PMT Signal Pickoff Circuit
IceCube 2007 Home
The original circuit design was provided to us by
the AMANDA (Antarctic
Muon and Neutrino Detector Array) web site at:
www.amanda.wisc.edu/karle/amanda2/pmt-bases/pickoff/PMTpickoffcircuit.html
Based on their design I devised the circuit board
layout shown here.
After constructing a rough circuit, shown here, we proceeded to test
it with our Optical Module (O.M.).
We tested the circuit using a supply voltage of 1800
volts. With both signal leads terminated in an oscilloscope we
observed the triggering event followed by a small reflection.
The two signals, when observed together, were indistinguishable with
our oscilloscope. We think the reflection results from a mismatch of
resistance at the transformer. The 130 ohms of impedance from
the twisted pair is an estimate. Even if the impedance of the
twisted pair was exactly 130 ohms the impedance ratio of 130:50 is
(1.612)^2 while the
turns ratio of 5:3 is 1.667. Some of the reflection may result
from this difference. A possible solution would be to increase
the
number of turns used in the transformer. This would allow us to
get closer to the 130:50 ratio. The time delay between the event
signal and its reflection is directly related to the length of cable
running from the O.M. to the splitter circuit. If we multiply the
time delay, 130 ns, by the speed of an electrical signal in a wire, 2 x
10^8 m/s (2/3 c), the result is 26 meters. Our cable is 13 meters
long and any signal that is reflected would have to travel 26 meters to
reach the oscilloscope.
With only one signal lead terminated in the
oscilloscope and the other left unconnected we observed some unusual
ringing in the signal. We expected the reflected signals to
increase in voltage, but they did not. Instead they increased in
frequency by what appears to be a factor of three. We do not know
the reason for the increased frequency of noise. I will be
testing the more complete versions of the pickoff/splitter circuit and
the splitter circuit to see how they respond to a single output
terminated at 50 ohms.
50 Ohm Termination
Tests
I wanted to see if my circuits produced the same
increase in reflection frequency as the first circuit I tested with one
output left without a 50 ohm terminator. First I connected the
Pickoff circuit inside our dark box and then connected the splitter
circuit outside the dark box. Using an input voltage of 1500
volts I tested the circuit with and without the 50 ohm terminator on
the second signal out connector. A graph of the data I collected
under both conditions is here.
Next I tested the Pickoff/Splitter circuit at the
same voltage with and without the 50 ohm terminator connected to the
second signal out connector. A graph of the data I collected can
be found here.
The difference in the time lapsed before the
reflection for the two circuits is due to the additional cable needed
to connect the Splitter circuit. It is interesting to see that
when the 50 ohm terminator is removed the reflection becomes
inverted.
Circuit Designs
It was decided
that we would create one circuit that contained the pickoff and
splitter components, a pickoff circuit with one output, and a splitter
circuit. For the outputs of the pickoff/splitter circuit and the
pickoff circuit we decided to use isolated grounds on the bulkhead BNC
connections. This was done to reduce the posibility of creating a
ground loop between the power supply and the oscilloscope. It was
also necessary to isolate the bulkhead PMT connection since it does not
go directly to ground.
Pickoff
and Splitter Circuit
Pickoff Circuit
Splitter Circuit
Impedence Matching
In the Pickoff Circuit picture there is a 561 ohm
resistor connected to the BNC output. This was done to reduce the
reflection caused by the resistance mismatch at the transformer.
Our first attempts to solve the problem of the resistance mismatch
invloved trying different ratios of turns on the transformer. We
compared averages of 1000 traces with a supply voltage of 1800 volts
(Gain = 1x10^7) and a scintillator to determine which turn ratio worked
best.
Impedence Matching with Transformers
With none of the transformer ratios giving us a
really good signal we needed to alter the circuit somehow. I
thought that it may be possible to change the ratio of impedences to
match the ratio of turns. This could be done by adding a resistor
either in series or parallel with the signal output of our pickoff
circuit. By adding resistance in series we can increase the total
resistance on the output side of the circuit and by adding resistance
in parallel we can decrease the total resistance on the output side of
the circuit. We decided to modify only circuits with the 5:3 and
the 11.5:7 transformers since they gave us the best inital
signals. To quantify our observations we took the peak voltage of
the primary
signal and divided it by the peak voltage of the reflected signal to
get a
ratio of the signal's peak voltage to the reflection's peak
voltage. By comparing these ratios we found the best results came
from the 5x3 transformer with a 561 ohm resister in parallel with the
output, turning 50.1 ohms resistance of our oscilloscope into 46
ohms. The data we collected can be found in the link below.
Impedence Matching
with Resistors
A transformer ratio of about 1.667 and an output
impedence of 46 ohms implies that the actual impedence of our twisted
pair is about 127.8 ohms.
twisted pair impedence, z = 46 *
(1.667)^2 = 127.8 ohms
It was suggested that we look at the signal output
without a scintillator and at a supply voltage that would produce an
average signal around 200mv. After a little experimentation I
determined that 1200v will generate average signals of
200mv. Using resistances similar to those in earlier tests we
observed similar results except the ratio of signal peak voltage to
reflection peak voltage is much higher than when we used a supply
voltage of 1800 v. To analyze our traces we normalized them by
dividing the voltage of all points in a trace by the absolute value of
the minimum peak voltage of that trace. From the graph combining
all of our traces taken with a supply voltage of 1200v makes it clear
that changing the resistances results in a change in the reflected
signal.
Individual
normalized traces.
I used the same method to compare traces as I did
with the 1800 volt traces. After dividing all of the signal peak voltages by their respective reflection's peak voltage it was clear
that the best ratio was with 43.95 ohms total resistance at 41.1.
Unfortunately the circuit with 43.95 ohms total resistance also
produced the largest positive peak. In an attempt to include the positive peaks in my
comparisons I added the absolute values of each reflection's peak with
their respective positive peak. The lowest total was 22.1 mV from
the 48.1 ohm circuit. The 48.1 ohm circuit also had the second
best signal voltage and reflection voltage ratio with 38.2. From this analysis the 48.1 ohm circuit should give
us the best over all signals.
For those interested I have included the non-normalized traces here.
IceCube 2007 Home
