A New Cube, or Two

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: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

A New Cube, or Two

Post by Liam David »

Over the last year, I've designed and built two systems that fall well within the "cube" category. I didn't post any intermediate updates for a variety of reasons, but I'm finally ready to share what I've been working on. Enough of the system has changed to justify describing the entire setup. Hopefully, having everything in one place will also help newcomers. I'll go through my rationale for various design decisions and some of the problems I encountered. Buckle up or skip around: this is a very long set of posts.

Part 1: Support Systems Overview

Vacuum System
The roughing pump is a Varian SD451 14.7 CFM dual-stage rotary vane pump with an ultimate pressure of ~1.5 mtorr. The exhaust is filtered by McMaster #9850K55 since I work in a small garage and often have it running for days/weeks at a time. A stainless steel mesh foreline trap reduces oil contamination upstream. Directly after the trap is a tee with a manual vent valve.

PXL_20220804_204830032.jpg

The rough valve is a KF 16 normally closed solenoid valve to protect the turbos in case of power failure (the SD451 internal anti-suckback valve is not sufficient, closing too slowly and leaking enough to pressurize the line to multiple torr immediately). Its conductance is not the best (3/8" orifice) but it has proven sufficient for my purposes.

PXL_20220804_204835495.jpg

The whole roughing line is KF 16 / KF25 stainless and I monitor pressure using a Granville-Phillips 275 mini convectron with digital display.

PXL_20220804_204958655.jpg

High vacuum is generated by two turbomolecular pumps: a primary TPU-062 and a secondary TPU-055, with pumping speeds of 62 and 55 l/s at the high vacuum flange. These are controlled with TCP-121 and TCP-040 controllers. These are used pumps with unknown histories (prior to the ~5 years I've had them) and so I run at 66% speed (standby mode) to preserve the lower bearings. Both are fitted with a splinter screen. The primary turbo is connected to the vacuum manifold + chamber through a UHV-type GV-1500 gate valve which I had to repair (viewtopic.php?t=14523). It acts as the main throttle valve to achieve workable fusor pressures.

PXL_20220804_204900886.jpg
PXL_20220804_204854993.jpg

The secondary turbo is used to differentially pump an RGA, model Vacscan Plus from Spectra Instruments (bought by MKS). It's from the late 1980 (literally among the first such systems sold) and has its own rackmount controller, and I miraculously got it working again following a slow death spiral. Details here: viewtopic.php?t=14523. The usefulness of an RGA cannot be overstated: deuterium purity measurement, leak detection, bakeout progression, secondary pressure measurement, etc. are all valuable data for improving a fusor. The RGA is connected to the vacuum manifold through a 2.75" conflat right-angle valve from MKS.

The vacuum manifold is a 6-way 2.75" conflat cross. It was my first fusor chamber. Viewed from the front, the RGA valve is on the left, the gate valve right, a VCR adapter in front, a Pfeiffer PKR-261 gauge in back, a viewport below, and the chamber above. The straight line of sight between the chamber and viewport allows me to monitor the plasma using a camera. The PKR gauge is a combination Pirani + cold cathode that measures from atmosphere to 3e-9 torr and serves as my high vacuum measurement device. I documented gauge cleaning and bakeout troubles here: viewtopic.php?t=14523, viewtopic.php?t=14519.

system CAD 2.PNG

All conflat seals are copper gaskets. Without the chamber attached and after a ~4 hr bake at 130C, the ultimate pressure is ~4e-9 torr, which is close to the limit of my gauge. With the chamber and a modest bakeout at 70C for ~24 hrs, I am able to attain ~5e-8 torr. There are no real leaks, meaning that the pressure is dominated by water vapor and can be lowered by further bakeout. I enacted strict cleaning and handling procedures to minimize contamination during the assembly, and I was rewarded with an ultimate pressure over 100x lower than my previous best. Procedures here: viewtopic.php?t=14519. I use two Omega CN 1611 temperature/process controllers, thermocouples, and two 312 W / 15 ft Amptek fiberglass heating tapes for bakeout, and there's an over-temperature alarm for when I bake overnight.

PXL_20220804_204852098.jpg

The whole system rests on a frame made of 3030 aluminum extrusion.

Deuterium System
Deuterium purity is of understated importance if you want the highest neutron rate. Even a fraction of a percent of the wrong molecules can cut your fusion by many multiples, if not an order of magnitude. My deuterium gas line is thus designed with purity in mind. Hence, all but one connection is VCR. A 50L lecture bottle is connected to a Harris 2-stage hydrogen regulator and then a 1/4" VCR diaphragm shutoff valve. The downstream side branches into two paths with a mass flow controller on one and another diaphragm shutoff valve on the other. Downstream from the MFC is yet a third such valve. Both paths then rejoin and connect to the chamber. Another tee junction connects to a teflon ball valve and micron filter for venting.

The valve parallel to the MFC is a bypass valve, included so that I can quickly pump out the gas line. Small orifices within the MFC absolutely kill conductance and so without the valve, the gas lines are never effectively pumped. The valve after the MFC acts as a shutoff since most MFCs have a small thru-leak (order 1e-6 torr l/s for mine).

The MFC is a Brooks model GF-100 with an integral shutoff valve and a 0-10 SCCM range. I reprogrammed it for deuterium and its mounting orientation using the RS-485 interface and Brooks software.

The gas line is heated with the rest of the chamber during bakeout and is pumped all the way back to the lecture bottle.

PXL_20220804_204938943.jpg
deuterium CAD.PNG

High Voltage System
Cathode high voltage is generated using a Teslaman model 2202 -100kV / 20mA switching supply. A standard XLPE cable (not coax) brings the HV to a 200k / 100W ballast resistor and then to the feedthrough. The connections are electrostatically shielded using 5" spun toroids.

PXL_20220804_204850148_mod.jpg

Neutron Detection
My primary neutron detector is a CHM-18 Russian proportional tube encased in a 3cm thick paraffin cylinder. It's driven at +1500V using a Ludlum Model 12 that I modified to provide 5V TTL pulses for every detection event (viewtopic.php?p=88956#p88956).

PXL_20220804_205009285.jpg

There are two other detectors worth mentioning, in order of increasing importance. First, I occasionally use a small CHM-56 Russian proportional tube with very little (~3mm) paraffin moderator to measure the neutron spatial distribution. With so little moderator, it is not very sensitive. I also drive it with a Ludlum Model 12.

The other detector is worth a post of its own: a completely custom Hornyak-style fast neutron scintillator based loosely on https://sci-hub.se/https://doi.org/10.1063/1.1770887. It is somewhat similar to Jon's. I measured the efficiency at ~2%, right in line with the linked paper, and I intend to better characterize it with my new system. I'll also use it to measure any anisotropies due to beam-target fusion. It's driven by a NIM system with a Canberra 3012 2kV bipolar supply, Canberra 814 preamp-amp-discriminator, and an Ortec 974 counter/timer.

PXL_20220804_205150525.jpg
PXL_20220804_205042868.jpg

I occasionally use a Ludlum Model 2363 to keep an eye on neutrons near the operator station.


Control / DAQ System
Monitoring and controlling everything from my remote operator station, which involves fiddling with a few potentiometers and watching the system diagnostics on a computer, is a massive upgrade over doing everything manually at the device. It improves the operational quality of life and ease of use, not to mention its high voltage and radiation safety benefits.

The brain of the system is an Arduino Uno clone with built-in ethernet, the Keyestudio W5500. It's nothing special, but it acts as an ethernet server to my computer client and interfaces with all sensors and counters. Two standard Arduino Unos act as neutron scalars and communicate with the main board via SPI. The cathode voltage, pressure, gas flow, etc. are read using three ADS1115 16-bit ADCs. The MFC is controlled using a 12-bit DAC to allow for closed-loop control. All measurements and control I/O are buffered with op amps and I added a latching circuit for additional high voltage safety. Everything is mounted in a box on the fusor frame. A small ethernet switch connects the Uno clone to an ethernet-fiber converter.

The 30m fiber, along with a 24-wire multiconductor cable pass through a braided umbilical to the operator station, where the fiber is converted back into the standard RJ45 plug and then connects to my computer. The multicore cable contains lines for the HV enable, voltage/current setpoints and readouts, MFC flow control, and a couple of other things. A small box contains the potentiometers, readouts, and converter to make a tidy package.

The Uno clone is queried at regular intervals, usually 100ms, by a Matlab GUI that contains all necessary readouts and indicators needed for my fusor to function well. A Raspberry Pi running MotionEyeOS captures video from a standard camera module and streams it to my computer.

PXL_20220804_204924641.jpg
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Finally, we arrive at the more interesting stuff.

Part 2: The Cubes
Cube V1.0

I designed my first cube in the latter half of 2021 and had a Chinese company machine it in December/January. There were many things wrong with it; some were poor design decisions or implementation, and others were negligence by the manufacturer.

The body is an 8cm cube machined from 6061 aluminum and bored out to a diameter of 5 cm. It has four 2.75" conflat ports and two endcaps since it was designed for a linear cathode. The vacuum connects on the bottom, HV comes in the top, and the remaining two conflats are for viewports. The threaded holes contain 1/4" x 28 Helicoils for improved thread performance. One striking feature is the multi-hole pattern on the viewport flanges. This was done in an attempt to eliminate the off-axis beams that I and some others have noticed (viewtopic.php?p=90557#p90557), although all it did was generate 8 smaller beams. One could have seen that coming...

cube 1.PNG
PXL_20220121_205806457.jpg

The aluminum endcaps are removable and are fully water-cooled. They push up against the copper endcaps that actually hold vacuum, which I call anodes for distinction's sake. Tightening the bolts seals both the vacuum and water. Water flows from one endcap to the other through two channels in the cube body, which are sealed to the endcap using gasket material, and enters/exits the cube through two hose barbs.

The feedthrough was constructed from 1.5" OD borosilicate glass and a 2.75" conflat to 1.5" compression adapter. The central conductor consisted of a 3/8" copper tube with a machined copper plug brazed onto the cathode end. One might notice that the glass butts up against a step in the chamber and does not extend into the chamber cavity. This was done for three reasons. First, it prevented the glass from being sucked in due to atmospheric pressure. Second, it avoided direct ion/electron bombardment. Third, it avoids the electric field enhancement that results from inserting dielectrics between conductors and hence reduces field emission/arcing.

PXL_20220126_234603530.jpg

Another striking feature might be the presence of a third conductor in the feedthrough. It is, in fact, a "second stage" that is biased at some fraction of the cathode voltage. The curvature of surfaces, and in this case the cylindrical center stalk, enhances the field and gives it a 1/r dependence. An intermediate electrode decreases the peak field for a given voltage between the cathode and chamber, provided certain conditions are met. Thus, field emission and arcing can be greatly diminished The inspiration for this addition was partially from https://www.tandfonline.com/doi/abs/10. ... T11-A12453 and https://www.worldscientific.com/doi/abs ... 35455_0023.

Problems with v1.0
All these things were great in theory, but as with most engineering efforts, problems soon arose.

Straight away, when I received the parts, I found that the copper anodes had been machined from brass instead of copper. Moreover, the main cube body contained through-holes that were supposed to be blind and many surfaces had a poor finish. Thankfully, the holes were in the water portion of the cube, and sandpaper exists. Not long after, I learned that copper sputters very easily, far more readily than aluminum or stainless, for example. This can quickly ruin insulators and cause other issues. Not only would my copper anodes have been problematic, but the brass ones I received were now doubly bad (copper + zinc). I decided to electroplate them with nickel as a temporary solution before the manufacturer sent replacements. While the plating went well and adhesion appeared decent, under plasma bombardment the nickel was quickly sputtered away, revealing the base metal again. Another plating fared no better.

PXL_20220123_224852354.jpg
PXL_20220131_171940043.jpg

The feedthrough termination method and issues with the tolerances ended up causing arcing at some 45-50 kV. The glass came poorly score-snapped, oval by ~0.5mm, and slightly oversized, and I resorted to using a diamond dremel wheel to square it up and reduce the OD. I believe that microscopic grinding remnants, as well as surface field effects caused by the ground surface finish, caused the arcing. Moreover, the interface between the glass and chamber formed an unshielded triple junction which did not help matters.

PXL_20220210_224529488.jpg

The second conductor in the feedthrough had several issues all by itself. First, I biased it using a 100 megaohm-scale resistor divider, which limits the current that it can source/sink. Under plasma bombardment, the voltage likely varied wildly, although I neglected to measure it directly due to extenuating circumstances. Second, the sharp end of the tube enhanced the field. This is something that you can really only fix using simulations and I did not take it into account. Third, the surface finish was poor. Fourth, the oval glass caused nonuniform and enhanced fields.

The imprecise glass, in addition to a slightly bent center conductor, resulted in poor cathode alignment despite my best efforts with a custom jig and coaxial camera. This problem was magnified by the overly-long glass pieces that I used. As a result, the beam quality was often poor and positioning was not repeatable.

I did not adequately clean the chamber and feedthrough, causing bad outgassing (I could even smell burnt oil upon opening the chamber post-run) and arcing.

Some of the o-ring grooves were improperly sized and inadequately polished.

I had aluminum galvanic corrosion (pitting) within the water channels, caused by having aluminum, stainless, nickel, brass, and hard tap water all in intimate contact. This didn't affect the plasma in any way, but it's unsightly and would certainly cause problems in the long run.

IMG_20220804_142732.jpg

The last problem worth mentioning was the high breakdown pressure of the device. The anode-anode centerline distance was just 62mm, and by application of Paschen's law (although its usefulness is limited in cube-like systems), the breakdown pressure is much higher than, for example, a 6" sphere. The comparatively small aperture of a linear cathode as compared to a traditional geodesic grid exacerbates the effect. Typical operating pressures were >20 and often >35 mtorr. This is not ideal for a variety of reasons that I will discuss below.

I managed to briefly (for a few seconds) eek out 2.5e6 n/s at ~60 kV, 8.56 mA, and 20 mtorr before arcing limited my voltage. Typically, I operated <1e6 n/s and sometimes at just 1-5e5 n/s.

Redemption - Cube V2.0
The first iteration can only be described as a failure of design and implementation, but it was not a total waste of time. Starting in late 2021 but with much more focus in 2022, coincident with my foray into "cubes," I dug deep into IEC and compact fusion research, high voltage engineering and general vacuum phenomena, taking advantage of journal access through my university. Combined with the hands-on experience, I gained a much deeper understanding of how fusors work and some insights into how they can be improved. None of these ideas are anywhere near revolutionary--they would be better described as simple tweaks, albeit with deliberate and careful engineering justification.

The new cube is slightly larger than V1.0, at 80 x 80 x 110mm, to reduce the operating pressure. All threads are Helicoils and I changed the cube alloy to 6061-T6. The multi-hole viewports are replaced with simple holes covered by a fine stainless mesh. Both endcaps are water cooled and the anodes contain zirconium-plated inserts that serve the same function as titanium: enhancing beam-target fusion. The now-aluminum anodes (~79mm apart) are designed to harden the deuteron energy spectrum at the cathode, although this simultaneously reduces the mean energy of negative ions at the anodes. The still-linear cathode has some key features that further enhance the device's efficiency. Replacing the glass feedthrough is a custom two-stage design that is engineered to minimize surface fields and other deleterious things such as triple junctions. It is designed to handle a minimum of 150 kV, although more should be possible. The second stage will be actively biased using my Spellman DXM70N600 -70 kV, 8.56 mA power supply. Unfortunately, I'm still waiting for the custom ceramic to arrive; in the meantime I'm using a standard 30 kV feedthrough. It handles up to 60kV in air with the toroid.

cube 2.PNG
cube 2 temp.PNG

I took utmost care in the cleaning and assembly stages, using the same general procedures I linked above. Furthermore, the cooling loop uses distilled water with "water wetter" additive that is typically used for car radiators. No metals other than aluminum contact the water.

PXL_20220804_205016968.jpg

Cube V2.0 Performance
I'm still running tests and making tweaks. I'll update this and make an announcement below when I have a better understanding of things. In the meantime, a metric I can give is that I've activated silver to >36,000 CPM.
Last edited by Liam David on Fri Aug 05, 2022 2:02 am, edited 2 times in total.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

I'll likely move this section to another thread.

Part 3: A Summary of IEC Theory
The popular view of how fusors work, most common among newcomers and true believers in "recirculation," is (perhaps unfortunately) not how they actually work. The visions of Farnswork, Hirsch, Meeks, and other early IEC developers revolved around ideas of recirculation and the generation of multiple virtual cathodes and anodes, the latters' existence of which was derived primarily from idealized theory and some interpretations of fusion product and x-ray distributions (single virtual anodes certainly exist, however). I want to make clear that I do not know much about who believed what and when. The history has already been explored by Richard, Paul, and other archivists here (e.g. viewforum.php?f=70), and I'll leave it to the curious reader to examine.

How It Doesn't Work
Why are recirculation and virtual cathodes/anodes myths?

First, consider recirculation. We should examine the cross-sections of various processes between D2+ (better written (D_2)+, by far the dominant species in most fusors) and D+ ions with the background D2 gas. For brevity, I will examine just one. A particularly devastating reaction is charge exchange (D2+ + D2 --> D2 + D2+), which prevents the electric field from further heating the original ion and generates a new, stationary ion further down the potential well. Using data from https://iec.neep.wisc.edu/usjapan/13th_ ... kajima.pdf, which was pulled from a now-dead Oak Ridge University database, we note that the cross-section of a 50 keV ion is ~1.7e-20 m^2. Let's now assume the background pressure is 10 mtorr (generously low) and that the temperature is 373 K (100C), roughly that of a hot chamber. Since fusors are weakly ionized (gas density >> ion density, by some 6 orders of magnitude), this gas temperature is a reasonable approximation. Using the ideal gas law, we obtain a background number density of 2.6e20 /m^3. The resulting mean free path (MFP) is 22 cm... not bad, but consider a typical 6" (15 cm) fusor. We see already that on average, an ion won't make a single transit before charge exchanging. Moreover, ions are obviously not instantly accelerated to 50 keV. The cross-section at 10 keV is already ~4.5e-20 m^2 and at 1 keV it's ~6.3e-20 m^2, meaning the typical MFP is <<10 cm.

It gets worse.

A large fraction of the ions are generated by electrons, which have a D2+ ionization cross-section that drops off sharply above ~1 keV. Therefore, most electron-generated ions are already deep in the well.

It gets worse still once you remember that this is only one reaction of several.

You might be tempted to say that we can just reduce the device size to the order of the total MFP, but a glow-discharge fusor will require a higher pressure to ignite, making it a moot point. These are the very same cross-sections that dictate the dynamics of that glow discharge. See also some of my early efforts to simulate these effects at viewtopic.php?t=14157. There are some tricks one can play to improve things, but that is for another time and place.

Perhaps I was a little harsh in dismissing multiple virtual cathodes/anodes as a myth. Indeed, some measurements have been performed that directly measure a more complex potential structure within the cathode, e.g. https://doi.org/10.1063/1.5107471. These multiple wells can indeed trap low-energy ions and electrons, at least until they are (quickly) destroyed by interactions with the background gas. The theory behind multiple wells (even infinite, in some cases) is based on perfectly symmetric, idealized models of electrostatics and ion flow in perfect vacuums. See, for example, https://doi.org/10.13182/FST01-A180. See also https://doi.org/10.1088/0029-5515/38/4/302 for an example (although with well-defined assumptions and limitations) of simple ion convergence and the dubious "transparency factor." Regardless, such virtual structures cannot contribute directly to fusion due to the low-energy and short-duration trapping. There is some merit in their ability to aid in space-charge cancellation, however.

(Approximately) How It Works

Some key mechanisms were hinted at in the previous section, and I'll attempt to flesh those out and add some more here, with a caveat: the number and variety of processes that occur in a fusor mean that I will have to greatly simplify. I'm not claiming to have the final word here, it is more of a working summary of what I've learned. Much is subject to revision and clarification, and much is unknown to everyone.

The typical positive ion, namely D2+, D+, and D3+, in decreasing order of their abundance (although this ranking is often permuted), is generated and destroyed by collisions with other ions. Electrons contribute to ion generation near the cathode and also recombine with ions, giving the characteristic purple/pink glow of a clean fusor plasma. The negative ions D2- and D- are also produced. Recombinations and charge-exchange reactions can generate fast neutral or slow D2 and D particles.

Positive ions accelerate toward the cathode, interacting frequently with the background gas and usually being destroyed before they can attain their maximum possible energy, which is based on their starting positions. The reaction probability is dictated by the cross-section evaluated the energy per nucleon, hence a 50 keV D+ ion is much more likely to fuse than any of the atoms in a 50 keV D3+ ion. The mean energy at the cathode is typically 1/5th or less of the applied voltage. Many ions impact the cathode, becoming embedded and quickly neutralized by the metal lattice. Depending strongly on the cathode material, surface finish, and ion energy, an average of 0.1 to >>1 electron can be emitted per impacting ion. These "secondary electrons" help sustain the glow discharge, although simulations suggest that ions play the primary sustaining role.

As they fall inward, the ions have a chance to fuse with background deuterium molecules in the so-called beam-gas pathway, or with deuterium embedded in the cathode (beam-target). Fast neutrals react likewise (neutral-gas, neutral-target). Fusion scales as (number density of reactant 1) x (number density of reactant 2), so if we assume the aforementioned ion/gas density ratio of 1e-6, we might expect one beam-beam reaction to happen for every million beam-gas reactions. Hence, there is effectively no beam-beam fusion as is so often described. Of course, we can also get neutral-neutral reactions with similarly low contributions. Fast neutrals may contribute upwards of 50% to the total fusion rate.

The negative ions gain energy as they travel outward, also fusing with the background gas and with deuterium embedded in the chamber walls. These ions, in addition to some fast neutrals, are why plated endcaps can really boost fusion rates. I lack cross-section data for negative ions, so their MFPs and hence lifetimes are unknown to me.

The glow discharge is initiated by a handful of naturally-occurring ions and free electrons and establishes within tens or hundreds of nanoseconds. Field emission and field ionization also contribute.

The star mode of geodesic grids and the single beamline of a linear cathode are thus emergent, collective effects rather than the results of recirculation. At high pressures (100s of mtorr), the MFP is so short that holes in the cathode are not favored over the cathode surface. Lowering the pressure to 10s of mtorr does not allow for a high-current discharge to be sustained between the cathode surface and chamber, but the longer path through the cathode allows for more ion multiplication and hence a stable discharge. It's closely analogous to Paschen's law. Electrostatic focusing is also at play.

The ratio between beam-gas and cathode beam-target seems to be very loosely based on the open vs. solid fractions of the cathode, as well as the total cathode surface area (likely a cooling effect). More important is the material selection for beam-target reactivity.

All reaction pathways, bar beam-beam, should scale linearly with cathode current. Intuitively this makes sense; in the absence of space-charge or other feedback mechanisms, more fuel = proportionally more fusion. No standard fusor will ever see super-linear growth with current. If you see sub-linear growth, however, there is something stealing your efficiency. That thing is almost certainly electrons--electrons generated by thermionic, field, or contaminant-induced secondary emission. Since electrons and positive ions both generate "positive" current, in that they impact surfaces of opposite charge, your power supply is measuring the sum of both. If 90% of those are electrons, you're wasting most of your power. They also make x-rays. The bottom line is that electrons are not your friend.

To give a few design criteria based only on what I've said, keep your cathode cold, smooth, and your chamber clean.

Comments, questions, and disputes are welcome.
User avatar
Jim Kovalchick
Posts: 637
Joined: Wed Apr 13, 2011 8:00 pm
Real name:

Re: A New Cube, or Two

Post by Jim Kovalchick »

Incredible work Liam. I am truly awed.

Jim K
User avatar
Finn Hammer
Posts: 289
Joined: Sat Mar 05, 2016 7:21 am
Real name: Finn Hammer
Contact:

Re: A New Cube, or Two

Post by Finn Hammer »

Liam,

Your ability to think and visualize in 3 mechanical dimensions, draw it in CAD and get it executed in hardware is astounding.

A hats off from here!

Cheers, Finn Hammer
Matt_Gibson
Posts: 365
Joined: Thu Sep 23, 2021 10:36 am
Real name: Matt Gibson

Re: A New Cube, or Two

Post by Matt_Gibson »

What purpose does all of the foil serve?

-Matt
Jon Rosenstiel
Posts: 1476
Joined: Thu Jun 28, 2001 1:30 am
Real name: Jon Rosenstiel
Location: Southern California

Re: A New Cube, or Two

Post by Jon Rosenstiel »

Liam,

An excellent progress report, thank you very much.

Jon Rosenstiel
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

All, thanks for the kind words. The inspiration for many design decisions comes from many of you.

The foil serves to more evenly distribute and insulate the heat from the bakeout tapes. I admittedly should have taken more pictures before wrapping everything up.

I may have cooked away my zirconium... time to break vacuum and check.
User avatar
Richard Hull
Moderator
Posts: 14039
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: A New Cube, or Two

Post by Richard Hull »

Fabulous posting!! First rate! A lot of peering into what I have always maintained as the many paths to fusion in velocity space (which you like to call background gas). Fusion is going on everywhere in what I have always termed the average spherical simple amateur fusor. Just enough to allow for a few million fusions per second. We are doing fusion regardless of dominant or nearly worthless fusion processes. All are additive.

Now, leave the simple spherical fusor behind in an effort to improve or take advantage of what ever this or that assumed better way to boost what might be termed a much more dominant fusion IECF mode, this is a research effort of the first rank, much as is brought into the above post with both additional expenditure, time and effort. It is to be applauded at the highest level, naturally.

Regardless, fusion does not like to take place. It is a game of quantum dice throwing in a casino where all the tables, dice, and overarching Quantum physics are against any and all efforts to allow for a true "winner". With each improvement in hardware based on new ideas, we just get a new pair of loaded dice and a different table to cast them upon. Still, "it's the game and the game's the thing!"

Definition: Velocity space is a term coined before me by Robert Hirsch in one of my 1999, in-person, interviews. He understood the fallacy of star mode fusion within a cathode. It took me until about 2004 to fully grasp the full import of his term. Work in this area long enough and do fusion with a fusor and you realize it is all about velocity space. It is where 90% of the fusion is done in a simple fusor and is exactly what limits it. I have posted many, many times on this in the past in these forums. An attempt to muse out the various multiplicity of adventures and misadventures in fusion within the device is laudable, but you will also miss many and if truly wise, leave yourself musing of many reactions within other reactions.

There are dogs with fleas who bite 'em and fleas with fleas which also bite 'em, and so on and so forth, ad infinitum....

Still one of the best and fully illustrative posts with ideas related to fusion in IECF above that I have ever seen...

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
Retired now...Doing only what I want and not what I should...every day is a saturday.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

While I attempt to diagnose and eliminate a few issues, here's an example of inefficiencies manifesting in a fusor. I've normalized the data to the maximum values in each respective plot.

efficiency normalized.jpg
User avatar
Dennis P Brown
Posts: 2693
Joined: Sun May 20, 2012 10:46 am
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: A New Cube, or Two

Post by Dennis P Brown »

Question on these plots last set of plots - you show neutron rate per sec increasing with current - so how does that square with your conclusion that electrons are not one's friend? Then you show neutrons per sec per ma and these values decrease with increase current (more in line with your statement.) Could you clarify the plots a bit more?
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Electrons aren't the only particles that contribute to the current. Both positive and negative ions do as well.

The ion current increases with total current, and hence the neutron rate goes up. However, at the same time, the fraction carried by electrons also increases. Perhaps an example will make it clearer (specific values do not reflect my or any system) :

Total current | ion current | electron current | electron fraction
1 mA | 0.5 mA | 0.5 mA | 0.5
2 mA | 0.7 mA | 1.3 mA | 0.65
4 mA | 1.2 mA | 2.8 mA | 0.7
8 mA | 1.7 mA | 6.3 mA | 0.78

In this example, the neutron rate from 1 mA to 8 mA would increase by 1.7/0.5 = 3.4. It would be 8x if electrons weren't in play.

In other words, the specific particles contributing to the current don't matter to the power supply--current is current, so why would you want to waste it on electrons?


In other news, my TCP 040 turbo pump controller decided to release the magic smoke. The turbo wasn't even running, but that didn't seem to matter. The number of factors conspiring against me never seems to diminish...
Matt_Gibson
Posts: 365
Joined: Thu Sep 23, 2021 10:36 am
Real name: Matt Gibson

Re: A New Cube, or Two

Post by Matt_Gibson »

I probably missed it, but is there a way to reduce electrons and increase ions?
User avatar
Richard Hull
Moderator
Posts: 14039
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: A New Cube, or Two

Post by Richard Hull »

Custom designs using custom design money, leaving the simple fusor in the dust.
4 to 8 Ion guns of good design would be a nice start, with carefully controlled differential pumping.

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
Retired now...Doing only what I want and not what I should...every day is a saturday.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Keep your cathode cold and your chamber clean.
User avatar
Dennis P Brown
Posts: 2693
Joined: Sun May 20, 2012 10:46 am
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: A New Cube, or Two

Post by Dennis P Brown »

Certainly good work and your summation seems very well thought out. You certainly spared no expense in building your very impressive system.

So, how did you measure ion current independent of electron? Or is this modeling?

Certainly makes sense that the electrons aren't directly useful compared to ions but then, no other way to create those necessary ions in most fusors.

Then by your overall conclusion, a sphere, actively cooled, would be your ideal cathode by your conclusion? How would size affect these aspects - a bigger cathode would run cooler but is there a competing issue that isn't obvious here? For instance, I have a 2 inch diameter (hollow) steel sphere - would that be superior to a similar sized grid?

As for equipment failures, you just joined a very well membered club here.

Aside: All that you posted makes a lot of sense but referencing where you obtained those details is extremely important to believe each assertion and trace it to the given experimental proof.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Upgrading to this cube cost some $1500 in machining and small parts costs. Once all the supporting systems are put together, testing different chambers is just a matter of swapping them with no other significant changes.

I haven't independently measured the ion and electron currents. My conclusion is based both on modeling and the less-than-linear and decreasing curves in the bottom two plots above.

Electrons aren't the only way ions are produced. In fact, simulations have led me to believe that in an efficient glow discharge fusor, ion multiplication through various stripping and charge exchange reactions is the dominant sustaining mechanism. This is an area I'm still investigating.

I do not know what cathode shape would give the best results. Perhaps it is a sphere, a simple ring, or some more complex shape. Cooling the cathode past what Finn has done with a hollow stalk and forced air cooling would be quite challenging, although Andrew has been successful there albeit in a spherical geometry. A couple of cathode improvements come to mind, namely a high thermal mass and large radiating (external) surface area. One could also increase the voltage (>> 70kV, say) while keeping the current low (few mA). Speculating will only get you so far of course... methodical experimentation is the way to go, funding permitting.

Equipment failures are nothing new to me; I've been dealing with them for as long as I've been here. Especially among lots of used equipment, the multiplicative nature of probabilities guarantees that something will break sooner or later.

Which claims would you like more sources for? I'd be happy to provide them. I just don't have the time to provide a full bibliography, which would necessarily contain hundreds of items. Perhaps I can find a way to distribute the IEC papers I've downloaded without angering the publishers.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Somehow, I managed to make things much worse...

efficiency normalized 2.jpg
User avatar
Richard Hull
Moderator
Posts: 14039
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: A New Cube, or Two

Post by Richard Hull »

Ya' just had to monkey with it, didn't you?

This stuff happens to everyone who experiments. Now, to find out what happened. That can be fun or a nightmare, but always a learning opportunity.
I hope it is something simple. With me in these situations, it is always my fault. I did something stupid

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
Retired now...Doing only what I want and not what I should...every day is a saturday.
Peter Schmelcher
Posts: 203
Joined: Sat Nov 13, 2010 1:56 am
Real name: Peter Schmelcher

Re: A New Cube, or Two

Post by Peter Schmelcher »

Great post Liam.

This 2010 paper on sputtering copper ions “Self-sputtering of an inverted cylindrical magnetron for ion beam generation” https://escholarship.org/uc/item/9fz127tc#main makes me believe an inverted magnetron pressure gauge can also double as an ion source and hopefully eliminate the need for glow discharge operating pressures. The copper ion source in the paper produces almost 100% Cu+ ions. The question is will this majorly eliminate the charge exchange problem.

Cheers
-Peter

Edit link corrected
Last edited by Peter Schmelcher on Mon Aug 15, 2022 4:35 pm, edited 1 time in total.
User avatar
Liam David
Posts: 367
Joined: Sat Jan 25, 2014 5:30 pm
Real name: Liam David
Location: Arizona

Re: A New Cube, or Two

Post by Liam David »

Peter, thanks for the compliment. I think you gave the wrong link: https://escholarship.org/uc/item/9fz127tc#main. Copper sputters very easily, much more so than aluminum, for example, which may be one reason they used it. However, I don't see how it's directly applicable to deuterium ion production. Deuterium and copper, and their reactions with themselves and the background gas, are nothing alike. Am I missing something in what you're saying? Something like a Penning source would be easiest: https://doi.org/10.1063/1.3054268.
User avatar
Dennis P Brown
Posts: 2693
Joined: Sun May 20, 2012 10:46 am
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: A New Cube, or Two

Post by Dennis P Brown »

Again, great post with a wealth of useful information - should be a FAQ.

One could cool using liquid propane - an extremely efficient refrigerant that is non-conducting. Any refrigeration system can run using propane (I built one recently.) Is propane an explosive hazard? Of course and a serious one if careless. However, far, far less than deuterium gas. Propane is nontoxic and easily obtained. Never considered cooling my cathode but if I did, this would be a simple, inexpensive method.

If I ever get a clean neutron measurement detector running again, I would consider looking into that area.

Peter, sputtering a metal like copper (a metallic electron system that has easy to strip outer electrons) is nothing like ionizing deuterium (tightly bound and quick to neutralize. I highly doubt any magnetron system would ionize deuterium very effectively.
Peter Schmelcher
Posts: 203
Joined: Sat Nov 13, 2010 1:56 am
Real name: Peter Schmelcher

Re: A New Cube, or Two

Post by Peter Schmelcher »

Lian apologies for the wrong link and thanks for correcting me.

I should have clarified that my thoughts were not about sputtering metal but about the fusor section of your post which I agree with but think we should also consider how ion guns can change the fusor operating picture. Below glow discharge pressures long ion path lengths are possible in a clean chamber.

The paper surprised me at how efficient the inverted magnetron was at producing a copper plasma of Cu+1 ions with very few Cu+2 ions, and then the possibly exploiting it for a D+1 ion gun.

From my memory, caveat emptor, all ions exit an inverted magnetron pressure gauge with something around 15eV of kinetic energy even while inside the gauge the electrons are accelerated across a 3.3keV potential difference. After gauge ignition (startup can take minutes at low pressures) the spiraling electrons build to several amps as electrons get stripped away from the atoms.

I intend to implement an ion gun using a copper capillary tube and spraying deuterium directly into my MPG401 pressure gauge. One catch is you need a very small fraction of an sccm. There are a lot of electrons in a mole of deuterium and my Spelman high voltage supply current is only capable of 8mA.

So, if I continuously flush a vacuum chamber with D+1 ions injected at the edge of a fusor potential well while pumping it down at maximum cfm, what is the resulting steady state mix of species.

Dennis for a inverted magnetron pressure gauge to function it has be able to ionize any atom. I looked up deuterium +1 ionization energy and it is 15.5eV so I would not expect any problems.
Post Reply

Return to “Images du Jour”