I'm back, and fusing

Current images of fusor efforts, components, etc. Try to continuously update from your name, a current photo using edit function. Title post with your name once only. Change image and text as needed. See first posting for details.
Post Reply
User avatar
Liam David
Posts: 518
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: PPPL

I'm back, and fusing

Post by Liam David »

It's been a few years since I posted in these forums, though recent progress on my fusor and a free summer have demanded I return. I have likely been grouped with the scores of youngsters who arrive here excited but fade away having failed to achieve fusion or lost interest. I want to remove myself from that group. Moving cross-country and a year in college can change things, but my interest in fusion remains. This rather long post will serve as an update on my progress, description of my setup, as well as a fusion claim that succeeds what I attempted a couple years ago.

System description
The main fusor chamber is a 6-way 2.75" conflat cross. The front and bottom ports are viewports, the back port has a right-angle valve which I will use to decrease cycle times, and the HV feedthrough drops down from the top. A tee on the right connects through a right-angle valve to a 50l/s turbo pump which is isolated from the system at the moment due to its rampant outgassing. The other side of the tee connects through a radius right-angle bend to a Pfeiffer PRK 251 full range gauge. Opposite the tee is a 4-way cross with a viewport, secondary HV feedthrough, and butterfly valve which connects to a RGA and a second 56l/s turbo pump. The plan, once I outgas the first turbo, is to differentially pump the RGA and chamber to allow for measurement of the gas plasma composition at pressures ~100x higher than the RGA can usually handle. Before I realized the extent of the outgassing I tested the system and found that I could maintain ~5e-6 torr in the RGA with the main chamber at 100mtorr. Both turbos are backed by a 2-stage rotary vane pump through KF16 fittings and hoses and the pressure is monitored by a 275 mini convectron gauge. The base pressure is in the low e-6 after several hours of pumping and the RGA spectrum confirms that I have no leaks. Water vapor comprises over 95% of the partial pressure so an eventual long bakeout should bring the chamber to e-7 or e-8torr, especially with dual turbos. Everything is mounted on a 3030 aluminum extrusion frame and some shelving brackets.

Image
The x-ray transformer bucket, Spellman supply, neutron detector, etc.

Those present in 2016 may remember the drama that unfolded with the batch of somewhat functional Spellman DXM70N600 supplies that showed up an Ebay. These units are capable of -70kV at a 8.56mA hard current limit and one serves as my primary power supply. It connects through a 60k 100W wirewound resistor to the HV feedthrough and is controlled by a custom box based on the design discussed in the power supply thread. Said feedthrough is loosely based on Andrew Seltzman's design and consists of several nested alumina tubes up to 0.75" OD and a central threaded conductor to allow for fast grid swapping. A 0.75" Swagelok cap serves as the HV connection and seals against the tube with an o-ring.

The primary grid (there's more than one) is of a different design than most here would opt for, consisting of graphite and tungsten in a cylindrical shape. It is 0.5" OD and about 0.75" long. The feedthrough side is tapped for an 8-32 thread and has four 19mil holes that friction fit 20mil pure tungsten welding elctrodes. The other graphite side has four matching holes and is bored at 3/8" to maximize transparency. I was having issues pushing too much power into two loops of tungsten wire and their tiny surface area and thermal mass rendered them white-hot at just 5mA, which, as has been discussed many times on the forums, reduces efficiency through electron emission. I was less worried about efficiency than melting the wires, however, so I opted for a more robust design. Moreover, I have run out of tungsten wire and GalliumSource LLC, which offered it at $8 for ~25ft, has long since gone out of business. Paying $150 elsewhere is a no-go. I rather liked Doug Coulter's design and decided to replicate it. So far the asymmetric but nonetheless functional grid has no problems handling the current. Fabrication was tricky since graphite is brittle and the parts have to be friction fit, but the design will no doubt improve as I make more.

Image
Feedthrough modeled in CAD.

Image
The graphite and tungsten cylindrical grid mounted in the chamber.

I have also constructed a mineral-oil potted x-ray transformer supply capable of 50kV. It arced over several years ago and destroyed one of the secondary windings, but I have rebuilt it completely and it works fine now. My intent is to use it with the other feedthrough and a secondary grid much to the same effect as Doug (I'm sensing a theme here...), but the full list of reasons is not worth writing here. Those interested can head over to his forums. The grid, which is the standard wire-loop design, acts as a Faraday probe and I found it to be floating at about -60V at -7kV and 5mA on the main grid. Unsurprisingly the potential rises with grid voltage. It's nothing particularly interesting, but perhaps it will serve some other purpose down the road.

The neutron detection system consists of a SHM18 1" x 12" 3He tube surrounded by 1" of paraffin wax moderator and coupled to a Ludlum model 12. It operates 1750V and the discriminator is set to about -2mV. Without a pulser calibrator the exact value is little more than an educated guess based on potentiometer turns, though given how the detector performed during a fusion run it will need to be tweaked. More on that below. The tube sits directly in front of the chamber with the central axis 4" from the grid.

X-rays beam out of the viewports over ~12kV so I have all three viewports shielded in lead as well as a lead vest to drape over the whole setup. They like to leak out of every tiny hole in the lead so I spent some time hunting leaks with a couple pancake GM counters. GM counters over-respond to x-rays of these energies by nearly an order of magnitude and make it sound much scarier than it actually is. A Ludlum model 9 confirms that the dose rate is on the order of 1-2mrem/hr about 8" from the grid as measured along the axis of the secondary feedthrough. I operate the device just a few feet from the action and have measured no x-rays there nor where observers can stand.

The deuterium system is pretty standard. A 50L lecture bottle is stepped down to 2psig with a 2-stage regulator and feed through drierite (which is not needed since the deuterium is dry, but serves as an up-to-air filter) a 0.05cv needle valve, and a shutoff valve into the chamber. The lines are all stainless since the original copper tubing coupled with the stainless Swagelok fittings proved leaky. The ferrules would spin on the tubing and fail to make a good seal. They make brass fittings for a reason...

Fusion claim
In late 2016 I first attempted fusion but had little success for a number of reasons. The Spellman supply, with its arc-detection and overcurrent circuitry, was difficult to control and I ended up making more EM interference than neutrons. Poor plasma purity, the low voltages I could maintain, and any neutron counts being drowned in noise compounded into a failure. Sure I may have fused a few deuterons, but no definitive proof = no fusion.

Yesterday, July 5, 2019 I achieved fusion and have the data to prove it. The parameters/procedures were as follows:

I pumped the chamber to about 5e-5torr with the turbo pump on standby (66%) and then mostly closed the butterfly valve to conserve deuterium. Achieving the full base pressure was unnecessary.

A Ludlum model 3 with 44-9 probe was placed on-axis with the secondary feedthrough for monitoring of x-rays and the neutron detector was turned on to establish a baseline of <50CPM. A Ludlum model 26-1 frisker was used to ensure no x-rays were present on the operator side at any time during the run and a lead vest was positioned to absorb rays emanating from small holes in the lead viewport caps. Since everything was covered in lead, the plasma was observed with a webcam and laptop.

I cracked the deuterium tank, put ~500psi in the regulator body, and then closed it. The volume in the regulator stem was more than enough for about an hour of fusion. The gas line was set to 2psig of deuterium and the shutoff valve and needle valve were opened. I established a stable pressure of 14-15mtorr (corrected for deuterium).

After removing the ground clamp from the feedthrough (the Spellman has some output capacitance) I raised the voltage first to 20kV and then 30kV as I got the hang of it. As mentioned before the safety features of the Spellman supply can be tricky to handle, and the plasma extinguished numerous times due to over-current and arc protection. I'm leaving out much of the nuance of balancing the pressure, voltage, and current, but years of tinkering with these parameters afforded me decent control of the system. There was leakage and somewhere in the HV line above 30kV so I had to call it quits after a couple runs for safety reasons. Once that's fixed I should have no trouble pushing 40kV at 14-15mtorr.

I removed the neutron detector from the moderator as proof of fusion once I established a stable plasma. CPM were measured over about 30s, both in and out of the moderator. The fusion rates are a very rough estimate based on the neutron moderator paper Carl Willis posted in (2008?). The formula I used is
S [n/s] = CPM * R^2 / (60 * eff * d * L * phi * (20+T)^2)
where CPM is as measured by the detector with background and noise baseline subtracted, R is the average radius from the tube to the grid (13.2cm), eff is the tube sensitivity to thermal neutrons (~0.5), d and L are the tube dimensions (2.54cm, 30.48cm), phi is the neutron fluence factor for 2.54cm of paraffin as extracted from Carl's paper (~9e-6), and T is the moderator thickness (2.54cm).

As for the actual data:

Run 1 (Noise baseline):
Pressure: 15.21mtorr
Voltage: 20.16kV
Current: 1.45mA
Power: 29.3W
CPM in moderator: ~500 (vast majority noise)
CPM out of moderator: not measured
n/s: ~0

Run 2:
Pressure: 14.03mtorr
Voltage: 28.0kV
Current: 1.97mA
Power: 55.1W
CPM in moderator: ~2000+-100
CPM out of moderator: ~500 (noise+background)
n/s: 24,600

Run 3:
Pressure: 14.03mtorr
Voltage: 31.4kV
Current: 2.31mA
Power: 72.64W
CPM in moderator: ~5000+-100
CPM out of moderator: ~500 (noise+background)
n/s: 73,800

Video:
https://youtu.be/goe6VyJwm5o

The rates are not too impressive, but for the input power and for a first attempt they seem reasonable. I need to increase the discrimination of the neutron detector to remove the ~500CPM of noise it detects, but a proper calibration is in order. Once I achieve a higher fluence I will activate some indium.

As per good practice please scrutinize my setup and ask for clarification on anything. I ask to join the neutron club.

Some pictures:

Image
Non-fusion plasma at low voltage (<10kV). At voltages below where I require x-ray shielding, the graphite has a distinct orange halo, but not the tungsten rods. Anyone have an idea what it could be?

Image
Fusion plasma and copious x-rays bombarding the camera. It has several dead pixels now.

Image
The setup while doing fusion.

Image
Bugle jet mode at high pressure.

More runs are planned for the next few weeks.

-Liam David
User avatar
Richard Hull
Moderator
Posts: 14991
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: I'm back, and fusing

Post by Richard Hull »

Great work! You will do better, I am sure. I have added you to the neutron club.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
chad ramey
Posts: 107
Joined: Tue Apr 28, 2009 10:45 pm
Real name: Chad Ramey
Location: Georgia
Contact:

Re: I'm back, and fusing

Post by chad ramey »

Hi Liam,

First of all, congrats and welcome back!

As a fellow owner of one of the ebay Spellman DXM70N600 supplies, I am curious about how you are connecting from the supply to your feed through. Could you provide some details about how you're physically plugging into the supply?

Thanks!
-Chad
User avatar
Liam David
Posts: 518
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: PPPL

Re: I'm back, and fusing

Post by Liam David »

Funny you should ask that. I 3D printed a connector out of ABS a couple years ago but once I pushed ~35kV there were arcing issues. It finally bit the dust just a couple days ago and left a carbonized trace. The socket cannot be run to any appreciable voltage unless there's a proper connector installed, and the printed plastic just didn't cut it even with HV cable extending almost the whole way in. I have a new UHMW (~2kV/mil) connector design based on the ones discussed in the old thread but unfortunately do not have a lathe and thus have to outsource its construction. Submitted the quote this afternoon and still waiting.... It will have o-rings to seal the perimeter and I'll likely pot the connector in RTV silicone or the like to give it no excuse to arc.

Image

The other end of the HV cable is soldered to a ring terminal which goes straight to the ballast resistor and feedthrough. At 30kV and above there appears to be significant charge buildup on the outer surface of the unshielded HV cable I use. It occasionally discharges and trips the arc fault circuitry so I might add a grounded braided shield. Speaking of arc sensing, have you been able to access the applet via ethernet? I've had no success on any Windows OS nor have I been able to communicate with it via MATLAB. The software allows you to program the arc sensitivity which should help with stability.
chad ramey
Posts: 107
Joined: Tue Apr 28, 2009 10:45 pm
Real name: Chad Ramey
Location: Georgia
Contact:

Re: I'm back, and fusing

Post by chad ramey »

Thanks for sharing, that is very interesting to hear!

I've chatted with Peter Schmelcher in the past about his designs for a 3D printed Claymount connector for these and even got his STL files so I can print one and experiment but I haven't had time to actually get around to it yet. Based off of my experience with 3D printing, I'd expect that all the little air pockets and voids that happen in the FDM process can add up to really unpredictable breakdown paths through the part. Perhaps an SLA printed version of these connectors would fare better? We've got a ProJet MJP 3600 at GaTech which can print in some funky materials which may be of some promise for this application.

Further, I've thought about 3D printing the connector as a positive and then pulling a silicone mold and casting the part in ceramic (via something similar to slip casting).

Anyways, too many ideas, too little time. I digress.

In a similar vein as the connector ideas, I've had lots of ideas for digitally controlling the supply itself but haven't had much of a chance to play around with it outside of the little applet. My first priority when I do get around to tinkering with the supply more will be to rewrite Andrew Seltzman's MATLAB stuff in Python and hopefully extend the functionality a bit more.

Thanks for sharing. Cheers!
User avatar
Liam David
Posts: 518
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: PPPL

Re: I'm back, and fusing

Post by Liam David »

FBM printing leaves too many irregularities and any hope of preventing breakdown would require 100% fill. That would really take some time to print. I suspect that SLA would work better, but I don't know about the dielectric strength of the resin. ABS is pretty bad at ~500V/mil, but that's likely much lower when printed.

While casting in ceramic sounds interesting, it would require several improving casts as you learn the technicalities and final dimensioning on a lathe to obtain a quality result. If you have a lathe, why not just chuck a piece of UHMW and turn a connector in under an hour? Pot it in silicone and there should be no issues. I, having no lathe and with my university unwilling to provide machine shop access unless actively researching under a professor, got a quote for a simple, single piece connector at >$200 from Protolabs. I guarantee one of their fancy CNC-lathes could turn it in under 10 minutes, but oh well. That's more than buying a connector/cable assembly at ~$150 directly from Claymount. Will have to see where I go from here it's preventing me from doing more fusion and pushing 40kV.

Was the IP of your supply 192.168.1.4?
Post Reply

Return to “Images du Jour”