Well, it has been three months now since the previous update, but I'm excited to announce the completion of my vacuum system. While the fusor as a whole still isn't ready for operation, I'm at the point now where the fruits of my labor are becoming apparent. Being able to see the completed vacuum system imparts a strong sense of motivation, as the ideas that I've been holding within my mind over the past few years are finally physically manifesting before me. So then, I suppose it's time to go over what I've done.
Pictured here are the majority of vacuum components laid out for cleaning.
My previous post had left off with the impeded completion of the system due to a damaged step-stud. I was very fortunate in this regard, as I only needed to lock some vice grips to the exposed edge of the stud to remove it. With all 1/4" studs replaced by 5/16" I was able to attach the conical adapter to the gate valve, after which the fusor chamber was ready to be mounted to the frame.
Pictured here are the components of the fusor that exist between the roughing line and the fuel line, excluding the HV feedthrough. The top front flange of the chamber is a glass viewport, and the top back flange is a zero-length reducer for attaching the KF-25 fuel and pressure gauges. The front bottom flange is a 6" to 4.5" conical reducer, leading to a manual gate valve, a 4.5" conflat elbow, and the turbomolecular pump. The back bottom flange is for the HV feedthrough.
The feedthrough wasn't connected until after the chamber had been mounted to its frame to prevent damaging the insulator. The chamber was attached at four points near the top and one near the turbomolecular pump. A total of four extended-length Grade 9 bolts were used in place of the standard conflat bolts, divided between the viewport and zero-length reducer of the fuel/gauge port. These bolts were inserted in opposition to the others, leaving their threaded ends facing outwards. Four 80/20 extrusions were then threaded to accept these bolts and the extrusions were mounted to the frame. The turbomolecular pump was mounted to a larger 80/20 extrusion using a u-bolt, with high-temperature rubber compressed between the two for vibration dampening.
Image of the frame
The frame had been moved to a new location before assembling the fusor. The location was selected to maximize the distance from which any person could unknowingly approach the device. In this new location, the X-rays from the fusor will be reduced below background levels at any distance one could approach from the outside. Protection within the room will be provided by a lead plate, which has not yet been installed.
Here are images of the completed vacuum system mounted to the frame
The roughing pump was mounted using high-temperature rubber standoffs. A KF-25 t-adapter was attached to the inlet of the pump. Right-angle manual valves were connected on either end of the adapter, one to seal off the roughing line from the pump, and the other to charge the pump's inlet back up to atmospheric pressure after finishing a run. The roughing line valve leads to another T-adapter, to which one end connects to a thermocouple gauge and the other to a 6" long bellows hose. The hose feeds directly into the exhaust port of the turbomolecular pump.
The zero-length adapter on the back of the chamber attaches to a 2.75" elbow, which in turn connects to a 2.75" to KF-25 adapter. The adapter then attaches to a five-way cross. The top flange of the cross connects to the DV-6 thermocouple gauge, the right flange connects to the inverted magnetron gauge, the left is a series of KF-25 straight pipes, and the bottom is blanked off. The straight pipes lead to a right angle valve, a needle valve, and then a KF-25 to barbed tube fitting.
More views of the fusor
The HV feedthrough was connected after mounting the system to the frame. The cathode grid of the feedthrough was made using 0.049" diameter 99.95% annealed tungsten wire from Thermofisher Scientific.
I had initially attempted to fabricate the rings by bending the wire around a 2" steel pipe, but all attempts at doing so resulted in the wire shattering. This was a bad start, as the bend around the pipe was at a much shallower angle than the 90-degree bend that would be needed on the ends of the loops. I eventually gave up on the idea of using the pipe, instead opting to form it by hand with jewelry pliers. Copper wire was formed around the 2" pipe to provide a gauge for visual reference.
Many more attempts were made to bend the wire by hand, but the wire would still always break at some point during the forming process. Through much trial and error, I eventually found success by forming the wire while it was submerged in boiling water. Though it was a pain to work over the stovetop for extended periods, the tungsten wires' ductility was noticeably improved, and no further breaks occurred while handling it.
Pictured here is a single loop of the tungsten wire after being formed by hand
Three loops were then inserted into a 1/4" copper tube, and the loops were positioned with an alignment tool I 3D-printed.
Once everything was positioned correctly, the copper tube was crimped to permanently retain them. The grid was then inserted into one end of a 1/4" stainless steel coupler.
The HV feedthrough pin had a diameter of 5/32". A 1/4" copper pipe was drilled out to accept the feedthrough, adapting the pin to fit into the coupler. The copper tube also had a cut made across its length. This served a dual purpose, first to allow the coupler's set screw to drive directly against the feedthrough pin and then to improve vacuum conductance by providing a channel for air to escape the coupler.
Pictured here is the grid attached to the feedthrough
Pictured here is the grid in the center of the reaction chamber. (Image taken at an angle)
With the entire vacuum system now fully assembled it was time to begin testing its performance.
Vacuum Testing
A pump down was performed after fully assembling the vacuum system. The reaction chamber reached a pressure of 11 microns as read by the DV-6 tube after roughly one minute of evacuating. The gate valve to the chamber was then closed, after which the pressure was observed to increase at a slow but steady rate. The gate valve was opened once again and the turbomolecular pump was started. The chamber pressure dropped to zero on the DV-6 gauge within twenty seconds. The inverted magnetron gauge was then powered up and the chamber pressure was measured with a voltmeter. After two hours of continuous pumping the system had reached a pressure of 2.6X10^-5 torr.
The gate valve was then closed and the rise in pressure was observed. The inverted magnetron went off scale within a few seconds, so I switched back to reading the DV-6 gauge to get an idea of the leak rate. After several measurements, a leak rate of 3 microns per minute was found. The system was left closed off for the next hour, but the chamber continued to gain pressure at the same rate, suggesting it wasn't a virtual leak.
I began the search for leaking connections by once again opening the valve and reducing the chamber pressure to 2.6X10^-5 torr. 99.9% anhydrous IPA was applied to all of the flange connections on the vacuum chamber, no effect was observed until the alcohol was applied to the DV-6 guage. Immediately upon application of the alcohol to the threaded KF-25 adapter, the pressure within the chamber sharply dropped. After around twenty seconds of improved pumping, the pressure began to sharply increase, surpassing the level from before the alcohol was applied.
The chamber was vented to atmospheric pressure and the DV-6 tube was removed. The threaded connection of the tube was cleaned with acetone and new thread tape was applied. The system was pumped down after the tube was reinstalled, and after another two-hour run, it reached a pressure of 4.2X10^-6 torr. The knife gate was again closed and the rise in pressure was measured with the inverted magnetron gauge. The measurement had started at a pressure of 4.2X10^-6, and after 5.57 minutes the pressure had risen to half a micron, indicating an outgassing + leak rate of just under 0.1 microns per minute.
The next day I performed a longer duration pumpdown of the system, lasting four hours in total. This time the chamber had dropped to a pressure of 1.5X10^-6 torr. After the run had been completed, I set about testing the vacuum performance of the chamber with reduced conductance. The manual gate valve I use between the turbomolecular pump and the reaction chamber makes five and a half rotations between the open and closed positions. The first five rotations progress the knife across the opening of the conical reducer, and the final half-turn compresses the O-ring. The gate valve was turned five rotations for the test, fully obscuring the chamber's opening but allowing gas to still circumvent the O-ring.
The system was held in this configuration for a few minutes, after which it had settled at a pressure of 3.4X10^-6 torr. This is likely the position the valve will be in when I eventually admit deuterium into the chamber, and thus it may represent the approximate background gas pressure. This should certainly be sufficient to maintain a 99% deuterium atmosphere in the reaction chamber.
Once this test had been completed I performed another leak check on the system. With the gate valve fully closed, pressure was observed to rise from 3.4X10^-6 torr to 5.0X10^-4 torr over the span of 22.7 minutes, giving an outgassing + leak rate of 0.02 microns.
After running this second test I became aware of the backing pressure of my pump increasing above previously seen levels. One reason was certainly the ambient temperature, it was a very hot day outside and the room had risen to 100F. The temperature of the pump was measured, it was still within operating spec but only just. Still, even if it were hot out the backing pressure shouldn't necessarily rise by 20 microns. I decided to perform an oil change on the pump. The old oil was collected and observed, no issues with color or viscosity were noted. The oil however was of a cheaper variety. I had originally purchased a 5-gallon bucket of it for use in cleaning the pump when I first purchased it. The pump had performed fine up until now so I never bothered to replace it.
Once the oil was removed from the pump I replaced it with the brand-name oil listed in the owner's manual. The pump was once again started and allowed to heat back up to its previously measured temperature. Upon reaching that temperature the pump was backing the system at 15 microns, over 15 lower than with the previous oil at the same temperature.
Upon the time of writing this post, it was finally my day off, so I went about performing a long-duration pump down of the system. After nine hours of continuous pumping, the system had dropped to a pressure of 5.5X10^-7 torr. I'm quite pleased with this result, especially considering the majority of the system was constructed with used eBay components and no bakeout was required. The rate of pumping quickly dropped off after this. The turbomolecular pump was allowed to run for an additional two hours, after which the system reached a final pressure of 5.0X10^-7 torr. The pressure was measured with the gate valve cracked a hair again, this time it settled around 9.0X10^-7 torr. A final leak test was then performed. The pressure rose from 9.0X10^-7 torr to 9.0X10^-5 torr over the duration of 27.8 minutes, giving a final outgassing + leak rate of 0.003 microns per minute.
Now for a few final notes on the vacuum system. There are currently still two connections within the system where leaks occur, one is in the roughing line and the other is the fuel line.
The roughing line currently makes use of an LCD-display HVAC thermocouple gauge. The gauge is connected to the system with an NPT fitting that leaks at a high rate. I've made multiple attempts to stop the leak but all have been unsuccessful. I've decided to replace this gauge with the DV-6 tube currently measuring my reaction chamber pressure. The first reason is to eliminate a potential future leak point on the UHV side, and the second is to allow myself to observe the roughing line pressure at the operator panel, as I currently need to walk to the other side of the system to read the gauge. I will be replacing the gauge on the UHV side with another DV-6 tube, but this one will have a welded KF-25 fitting instead of a threaded adapter.
The second leak originates from the needle valve on my fuel line. The needle valve has a threaded adapter on each side for connecting KF-25 fittings. The open end of the valve was blanked off and the leak was confirmed to be originating from the threaded connections and not from gas slipping past the needle. I will have to try and seal the valve before I attempt any runs with the system. It's possible that I may even change the valve out for something else entirely.
Well, that brings everything up to date with where I'm currently sitting on this fusor. My efforts for the remainder of this month will be directed toward the completion of my custom power supply. Once completed I will finish up my metering equipment and get everything grounded, after which an air plasma test run can be performed.
