Fig. 1 Single photon detectors in its enclosure. We have about 300 s-1 dark counts.
The Detection System
The photon detectors need to have high efficiency because coincidence detection depends on the product of the efficiencies on the two detectors. Photomultipliers are not good enough because their efficiencies are low in the near infrared. For this reason everybody uses avalanche photodiodes. Currently there is only one vendor of these modules (Perkin Elmer/EG&G Canada/Pacer-UK).1 The modules, though expensive, take care off the electronics, producing convenient TTL pulses. Figure 1 below shows our two detectors.
Fig. 1 Single photon detectors in its enclosure. We have about 300 s-1
dark counts.
Because the detectors are very sensitive to background light, we have to either do the experiments in complete darkness or cover the detectors. As can be seen, we do the latter. We made a light-tight enclosure that has two openings covered with red filters.
Since the down converted light is broad band we need to filter it for doing interference experiments. We bought band-pass filters from Andover Corp. with bandwidths of 0.1, 1 and 10 nm. In addition, we need to focus the down converted light with a lens, because their active area is circular with a 0.175-mm diameter. The irises shown in the figure are really not needed.
The next step is to detect the pulses produced by the photon pairs in coincidence. We could use coincidence circuits, but then the photons need to travel the exact same distance. Since we want to send one of the photons through an interferometer, their corresponding pulses will be delayed. For this reason we use a different but also conventional approach. We use a Time-to-Amplitude converter (TAC/SCA),2 which can count pulses delayed by a certain amount as a single event. When we use this method we need to set up the window of the TAC/SCA, so we also need a multichannel scaler. Our electronic components are NIM modules. Figure 2 below shows our setup.
Fig. 2. Electronics. We use a TAC/SCA (left-most), a multichannel scaler (blue)
and three
counters. All fit in a NIM mainframe.
Since then we have built our own counters using the "stamp" chip, which outputs to green displays (because the light from the red displays gets into the detector box).3
Below in Fig. 3 we show a plot of events as a function of the delay between the detector pulses. Because of the insertion delay in the TAC/SCA one of the cables was 10 ft longer than the other one. The peak clearly shows the signature of down conversions.
Fig. 3 Number of detector pulse pairs as a function of the delay between them.
These data were taken by Naomi Courtemanche '02.
Another signature of down conversions is the variation of the counts as a function of the position of the detectors. This is shown in Fig. 4 below.
Fig. 4 Counts recorded when we move the signal detector. The singles counts
of the signal (pink) and idler (white) vary smoothly, but the coincidences are
much more localized (yellow) due to the momentum correlation of the pairs.
Data taken be Naomi Courtemanche.
In the early days of our project we recorded the reading of the detectors by hand. After a while this became quite cumbersome so we decided to automate it. We bought a pulse counting module from National Instruments (PCI 6703+BNC2121) and used a program written in Labview. Figure 5 below shows the screen of the computer of our current version.3 It records up to four streams of pulses for a certain amount of time and then plots them. Normally we use only three inputs: the singles of each detector and the coincidences, which are the output of TAC/SCA.
Fig. 5 Data acquisition system. The program was written
in Labview by Matt Pysher '04, Justin Spencer '05 and
Dimitar Simeonov '07.
When we do interference experiments we apply a voltage onto a piezo electric that changes the length of one of the arms of the interferometer. The program supplies this voltage through a digital interface (Stanford SRS245). This voltage is further amplified by a voltage amplifier (Trek model P0516A-1). The program then takes data in a scan mode, where it steps the voltage that goes to the piezo.
1Perkin Elmer model SPCM-AQR-13.
2Canberra Industries model 2145.
3We will be happy to supply the diagram/program upon request.
E.J. Galvez/Colgate U.