Set up to test detector in vaccum with low energy electrons. Tried using the UV lamp-chrome window electron source. The dust detector entrance was placed as close as posible to chrome window to minmize gyro radius problems. Screen wires were disconnected from detector circuits and run external.
There was response from the gun in both the high gain AC coupled DUST channel and in the low gain DC coupled LF channel. The screen voltage modulates the output but the response to different gun energies isn't as expected. There is a large background and there is significant effect from the screen voltage even with the gun at 30eV (well above the the screen voltage).
A possible explanation is that the UV shining into the detector is generating photo-electrons at the screen and on the inside walls of the detector.
The chrome window was put at +30V to eliminate it as a source of electrons. The voltage on the modulation screen was varied slowly from -5V to +5V and the DC current collected was measured using the LF output.
With no electron source (black trace), the detector still sees a large input current that is affected by the screen voltage. The response is larger when the screen is negative. This is consistient with the screen as a source of the electrons rather than rejecting electrons coming from outside the detector. (The background level goes to zero when the UV lamp is off).
With the gun as a source at -2V (red trace) there is an increase in current as the screen voltage passes the -2V level. This is consistient with a source of 2eV electrons from outside the detector being added to the background. With the gun at -4V (green trace), the voltage where the increase is seen moves as expected.
The gun current contribution is smaller than the background. This doesn't fit as well with the screen being the source for the background. The screen is 90% open area compared to the full area chrome window source. Both are illuminated with similar intensity UV.
There is no good calibration data for the absolute output of the chrome window source. Count rates seen during calibrations of particle detectors (gun at 4KV, same lamp intensity) are equivalent to about 2pA/cm^2. The dust detector area is 5.5cm^2 so 11pA would be resonable. It looks like the gun contribution is about 15 or 20pA. This is in within the huge error bars for gun behavior.
Set up to try a hot filament gun as the low energy electron source. The dust detector entrance was placed close to the gun output to minmize gyro radius problems. The small diameter beam was aimed down the center of the dust detector. Screen wires were run external so that the RPA experiment could be tried for this gun.
This gun showed no background and nice sharp current increase as the modulation voltage is swept. Modulation voltage sweeps were done at filament potentials of 0,1,2 and 3 Volts. The filament heater current was adjusted before each sweep to give roughly 20pA. Output flux is a very sharp function of heater current so the adjustment is coarse. Lower energies require more current.
According to this data the gun peak energy is about 1eV higher than the filament potential. The calibration of modulation voltage and filament potential are accurate to within 5%.
The voltage drop across the filament should not contribute much error to the energy. The potential used was the average of the voltage at each end of filament. Filament drop is 1V or less. The gun is designed so that the beam is emitted from the center of the filament. Errors from filament drop should be < 5% of the drop or .05eV. The max possible error is only 0.5eV.
This energy offset is repeatable and seems independent of focus grid potential. The focus grid was run at +8V relative to the filament. This voltage needed to be above about +2V to get any output from the gun in this energy range. Changing this voltage affects the flux but doesn't have much affect on the energy.
Changed setup so screens can be driven from the detector electronics. This puts -1.2V on the rej screen and a +-11V 1KHz square wave on the modulation screen.
The high gain DUST output shows the expected modulation for filament potentials from 1 to 9 volts. Currents of 1pA are easy to see above the noise. (The 1pA is based on nominal gain. Actual gun output is unknown.)
The modulation is visible with the gun below -1V and disappears rapidly at about -10V. This seems to indicate that the 1eV energy offset is still present at these lower flux levels.
A check of the phase delay between the SAMP timing pulse and the output signal gives the same answer for optimal delay as the signal inject bench test. This seems reasonable since the transit time for electrons should be very fast compared to the delays in the circuit. Phase lock is solid and repeatable.
A sweep over gun energies shows a nice cutoff near 10V as expected. Most of the decrease at low energies is due to a decrease in gun output not a lack of response from the detector. There should be a detector cutoff at 1eV but with the gun's 1eV offset in energy and low output, this isn't visible.
The modulation seen at -11V extends to -20V (and probably higher). It is about 15% of what is seen at -8V. The modulation seen above 11eV is 180 degrees out of phase. More current is collected by the anode when the screen is at -11V than when it is at +11V.
A check of the LF channel shows that for gun energies above 11eV, the average current doubles. This indidcates that current is being collected on both modulation phases as expected. This channel has 100Hz cutoff so it doesn't see the 1KHz modulation. This data can't resolve if the modulation at higher energies is caused by an enhanced flux at -11V or a reduced flux at +11V.
Changed back to external control of screen voltages.
Successive runs were made sweeping the gun filament potential with different fixed voltages on the modulation screen. The gun intensity drifts with time but these were done in rapid succession. Error due to intensity drift could be 5% or more but the fact that -10V sees more that +10V at high gun energies is repeatable.
This data shows that the effect is steady state and not some transient effect at 1KHz. It also indicates that compared to 0V, both enhanced flux at -11V and reduced flux at +11V contribute.
To check for scattering from the anode, data was taken with the detector in ion mode. This changes the anode voltage from +3V to +0.3V. There was a significant loss in collected current at the reduced anode voltage.
At higher gun voltages the effect is larger. The difference for a -2V gun was a factor of 1.5. At -30V gun it is a factor of 3.5. At -60V it is factor of 55.
These plots show the change in flux vs. modulation screen voltage relative to the flux collected with the screen at 0V. The total flux is lower at 0.3V anode but the percent variation is more.
The flux enhancemnt at -5V fits with electrons that back scatter from the anode but are forced to return by the -5V screen.
The reduction in flux at +5V doesn't fit well with this theory. Any back scatter that escaped the anode region would not return at 0V. Increasing positive voltage would have no effect on the number that return to the anode.