From tigwiki

[edit] What is it?

DANTE (Di-Pentagonal Array for Nuclear Timing Experiments) is a collection of Barium Fluoride (BaF₂) detectors for use primarily with the 8π.

[edit] Hand Assembly of Detectors

Scott Williams is the resident expert on hand assembly, however a short precis follows:

• The photomultiplier tubes are Photonis XP2020/URQ type which come from France. The BaF₂ crystals are manufactured by Scionix in Holland.
• The PMTs come from the factory with a test ticket (see here for some older ones) and (optionally) a cover for the quartz front face of the tube. When the PMT is first unpacked, it will have a protective tape over the quartz face which you can peel off. Do not touch this front face. Make a note of the serial number of the PMT and make a label to be attached with the label maker.
• The tube is then wrapped with a black onionskin paper and this paper is taped onto itself with black electrical tape, somewhat like how you would wrap a Christmas present. Then take a roll of black electrical tape, and wind tape around the PMT, starting from just at the edge of the quartz face until you reach about a quarter of an inch away from the pins. This layer is never removed. Attach your serial number label to the bottom part of the tape. (see here for an unwrapped PMT. Note the white label which will have the serial number on it. I don't have a picture of a partially wrapped one at the moment.)
• At this point, clean the surface with any of methanol, ethanol or isopropyl alcohol to remove the residue on the front face. This residue will either come from the protective tape or old optical gel from a previous assembly of the detector (see this picture for what the front face should look like after cleaning - you should be able to reflect a fluorescent lamp in the ceiling or another light source with no blurriness, and you should be able to see your face in the tube without trouble).
• Unpack the BaF₂ crystal and clean the bottom face of its surface, also with alcohol.
• Wrap white Teflon gauze securely around the conical and top surface of the crystal unless it has been precoated with a metallic surface. The Teflon "sticks" but can come off easily. Take care when doing this part of the wrapping.
• Find a container of the specialized optical gel. Do not use the Dow Corning gel from TRIUMF stores. The special gel has a UV cutoff at just below 200 nm, which is what we need to pass the fast component of the light output from the crystal (at 220 nm).
• Stand the photomultiplier tube straight up on the special mount provided for the purpose. Note where the serial number has been placed per the unpacking and initial wrapping instructions, and the square foam padding from the PMT box underneath to cushion the tube if the metal chain should come loose.
• Place a small amount of the optical gel on the front face of the PMT.
• Carefully lower the crystal onto the front face of the tube, and use the weight of the crystal to help spread the gel evenly over the front face of the PMT. (As shown in this picture)
• Using black electrical tape, carefully ensure the crystal remains in position as you tightly secure it to the PMT proper (initial stage shown in this picture). Ensure that the tape layer is light-tight! You will have to do this part of the taping twice for this reason.
• Then wind 'round the PMT cylindrical part again from the top of the cylindrical part until your tape reaches the label you attached previously. See this picture for the final result. Note again where the serial number label is.
• That's it, you're done.

[edit] Operation of the Barium Fluoride Detectors

• The photomultiplier tubes go into a voltage divider base (shown from the side without the protective cage), which has connections as shown on this picture of the base's rear. HT is high voltage, and "anode" and "dynode" are self-explanatory. The anode pulses go to the CFD inputs for timing and the dynode pulses go to the preamp boxes. The PMT's pins go into this part of the base. Note the notch in the center of the circle, which matches the equivalent notch on the bottom of the photomultiplier tube.
• For those who are familiar with HPGe detectors but not scintillators, the anode output is the equivalent of the output from a TFA, as both are used for timing signals. The dynode output is the equivalent of the raw output from an HPGe detector, but with a reversed shape (so instead of the pulses going under the baseline, they go over the baseline, as shown here).
• When you have the base secured (either in the test rig or in the 8pi proper), the cable connections look like this. One of the Surrey BaF₂ detectors used for a reference is shown in its test rig, and the roll-our-own models are shown in the 8pi, with mu-metal shields, here.
• After assembly, conditioning the tube is not really necessary but I have read references that state that once powered on, the tube should be left to stabilize for ~half hour before taking any measurements that are of some importance. The specifications from Photonis are in this PDF file. If you're just doing some quick checks, they work almost right away. Do not exceed the maximum safe operating voltage (in fact, I would hesitate to go over 3000 volts). Nominally the minimum is about 2200 volts, but in fact I have operated these at 2150 V without adverse effects, though I imagine the signal to noise ratio might be out of spec. There might also be hysteresis effects but I haven't investigated this too closely. The bottom line is, once you find a good operating voltage, stick with it if you can.
• In that vein, to find the operating voltage take anode pulses with an oscilloscope, as shown here, for example. The depth of the pulse should be about -7 volts for the 1408 keV gamma from a ¹⁵²Eu source, or a 1332 keV gamma from a ⁶⁰Co source. This number was more or less arrived at ad hoc based on the following:
  • CFD inputs range from 0 to 10 V.
  • We want to have some 'head room' for higher energy gamma rays than the sources allow - i.e. we might see a ~2 MeV gamma and if the CFD has to clamp the voltage then the timing may not be accurate.
  • The gain is nonlinear with respect to applied voltage. There is roughly a parabolic relationship, and while I haven't plotted the curves for all the detectors, I've done it for at least three and they all are parabolic. So depending on the initial voltage, you may find that a relatively small change (say, 25-50 volts) will greatly enhance the signal output at the anode.
  • Obviously, feel free to adjust this as necessary for your experiment. In particular if you wish to spread out the spectrum to focus on very low-energy gamma rays then you can go ahead and crank the applied voltage upwards.
• Current applied voltages on the BaF₂ detectors are all around 2200 V. I will put together an actual table with DANTE01 to 10 later on. Since the voltages are right near the minimum and still yield approximately -7 V (caveat: the termination of the oscilloscope can alter these readings. Make sure you're using the 50 ohm scope input), there is ample room for cranking up the voltages. This is where I suggest using a ²⁴¹Am source, or just use a source with known strong X-rays (¹⁵²Eu has two strong X-rays at around 30 keV).
• Equipment:
  • Ortec 855 Dual Spec Amps are used for energy signals. Currently, coarse gains tend to be 2 or 4, with the exception of DANTE07, which has a wonky pre-amp. I had to raise the coarse gain to 20 or 40 to get proper gainmatching. Be aware of this if you need low-energy gammas. Don't bother trying to pole zero. The unipolar output actually 'looks' bipolar and I suspect this is just a function of the shaping time of those amps.
  • Ortec 935 fast Quad CFDs are used for timing signals. The basic procedure for adjusting thresholds and walks is identical to that for HPGe type detectors. Trigger off the logic output of the CFD and monitor either the associated amplifier or the CFD internal monitor. Note that unlike the HPGe Ortec 583B CFDs, there is only a lower level threshold and this is adjusted by a screw. Walks are also adjusted by a screw. Walks are particularly tricky compared to HPGe, and it takes a bit of playing around to get a feel for when the walk is best adjusted. Basically read the manual and follow those instructions. You'll see where the pulses suddenly 'snap back' and look really good. However there will still be a little room for improving the CFD walk and this is where you'll have to 'zoom in' with the scope by going from, say, 500 millivolts/div vertical to 100 or whatever. Basically you turn the screw until you see the 'walk area' get as small as you can make it.
  • The amplifiers go straight to the ADCs are are then read out in the FERA chain.
  • The outputs of the CFDs go all over the place. One of them goes to the TAC modules (Ortec 566) to provide the "start". Another goes to what used to be an octal logic set to provide "stop" signals. Since then this has been deleted in favor of several fan in/fan out modules and Ortec logic modules which have switch settings allowing for self-timing or non-self timing modes. These modules now do the OR logic which provides a 'stop' to the associated TAC if any two detectors fire. The TACs are then read out by ADCs as well in a FERA chain.
• Checking energy resolutions:
  • The "standard" for a BaF₂ detector is to use a ¹³⁷Cs source. The placement of the source is not critical unless you are doing efficiency measurements. Place the source near the detector(s) of choice, and then acquire data until you reach an integral of at least 10000 counts under the photopeak. You do not need to write a .mid file in the 8pi DAQ; the .root histogram alone is enough to get what you need.
  • The way I fit the peak is to extract the appropriate energy spectrum from the .root file using the program root2spe. One way to use it is to log in to isdaq08.triumf.ca and issue commands as follows (keep in mind this is a sample session based off of a ¹⁵²Eu experiment being done at the time of writing. Your exact .root file name and directory will be different. The ROOT histogram name is of the form EBF(detectornumber) which ranges from 00 to 09. Since we number the preamp boxes and the detectors themselves from DANTE01 to DANTE10, you will need to be careful not to mismatch detectors when you extract RadWare spectra.

[dcross@ibm00 ~]$ ssh tigress@isdaq08
tigress@isdaq08's password: (password here)
Last login: Thu May  1 13:34:51 2008 from vistemp60.triumf.ca
Fri May  9 13:03:25 PDT 2008
1:03:25pm tigress@isdaq08:~% cd root2spe
1:03:28pm tigress@isdaq08:~/root2spe% mkdir ResolutionCheck
1:03:37pm tigress@isdaq08:~/root2spe% cd ResolutionCheck
1:03:40pm tigress@isdaq08:~/root2spe/ResolutionCheck% ../root2spe /data3/8pi/152Eu/his00539.root EBF00 DANTE01.spe
Hist is 8192 by 1 by 1

• After that, I send the .spe file back over to my IBM00 account where I have RadWare's gf3 installed and I fit the 662 keV peak and quote the resolution in terms of the FWHM divided by the centroid (e.g. a 10% resolution would come from a FWHM of 300 channels divided by a centroid of channel #3000 - this neatly gets rid of the problem of imperfect gain matching). An error analysis is possible, but if you have well over 10000 counts the error will be negligible in percentage terms. A resolution of much greater than about 12% is definitely unacceptable. It is a judgement call if a detector is "borderline" at around 12%. The best resolution possible is 8 to 9 percent so do not expect miracles here. A caveat must also be introduced here that to some extent the resolution may possibly be affected by the applied voltage and gain matching, in a more than negligible fashion. This has not been investigated exhaustively but the usual practice has been to make sure the 662 keV peak is between one-third and two-thirds of the way across the spectrum. (i.e. from about channel # 2000 to 5000 or so).