Farnsworth Fusor   -   2016/01/15Viewed 2027 times this month, last update: 2017/02/06



Fusor central

Pretty star-mode

"I think I'll build a fusion reactor" - something I probably said a few months ago.

A Farnsworth Fusor is a nuclear physics demonstration device; (probably) not a practical means of generating power, but a way to achieve real nuclear fusion of hydrogen atoms into helium. In the process x-rays are produced, a very small amount of neutron radiation is created, and a fantastic light show is made, so fantastic to see it has been called a "star in a jar".

My intent is to demonstrate some fusion, take some great pictures, and try not to electrocute myself, injure myself with an implosion, or burn the house down; for fun!

Theory of operation:
A Hirsh-Farnsworth Fusor is an intertial electrostatic confinement device intended to fuse hydrogen into helium. It functions by ionizing gas of the hydrogen isotope deuterium, then attracting the deuterium ions toward the center of a chamber by means of electric fields. The outer casing of the chamber is positively charged, which repels deuterium ions. A "grid" of wires arranged in a sphere in the center of the chamber is negatively charged, attracting the ions, but allowing them to zip right by toward the center, out the other side, and accelerate back again. The difference in potential between the grid and case causes Paschen Discharge, ionizing some of the deuterium gas. When the deuterium ions are accelerated fast enough through the inner grid, some smash together in the center of the reactor with enough energy to fuse. In the process, the deuterium ions emit light, causing an eerie glow from the center of the chamber, as well as x-ray radiation due to deceleration and acceleration forces. If fusion is achieved, some neutron radiation is also released, as well as high-energy helium ions.

My fusor equipment is arranged this way:


Build Log:
I started with a very old, very rusty steel propane tank found on craigslist for $100. It took me a few days to cut off unnecessary structure, and remove all the rusted-on plugs. The tank is 0.25" thick steel, and weighs several hundred pounds.
The next step was to cut down the tank to a more managable size. This required two long circular cuts with the plasma cutter to remove about two feet of total length.
The tank could then be stood up on a rolling stand, and welded back together.
A frame was then built all the way around the tank out of two-inch square tube with angular mount structs, intended to give anti-crushing strength to withstand vacuum. I then cut out a large door hole. I was concerned that the door hole would be a weakness in the structure, causing it to fold like a beer can under vacuum, so I welded additional square bars along side the door openings. These also provide the bracket function for the door bolts.
The plate cut out of the door hole then gets a flange, and a window. This was very tricky and had to be cut off and re-done in order to get the flange to line up with the tank door. The tolerance is about half the width of the o-ring cord used. 3/8" extra-flexible cord helps.
All welded up, cleaned and ready for painting. For window glass, I'm using 0.5 inch thick polycarbonate from McMaster Carr. The pressure of the door under vacuum compresses a 3/8-inch o-ring seal made from o-ring cord stock.
The tank holds vacuum very well (no measurable loss in over a week at full vacuum) but to do real fusion, a simple piston vacuum pump will not suffice. To that end I purchased this large diffusion pump on ebay.It is an Edwards Diffstack-160 with 6-inch orifice, valve header, and 1300 watt heater.
A hole is cut into the tank for the diffusion pump, and a matching flange is welded on. All 0.25-inch thick stainless steel from Industrial Metal Supply.
Starting with an elbow section of 6-inch stainless steel pipe I got from Machinery and Equipment, I created a flanged elbow pipe to mount the diffusion pump to the tank.
Here's the complete vacuum system, holding vacuum very well! In order to present a uniform ground potential, the inside of the tank was ground down to clean metal, then painted with a conductive graphite paint. This required me to spend a few hours curled up fully inside the tank, holding a 2-horsepower 9-inch angle grinder. That is officially the worst, loudest sound in the universe.

Fully plumbed the vacuum and water cooling ports of the diffusion pump, and tested the resulting maximum vacuum. It's difficult to precicely calibrate my vacuum guage, but I estimate I'm getting down to about 4 millitorr.



The high-voltage power supply. The circuit design is based on fusor.eu. The parts are mounted to a frame of acrylic plastic, is placed in a grounded steel container, and covered in transformer oil. This transformer outputs 40kv at full input voltage, which is rectified to DC by the big diodes, and smoothed out by a capacitor. A variac is used to vary the input voltage, and thereby control the output voltage. I added a few important safety features, as 40kv is enough to kill a person instantly: the case is conductive and grounded, the HV cable is covered in grounded braided wire and though most of the HV components are rated for use in air, everything is submerged in oil which protects against arcing and over-heating.

The grid sphere I made of 0.029" stainless spring steel wire. Spot welded together with a home-made spot welder as in this Instructable.

Of the elements, deuterium is the easiest to fuse, but it turns out that acquiring deuterium is almost impossible for a private person living in the USA. However, heavy water, which is D2O can be purchased from United Nuclear, and the deuterium can be separated from the oxygen via electrolysis. My equipment for doing this is a tiny fuel cell/electrolyser from Fuel Cell Store, a balloon to collect the deuterium, and some desiccant beads to dry the gas. The fuel cell is powered from the 3.3vdc rail of a AT power supply.

First plasma! This is the first time I got real Paschen Discharge, rather than just destructive arcing within the vacuum chamber. Most of the discharge is between the left side of the grid, and the wire mesh I placed over the intake pipe to the diffusion pump. I put the screen there to keep the pump clean, but it's obviously unbalances the ground plane within the chamber, so my next test will be without the screen, with a smaller grid, and more effort made to center the grid within chamber. I will also insulate the rest of the hanging wire, as there is significant discharge there. This is a about 4mtorr and 30kv with just air remnants in the chamber.
Grid12cm dia, 0.74mm stainless wire
Vacuum1.010 v
GasResidual air
Voltage20vac in
X-RaysAmbient
ResultWild Paschen discharges

My first contained plasma. This is air plasma at about 20kv, after letting my diffusion pump run for about an hour. The plasma ball is fairly large and low density, likely due to the low voltage. Something in the electrical system failed shortly after I took this picture, so some repair work might be necessary before the next test run.
Grid12cm dia, 0.74mm stainless wire
Vacuum0.986 v
GasResidual air
Voltage30vac in
X-RaysAmbient
ResultDiffuse plasma containment

My second run explained the final failure in the first run. With high voltage on, and the diffusion pump running, there was about a 3-minute window where the ball of plasma inside the grid was visible, then got dimmer, and dimmer, until it was not visible at all. Turning the voltage up only caused a arc in the HV wire. I believe this was due to too little gas remaining in the tank to ionize, leaving nothing to see, and nothing to conduct the voltage, until breakdown of one of the other components. On my totally un-calibrated vacuum sensor, this occurred at 0.986 volts, the same as the first run. That may be my effective zero.
A post about this run on the Fusor.net forum put me in the Plasma Club!
Grid12cm dia, 0.74mm stainless wire
Vacuum0.986 v
GasResidual air
Voltage30vac in
Amperage9.2a in, ~10ma out
X-RaysAmbient
ResultHigher density plasma, one electron beam.

This run progressed very similarly to the last, with a nice plasma ball and one electron beam, evolving from a column shape to a trumpet shape. After the plasma dimmed with higher vacuum, I kept the input to the HV supply between 9-9.5 amps, which would allow the voltage to rise. This kept the plasma ball visible for a longer time, but eventually something blew within the HV supply, likely an arc across one component or another. I suspect I will have to reduce the grid size in order to get energy density higher without having to push to unattainable voltages. I also saw some arcing around the window. This may be due to poor grounding of the door compared to the rest of the tank.
Grid12cm dia, 0.74mm stainless wire
Vacuum0.985 v
GasResidual air
Voltage35vac in
Amperage9.4a in
X-RaysAmbient
ResultHigher density plasma, one electron beam "trumpet".

A new grid was constructed, using the same material, but smaller overall diameter. The result of this was another plasma ball with electron beam, but the plasma ball and beam were much less well defined and energetic than the previous runs.
Grid7cm dia, 0.74mm stainless wire
Vacuum0.985 v
GasResidual air
Voltage35vac in
Amperage9.0a in
X-RaysAmbient
ResultLess well defined and energetic plasma ball



In an attempt to increase energy levels, I added a coil of copper tubing around the inner grid as an outer grid to shorten the distance from the grid to the anode. Indeed, higher discharge brightness was achieved at lower voltages, but the overall behavior was the same.
Inner Grid7cm dia, 0.74mm stainless wire
Outer Grid30cm dia, 0.25in copper tubing
Vacuum0.985 v
GasResidual air
Voltage20vac in
Amperage3.5a in
X-RaysAmbient
ResultLess well defined and energetic plasma ball

A new inner grid was constructed, this time with thicker wire (which is easier to form) and more spherical than the last. The HV supply was better insulated to allow for higher voltages, and output voltage and amperage instrumentation was fixed. Unfortunately, my vacuum sensor seems to be stalling at higher and higher voltages, so no accurate vacuum reading could be attained. After the plasma ball was extinguished due to high vacuum, voltage continued to rise to 60kv, at which point my radiation detector started beeping, showing up to 1uSv/hr for about one minute.
Inner Grid8cm dia, 1.3mm stainless wire
Outer Grid30cm dia, 0.25in copper tubing
GasResidual air
Voltage60kvdc at plasma extinguishment
Amperage6mA peak at plasma ball stage
X-RaysAmbient
ResultWell defined plasma ball with high energy electron beam. 1uSv/hr x-ray radiation at 600kV after visible plasma was extinguished.

Removed copper coil outer grid, replaced thick wire with thin wire for inner grid. Not much change, but saw HV breakdown failures of the HV cable, and drop-wire insulator.
Inner Grid7cm dia, 0.74mm stainless wire
GasResidual air
Voltage64kvdc at plasma extinguishment
X-RaysAmbient
ResultWell defined plasma ball with high energy electron beam.

Removed the drop-wire insulating tube, and added a small balloon of un-dried H2 gas straight out of a fuel-cell electrolyser. The dump valve let in significant air, but the plasma does seem to glow a little more in the red.
Inner Grid7cm dia, 0.74mm stainless wire
GasWet H2
X-RaysAmbient
ResultWell defined plasma ball with high energy electron beam. Slightly red.

First test with dried H2, and new calibrated vacuum gauge.
Inner Grid7cm dia, 0.74mm stainless wire
GasH2 dried
X-RaysAmbient
ResultWell defined plasma ball with high energy electron beam. Slightly red.

Got down to 0.0 mtorr, so pressure was controlled with H2 through a dessicant chamber. Pressure shot up with each plasma ignition, likely due to the plastic HV pass-through burning. Saw copious x-rays generated at >65vac input.
Inner Grid7cm dia, 0.74mm stainless wire
GasH2 dried
X-RaysUp to 9 uSv/hr at 65vac input
ResultNice plasma, uncontrolled offgassing.

2016-03-13
Switched out the PVC HV passthrough to a pyrex glass tube. Switched to dried D2 gas from heavy water electrolysis. Little behavior difference from previous runs. Managing a stable pressure is proving to be a challenge, especially during plasma. One run produced one bubble in the neutron dosimeter, but I was unable to produce any more bubbles. Substantial movement of the grid occurred, initially attributed to the gas flow, but turning off the gas inlet still produced movement as voltage increased.
Inner Grid7cm dia, 0.74mm stainless wire
GasD2 dried
ResultNice plasma, x-rays, and one bubble

2016-03-14Getting good vacuum is always a challenge for Fusors, and doubly so with equipment such as this, which was not purpose built for high-vacuum applications. I am working on improving my vacuum with the following steps:
  1. Replace old, squished door o-ring with new stock, cleaned and lubricated with silicone spray.
  2. Testing for leaks with helium gas (didn't find any)
  3. Keeping the tank at full vacuum for an extended time. (Up to about a week) This allows any water to evaporate, and any oils or other volatile to evaporate.
  4. Run the diffusion pump at least daily to remove evaporated volatiles.
  5. Track leak rate over time. I'm using a spreadsheet with pressure and time readings to calculate the leak rate
I am currently down to a leak rate of 30 mtorr/hr, which comes out to about 1.6 sccm/min. Some of this is likely air leaking into the tank, and some is offgassing of materials within the tank. The leak rate is slowly going down, so I should see a plateau where everything has finished offgassing.

2016-03-23
In previous tests, the hanging wire within the HV feedthrough was vibrating against the glass wall, likely losing current. The wire was replaced with a 0.25" stainless steel rod. At maximum vacuum, I was able to increase input voltage to 110vac without arcing. However no neutrons were detected across a range of voltages and pressures.
Inner Grid7cm dia, 0.74mm stainless wire
GasD2 dried
ResultNice plasma, x-rays

2016-03-26
This project got featured on hackaday! To address some questions from the article and comments:
  1. My target electrical OUTPUT is 40kv dc at ~30 milliamps. Most readings above are in voltage/current IN to the HV transformer, as until recently my output-side instrumentation was lacking. I am power limited by the 10 amp fuse in my 120vac variac, so total power cannot exceed 1200 watts.
  2. For xray detection I'm using a Geiger counter, seen to the left sitting on top of my camera lens. It beeps angrily at input voltages over ~60vac.
  3. For neutron detection, I am using a bubble dosimeter from Bubbletech. You can see it in action in the picture to the left. The one bubble in this picture was probably produced by something other and successful fusion. :-(
  4. Regarding size: The vacuum vessel I am using is indeed huge (I wanted a large chamber for a variety of projects), but the grids, voltages, and currents I am using are in the same range as smaller Fusors seen on fusor.net and elsewhere. In my most recent tests, the innter grid is only 7cm in diameter. Though having a large chamber has the advantage that I have a great degree of freedom in experimenting with the effects of different grid sizes.

2016-03-26
After putting an oscilloscope (DSO Nano v2) on the voltage sense leads from the HV supply, I found that what should have been smoothed half-sine-wave, was closer to AC. So I tested my HV diodes, and found one to be non-functional, and one to be partially functional. I replaced those with a string of lower-voltage rated diodes on one leg of the transformer, and shorted across the HV capacitor to see what difference a simpler waveform would make. I also removed the analog voltage gauges from the voltage and current sense lines, which allowed me to get accurate readings with my remote multimeters. Lastly, I replaced the 1/4" copper tube outer grid, in an effort to allow higher voltages with less current. Still no neutrons, but much higher voltages with lower currents were attained, until this replacement diode also blew. New diodes are on the way.
Inner Grid7cm dia, 0.74mm stainless wire
Outer Grid30cm dia, 0.25in copper tubing
GasD2 dried
Best Result Input Power40 vac, 5.6 amps
Best Result Output Power4160 vdc, 77 ma
Best Result Pressure3.7 mtorr
ResultNice plasma, higher voltages attained, red-hot hanger wire.

2016-05-25Still working on getting voltages up to fusion levels (>10kv RMS). I suspect part of my trouble is due to leakage in the chamber, so I have been re-sealing everything, and checking for leaks with helium. Biggest leak I've found so far is my gas inlet valve, which was not cranked closed tight enough...
2016-12-12


In my effort toward higher voltages, I built a new HV feed-through. The old glass-tube feed-through could only get me up to about 30kv before arcing at various points. So the new feed-through is constructed of a industrial ceramic insulator and a 6-inch conflat vacuum flange.
The insulator, should get me to much higher voltages, as well as being less of an implosion risk if it does arc.
The flange should provide a better vacuum seal, as well as greatly reducing the offgasing potential during ion bombardment during operation as opposed to the old PVC-and-epoxy seal of the old feed-through.
2016-12-12
My first view-port was a 12x12-inch sheet of 0.5-inch thick plexiglass. This was more than strong enough, and being far away from the grid, was at very low risk of being melted by ion beams. However, being plastic, it could offgas during operation due to ion bombardment, so I replaced the old plastic viewport with this 6-inch diameter 3/4-inch thick pyrex glass disk, mounted in a stainless steel flange. All surfaces are sealed with Buna-N o-rings, but not directly exposed to plasma. With this, as well as a few plug replacements, there are no more organic materials in the chamber exposed to plasma energies. I hope this will lead to a cleaner environment for fusion.
2016-12-13
First plasma with the new feed-through and view-port. Low voltage, and only residual air, but so far, so good! A nice deep vacuum after several days at <1torr to allow all the materials to offgas.
2016-12-13


Pretty pictures
2017-01-28
The whole Fusor took to the road to do a talk and demo about fusion at the Sparklecon 4 event ad my local Hackerspace 23b Shop. John Essex took and posted a video of my talk.
2017-02-06
After using a small fuel cell to split water for more than a year, the fuel cell got gunked up, and I was frustrated with it's slow operation anyway, so I bought and setup a Hoffman Apparatus to do the electrolysis work. Same color-change desiccant dryer, but this guy pulls upwards of 60 watts, needs active cooling, and can fill a party balloon in about a day.

Comments:
Mike Massen (2016-03-26): Well done, a passion with focus, application and determination, good stuff. Interested to know the instrument you used for neutron determination and its relevant details, age, when calibrated, condition etc ?
Thanks


AJ (2016-03-26): I've demonstrated homemade speakers and welders in class, accompanied with your videos, but I don't think this is a practical one to try to build/demo. Great gallery. Be sure to let us know if you ever get it to sustain a reaction like Farnsworth claimed he did.

rp (2016-03-26): God write up. On achieving a high vacuum: get some vacuum grease for your o-rings. It's designed to have low vapor pressure so it won't out-gas into your chamber. Always leave the chamber under vacuum when your not using it. Also, consider going over the chamber with a heat gun while under high vacuum at least once to cook off some of the the volatiles that may be adsorbed onto the inner wall. Get it as warm as you can without damaging seals etc. and it will greatly help your max vacuum level. With that diffusion pump you should be able to get to 10^-5 Torr.

Erik (2016-03-26): Mike: I added an entry to talk about the neutron detectors I am using.
rp: I use all those techniques to improve my vacuum level, except heating the chamber. I generally keep the chamber at full vacuum all the time, except when I need to swap out grids or pass-throughs. My vacuum gauge can read down to 0.01mtorr, and I see 0.00 if I let the pumps run for a while. I figure I'll never get rid of all leaks, so my plan is to keep the leaks down to a tiny fraction of the fresh deuterium feed rate.

Moor Bo (2016-05-24): Hey there, I'm currently an engineering student in germany, and I'm working on a fusor, and found this project through hackaday...I see youv'e done some experimenting with your inner cage radius compared to a constant outter cage, Have you gained any efficency compared to a large ratio, or a a smaller? at the moment I see you converged around 7.5/30, or rather 1/4. Thank you for your time!

Erik (2016-05-24): Hi Moor, defining "efficiency" is hard, since I have not yet achieved fusion. I have not observed a significant difference between using an outer grid or not, when looking at power consumption, voltage/current ratios or plasma behavior. On advice from the Fusor forum, I am currently running without an outer grid (chamber walls are the outer grid in this case), but I am unable to maintain plasma at >7kv. In the future, I may add in an outer grid, and shrink the size of both. A good reference is here:

http://www.fusor.net/board/viewtopic.php?f=24&t=8765

Your Name:

To add a comment to this page:
1. enter your name...
Your comment:
2. Enter your comment
Anti-spam check:
5. Press "Post Comment" button.

Email me at erik@erikburrows.com if you hate the new anti-spam checker.

permalink