CDMS II Blue Light Pulser Project
Overview of the Pulser Project
- The Pulser is a device used to actively test the Moun Veto Shielding which surrounds our detectors
(go a page back to see a diagram of the shield). Each panel on the shield has a photo-multiplier-tube
(a light sensor) attached to it. These PMT's are only sensitive to blue light, so we have to use this
color when we run a test. The way we test the panels, is by flashing the light created by the Pulser
into them and then reading out the response through the electronics. If a glue joint breaks, or a PMT
goes bad, we will see it as a decrease in output from the damaged panel. Every care has been taken so
that the Pulser does not interfere with the experiment. That is why the pulser isn't even in the same
room as the detectors. The only contact the Pulser makes with the sheild is through 42 fiber optic cables
that run through the wall. This technique is almost completely experimental, experiments in the past
have used LED's to actively test scintillators, but not to the accuracy that we are trying to achieve
here. The main difference is the fact that we have no additional electronics attached to sheild so we
are not adding a potential source of noise in the sheild, and we will also care about minute changes in
output from the PMT's.
Tony Piro's Pulser Page
- This is the original Pulser Project
web page. There are some nice digital photos and general
explanations. Various LED, optical fiber tests can be found here as well.
Pulsing Circuit (postscript)
- This is the final circuit that is being used in the Pulser at Soudan. With
this circuit we are able to get a pulse width from about 70ns to several microseconds,
which is about 100 times faster then before! This is actually quite impressive when
you consider that we are trying to switch 6 amps that fast. We can change the hieght
from around 2 Volts to a maximum of 8 Volts. The frequency of the pulse train varies
from a few kilohertz to 10 kilohertz. The pulses start out as a square train, but do
to the slew time of the MOSFET they get warped into rounded spikes at small pulse widths
(smaller then 100ns). Various tests have shown that this is rather stable, and doesn't
effect the final result. In fact a narrower peak is better for the life of our LED's so
that they are not hit that hard, that fast.
- Blue Light Uniformity Tests
- There's a concern that the pattern of
blue light emerging from the end of the light cone (see
this photo for example) will give highly varying light
levels (from fiber to fiber) as seen by an 8x8 grid of optical
fibers coupled to the
end of the cone via this collimater.
There's also a concern that any given fiber receives most of its
light from a single LED in the array. Our original thinking was that
the light cone would mix the output of all the LEDs and thus
any given fiber would sample light from a large number of sources.
In this way if a single LED becomes faulty there will be little
affect on the light seen by the fibers (i.e., one fiber, or even
a small group of fibers won't suddenly become dim because a single
LED has burned out).
The following plots represent a series of tests made to assess and
quantify these two concerns. They seem to indicate that neither
is much of a problem.
LED Failure Test
- Here is the uniformity of the light after it is run through
the Pulser housing and recorded relative to the face of the box. There are only 56
points on the box face and we will need only 42 of them to couple to the scintillator
panels. Which gives us plenty to spare. This plot differs from the other uniformity
plots, in that the drastic changes in color represent standard deviations away from the red
mean. The plot only shows up to three sigma. There are three bad points on the plot
which were do to those fibers breaking during the epoxying process. We are not that concerned
about them because we still have enough points to spare.
Front Panel Uniformity
Panel Charge Spectrum Analysis
- When we run the Pulser during a test of a scintillator panel, we wish to see if anything has
changed. To do this we are going to integrate the charge spectrum that is produced by the readout
of the photo-multiplier tubes. How much this number changes, in time or relative to other panels
will give us an early warning sign that there is something wrong. These plots which are created by
taking the sum total of all signal counts (gamma-ray's, muons, and the pulser) and then graphing
them versus the charge at each count. We let the ADC's run through 2000 counts to get enough data.
As it is fairly obvious, the pulser's charge spectrum dominates all other events. Here are two plot's
using different amplitude pulses.
8 volt Pulse amplitude
6 volt Pulse amplitude
The Gamma-ray's are the section of counts near 30 picoCoulombs. In previous spectrum plots, these
are the dominant count's as you can see from this
sample plot taken during the September 2001 Soudan trip.
The Pulser peak can be shown to be almost a perfect Gaussian distribution (which is the red curve)
, of which we can determine it's peak to within a picoCoulomb! Which happens to be better then
our equipment can resolve.
- Pulser Stability
- There is also a concern about the pulser drifting with time. In order to understand this effect I
have set up the original LED array with a fiber optic cable pointing at it in such a way to give us
similiar results to the pulser which is at Soudan. Now in these tests conditions are not ideal. We are
using a bad high voltage supply (that drift's randomly about 50 volts) and the actual set up has
introduced some randon error, which we dont understand yet. The main purpose of this test is to
give us a handle on how the Pulser is actually going to be implemented in the shield, and how we are
going to recieve and analyse data. For the time being I have been doing the analyses through the use of
Matlab. As a word of caution, we dont know what this means yet because I have not taken enough data.
One thing to note is that the x-axis on the plots are not even increments of time, but they are
Here is the track of the drifting peak.
Here is the calculated number of incident photons.
PMT's 10 and 11 are powered by the same supply, while the test panel is powered by our faulty LaCroy
supply. The important thing to notice at this point is that the two monitoring PMT's track the same.
This means atleast that whatever is causing the drift, is independant of the PMT's, otherwise they
would not track so nicely.
- Periodic High Voltage
- The purpose of these tests are to see if there is any correlation between the high voltage power to our
PMT's and the charge spectrum peak we read out of the PMT's. The first test had it's flaws, but after careful
consideration of montoring techniques, we have developed a method to simultaneously monitoring our high voltage
while taking a data set. Look at the top of the table to view a summary of the results of all the runs. I
will update those when ever I add another run to the set. However, I only plan on doing one more long term run
since the correlation is already apparent in these first few runs. Which corresponds to about 60 hours of data
- Here is a diagram of the current setup (or in
- And here is the run data.
||High Voltage Drift
The Pulser at Soudan, MN
- Now that the pulser has been moved to Soudan mine we would like to know if we can
find a suitable range for the pulsing circuit that gives us values within the range of our
readout electronics. The three inportant panel geometries are the bottom, top, and side
panels. When a sample from each set of geometries was tested with the pulser we found
satisfactory values in all panels. Our most sensitive panels are the angled side panels
which seem to be at the upper end of our spectrum. The largest top panels fell around 100pC
with the bottom panels coming in around the middle.
Here is a charge spectrum taken
during calibration. Notice the absence of a moun peak (it would be around 80pC) which is one
of the reasons we are at Soudan Mine.
And here is another uniformity plot taken
across the pulser box face using the charge peak for reference.
- More Digital Photos
last updated: 04/05/2002 by Robert Nelson (email@example.com)