Mark3 operation 3e-4 torr / quad ion source @ -40kV, 1mA-9mA

[Andrew Seltzman]

Some power supply upgrades are coming to the fusor allowing increased voltage operation. Earlier this year I got a second spellman PTV power supply on ebay for $200 with the intent of paralleling the outputs for increased current
http://www.ebay.com/itm/Spellman-Lunar- ... 7675.l2557
Unfortunately the seller had one positive and one negative polarity and sent me the positive polarity one, which had to be returned. Kuba got the other one that had the negative output.

Fortunately Bern spotted a spellman x-ray supply on e-bay which I managed to win. It's a -70kV, 8.56mA spellman power supply(DXM70N600X3547)
http://www.ebay.com/itm/Spellman-Power- ... 7675.l2557
The manuals are listed here
http://www.spellmanhv.com/DXM
http://www.spellmanhv.com/~/media/Files ... XMMAN.ashx

It has digital control of the output(serial, usb, and ethernet)

though it does use an unusual connector, a Claymount Mini Federal Standard X-ray connector, identified as a model CA11
http://www.claymount.com/en/products/hi ... /connector

I managed to find some short cutoff cables for testing of the supply
http://www.ebay.com/itm/Claymount-Xray- ... xyMZ5RnAlY
But I am looking for the screw ring for the connectors and longer cables that I can adapt to my system if anyone knows where to get them.

My grid system is in the process of being upgraded to prevent arcing from the tip of the ceramic insulator and prevent outgassing by replacing the boron nitride with macor and alumina. With the grid removed the fusor will easily pump down into the 1.5E-5torr rang and into the high E-6 range with help from the bakeout heaters and ion sources to desorb water from the walls, so the grid BN ceramic was definitely causing the vacuum problem. With the bakeout heaters on and the fusor shell in the 80-90C range, the vacuum will hold in the mid E-5torr range, so it should still have 99% deuterium purity when operating in the mid E-3torr range. Leakage through the few remaining viton o-rings still lets in some water, but the vacuum is getting progressively better.

[Andrew Seltzman]

The old grid was modified to remove all boron nitride components that were causing vacuum problems due to their porosity and water retention.
Deposition on old boron nitride grid insulator

Arcing damage to top of old grid insulator cracked the quartz tube

New grid system with macor deposition shield, alumina insulator stalk, and molybdenum field control ball

molybdenum field control ball fits over the tip of the alumina stalk preventing any sharp edge from being exposed to the plasma like on the last grid design. Sputtered metal deposition will eventually make the lip of the insulator conductive leading to arcing. The molybdenum ball fits like a cap over the tip of the ceramic rod presenting only smooth surfaces to the plasma.

The new macor bushing adapts the quartz vacuum feed through to the alumina rod and includes a deposition shield, the bushing has a machined in pump out duct to vent the inside of the grid assembly to prevent a trapped volume

Design of bushing and molybdenum cap

Operation of new insulator

Unfortunately the new design arcs at 25-30kV at 10mTorr at the base where the alumina rod meets the macor bushing. This only occurs when exposed to plasma, in a neutral gas at 10mTorr with the ion sources off to prevent any plasma breakdown the insulator will hold off 40kV. This design needs to have additional bushings outside the alumina stalk like on the boron nitride grid. These will be machined out of macor and added to the system.

The vacuum performance of this grid is perfect with no outgassing or water retention. The system will easily hit the mid 5E-5 torr with the grid installed, macor is definitely the way to go.

Preliminary tests are being conducted on the new Spellman DXM supply. It turns out the supply is brand new, with no logged hours on the HV on hours counter. The supply can be controlled and monitored over an ethernet connection, or with analog inputs.

Matlab program generating a 5kV ramp waveform of the voltage output

[John Futter]

Andrew
I have had various results with the ethernet control program as supplied by Spellman with our 6kW 140kV unit
it disappeared from our intranet when we needed to shut it down urgently due to excessive x-ray production in our experiment ie counters went over 300mSv.
So we now have a kill switch wired through the interlock connector.
Not sure why you are having so much trouble with the BN insulator.
As for o-rings some of our equip has nearly a 100 meters of viton o-ring and it sits at around 2x 10^-8 millibar base pressure with a single 300L/S turbo, the RGA does not show appreciable water signal we see a weak air signal N2 O2 H2 Co2 and heavier hydrocarbons probably fingerprints and grease ,backing pump oil etc.
Pyrex makes very good shields as it is slightly conductive and defines the "E" field with a nice gradient.
I do not like MACOR for this as it is too insulative and the field can concentrate in the wrong places IMHO

[Andrew Seltzman]

John,

I definitely plan to have a hardware interlock/kill switch on the final setup.
The problems with the BN insulator were mainly due to it having 7% porosity and a boric oxide binder, which is hygroscopic. It soaks up water vapor and then released it when it starts to get warm.

I'm trying to avoid any non-quartz glasses like pyres since they benin to get conductive when heated by direct ion bombardment. This is especially a problem at the lip next to the grid. The quarts has much less problems and can tolerate much higher temperatures, but still needs to be shielded from direct bombardment at the top lip. It's also kind of brittle and can fracture during an arc. Alumina will not fracture like quartz, but has much lower dielectric breakdown strength.

The macor is something new I'm trying due to it's machanability, I hope to make something similar to the BN insulator but with macor shields, an alumina core, and a molybdenum field control ball at the top to prevent any ion bombardment if the insulator lip. I really think the re-entrant molybdenum ball will greatly increase grid insulator life by preventing conductive deposition on the insulator lip by the grid from initiating field emission discharges.

Do you have any experience using macor under intense ion bombardment, like near the focal point of a fusor? Does it begin to melt/degrade/outgass etc...?

[Andrew Robinson]

Great to see you're still making progress on this design. We're running into some of our own operational challenges with our liquid feed through. Always great to see the challenges you are running into. It has helped us immensely in the past to avoid some costly mistakes. Thanks Andrew!

[Andrew Seltzman]

Latest upgrades to the mark 3:

A complete rebuild of the grid cooling system has been completed.
The old system was removed

and replaced with an upgraded version consisting of a compact thermometric heat exchanger designed around an all in one waterblock pump combo (swiftech apogeedrive 2) http://www.swiftech.com/ApogeeDrive2.aspx

as well as a new radiator and low noise fan

An acetal block is machined to hold a second copper waterblock that fluorinert flows through

A thermoelectric cooler transfers heat from the fluorinert to the water cooling loop

A CPU retention plate is outfitted with extension standoffs

Allowing the apogeedrive cooler to clamp down to the thermoelectric cooler and fluorinert waterblock

Dimensions of the acetal block

The newer system is considerably more compact

A stainless reservoir stores fluorinert for the cooling loop

Complete system mounted in reactor frame

[Andrew Seltzman]

Upgrades to the grid system have also been completed

The arcing problem on old grid design was traced back to the area between the alumina insulator and the macor bushing

This was only a problem when exposed to plasma bombardment. In a vacuum it would hold off 40kV, but arc at 25kV under plasma bombardment

A new design with a macor insulating cap and new bushing with a overlapping interface that prevents any scattered plasma from forming a conductive path

The new version will operate without arcing at 40kV in a plasma and will not produce any outgassing(unlike the boron nitride version) allowing continuous operation of the fusor

[Andrew Robinson]

As always, fantastic work Andrew! Both myself and my team always look forward to updates on your grid design and have been following them with great interest.

You have caught my particular interest with Macor. Will need to read up on this ceramic. Thanks!

[Andrew Seltzman]

Latest operation data from the new grid and cooling system:

The new macor insulated grid allows significantly better vacuum performance due to it's zero porosity and lack of binders. There is no discernible outgassing as the ceramics heat up, allowing continuous operation.

Latest operation at 7.2mTorr, 40kV, and 8.8mA

Generated 70 bubbles (30bub/mrem) in 60 sec at 3.5"(8.89cm) from the center

Current best neutron output is 1.08E6n/s

Next on the to do list is connect the second PTV supply in parallel to run at 40kV, 17mA, and hook up the new 70kV, 8.5mA spellman supply for higher output.

[Richard Hull]

Fabulous report and great work across the board. Few folks who have done fusion here have hit the mega mark. Probably as few as 10%. You kinda' have to stick with it once you have done fusion to get numbers up. You system is clean and very well executed. You can now easily do activation on a significant scale, far from any need for satisitcal back up.

You might try to calibrate a fixed silver activation set up against your BTI bubble detector, if that is your only neutron indicator. When the bub dosimeter craps out, you can rely on the silver activation as a crude but calibrated indication of future performance. (assuming you don't want to blow the coin needed to replace the BTI when it goes south.)

Richard Hull

[Steven Sesselmann]

Andrew,

Great engineering and incremental improvements and it's obviously working. Your fusion rate and quotient is improving with every iteration, you are now sitting somewhere between Richard Hull and Jon Rosenstiel on the list, despite running at lower voltages then Jon. I shall hold off putting name on the list until you report back again after a 70 kV run.

Steven

[Andrew Seltzman]

Best run to date is 1.4e6 n/s, at a slightly increased pressure of 7.8mtorr, 40kV and 8.8mA. 1e6n/s is now easily achievable with 1.4e6 n/s possible with some tweaking. Best to add this Q value to the list for now, it might be a while before the system is capable of 70kV operation.

The molybdenum cap under the grid is also heating up during the runs, getting into the dull red range. I might switch to a stainless cap and braze it onto the grid tubes for better heat conduction to the coolant.

Has anyone noticed how the bubbles seem to be clustered towards the center of the dosimeter? Is anyone else seeing this of their BTI?

Here is a list of Q values for university fusors, mine is currently at 4e6 n/s/kw

[Richard Hull]

Keep pushing that pressure!! Old hands who have been here for years, and still working with their fusors, will tell you.......Once you have a fusor working over 35+kv, GO FOR PRESSURE!! My very best runs are always over 10 microns and my best run, ever, of 1.58 e6 neutrons was at 42 kv applied and 15 microns! Right after doing fusion, the secret is to run and run and run and learn with each run. Learn operational technique!

We are working on a ragged edge, bull-heading fusion in a very narrow zone not found or to be had casually for the finest results. Even the finest constructed systems from the best materials have to be considered crude based on our modus operandi in doing fusion. In the end, the science allows it to happen, but operational artifice makes it happen. Improved artifice makes it happen better.

Newbs who come, see, marginally do and then leave, never develop the artifice for any number of reasons. Yes, the fusor, as we typically configure it, can only go so far, but pushing it becomes an art and craft.

I have trumpeted this for years here in numerous postings. It is good to see someone pushing their individual envelope in continued work with a good fusor system.

Richard Hull

[Steven Sesselmann]

Interesting, I just looked up some information on the SIGFE reactor and I never realised how close the construction is to my current FICS design, as in FICS the cathode is semi enclosed with ions entering through holes, forming almost a hollow cathode, it also has electron suppression. SIGFE and Hirsch design clearly stand above the rest in terms of efficiency.

The only critical mistake they made IMHO is to accelerate the ions into the cathode, obviously they believed ions need speed to overcome a coulomb force and fuse, quite to the contrary I think the ions need to stand still before they can fuse, but time will show.

Steven

[Andrew Seltzman]

Latest run updates:
2.4E6 n/s at 39.8kV, 17.68mA, 9.24mTorr

This higher current run was made possible by paralleling 2 spellman PTV power supplies to double the output current

A new plastic insulator was machined that allows 2 HV cables to be plugged into the grid via a splitter adapter. This insulator will also allow use of the 70kV supply that has a larger cable.

A control connector is then made to allow the master supply's current monitor line to drive the slave supply's current program line

Both supply's are then run in constant current mode at slightly below the voltage setpoint

There are definitely heating issues on the moly cap, with it now becoming bright red. The macor is still holding up very nicely with no discernible outgassing, even when in contact with the red hot moly cap. I'll probably have to switch to a stainless cap brazed to the grid structure if I want to continue at these current levels.

[Steven Sesselmann]

Andrew,

You are really cooking now, heading towards the 5 meganeutron mark, I trust you are keeping a safe distance.

Even at this emission rate the efficiency has remainded roughly the same.

Steven

[Andrew Seltzman]

Runs are kept short for safety(and thermal issues)
x-ray rate at the viewport surface is 10-15mR/h through the lead glass shield
neutron rate at the surface of the vacuum vessel is 140mR/h at 2.4E6n/s
Runs are 30s with the two parallel supplies, about 60s with the single supply, and I only run long enough to get good counting statistics from the BTI dosimeter, so about 2-3mR of neutrons at the surface of the fusor(~3.5" from focal point) per run. I'm usually about 1-2ft from it so it drops off a lot due to inverse square law.

[Steven Sesselmann]

Andrew,

I know only too well how fast the minutes pass when one is fiddling with the knobs on a fusor, big difference between 30 seconds and 5 minutes. At those emission rates I would want to be further away than 2 feet.

My new fusor may not achieve 5x10^6 neutrons like yours, in any case I have built it for remote operation with a 10 metre fibre optic control cable extending to my laundry behind a 2 foot sand stone wall.

Steven

[Andrew Seltzman]

Agreed, adding more shielding is next on the to do list. In this case there isn't really any way to push the runs any longer, the difference between 30s and 40s of operation is the moly cap not glowing and glowing bright red, it's a pretty reliable timer.

More automation is also in progress to allow remote operation, I bought a labjack T7
https://labjack.com/products/t7
to monitor and control the fusor. It has an ethernet interface and numerous digital and analog outputs for control.

[Richard Hull]

Great Work! I guess the best of all current operating fusors!!! Very, very clean system and star. Those ion sources are making a real impact. Plus ultra all the way.

Richard Hull

[Andrew Seltzman]

I've tracked down the source of x-rays coming from the fusor. I had initially thought the 15mR/h was some of the higher energy x-rays coming through the lead glass, or stainless shell, however it turned out to be coming from the viewport near the vacuum gauge. Some of the secondary electrons from the grid were getting up into the conflat cube and generating x-rays. A lead foil cap over that viewport has now eliminated any measurable(on a ludlum 17 ion chamber, minimum resolvable rate ~0.5mrem/h) x-ray flux from the fusor. I'll still need to add lead shielding around the vacuum chamber when I start running at 70kV, but for now the x-ray issue is solved.

[Richard Hull]

The shell starts to go transparent on the standard .060 inch thick SS shell at about 35kv applied. At the shell, working at 40kv, using a victoreen ion chamber, I get about 10mr/hr. It will only go up from there. Any window or view port will be much higher at lower voltages as the lower energy x-rays produced will pass through the glass as electrons strike the shell on the opposite side. This is much the same as old x-ray tubes.

The lead cap trick over the view port was used by Joe Zambelli in his 2001 fusor. I avoid it by pointing my view port straight down into the concrete floor through the video camera. View ports are the most dangerous emitters of radiation in the amateur fusor and all that danger is x-radiation. Neutrons from fusion are of no real consideration for the average successful amateur fusioneer as few will use more than 40kv applied.

Richard Hull

[Andrew Seltzman]

I recently scored an URSA multichannel analyzer on ebay and am hooking it up to do neutron activation analysis

The system was calibrated with a Cs137 check source

And an Am241 source(smoke detector)

All of which showed distinct peaks.

Spectra was also taken of the x-ray emission of the fusor
At 25kv (with the lead shielded cap off)

At 30kV (with the lead shielded cap on)

At 40kV (with the lead shielded cap on)

It's interesting to see how the 30-40keV x-rays are just starting to get through the 1/8" stainless shell, while the 25keV ones are blocked entirely unless the lead cap is removed.

[Richard Hull]

Great work Andrew! I have noted this above and for some years. Your work puts a real fine point on this. Thanks for the excellent report.

Richard Hull

[Andrew Seltzman]

Additional progress is being made on the liquid cooled grid to address the heating of the field control cap that covers the lip of the ceramic by brazing it to the cooling lines on the grid. Cusil brazing alloy (72% silver and 28% copper eutectic) was used in the brazing with a boron modified flux(Superior No. 601B/3411)

The allow has reasonable wetting properties on stainless and only contains vacuum compatible materials(no zinc or cadmium)

A test braze was completed on an older grid, shown after sandblasting and some cleaning

The braze seems structurally and thermally sound and generally wetted ok on the stainless, though the gap may need some tuning to get good flow down into the area between the cooling lines and the cap.

Additionally, the neutron detection system is being upgraded to use He3 detectors instead of the older BF3 detectors that were previously in use. The new detector is about 7x more sensitive then the BF3 tube on the system. This should allow the time constant on the constant neutron flux control system to be reduced for tighter regulation while still remaining stable.

[Steven Sesselmann]

Andrew,

It's really good to see you make these small incremental improvements it means the technology is being refined and becoming more mature.

By the way I have one of those LND He3 tubes, and I'm interested in knowing where you got the connector. Does it come like that with the threads or is it something you have custom made?

Steven

[Andrew Seltzman]

The connector is a TNC bulkhead mount adapter to clamp on coax, the clamp on section has the same 7/16-28 as the TNC and the LND tube. the other fitting is an Amphenol 79675 Adapter TNC to BNC adapter, a small spring connects the pin on the end of the He3 tube to the center pin on the other adapter

[Andrew Seltzman]

Pictures of the molybdenum cap glowing under ion bombardment after about 40sec of operation at 17mA

Pictures of the molybdenum cap glowing under ion bombardment after about 40sec of operation at 17mA

The old grid was removed and inspected prior to the upgrade to the version with the brazed cap. The macor ceramic has held up extremely well with no signs of thermal or arcing damage.

A cooled grid with a brazed field control cap has been installed and is currently being tested

The new grid was tested for several minutes at ~2.5e6n/s with no discernible heating of the field control cap

The current limit for operation time is now the heating of the vacuum vessel which will require some additional fans to keep cool .

[Richard Hull]

I have never really worried about the shell heating which tends to reach about 200 deg F after 1.5 hours of running. (electron bombardment and neutral crash area).

Richard Hull

[Andrew Seltzman]

Mine seems to have localized spot heating near the ion sources. I think the fraction of the deuterium beams that is neutralized or charge exchanged at the focal point is being accelerated outward and striking the shell near the injector ports on the opposite side.