Page 1 of 2

RMR-Farnsworth Fusor 5

Posted: Thu May 02, 2013 10:55 pm
by Martijn Rambonnet
After our first neutrons (viewtopic.php?f=6&t=3128&p=12638&hilit= ... f76#p12638) with our first working fusor, the RMR-FF4, we pretty much completely blew up the x-ray transformer we used as a power supply. Since we were and still are working towards recreating a polywell it became clear that the next iteration of our fusor would have to be a little bit more professional. We had some great luck finding parts and finally completed the 5th iteration of the fusor. This iteration is basically a test of the different pieces of equipment in preparation for the polywell. It will be used to test the high voltage supplies, vacuum system, fuel system, neutron and x-ray radiation measurement and data logging. Since we are already in the neutron club we will mostly concentrate on describing the current setup and experiences building this system.

System overview
The setup is pretty large but the two racks help a lot with keeping everything organized. All power related stuff is in the left rack and connected to the left side of the chamber. All measuring equipment is in the right rack and connected to the right side of the chamber. Everything is on wheels so it can be easily moved around to access the back or sides.

Vacuum system
The vacuum chamber consists of a 300mm sphere with a large amount of conflat flanges. To the side is an 8" conflat tube with several KF and conflat connections. The backing pump is an Edwards E2M5 and the high vacuum pump is a DC705 charged Edwards diffusion pump with an estimated rating of 300L/s. The diffusion pump has a water cooled internal baffle and a tilting disk valve on top. The aluminium table where the setup rests on acts as a conversion flange between the custom Edwards pump flange to 8" conflat. For pressure measurement we use a PKR-251 pirani/cold cathode gauge and a Granville-Phillips 274 hot cathode gauge.
As already mentioned by others the biggest problem of a large chamber was sourcing all the blank off flanges. We ended up making an aluminium 8" blank because of the high (shipping) cost of such a flange and because we use a rubber o-ring on it since we also use it it as the main access port. The backing pump gets the vacuum system down to low in the e-3 mbar. The diffusion pump currently gets everything down to e-7 mbar without a bake out of the chamber but we still have a small leak somewhere (but on the RGA the N2 is relatively small compared to the H2O peak at that pressure).

We use DC705 in the diffusion pump but we also had good experiences with DC704 which can be easily found on ebay. If you look at the data sheet for DC oils you will see that it can easily withstand accidental exposure to atmosphere at working temperature and it has no dangerous break down products.

The current valve on top of the diffusion pump is a little bit to rough for easy control. We have to keep it pretty much completely closed at pressures necessary for the Farnsworth fusor, because when we increase the fuel flow too much the backing pressure will rise too much (we try to keep it at 150e-3 mbar maximum, most manuals advise anything between 150e-3 and 500e-3 mbar as a maximum backing pressure).

The PKR seems to be pretty reliable compared to the hot cathode gauge although our backing pressure gauge is also a PKR that is pretty contaminated (from when we got it) and it indicates much too low in ion gauge mode. They can be easily cleaned but it takes some effort.

The RGA is a Balzers QMG 064 0-64 a.m.u. We are still having some problems with contaminated grids in the ion source, after trying acetone, ultrasonic, sulphuric acid, muriatic acid and nitric acid cleaning nothing seemed to work. Eventually we tried cleaning it in plasma which made a huge improvement but not yet up to normal levels (still indicating pressure ~22x too low).

Following the advise from a vacuum technician at our university we placed all the measuring equipment at a right angle with respect to the main chamber plasma to prevent interference. We also had a small accident where we accidentally caused some sputtering from a copper grid which seemed to coat everything in straight view of the main plasma. The high voltage or plasma does not seem to affect any of the gauges.

For the conflat flanges we try to recycle the copper gaskets as much as possible (so don't tighten them too much if you want to reuse them, you can pretty easily see by the groove depth whether the gasket can be reused). After replacing a flange we do a quick leak check with some acetone or alcohol to verify if the gasket is still good and sufficiently tightened.

High voltage
We initially planned on using another x-ray transformer as high voltage supply but found a Glassman WG-50P6 50kV, 6mA supply. It apparently came from a spectrometer setup but it did not have voltage control on the front. After some trial and error it seemed that there were several preset voltages that could be selected from the back. This was removed and a multiturn potentiometer was installed which provided normal voltage limit control.

Since it was a positive voltage supply it also had to be reversed. The voltage multiplier is pretty simple and only the diodes had to be reversed. There was an extra wire on the connector which turned out to be the negative voltage feedback. We could not determine the negative voltage current feedback on the control board and just inverted the current feedback signal. A charge pump ic was used to generate the -10V which powers a simple inverting opamp circuit.
On advise of a high voltage technician at the university we added a series resistance at the feedthrough to prevent damage to the feedthrough ceramics. The capacitance in the supply is protected by internal series resistors but combined with the cable capacitance this can still add up to damaging current for the feedthrough. Currently we use 80kOhm (because that was the one we had available capable of handling high voltages, in case or a short circuit it will have to resist the full voltage), this should keep the short circuit current limited to <1A and only gives a voltage drop of 500V at maximum supply current. A 10kOhm would be more suited but this works too.

The grid is approximately 60mm in diameter and consists of several 0.6mm tungsten rings with metal legs that stick into a stainless steel rod with some additional tungsten rings suspended in the grid. Everything is held together at the stainless steel rod without any welds. A ceramic tube was later added around the stainless steel rod to prevent some arcing we experienced near the feedthrough. Our initial grid seemed to be a little bit too open but after adding some extra rings it produces a nice star mode.

Fuel electrolysis
We initially worked with a normal electrolysis setup with platinum electrodes in an oil bubbler with a large silica gel filled dryer that gave pretty pure deuterium (on the RGA). We later replaced this with a fuel cell which we run of a lm317 switched as a current source capable of supplying 4.5V or 0.5A max. This produces several ml of D2 per minute which is stored in a glass syringe that triggers an automatic power cut off when it is full. The syringe plunger is covered in some backing pump oil for smoother operation.
The D2 is added to the vacuum chamber through a Granville Phillips 216 metal face seal valve that provides very precise regulation.

Below are some graphs from the RGA at 5e-5 mbar (with the backing pressure similar to during fusion runs, thus approximately same D2 flow). There are some D2O, DH and HDO peaks but they are really small (less than 1/50), we might add some silica gel in the future to dry the D2 gas.

Radiation detection
We are currently using a LND25169 He3 tube (at ~1000V) and a Centronic SP-9 (at ~850V) spherical He3 tube (from a Nuclear Industries remball, thanks to George Dowell for the heads up in the Trading Post!). As a moderator we used an 8cm thick HDPE remball shell. It is connected to a nim rack with a Canberra TSCA. Getting the TSCA to work we had to replace some internal capacitors which had gone bad over time. Some other of our nim modules had suffered similar fates where capacitors had gone bad or the calibration was slightly off. However for most modules manuals could be found and Canberra was nice enough to send us a pdf of a manual we needed.

As a pre-amp we use an AD8510 based charged sensitive pre-amp circuit that came with the LND tube from ebay (thanks Adam Ingle for the heads up in the Trading Post and Carl Willis for his report on the tubes). We still have to finish the improvised high voltage supply for the detector and have to tune the LLD and ULD of the SCA. Also we will make a separate moderator for the cylindrical LND He3 tube and restore the remball with a direct pulse output to our nim bin.

Control and logging
In preparation of the much more difficult polywell setup it was important to be able to easily log and control the different parameters, especially if it should become necessary to operate it in pulsed modes. We use an EURO/VERO rack which contains an arduino, ethernet module, custom PKR pressure read out card and custom valve control card. This will be extended with additional interface cards to read out things like temperatures and control things like the electrolysis setup and D2 valve. All the interface cards communicate with the arduino, mainly via SPI. The arduino in turn communicates via ethernet with the control computer in the rack.
The Glassman power supply has a separate arduino with ethernet module which is currently used to read out current and voltage but will later also used to control voltage and current.

The control computer runs EPICS which reads out the two arduino's, the nim bin dual counter, the Balzers RGA and the multimeter. All are interfaced via custom ASyn/Streamdevice protocols.

The data is logged via ArchiveEngine from Control System Studio to a database and we also use CSS to easily view, plot and control the data. It takes some effort to setup EPICS and CSS but you can really see it is made for this kind of stuff. We initially tested some other open source SCADA systems but in the end EPICS suited us best. CSS also allows us to easily export plots and screenshots from measurements to our elog electronic logbook in which we try to document the work done and results.

Initial results
We did several runs to test the different neutron detectors up to a maximum of 35kV grid voltage. Initially our count rate and D2 flow was a little bit to low and the star mode was somewhat whiteish, increasing the D2 flow a little bit returned it back to a nice purple. Both the He3 tubes gave a count rate of approximately 0.5-1 cps (1-2 cpm background) at >25kV at 300mm distance from the grid center using the remball shell as moderator. 0.5-1cps seems a bit on the low side, our best estimates give approximately 20k fusions/s but it could be that the gain, LLD and ULD are still off from the correct settings. Backing pressures in the graph below 8e-3 (when it automatically switches to cold cathode ion gauge mode) are incorrect due to contamination of the gauge.
Some nice pictures from the plasma, IR glowing of the grid and the associated x-ray's which show up nicely on the webcam (which is of course shielded with lead at the outside).
I would love to upload higher resolution photos but the forum does not seem to thumbnail them.

A little bit more information can also be found on our website:

We would like to thank Jeroen Vriesman for all his help thus far and ofcourse everyone on for the valuable information.

- Bent & Martijn

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 3:01 am
by Frank Sanns
Really nice setup and work. Top rate. Congrats and thanks for sharing.

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 3:16 am
by David Geer
I agree. VERY impressive setup! Though from the images of your grid, might need to use something a bit more durable to thermal stress as the grid shape looks to be distorted when in operation. Other than that, everything is very well done.

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 10:36 am
by Martijn Rambonnet
The grid does seem to look a little bit warped in the two plasma photos but that is not something we observed in other pictures. The alignment of the rings is not perfect but the inherent construction and high stiffness of tungsten makes this very difficult. Ideally the grid should ofcourse be spot welded but that is something for the future grids.

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 2:28 pm
by Richard Hull
Simply fabulous! One of the finest and most complete expositions of a fusor system that I have ever seen. I can't complement you enough. Your professionalism is apparent and appreciated.

You have a lot of talent gathered around this system. What some individuals lacked in skill and knowledge, you found in others who helped in the effort. College based efforts are always good as there is always a surplus of skill sets and bright eager talent on hand from the mutli-disciplined course offerings in house. What any college may lack in funds for the effort is more often offset by the "can do", eager efforts of youth on a mission. When colleges and department heads see this, trees can often be found and shaken from which money falls upon the effort.

All the best in your quest to cobble up a polywell. That is a super big job and you are approaching it correctly.

Learn about fusion by doing it first in a far simpler system. In doing that you know fusion's signature in instruments already in hand and that you know to be functional. You know the basic difficulties and paths to fusion. You have done fusion and will proceed from a good basic knowledge base in which all niave, simplistic, preconceived ideas about fusion have been wiped out by the "doing".

Doing fusion is really easy. However, increasing one's yield in excess of a very good fusor operated at 60kv or more, goes up in difficulty by 2nd power of complexity and the 3rd power in cost, it seems, should other devices and methods be used.

Richard Hull

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 8:19 pm
by Martijn Rambonnet
Thank you all for the kind words. So far the project has taken a lot of effort but has also proven to be a very good and fun learning experience for us as engineering students with regard to particle physics, electronics, machining, vacuum technologies, radiation physics and even a little bit of chemistry.

Although the project is not affiliated with our university in any way people have always been most patient in answering our questions, showing examples of technologies or tricks used in their experiment setups and sometimes even providing us with some parts.

We will continue finishing up the neutron detection systems and hopefully soon start the construction of the first polywell test setup.

In the mean time we are also constructing a Langmuir probe (we are aware that past experiences with Langmuir probes in Farnsworth fusors weren't really succesfull) and are going to attempt to measure ion temperature using doppler broadning of spectral lines. Both are relatively simple diagnostic methods we also want to try on a polywell setup.

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 9:28 pm
by Richard Hull
Way back in 1999, before the turn of the century. I was involved with an Ocean Optics spectrometer used on my Fusor II When I visited Doc Bussard and Tom Ligon in Manassas VA. (Old location of EMC squared.) We used the resulting spectrogram to measure the doppler broadening of the H beta line. At low potentials 4-8kv the ion temp-(velocities) were just about a match for the accelerating voltage. Bussard was happy with the results, but later it was found by Tom Ligon that the higher the voltage the less was the gain in ion temps-velocities. So it seemed non-linear as the broadening did not pace the increase in voltage.

There are ancient postings on this work on the old Songs, Intranet and recently abandoned platform.

Richard Hull

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 9:47 pm
by Martijn Rambonnet
Thanks for the information Richard, that sounds most interesting and I will try to look it up. Our idea was initially based on a paper by the University of Sydney (attached) who also did doppler shift spectroscopy on an IEC device. We have a monochromator in the mail that is factory spec'ed up to 0.2nm band pass with the right slit size and we hope we can push it to even a little better resolution (and maybe be convert it to a spectrometer using a linear ccd).

It will also be interesting to compare the results of the Langmuir probe with the spectroscopy information.

Whether or not a polywell is a feasible fusor concept is of course the big question but for us it is mainly about the challenge to recreate at least some of the basic results/phenomenons reported by Bussard.

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 11:33 pm
by Peter Schmelcher
Hi Martijn,

Nice work, setup, and thanks for sharing your results.

Did you degas the heavy water before you put it into the PEM cell? I see the O2 peak is almost gone but I suspect CO2 will be the major contaminant. Some literature suggests that activated carbon will remove CO2, however, I can't find any hard ppm data. The lack of performance specifications might be to reduce Darwin award tryouts, or alternatively activated carbon does not actually remove any CO2.

Best wishes,

Re: RMR-Farnsworth Fusor 5

Posted: Fri May 03, 2013 11:54 pm
by Martijn Rambonnet

We did not degas the D2O but I do know that we usually don't find any significant peaks above 32 a.m.u. I also went through our logbook and I don't see any significant peaks around 44 a.m.u. in any previous scans we stored (the small leak or diffusion through o-rings, indicated by the N2 peak, of course also adds to the CO2 content).

When we do the next run we will do a full 0-64 a.m.u. scan to check for CO2.