High School Fusion Reactor: Fusor 0 Rebuild and Tests

Max E.
8 min readSep 13, 2021

Since early August, all of my fusor progress has been made in the research and design phase of Fusor 1 as both the grid and bottom plate of Fusor 0 had been severely damaged. Last weekend, however, I finally had the time to revisit my plan to repair Fusor 0 as I have finished the design of Fusor 1 and am now waiting for parts to arrive.

As I said in my previous post, the three problems with Fusor 0 were that a solder joint on the grid failed, an arc regularly formed between the lower part of the grid and a specific point on the bottom plate, and there was no accurate metering of voltage or current. This weekend, my goal was to fix all three problems.

Note: In the course of events cataloged in this post, I make a very stupid mistake. Do not emulate me.


First, I decided to disassemble and clean the chamber, replace the grid with a new one of a different design, cover the damaged point on the bottom plate with epoxy to “insulate” it, and reassemble the chamber.

I began by removing the top aluminum plate. Upon examination, I noticed six large brown spots (and a number of smaller ones) where electricity from the grid had arced to the plate. When I cleaned the plate with a paper towel soaked in isopropyl alcohol, parts of the residue rubbed off, but five discolored spots remained. I was able to scrape some off with a knife, but stopped as I did not want to scratch the plate.

The top plate after it was removed from the chamber before being cleaned.
The top plate after being cleaned with alcohol. The rubber grommet was removed to be cleaned separately.

I then removed and cleaned the borosilicate tube that forms the chamber body. The sputtering on the inside was scraped off with my fingernail.

The borosilicate tube with black sputtering residue on the inside.
The same tube after I cleaned the inside with water and isopropyl alcohol and scraped the sputtering residue off with my fingernail.

Then came the time to clean the bottom plate, which had a significant mark where it had been arced to repeatedly. I hypothesized that that point had been damaged such that electricity would always arc to it, so I covered it with more epoxy after attempting to scrape away the discoloration (which partially worked).

The bottom aluminum plate with a single, large discoloration where it had been arced to repeatedly. I believe that something about being arced to caused that point to be more susceptible to arcing in the future as it was the only point on the bottom plate to be arced to.

After the entire chamber was disassembled and cleaned, I inspected the broken grid and saw that while there was some solder on both parts of the wire, the connection had broken, presumably because there had not been enough solder in between.

The old fusor 0.1 1.5" diameter geodesic grid with a broken solder joint. I think that a combination of heat and too much flux lead to the disconnection.

While I could replace the grid with an identical spare that I had made at the same time as the damaged one (or attempted to repair the broken joint), I decided to update my design for two reasons. Firstly, according to Richard Hull’s ratio,¹ my grid was too large for my chamber (at 1.5" for a roughly 3" chamber), and secondly, my design had too many solder joints that I worried would act like breakout points and destabilize the grid containment. I had seen a multitude of non-geodesic grid designs on fusor.net such as a tungsten loop design² or a graphite stacked prism design³ applied in fusors and demo-fusors alike with great success, but decided to stick with a standard geodesic grid as it would be most similar to the tantalum grid I had just purchased for Fusor 1.

A fusor grid made of a tall tungsten loop.
A grid made of two graphite discs.

Having decided the general design of my grid, I still had a lot of difficulty determining the finer points of the design and how it should be constructed. The grid I had purchased was made of a single tantalum wire bent into three concentric loops and connected to a long feed-through, and I wanted to emulate that in some regard, but I was not confident in my ability to accurately bend three 1" loops out of one wire. I decided to instead make a grid out of two continuous loops of wire soldered together at the feed-through connection so that there would still be only one solder joint.

The grid I am going to use for Fusor 1. This design influenced my decision to go with solely vertical loops instead of two vertical and one horizontal to limit solder joints.
Another grid design that gave me the idea to use individual loops of wire connected at a singular joint.
A quick sketch that I made of my new grid design. Although I tried to build a grid with three loops, I could not get it to work practically. I do not believe that this caused any issues.
My finished grid for Fusor 0. The grid loops were connected to a screw that threaded into an alumina conductor seated within the ceramic feed-through I made. The two loops were centered before insertion.

I reassembled the fusor with the new grid, but as it was late at night by that point I decided to test my repairs in the morning.


I began testing the repairs I had made by doing a “vacuum run” where I connected the vacuum line to the pump and pulled a vacuum for 10 minutes to test the chamber’s resilience. I then turned off the pump and observed how long the many leaks in the chamber took to re-pressurize it to atmosphere. To my surprise, the chamber maintained vacuum remarkably well (compared to before). I believe that this is because I replaced the plastic cap on an unused port with a metal one which sealed the system much better.

I then added (rudimentary) instrumentation using a current clamp and a high voltage probe. While I believe the high voltage probe to be a good way to measure voltage moving forward, I know that the current clamp lacks the resolution to truly be useful in this application. As I did not have a high voltage resistor over which to measure current on hand, however, this was the best I could do.

This metering setup was not ideal, but it worked well for what it was. The current clamp measured the AC output from the transformer as I do not have a DC clamp and the voltage probe was attached to a ring terminal as I wanted a more electrically sound (and safer) connection then just poking the rectifier output terminal.

While the new grid worked perfectly, the layer of epoxy on the bottom plate did nothing to prevent arcing and, while I got some plasma, I still had a significant issue to repair.

A frame taken from the video monitoring (now with a mirrorless camera) of my first Fusor 0.2 run.

With this test, I had conclusively, experimentally shown that the distance between the grid and bottom plate was too small (I am sure that there is a way to prevent arcing at such a small distance, probably with higher vacuum, but I could not apply it to this chamber). As I did not have any more aluminum from which to machine another plate and the feed-through was immovably inserted with epoxy, I thought that I would have to give up on Fusor 0 and wait until the Fusor 1 parts arrived to start working on fusion again. By sheer luck, however, the other end of the feedthrough was the exact length that I needed (literally almost exactly 1") and that side of the plate was smooth enough to use.

I disassembled the chamber again; thoroughly cleaned the entire thing, again; threaded the grid onto the other side of the bottom plate; reassembled the chamber, again; reattached the vacuum system, again; rewired the fusor and metering system, again; tested the chamber with a vacuum run, again; and set up my camera, again so that I was finally ready to test the slight modification that I made… again.

Unfortunately (again), the grid was now too near the sharp edges of the vacuum line and gauge connections and when I activated the chamber, arcing occurred — to a significantly lower degree — from the grid to the top plate. This was still a great improvement over the arcing to the bottom plate as the point to which arcs formed switched around instead of staying fixed, but it was quite disheartening to have done so much work and be left with little progress as the energy lost to arcing prevented plasma from forming at the same level as it had before the grid had failed.

When I resembled the chamber, arcs formed to the top plate as it was too close to the grid. This was disappointing, but significantly better than arcing to the bottom plate.

Unfortunately, there would be little improvement (to put it lightly) in a subsequent run as the chamber unexpectedly re-pressurized with atmospheric gasses during shutdown, leading to catastrophic grid failure. While I initially thought (and explained to my Dad who had been helping me but was not present during the failure) that either the chamber itself or the pump failed, re-examination of the footage taken suggests that I likely inadvertently de-activated the pump while the transformer was still supplying power to the grid, leading to the chamber re-pressurizing before the grid was shutdown.

This severe mistake on my part caused the grid to remain energized in a pressurizing environment, which caused it to draw more current from the transformer, which caused it to immediately begin melting before I could hit the main cutoff switch.

As the chamber began to re-pressurize, more current was supplied to convert the extra gasses to plasma, this heated the grid significantly.
When some critical point was reached in the re-pressurization process, the plasma disappeared and the grid began to melt, ejecting what I believe to be accelerated, molten steel toward the top plate.
When the current draw became too high, the steel grid quickly began to melt. This entire process took under a second.
The final result of my mistake was a partially melted grid. Luckily, the damage to the top plate seems minimal and is likely reparable.

Such a mistake should not be possible going forward, so I have amended my operating procedure such that the switch that turns the pump off also cuts power to the variac (which should be turned off already, but that cannot be assumed as I showed).

This mistake has served as a much needed rude awakening to the significant danger of the high voltage that I am working with. Luckily all that was destroyed was the easily replaceable grid and not the less-easy-to-replace me. Levity aside, I am very lucky that my lapse in concentration did not end in serious injury and I am going to take more precautions to avoid such incidents happening again. The fact that I was alone, operating the device without (as I am now aware but was not at the time) my parents knowledge, makes this incident all the more concerning. I am still excited to work on this project, but now that excitement is tempered with well-needed caution.



Max E.

Hi, my name is Max and I’m a freshman at Columbia SEAS! This blog is where I document my progress designing and building a FarnsworthHirsch IEC nuclear fusor :)