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

[Daniel Firth]

I don't have much to add... just wanted to say it's been interesting watching your progress in this thread.

I think this got buried in a thread about my construction, but I based my grid off of yours. It's not water cooled, but I used your bending method with the slotted pipe.

Also, my BTI bubble detector was done in about a year. I got it in 2012, and now it has many gigantic bubbles. Do you keep yours stored compressed, or uncompressed?

[Bob Reite]

If you got a full year out of your bubble detector, you did well. Mine was gone in 9 months. They are supposed to be stored compressed in the factory supplied tube between 15-20 degrees C. I was shocked when I took mine out of storage and found large bubbles in it, even though it was stored compressed. That seems to be the failure mode. The bubbles from the last measurement "grow" even while under compression.

[Andrew Seltzman]

I think there might be a lot of variation based on how well the dosimeter seals when assembled.

They used to manufacture the dosimeters with an anodized aluminum cap and compression system, but then switched out to a plastic one which seams to seal a lot better, probably since the epoxy that they fill the threads with bonds well to the plastic. The first one that I had was made with the aluminum handle and always had a "fishy" smell inside the tube. It lasted about 1year range before it was leaking and would not recompressible. By 2 years it had huge bubbles as well
viewtopic.php?f=13&t=6121&p=34565#p34565

The second one that I bought had the plastic handle and it looked very well epoxied in place. It never had any hint of a "fishy" smell in the storage tube. I got it in 2009 and it would recompress until 2012, with the built in compression chamber, and after that it would still recompress by placing a mechanical pencil eraser in the compression chamber before tightening it down to increase compression.

I always stored both in the tube, compressed. It's possible there is a lot of variation in life due to problems with the sealing system.

[Andrew Seltzman]

Here is an excellent publication on leak rates for viton/kalrez/teflon o-rings for UHV systems(atom optics traps), the vacuum tests start on the 3rd page of the publication

viewport-oring-leak-rate.pdf

(440.42 KiB) Downloaded 737 times

And a reference of various vacuum compatible ceramics, I'm thinking of making part of the grid insulator out of macor, or LAVA, does anyone have any experience using these in a fusor under ion/electron bombardment?
http://www.technicalproductsinc.com/ceramics.html

[Andrew Seltzman]

All 4 ion sources have been cleaned and rebuilt and baked out at 200C before instalation. They now use the angled anode rings and SmCo magnets instead of NdFeB magnets for compatibility with Deuterium with out disintegrating, and to allow high temperature operation(up to 300C) without loss of strength. I decided to go with the original style pole piece for now since it provided a near parallel beam with the angled pole piece but may switch over to the other style later. The SmCo magnets have lower field strength though and the ion sources require higher voltage to operate and produce a less intense beam which is fine since they are still significantly over powered for the fusor.

Ion sources before cleaning

In addition, one of the glass disk viewports was replaced with a conflat viewport eliminating one more viton o-ring (5 removed in this modification)

They now use copper o-rings instead of viton and have pump out ducts to avoid virtual leaks.

One of the swagelok valves was outgassing something when it was getting hot and was also replaced, and a 0.5um filter was added to the dry air admit line for venting the fusor to atmosphere for servicing

Some burning on the inside coating of the ZnSe viewports was observed when running with deuterium(doesn't happen with air). There seems to be some charged particle coming from the grid that is not readily deflected by the magnetic beam deflectors. I'm assuming it's a negative deuterium ion

I've doubled the field strength on the beam deflectors by adding a second magnet but it doesn't help much. I may have to remove the ZnSe ports for deuterium runs or re-design them or the grid to deflect the ion beam away from that area.

A feedback control circuit has been built to regulate fusor pressure by feeding back from the vacuum gauge analog output, It will stabiily control pressure from the 1e-4torr range to the 10s of mTorr. This considerably reduces workload during fusion runs and increases neutron output by tracking the optimal pressure. It's a very simple circuit.

Pressure controller circuit

An NP10 boost converter similar to this, or other boost converters used to drive nixie tubes:
http://www.ebay.com/itm/SHV5-5Vin-400Vo ... SwwE5WVDQB
provides 87V from the 24V supply, then an LR12 adjustable regulator steps the voltage down to 70V to feed the positive side of an OPA445 high voltage op amp which is configured as a gain 21 non-inverting amplifier. A recom RS3-1212D DC-DC converter provides -12v to the OPA445 and an AD826 that is used as an input differential amplifier. The vacuum gauge output(quattro 999, 0-10v, logarithmic to pressure) is connected to the negative side, while a potentiometer supplying 0-12v is connected to the positive side. The valve will now accurately control fusor pressure. There is still some drift and overshoot since it's only a proportional feedback controller, but the final version will be a full PID with valve dither ontop of the analog signal to improve performance.

The fusor will now pump down to the low E-4 range in about 5-10 minutes and max out at about 5e-5 torr after several hours.

The cleaning of the ion injectors and removal of 5 viton o-rings seems to have really helped, though the pressure still climbs when running as the grid insulator heats up. It's made of boron nitride with a boric oxide binder which is fairly hydroscopic and will soak up water every time it sees atmospheric air. It will have to be replaced with a combination of macor/alumina/fused quartz.

Fusion rate is now at 3e5 n/s

Star mode at ~3mTorr during fusion fun

[Steven Sesselmann]

More nice work Andrew, efficiency is improving you are now in the E-9 range and it should get better with use. If you can get unto the -65kV range I thing the numbers will really shoot up.

If I understood correctly you are using feedback to control a vacuum valve, another possibility is set the vacuum valve and use a combined pressure and mass flow controller, to regulate the pressure with the gas inflow. Something like an MKS 649 or similar.

I look forward to see the more run data.

Steven

[Andrew Seltzman]

Actually it controls the inlet gas flow by controlling the voltage on the piezo valve.

[Richard Hull]

Getting "betterer and betterer"!
Good work

Richard Hull

[Andrew Seltzman]

The gas control system has meed modified to include a full PID feedback system and in instrumentation amplifier to buffer the input.

This results in better long term pressure stability and lower overshoot

A dual layer circuit board has been designed in expressPCB for the final configuration of the control system

The circuit board holds 2 independent PID control circuits, for the gas valve control only one will be used, the version with both in use will be used to control the HV power supply and in injector current. The one connected to the power supply will compare the grid current to a setpoint and adjust the ion sources accordingly to allow operation at any pressure/voltage/current point with all 3 adjustable independently.

The ion source control will use this circuit to control the input voltage to a set of emco F40 proportional HV supplies driving the ion sources:
http://electronicdesign.com/power/simpl ... ers-output

And assembly is in progress

The boards will be housed in Hammond 1590BB boxes, one for the power supply/ion source control, and the other for pressure control

[Roberto Ferrari]

Hi Andrew

Your comment about the SeZn screen and its damage when running on D2 is consistent with the reductive power of hydrogen.
Please comment on any chemical alterations.
Congratulations for your excellent work!
Roberto

[DanielSandner]

Hello,
there is something very effective against long pumping times with a Turbo.
A Turbo can pump heavy molecules like hydrocarbon very fast in contrast to light molecules like D or He.
So if you can i would try to fill the chamber with an heavy gas and 1% D. The Turbo will remove the heavy molecules much faster and after 1-2 hours pumping you should have only Deuterium at a very low pressure.
kind regards, Daniel

[John Futter]

Daniel
your post seems to be arm chair posteuring.
most pumping time on vacuum chambers is getting rid of gas and water molecules that have stuck to internal surfaces
with water vapour being one of the hardest to remove as the molecule is polar and sticks to the chamber walls with great tenacity.
Just backfilling the chamber with a dry inert gas ie argon will save hours of pumping when restarting a vacuum system

[Andrew Seltzman]

Roberto,

I'm not sure about any chemical changes to the lenses, It looks like beam burn in there the D- beams hit the window surface. There is some evidence of the same thing happening to the pyrex glass viewports as well, though to a lesser extent. I plan to remove the ZnSe ports and may replace them with germanium viewports at a later time to see if that makes a difference.

[Roberto Ferrari]

Thanks Andrew, we wait for future reports.

[Andrew Seltzman]

Final construction and tuning of the PID gas control system is complete

The control system is now in a complete box,
The controls in the front are(from left to right), power LED and switch, valve purge button and internal/external pressure setpoint reference, Direct output voltage control pot and direct/feedback control switch, internal pressure setpoint pot, process variable and output voltage displays

On the back panel, valve output connection, power input, error voltage monitor, external pressure setpoint, pressure input from gauge

Final circuit values

The final circuit is a full PID controller with output dither, there is a 6v, 90Hz triangle wave superimposed on the output to improve accuracy

Tuning between variable pressures

Holding constant pressure

step response, achieves final value in about 8 seconds

The pressure control system holds pressure very stability, now onto the automatic control system for the power supply and ion injectors.

[Steven Sesselmann]

Nice work, so you basically made your own pressure controller from a vacuum gauge and a piezo pressure valve. I guess this would be not so different from using a ready made MKS pressure controller with a built in baratron gauge. My only concern about using a pressure controller is accidental loss of gas if for some reason a gate valve is opened and the pressure drops rapidly, causing the gas to vent too fast.

Steven

[Andrew Seltzman]

By limiting the maximum valve voltage, the maximum flow rate can be limited, preventing a runaway in case of control system failure.

The maximum conductance of my vacuum system is 0.3L/s, with only an open and closed valve

[Richard Hull]

Fabulous coup on your pressure controller, Andrew. A real win on that one.

Richard Hull

[Andrew Seltzman]

The ludlum model 12 counter has been modified to provide a 0-10v output based on meter movement to allow the power supply control system to feed back on the neutron signal and provide a constant neutron flux

A boost converter provides +12.5v to a rail-rail opamp

Amplifier board converts 0-1.25v signal to drive meter to 0-10v signal on output

Boost converter and amplifier installed

Voltage and current control system construction in progress

Right now it can display power supply voltage and current and control power supply voltage. The final version will have a 2 channel PID controller for current and voltage control. The system will either directly set output voltage, or control voltage using the neutron flux for feedback. For current control, it will look at the power supply current monitor output and adjust ion source current to track a specific grid current. This will allow the system to accurately track a given voltage/current/pressure operating point with minimal operator tuning.

[Richard Hull]

Wow! Shooting for a constant neutron source in a fusor. Lots of feedback loops here. I hope they don't go "slinky" on you. Fine tuning such systems can range from moderately simple to near impossible. Damping system wide swings can be the real win on such efforts. Good luck.

Richard Hull

[Steven Sesselmann]

Andrew,

Nice work and AFAIK a first here on fuser.net, I echo Richards concerns, you have quite a few loops happening there, so I guess you need some kind of hysteresis to compensate for delays. If anyone can do it, I am sure you are the one.

Looking forward to your next run..

Steven

[Andrew Seltzman]

I have the control system successfully holding a constant neutron flux based on feedback from the neutron detector. At the lower rates that the fusor is producing the integrator time constant has to be fairly long to prevent the system from oscillating. For these tests the ion injectors are just holding at constant emission without feedback control from current since that part has not been built yet.

Video of operation
https://youtu.be/BiEOAbmksow

Circuit schematic

Control system close up

Control system and neutron detector

[Richard Hull]

This appears to be a first in electronic fusor control that appears to be rather successful. Do you have a current fusion level at which the system seems stable over about 20 minutes or longer? What is your longest stable, controlled run? Just curious.

Really a super effort.

Richard Hull

[Andrew Seltzman]

Based on comparison between the old BTI(it finally did fail) I had and the BF3 detector, I would estimate neutron rate in the 2e5 n/s range. More fine control will be possible with higher neutron rates as the effect of a single count becomes less. As of now, the control system is running with mostly integral feedback with a very long time constant to prevent oscillations. If the neutron rate was higher, the time constant could be decreased.

At high proportional gain or lower integral time constant the system does start to "slinky" rather noticeably, but with low Kp and long Ki, it runs stably.

The longest run would be in the 2-3min range. I'm still having problems with out gassing of the boron nitride grid insulator and possibly the ZnSe viewports as things heat up which limits run time as the pressure controller reduces gas flow to maintain constant pressure until no deuterium is being admitted to the chamber. Removing the ZnSe ports and upgrading the grid insulator design is on the to-do list.

[Dennis P Brown]

I am a bit concerned by the ZnSe window - powerful electron bombardment could release Zn into your system and cause long term contamination - Zn, if it does deposit will re-vaporize again and again - not good for electronic sensors, I'd think. I assume you need that ZnSe window as an IR clear window but if you aren't using it at the moment for that purpose I'd suggest replacing it with a glass window for now. Why keep an unknown that might be causing issues until you have the system fully operating in the manner you want and then installing the ZnSe window? Just a thought.