Page 1 of 3

A semi-DIY preamplifier for radiation detectors

Posted: Fri Feb 15, 2019 6:25 am
by Rich Feldman
Yay! Got some own work to report. One more step toward counting neutrons at home. I scampered & cut corners last weekend. Then my birthday came and went, and project still needs work.

My BF3 tube wanted a Charge Sensitive Preamplifier (more about CSP's later).
Following in the footsteps of Dan Knapp, I started with a Cremat brand hybrid circuit module called CR-110.
Chose not to buy their almost-ready-to-fly circuit board (CR-150) or matching aluminum box. Partly because of wanting to make component substitutions. Mainly because of habit: the sacrifice of hours of "making" time, on the altar of low out-of-pocket cost.

You couldn't get the amplifier much closer to the detector tube.
The circuit differences between my build and a CR-150 rev5 are highlighted in schematic diagram:
First: both HV caps are rated for 3 kV. Online pictures of rev4 and rev5 show 2 kV capacitors, but Fred from Cremat told me his latest batch was assembled with 3 kV parts.
Second: In email discussion about resistor voltage ratings in bias circuit, Fred warned me about a more important kind of fault that can destroy amplifiers, when detector is a HV gas tube. Not sparking bias resistors. A sparking detector (or detector connector?), can cause C1's HV charge to suddenly appear on amplifier input pin. I followed his suggested mitigation: Add series resistor (100 to 200 ohms) and reduce C1 to 1000 pF -- which is OK when detector node capacitance is very small.
Third: Expedient power source is a pair of 9 volt batteries. Originally they were going to be inside the box, but I decided to skip that in order to see pulses sooner.
Fourth: In other places I used components that were handy. Didn't need to use the HV version of 10M resistor. CR-150 is OK with regular (non-HV) resistors in bias circuit, though they see full voltage momentarily if bias is switched on suddenly, or when there's a spark discharge in detector circuit. The CR-150 design had chip resistors in rev 4, changed to regular axial lead resistors in rev 5.

Circuit was assembled on a scrap of perf board with copper pads on one side, copper ground plane & antipads on the other. The hybrid module is in back corner of picture, leaning like a shed roof, in a bent-over SIP socket.
Got a setback last night. Connected bias supply for a proof test at 3000 volts, before plugging in the expensive CR-110.
I'd cleared lots of metal around each HV circuit node, but not enough. A couple times a minute there were little snappy sparks next to one or the other of the HV nodes. Good thing the bias supply is very weak on purpose -- equivalent resistance of a couple megohms, and output capacitor that stores only 0.045 joules. My response was to remove even more metal -- already done when the picture above was taken. But soldered components interfered with doing that properly around the detector node connections. I think the best course is to rebuild the circuit on a perfboard that has much less metal to begin with.

In the meantime, with the bias input cable removed: Plugged in CR-110. Powered up without loss of smoke. Exercised the Test function, a valuable feature for which I applaud Cremat or whoever they got the idea from. We can verify that the amplifier is in good working order, without needing HV bias or even a detector. It worked as expected on the first try!

Re: A semi-DIY preamplifier for radiation detectors

Posted: Sun Feb 17, 2019 8:03 pm
by Rich Feldman
Progress report.
1. Sparking was solved by removing the R's and C's in bias circuit, so perf-board metal could be cleared in a wide margin on both sides.
With the R's back in place, circuit sustained 4500 V for 5 minutes with no spark or series-NE2-lamp visible glow in the dark.
Then with both high voltage C's back in place, and another NE2 in place of amplifier input, circuit sustained 3300 V for 5 minutes. Lamps glowed only when HV was ramping up or down. I didn't attempt a more quantitative meggering.

With 10 megohm DVM at detector connector, and HV cranked up briefly as far as 1500 V, voltage division ratio was roughly as expected. This test took the four 33M resistors well above their voltage limit, and hinted at some nonlinearity (more current than expected for the higher voltages).

2. All back together, including BF3 tube, the amplifier Test function worked as before. This time on my home oscilloscope.

3. Next problem became evident when HV bias was cranked up to 500 or 1000 volts. Got a couple volts of 60 Hz ripple at amplifier output. I had never modeled or measured the ripple voltage of DIY bias supply, except to know the RC product of output cap and kV meter resistance: 300 millseconds.

Went to measure bias ripple by cabling "detector" connector to scope channel (1 megohm input) and observing expected divided voltage (6 V when bias meter said 1000 V). Flat line at that scale, so I switched scope to AC-coupled mode. Unexpected behavior led to an uh-oh moment: AC coupling removes the 1 megohm DC load, and consequent attenuation of 1000 V to 6 V. Time to call it a day.

4. Simple remedies may be a combination of: Increase bias supply output C. Make a better lowpass filter (inductive choke even?). Change rectifier from half-wave to bridge. Reduce C1 in amplifier circuit. AC-couple the output of amplifier circuit, with time constant small enough to suppress 60 Hz or 120 Hz.

Suggestions are welcome.
[edit] It's now obvious where to start. Simulation confirmed quick analysis: go with the bridge rectifier to double the ripple frequency. The existing circuit has 3 stages of low-pass filtering, so ripple amplitude will come down roughly by a factor of 8 (well maybe 4 at amplifier output, because of differentiating effect of HV blocking cap).
Unsymmetrical 60 Hz waveshape on scope looked like the blue curve in this simulation.

Re: A semi-DIY preamplifier for radiation detectors

Posted: Mon Feb 18, 2019 6:00 am
by Jerry Biehler
You could use a cheapish isolated DC/DC converter module. They have basically no ripple. You definitely wont see any 60hz!

Re: A semi-DIY preamplifier for radiation detectors

Posted: Mon Feb 18, 2019 7:43 am
by Rich Feldman
Well said, Jerry. I could probably find one or two, perhaps a laser power supply module, in one of many HV junk boxes in the garage. Or swap the transformer in a cheap low-voltage regulated DC power supply, with flyback topology, like we find in appliance controller boards these days.

The primitive bias supply design attracted me. In fact it's one 1X2A tube short of being semiconductor-free.
1x2a~~1.jpg (9.75 KiB) Viewed 5421 times
A tube like the one in picture was measured to have a lower forward voltage drop than the 30 kV silicon diode it would replace.
Of course vacuum technology would make the change from 1-to 4-diode rectification much less trivial.

Wonder what Nancy Wood used to bias her BF3 tubes in the Manhattan project?

Any other retro electronics fans out there? I forgot to bring, to HEAS, my red laser pointer (commercially manufactured) with a He-Ne laser tube and 9-volt batteries.

Re: A semi-DIY preamplifier for radiation detectors

Posted: Mon Feb 18, 2019 9:35 pm
by Richard Hull
I am about as retro as anyone in electronics. A smaller and cooler tube rectifier was used in all Textronics O'scopes for their anode supplies for years. It was a flying lead diode and could easily handle 10 KV with less filament current. I have seen them in a few early portable GM counters. The smallest in the image below is from a Sears and Roebuck "tower" brand scintillation counter of the early 50's.

As such GM HV supplies only needed mere microamps of current, portable counters quickly moved over to long tubular selenium pellet stack HV diodes.
Most were 2" long, slender 3/16" diameter stick phenolic bodies filled with tiny bonded, stacked Se pellets.

Richard Hull

How retro am I?..... go to my old post here


Data on the tubes below...

Re: A semi-DIY preamplifier for radiation detectors

Posted: Tue Feb 19, 2019 5:31 pm
by Harald_Consul
Very interesting work Rich!

Just for the stupid guys. How much MHZ/ nanosec rise time is your from-scratch-preamplifier?

Re: A semi-DIY preamplifier for radiation detectors

Posted: Wed Feb 20, 2019 9:52 pm
by Rich Feldman
Thank you, Harald.
I don't have meaningful risetime measurements yet. The CR-100 is supposed to have an impulse response like:
with risetime of about 7 ns and decay time constant of about 140 us (you could work it out from the test mode waveform picture).

I think it's common for CSP's to be followed by shaping amplifiers that get rid of the long tail. Here is a picture of two long-tailed events piled up, and a signal that can be recovered from that waveform using trivial analog electronics.

Re: A semi-DIY preamplifier for radiation detectors

Posted: Thu Feb 21, 2019 8:13 pm
by Dan Knapp
There are two companies that use the Cremat modules in their preamps for which they post the manuals online. These user manuals include circuits and other useful information: ... /csp1x.pdf

Re: A semi-DIY preamplifier for radiation detectors

Posted: Mon Feb 25, 2019 3:47 am
by Rich Feldman
Thanks, Dan.

Here's a question for all, about turning HV bias on or off. The cited manual from Fastcomtec says "Increase or decrease the high voltage only at a very slow rate. Observe output during bias voltage change with an oscilloscope. Do not allow the output to saturate during change of high voltage."
Do people really do that, when dealing with three- and four-digit bias levels for gas tubes? In my experience it takes tens of seconds to get from 0 to 1000 volts.

The numbers make sense. Manual for the CR-110 hybrid module recommends AC coupling for detectors with bias current exceeding 10 nA, to avoid saturation. A steady 10 nA produces 1 volt at output of the first amplifier stage, and 2 volts out from buffer stage. That's within a factor of three of output clipping levels.
If external HV blocking capacitor were 10 nF, as found in all three products cited in this thread, a HV ramp rate of 1 volt per second would produce 10 nA at amplifier input. 3 volts per second would surely saturate the output. I'm using a 1 nF capacitor, so it can charge at around 30 volts per second without pegging the amp.

Why is saturating the amplifier bad? Reminds me of oscilloscopes in AC-coupled mode, when you put the probe on a signal with large DC offset, and trace disappears offscreen for a few seconds. If my HV were ramping from 0 to 1000 V in one second, a whole microampere would be flowing into the circuit module input. Rf and Cf can only pick up a small fraction of that, so the rest will find what I expect is a nondestructive alternate path to ground. Unless the module's electrical performance requirements call for a bare MOSFET gate and no protection diodes.

Re: A semi-DIY preamplifier for radiation detectors

Posted: Mon Feb 25, 2019 5:42 am
by John Futter
Yes Dan
we increase voltage very slowly especially on solid state detectors (conditioning)it allows the depletion region to move into place ie extremely high impedance without puncturing (read expensive and now a low voltage zener not SSD)
we use ortec quad bias supplies