Needing more than a spark test?

homebrewed said:
It would be interesting to experiment with. There are some indications that plastic scintillators may not be very good for detecting low-energy gamma/x-rays -- take a look at this: open-physics plastic scintillator test. They describe an attempt to see the 59Kev gammas from americium, which was a bust. Doesn't bode well for the ~6Kev photons we're interested in, although the plastic should be somewhat more absorptive compared to the primary gamma (a built-in energy filter???). A composite detector, plastic + high-Z scintillator powder, might work better. It IS interesting that they are using a reflective wrap around the scintillator to increase the light collection efficiency. Their approach looks simple but effective.....right up my alley.

Don't buy PEN from Goodfellow -- they want an arm and a leg (maybe more) for their stuff. I can't believe how expensive EVERYTHING is there. They sure aren't selling to hobbyists.
Click to expand...
Re: The Si(PM) with plastics has very interesting links to the detector amplifiers and diodes.
I note they wrapped the plastic in foil to increase the probability of capture.

The scintillation efficiency of Na(Tl) in producing about photons of visible from a photon of X-ray is around 40%
This compares to LYSO and plastic scintillators at about 8%.

I agree we need to take care of background noise, and diode noise as best we can. Fortunately, very good op-amps and other ICs are available at low cost.

Sloppy Arithmetic
In the context of using PMTs, I am still kicking around some loose back-of-a-napkin style calculations
Start with even a "loud" X-Ray photon, say about 50KeV. Even at 100% efficiency, that can only deliver 50,000 electrons.

Now lose 60% of the energy in a Na(Tl) crystal. We are at 20KeV distributed in some new number of visible photons, perhaps 20 or 40, but let us imagine that is the lump we have. There is the photocathode efficiency, and I have no idea what that might be. Suppose we say it is also 0.4. It might be 0.2, who knows?

We are left with as few as 8000 electrons into the first dynode. This does not sound like much, but a "blue" light photon is only 3 electrons worth. The trip to each dynode speeds them up with about 100eV of kinetic energy, and every collision yields maybe 4 to 7 new extra electrons. The whole 12-stage PMT might have a gain of 10^8
8000 electrons is 1.28E-15 Amps, or 0.00128 pico-Amps. At the other end arrives 0.128uA, which is more like something we can use.

I would use a low noise transimpedance op-amp to capture this. It is small, but OK. Easier perhaps than a Ph meter.

Yes - I know this is a sloppy first order shot, just to get a feel for it. It is a long time since I did op-amp design with currents so small it was easier to use electron counts. For those who can appreciate that a single electron can take a week to get around a circuit, but the energy arrives at a large fraction of the speed of light - I think the analogy of hitting a cue ball into a straight line of placed snooker balls helps.
 
homebrewed said:
Silica drying beads can be rejuvenated by baking them in an oven @250F for a few hours. I've done it and it works (you can tell because the beads usually have a little bit of cobalt chloride in them, which changes from blue to pink as it transitions from "wet" to "dry").
Anhydrous Cobalt Chloride is blue, hydrated, it is pink.
Click to expand...
 
According to the folks at Angel Guilding, the thickness of their chemically-deposited silver is around 50 nm. I assume this is somewhat dependent on deposition time, but if it's like electroless nickle, once the surface is coated the dep rate goes down by a large factor.

The silver layer would be pretty delicate. A thin coating of clear-coat (no pigments please) wouldn't absorb x-rays to any significant degree. And it might help the exterior stay a nice shiny color, too!
 
A couple of things. First, regarding silver-plating scintillator crystals, the small mirror-coating kit from Angel guilding already includes paint that is meant to be applied to the exposed silver to protect it from scratches. Applying that soon after the deposition is completed should make it a lot easier to handle coated scintillator crystals.

Second, I've been running simulations of variations on the Theremino pulse processing filter scheme. An AC sweep of the Theremino filter reveals, as expected, really poor performance -- to start with, the filter attenuates the input pulses by 10dB, starting at 10Hz! And the filter slope isn't all that great, either. As a result, the output of the filter for a 1 volt, 100usec wide input pulse is only a few micro-volts. While that is amplified by the transistor amplifier by about a factor of 100, the output signal still is down into the tenths of a millivolt range -- hardly the optimum situation for an ADC whose full-scale input is a few volts. No wonder they want a 16 bit ADC. I think they're also trying to attenuate 60Hz pickup -- but the cost is pretty high, because their filter attenuates the pulse frequencies even MORE (not the right way to improve SNR, if you ask me). Amplifying a low-level signal will also introduce thermal noise so I think they've painted themselves into a corner, so to speak.

I also simulated three different 2-pole low pass filters -- Bessel, Butterworth and Chebychev, each with a -3dB cutoff frequency of 1KHz. The output signal amplitude still is low, on the order of mV, but it IS much higher than the Theremino filter. All of them appeared to perform about the same w/regard to a 100uS input pulse -- there were only small variations in the peak voltage levels.

For those who are not familiar with the filter types, they differ in the sharpness of their filter response. Bessel is the least-sharp, Butterworth is the best you can get without any peaking in the frequency response, and Chebyshev has a small peak just before the cutoff frequency. The use of one vs another depends on whether or not you want good phase response, maximally-flat response, or faster rolloff above the cutoff frequency. For filters in this "slow" regime it probably doesn't matter much which one we want to use. So to eliminate high frequency noise I'd probably just go with the Chebychev.

Also, out of curiosity, I checked to see how the output amplitude scaled as the input pulse height varied. As expected, since the filter is a linear system, there was a nice linear relationship between the input and output pulse heights -- F(A+B) = F(A) + F(B). For the model I used, a 1 volt change in the input pulse height resulted in a 2.3mV change in the filter's peak output voltage. That's without any subsequent gain.

Not much to do with machining here but there is a lot of "magic" in the background that will be needed to get to that "more than a spark test" goal.
 
I was searching the 'net to pursue some other ideas on XRF when I came across a post on the OpenPysicsLab web site. The entry describes using a silicon photodiode array to directly detect x-rays, no scintillator needed: this. The detector runs about $110 from Mouser, although they don't have any currently in stock. The manufacturer also sells their detector integrated with a CsI scintillator but that's gotta cost more. The detector technology looks a lot like the SiPM approach (but the intrinsic layer may be thicker to improve gamma absorptivity compared to visible light detection).

According to the datasheet, the X100-7's gamma absorption efficiency at 5KV is close to 90%, besting the diode/scintillator at the low energy end (the region we're most interested in).

I found a Hackaday post here which suggests that the detector's pulse height doesn't vary much with photon energy, but the pulse width does. A charge amplifier would take care of this issue, although it might be necessary to incorporate a reset circuit to accommodate higher count rates.

I also found an Instructables account of someone's attempt to use one of these but there are some aspects of their circuit schematic that just don't look right to me. Here's that entry. The most obvious problem is the charge amplifier -- the noninverting input only has a 1pF capacitor connected to it, so eventually its input bias current will charge the capacitor up to the point where the charge amp stops working. This one looks like a beginner's attempt at circuit design, and not implemented very well.
 
Hi @homebrewed
You have been so energetic! At least, I mean in the research sense. Because of the vulnerability of two in our household, I have been shielding here in the countryside, and my meddling with PMTs has be somewhat displaced by struggles with a tree stump right where my shop/outbuilding/man cave hard-standing is to go, and a whole pile of other stuff.

I will be playing some with my (working) Ukrainian PMT. The scintillator has not arrived yet. Even so, I think that an affordable alternative gadget that reduces the amount of exotic material components is exactly the way to go. Direct pickup of the X-Rays for the low energy range we are interested in seems obvious, so long as the efficiency of the PIN diode is high enough. We are used to photodiodes delivering for visible light. One hopes it can be near as good for X-Rays, skipping out the scintillator.

Re: Your filter researches.
You are right that when you see a 1pF capacitor, you know that it would only be appropriate for either very high frequency RF or picosecond pulses, or be involved in extreme high impedance circuitry.

In passing, in that you design and simulate with electronic stuff, you will have come across, or deployed active filters. There is no need to allow depletion of original signal information through a passive filter placed on, of all places, a low level signal input, as in Thermino (and a few others).

One can even get multi-stage OP-amps with places to connect R, C, and sometimes L in various networks, often in the feedback circuit, to synthesize all the filter types, band-pass, low-pass, notch Chebyshev, Elliptic, Bessel, etc.

Generally, the first device should supply a whole lot of gain without contributing a significant proportion of noise from it's own input, and have enough dynamic range not to saturate it's own output. In this case, the noise from PMT's, diodes, etc. is way more than the input NF anyway, so the first stage can be an impedance buffer also, at the expense of some gain. The first device amplifies the signal AND the noise, to the point it is hard to add any more noise from subsequent circuitry. The S/N ratio gets preserved.

The best way to stop 60Hz hum is to prevent it getting into the information in the first place. Not sharing common mode current paths, star point design, grounding, shielding, etc. can do this completely. If some gets in, it can of course, be stopped by a (say) 90dB op-amp active notch filter only a few Hz wide.

About shaping
I draw a distinction between the "shapes" of scintillator afterglows, including waveforms produced by slow, capacitively loaded stuff, and the shape of the diode pulse from electrons returning to their stable energy levels. We need the magnitude only, with a gated peak detector so steered that if a "new" bump happens to lift the waveform somewhere during the first pulse decline, it too can be captured, whether it be smaller, or even bigger, than the first. This can be done with analogue circuits. I think it might be better done in numbers, digitally, after the A/D conversion, but I am still thinking about that.

Thanks much for the links. I have to park them for a bit because I have to deal with boring paperwork. (It's accounts and tax time - yuk!)
 
Last edited:
Understand about other things getting in the way of 'fun'. Our choir director had a saying to cover that -- Life Happens. For us right now, the veggie garden is the Happening place -- so I'm getting a pretty good farmer's tan, what with all the weeding. Gotta wear a hat, though, for some reason the sun finds the skin part of my head a lot more easily these days :D

Regarding your comments about filters, I've been thinking that the Theremino approach doesn't really throw away any more information than the so-called "classic" MCA approach. The detector output, which is indeed a current pulse, is immediately integrated by a charge sensitive amplifier (CSA). Just a fancy term for a fast integrator. The "zero-pole cancellation circuit" is there to compensate for the large difference in time scales between the risetime of the CSA due to the fast current pulse, and the slow exponential fall due to the CSA's long RC time constant (even 1pf takes awhile to discharge if shunted by 10^8 ohms).

After writing my previous post I found a module offered by Sparkfun which combines an X100-7, a bias generator to run it, a CSA and pulse shaper + comparator. It's touted as a highly sensitive geiger counter, but that's because the comparator squares up the pulse. All yours for $69.95 USD!! It also has a connector on it that make the shaper output and bias voltage easily accessible, so it may not even be necessary to do any hardware hacking to the board. I haven't seen a boost circuit quite like the one used to generate the bias voltage, but I note that it is pretty heavily filtered -- there's a 100K resistor and 1uF capacitor on the the output -- so it looks like it should be pretty quiet, just what's needed for an MCA application. It probably mostly depends on how well the board is laid out, more than any design shortcuts. I may have to buy one of these.....
 
ericc said:
Isn't the time constant of that RC filter a little long for this application?
Click to expand...
Eric,
Can you be a little more specific on which filter components you're referring to? Is it in the CSA or 2nd amplifier circuit?
 
You must log in or register to reply here.
Back
Top