrons, and so everything else just adds to the list of unknown factors.

The e-gun was running normally, giving us readings of about -50VDC on the oscilloscope.

The glow from the hot cathode

For the first shot we did a control. We hooked up one probe to the center langmuir, and one the the shunt resistor on the on power supply, and we got a small well.


So everything was working as expected, despite the unusually high pressure. This is good, but also strange in light of the last test results, in which the charge at charge at the center of the core became less negative when we fired the coils.

Then we switched the oscilloscope probe on the coil power supply to the left langmuir probe, and fired.

Top: Left langmuir probe
Bottom: center langmuir probe

Nothing on the left langmuir. We tried again with more power going into the coils.

Here’s what we got

The charge at the leftmost extreme of the well is about -3VDC, and the charge at the center is about -10VDC. Not surprisingly, electron density has some relationship to distance from the center of the core. We intend to eventually define this relationship precisely, but to do so would require much more data.

Notice how even before the coils were fired, the top line is a slightly below its zero point (indicated by the crosses at the left of the screen). This means that for some reason, the left langmuir is brought to a slightly negative potential by the electron gun, even though it’s not pointed anywhere near the probe

We then switched the oscilloscope probe on the left langmuir to the right langmuir, and fired the core again. The moment we did, we heard a metallic noise from inside the chamber, like a coin dropping onto a metal surface. Can’t be good. This was the readout on the oscilloscope

Not especially meaningful to us.

We rightly assumed that the noise meant our trial was over, so we opened up the chamber.

We found the accelerator anode laying in the bottom of the chamber. The heat from the filament melted the plastic enough for the screw mounted in it to come loose. Everything around it was coated in a thin film of blue plastic, and much of the wire insulation was burnt as well.


Obviously, ABS plastic and rubber insulated wires just aren’t right for this experiment. They can’t take the heat of the cathode, and they out-gas so much that they ruin the vacuum.

Back to the drawing board.

Domenick Bauer

Safer Coil Power Supply

3 08 2012


All Photos (1/2)
All Photos (2/2)

The capacitor bank which powers the coils is very lethal; when fully charged, it dumps more than a megawatt into the coils. Needless to say, If I were to accidentally complete the circuit through my body, it would rearrrange my insides. To decrease the likelihood of such an occorence, I’ve done my best to idiot-proof it. Here’s how it looked before:

Pretty easy to thoughtlessly reach in and touch the dangerous parts.

So first, I took a broken shipping pallet that someone across the street was throwing away, and made a box out of it

The box fits nicely on the crate that we’ve been using as a controls stand


I also left the front partially open so they we could get at the controls

With this setup, the only thing you can touch is wood, plastic controls, and the grounded metal face of the box, so it would be pretty hard to blow your hand off.

The next step was making the HVDC out wires safer. (the fat green and yellow ones in the first pic) The way we had it, the only way to dissconnect them was to reach into the supply and unscrew the ring terminlas to which they were connected. Not great. To fix this, I installed a high current outlet onto the back of the box and connected some shorter wires.

The other hole is where the cord which plugs into the wall goes

Took the wires we had and connected them to a matching high current plug

Assembled (and labeled):


So now it’s pretty safe. In fact, it’s so safe that getting into the power supply at all istoo much of a pain. We don’t want to have to move this box every time we need to reset the breaker. To make access easier, I put hinges on the bottom of it (see above), and chains on the side, so it opens like a tool box.

MakerBot mailed me an iron-on patch with some supplies I ordered, so I taped it on for good luck. (top right)

Look’s pretty badass, kinda like a ten-cylinder car engine.



Here’s one more safety feature I added:

If you open the “hood”, the power cord is automatically pulled out. This way, it’s impossible to open the power supply while it’s plugged in.

The coil supply is idiot-proof, but isn’t so locked down that minor repairs and upgrades are too much of a hassle. This is one more step towards a streamlined, repeatable, experiment procedure.

Domenick Bauer


Symmetry Test

27 07 2012

All photos

This project is moving into a new phase. Now that we have a working electron gun/polywell setup, it’s time to begin studying the way they interact. To do this, we need a more comprehensive measurement apparatus.

In light of last week’s murky trial results, It’s apparent that a single langmuir probe in the center of our polywell core does not give us a clear picture of the whole well. I think a good first step would be to try to understand how the differs at different distances from the center.

In other words, how is the well different at different points on the red line?

Before I try to answer this question,  I want to make sure that each of the six coils behaves in more or less the same way. In other words, I want to make sure  If the red line above was pointing in the direction of the left coil instead of the right one, it would made no difference. I’m going to call this a symmetry test, because it will reveal whether or not the well is electromagnetically “symmetrical” about the x, y and z axes.

The symmetry test will work like this: Two more langmuir probes into the core, one on the extreme right of the chamber, and the other at the extreme left.

Right Langmuir

Left Langmuir

If our oscilloscope readings from the two probes are the same, then we may safely assume that the faces are interchangeable.

We’re rapidly running out of feedthrough pins

Core, from the right.

I can’t imagine that the thing’s not symmetrical, because the construction of each coil is identical to that of all others. However, thoroughness never hurt anybody. Also, I’m going to leave the center probe in, so we will be able to compare the center of the well to the edge, which will be interesting in its own right.

Domenick Bauer

Armature Upgrade and More Trials

24 07 2012

All photos(1/2)
All Photos(2/2)

We weren’t satisfied with the layout of the inside of the chamber last time. The alignment between the center of the polywell core, the electron gun and the lagmuir probe wasn’t very good, so I upgraded it.

The main advantage of this version is simplicity. It puts the filament, accelerator, and probe, and polywell core all on one flange, rather than two, which means the whole thing can be assembled on a desk and then put into the chamber Much easier than trying to make electrical connections and get alignment right with the thing half way in — something which, rest assured, is a royal pan in the ass.

Another advantege of this one is strength.

Dosn’t look like it, but that’s actually a very strong connection

A big problem last time was that we were unable to get the whole assembly into the chamber without accidentally bumping it a little, and ruining the alignment. This version is more securely attached to the flange, making it easier to keep all the components in the right place and pointing the right way.

Another flaw in the earlier version: The screw which attached the accelerator to the armature extended pretty far inside the copper sleeve which we were using as the accelerator anode, so it was partially blocking the beam.

In this version, that connection has a much lower profile.  I also switched out the old copper sleeve – which was too big and full of holes — for a new smaller one.

While trivial things like holes in the accelerator probably don’t matter, I’m trying to correct them because at this stage of the game, our goal is to eliminate as many variables as possible before we really start collecting data on a large scale. The cleaner the setup, the better.

We also took Remy Dyer’s advice and grounded the positive side of the DC output going to the hot cathode, in order to maximize the potential difference between the accelerator anode and the cathode.

Then, I pumped down the chamber and tested the electron gun.

If you look closely, it’s clear that all the components are aligned pretty well

The lower line is the voltage picked up by the Langmuir probe. The little cross shaped marker on the left side indicates the zero point. Every box in the y direction is equal to fifty volts, so our electron bean is delivering -50 volts to the Langmuir probe, with almost no AC disturbance! Not bad, and this is with the voltage across the hot cathode at about 60 volts out of a possible 120, so it could get even higher.

With this bigger, badder electron gun, we ran another set of trials.

We pumped down.

Hooked everything up.

And took our first shot.

As before, the lower line indicates the voltage on the Langmuir probe at the center of the Polywell core, where the potential well should be. Instead of a well, we have a hill! The magnetic fields generated by the Polywell are supposed to compress all the electrons within the core into its center, so the voltage detected by the Langmuir probe should go even lower. Instead, it goes up, from about -50 to -25. Very confusing.

We rant it a few more times, increasing the current sent through the coils each time, which translates to stronger containment fields. The results were similar.

Here’s a strange one where the center of the core seemed to become very positive. We suspected an arc.

Sure enough, we couldn’t get the core to discharge after this, so we worried that we fried the coils. When it was safe to do so, we opened the chamber and saw that there was an arc, but thankfully not on the core.

Evidently, there was a bad connection between this din rail connector and the core feed throughs on the inside of the chamber, and so an arc occurred, and broke the connection entirely.

So that was that for our trial.

It’s left me confused. What on earth could be making our well positive instead of negative, whereas in the last trial, we got good, negative wells?

Domenick Bauer

Discrepancy Between Circuit Simulation and Reality

29 09 2011

Previously we modeled the polywell coils and power supply in SPICE.

Today I returned to that model.

All resistance values in the simulation are based on real world measurements with the exception of coil inductance (code).

Starting with an estimate for coil inductance of 0.1 mH the discharge current looks like this:

The simulation’s peak of 1.5 kA is nowhere near the 2.3 kA we are getting in the real world:

OK. Maybe the value for coil inductance is off?

I played around with the value for coil inductance but the simulation would not match reality.

As a control I replaced the simulated inductor with a 1 mΩ resistor (code). Looks like this:

The simulation predicts ~1.8 kA but in reality we see 2.3kA!

Where does this discrepancy come from?



UPDATE: Reader Andrew solved the mystery:

You could try changing the ON resistance of your switch/SCR to something a bit lower than 100mOhms

.model MySwitch SW(Ron=.1 Roff=1Meg Vt=3 Vh=0)

I can’t see the part number of your SCR but 2mOhms would seem reasonable.

I didn’t notice that rather high resistance lurking in the SCR model.

Now the simulation matches reality very closely with 0.06mH coil inductance (code):

Good work Andrew and the rest of the internet brain!

Coil Power Supply Upgrade

26 09 2011

All photos.

I completed some nice upgrades to the coils power supply for safety and quality.

I added DIN rail terminal blocks and rearranged the parts to emphasize discharge path:

Added an internal variac:

Added terminal blocks to the trigger board:

These pins on the SCR never soldered well because they are mini-fastons. Fixed:

For safety, I added an AC switch:

I replaced the trigger’s battery with a 9V wall wart:

And added a DIN rail 0.5 A circuit breaker:

Welding 3D Printed Steel

16 09 2011

All photos.

My shopmate had a TIG welder here the other day.

I took the opportunity to try welding the 3D printed metal parts I made last year:

These are intended to be coil holders, so I installed a 40 turn coil prior to welding.

Mike welding the halves together:

We used no filler rod on the theory that the infused bronze would melt and form a braze of sorts.

It worked very well:

The coil insulation didn’t survive: The coil is conductive to the casing.

I’m encouraged by the weldability. I am ordering more test parts to keep pushing this approach.

2.36 KiloAmps

6 09 2011

All photos.

Today I tested the newly rewound polywell coils and thicker leads. Best shot was 2.36 kA.

Now we are in the right neighborhood.

Terrifying Power

4 09 2011

All photos.

Tonight I really experienced the power of the coil power supply. Whoa.

I’ve been working to increase the coil current from ~1.2kA to ~2.5kA.

Previously I discovered the coil discharge path had more DC resistance than expected.

I rewound the Polywell coils with 16 gauge wire (previously 18 gauge).

The 16 gauge DC resistance is 144 mΩ compared to 227 mΩ for 18 gauge wire.

I beefed up other wires on the coil discharge path (4 gauge):

Lets test the wiring with the dummy coil:

I took the power supply up to 100V… a small test charge…

When I fired, the noise from the coil made me flinch. It was never that loud before.

Lets turn up the power!

300V for second test.

When I fired the coil there was lightning! HOLY CRAP.  Look what happened:

The coil fucking wrapped itself around the transformer (electromagnetic forming). Then it discharged to ground:

So I haven’t measured it yet, but I think we are getting more current to the coils.


Get every new post delivered to your Inbox.

Join 630 other followers

%d bloggers like this: