Needing more than a spark test?

[graham-xrf]

Hmm - The thought that passes is that the circuit may be storing energy in low-pass filter mode, and delivering a pulse of longer duration than the integration deserved . This is not quite like "pulse stretcher", which at least seeks an amplitude, but I think you get what I mean.

In the most extreme form, the "peak-hold" detector, they give up on the pulse duration altogether, and settle for the amplitude only
This need not be too bad. The calibration exercise will discover "buckets" of counts in groups anyway.

It is possible that the "apparently 161keV" pulse actually started out as something shorter duration, provoked perhaps by 38keV to (say) 50keV originally. The circuits we are trying for do attempt a "real" measurement of photon energy. Perhaps unwisely. It might be considered a bit OTT, but it will give a very fine resolution , and can potentially hike the performance if two pulses arrive a bit together.

SparkFun Import Fun
I have untangled what happens. The Royal Mail is a spun off "privatized" company, forced by statute charter to deliver everywhere in UK, regardless how low the population, allegedly "in competition" with other carriers. Then starved of cash while having to employ postmen. Basically, the costs went way up, and the level of service went down, from all directions.

It turns out that, without a "trade deal", import duty on anything over £39 will happen. This amounts to £10.97 on the $96.95 value listed. If I were a a trader, selling it on, this tax is recoverable. But if I am the end user, it gets paid. The flat £8 charge on top is the fee to the Royal Mail, for the task of delivering it to me as a postal parcel. So - if it is from the USA, only buy if less than £39!

I shudder to think what will happen after Dec 31st, when tariff-free trade with the EU ceases. The present situation is that a controversial condition was set that the European Union could never agree to, looking very much like a deliberate deal-breaker. We will see some lip-service gestures, but I fully expect that it will be a bust, and a big mess. The pre-owned South Bend compound part in USA, $125 asked, is going to stay in the USA. I don't know if used, worn, pre-loved, recycled stuff gets a customs charge, but I decline to find out right now.


[Duh! The caliper (empty box) was still in the picture].

 

[graham-xrf]

Although the Pocket Radiation Counter hails from 2012, there have been code updates as recent as 8 months ago.
This includes Python coded plotters, pulse "fairness" detection, "wild gamma ray" detection, etc.
https://github.com/MonsieurV/ArduinoPocketGeiger

APK program to have it work on Android Phone, etc. which would be the pages you likely already got to
http://www.radiation-watch.org/2011/05/professional.html#emb

Vibration Microphonics
I won't be doing any of that, but the thing that is very concerning is the extreme sensitivity to vibration, and the warnings everywhere about it, and the extra circuits and coding to allow it to reject readings.

The key thing I see is that the energy polarity of the photon-inspired pulse is in the opposite direction to a vibration noise pulse.
That is what allows the noise pulse comparator to ouput HIGH when vibration is detected.

Since yours is already powered and running, do you see vibration piezo noise?

 

[homebrewed]

Although the Pocket Radiation Counter hails from 2012, there have been code updates as recent as 8 months ago.
This includes Python coded plotters, pulse "fairness" detection, "wild gamma ray" detection, etc.
https://github.com/MonsieurV/ArduinoPocketGeiger

APK program to have it work on Android Phone, etc. which would be the pages you likely already got to
http://www.radiation-watch.org/2011/05/professional.html#emb

Vibration Microphonics
I won't be doing any of that, but the thing that is very concerning is the extreme sensitivity to vibration, and the warnings everywhere about it, and the extra circuits and coding to allow it to reject readings.

The key thing I see is that the energy polarity of the photon-inspired pulse is in the opposite direction to a vibration noise pulse.
That is what allows the noise pulse comparator to ouput HIGH when vibration is detected.

Since yours is already powered and running, do you see vibration piezo noise?

I did see some very large negative pulses coming out of the 2nd analog stage. I didn't think about microphonics, perhaps because I didn't notice a ring-down (but maybe the scope's time base wasn't slow enough). I had the trigger mode set to Normal so I'd only see the front end of a really slow pulse. The pulse waveforms I showed were taken with the board held in a vise, and connections to the analog-out were via a soldered-on connector, so I wasn't even touching the board. It would be easy enough to slow the scope down to get a better idea of what those giant pulses look like. I'll post a photo or two when I have an opportunity to fire the board up again.

If microphonics are an issue that's another argument for some good pulse-discriminator S/W.

I do recall thinking I was happy to have added some diode clamps to the output of my signal-conditioning board so no big negative excursions would find their way into my 0-3.3V ADC. But now I'm starting to wonder if I should have put some in my 10X gain stages so their recovery time from a large pulse won't be too slow. Well, there are reasons for assigning the board a "version 1" title.

I did find the Android APK but, since our particular application requires access to the analog pulse, I haven't installed it. If this thing doesn't work out as a detector for an MCA I might just use it that way.

****OT****
On this side of the pond we get a well-filtered story regarding the England-EU split. We often watch the "BBC America" broadcast on our public broadcasting channel but, even so, it is painted with a very broad brush.
\****OT****

Now, just where is that $125 SB you mentioned?

 

[graham-xrf]

If microphonics are an issue that's another argument for some good pulse-discriminator S/W.

Agreed! So far, we need..
1. The "It's bonkers size" rejection
2. The "It's too long duration" detector.
3. The "It's the wrong polarity" detector

Microphonics
The comparator in the supplied circuit only trips for vibration pulses going in the opposite polarity as the photon pulses.
I don't know if this is some magic piezoelectric property of the chunk of silicon that it obligingly goes microphonic in one direction only.

Crazy mad theory!
I speculate that if the chunk of silicon is unbiased, and gets a thump in either direction, it can produce the microphonic pulse either polarity, but if reverse bias happens to be applied, the chunk of silicon stresses, and (in theory) deforms a little. Very very small!

In that state, a microphonic thump in direction assisting the bend would deliver, but in the opposite direction, would have to be hard enough to bend it back enough to overcome the bias before it could make the material give a voltage. The circuit is clearly designed as if vibration noise was always going in one direction.

If this is so, then we have an extremely simple fix. A variation of the "Absolute Value" or "Perfect Rectifier" op-amp circuit.
A diode + very low value resistor across the feedback resistor of one of the stages will automatically have it only deliver the wanted polarity pulses. The non-linear diode characteristic is divided by the extreme open loop gain, and it becomes very linear indeed.

In general, this vibration sensitivity is concerning. We should consider de-coupling the diode mass from the board, onto something soft and damped, and make the connections with audio headphone/mic wire lifted from a earplug-type audio lead. That's the stuff designed not to ruck up with multiple flexing, where the main strain is fine nylon threads, and the conducting bits are foil spirals. I have just raided an old headset here. Stripped of the outer jacket, the little wires going to the transducers are very thin, and soft.

Whatever is is, we try not to carry vibrations onto the sensor. Maybe a miniature variant on one of those 1920's Radio City style microphones which has a mic suspended in the middle of a circle frame on rubber bands?

I do recall thinking I was happy to have added some diode clamps to the output of my signal-conditioning board so no big negative excursions would find their way into my 0-3.3V ADC. But now I'm starting to wonder if I should have put some in my 10X gain stages so their recovery time from a large pulse won't be too slow. Well, there are reasons for assigning the board a "version 1" title.

Again agreed. Crude clamps need not be used. We can use the full speed, classic op-amps circuits that precisely limit the signal excursion.

I did find the Android APK but, since our particular application requires access to the analog pulse, I haven't installed it. If this thing doesn't work out as a detector for an MCA I might just use it that way.

My son once made a little app for me, and went the route of making an APK. OK - he designs satellite data communication, tracking, and control software, so he has an advantage. At some stage, we are going to think about how to get this onto a phone, and I might well just ask him what was involved - but that is down the road some.

****OT****
On this side of the pond we get a well-filtered story regarding the England-EU split. We often watch the "BBC America" broadcast on our public broadcasting channel but, even so, it is painted with a very broad brush.
\****OT****

Now, just where is that $125 SB you mentioned?

Oh yes - that was from the other thread --> Re: Compound Ouch
DAT510 supplied a link to Lost Creek Machine
It's the thirteenth down the set of listings, where one of the images has a "A" text in it.

 

[homebrewed]

Crazy mad theory!
I speculate that if the chunk of silicon is unbiased, and gets a thump in either direction, it can produce the microphonic pulse either polarity, but if reverse bias happens to be applied, the chunk of silicon stresses, and (in theory) deforms a little. Very very small!

In that state, a microphonic thump in direction assisting the bend would deliver, but in the opposite direction, would have to be hard enough to bend it back enough to overcome the bias before it could make the material give a voltage. The circuit is clearly designed as if vibration noise was always going in one direction.

There are other components in the signal path that could contribute something w/regard to microphonics. I speculate that the PCB probably is FR4, which is notably bad with regard to very high impedance circuits. Flexure might produce signals just like the wrong kind of coax would. The other thing is the copper-foil shield over the detector. There's a good-sized parasitic capacitor just waiting to wiggle some -- and, like the venerable capacitive (or electret) microphone, should be quite sensitive to vibration.

To resolve the copper-foil issue it would be helpful to know if the top of the detector is the cathode or anode....if it's the cathode, the signal coupled into the detector would be going into the relatively low-impedance bias voltage source. And therefore wouldn't contribute so much in the way of microphonics. Maybe First Sensor has some app notes that might shed some light here.

I tend to think some of these effects are larger than what would change the operating characteristics of a back-biased diode; but I could just be plain wrong in the case of a really large diode like this one.

 

[graham-xrf]

Simulation Limitation Overcome
There are some default settings for tolerances, and minimum value hidden components in default settings, put there to avoid division by zero, and to help convergence for some calculations.

When I find very silly things happening, I had to look for why. I know that the purist way would be to start with a keV energy, which just suddenly "appears" in a circuit, (from a photon arrival), and then spreading itself around in the found capacitance, leaking away into the 40MEG resistance all the time, and some going into the charge amp. It's a bit laborious to model all that, so I take the shortcut, just enough to let me get on with the amp design.

I thought to just cook up a representative pulse that had a similar shape to those we have seen in the work of others. I could get close enough to the energy range by integrating afterwards. I only wanted an approximate maximum, and minimum, and I was interested in how much of it was dark current vs bias. Nothing I could do made sense! Bits of waveform turned back on themselves, Supposedly fixed, instantaneous sources would only get to the settings after ramping. A 2nA peak on a pulse was a mess, that suddenly comes right if you set to 20nA.

The answer is in Control Panel, where one can set the tolerances. Also, capacitors have an automatic minimum conductance across them. etc. It's a bit inconvenient that the settings are "per circuit", and keep going to defaults - but anyway..



I can get picoAmp plots, and femtoVolts etc. by adjusting the tolerances until things snap into shape.
Given the Pi4 can run it's SPI at 125MHz, I was thinking of using a A/D converter with only the serial wires going to it. It does not have to go that fast, just fast enough to gather (say) 20 samples or so over a pulse at the max speed of the converter.
I know I am still circling around solutions, but I will get there soon.

One little question in the mind is about the photon excitation. In the classic photo-electric effect, photons arriving at a metal, the fate of the excess energy after the electron has been liberated is simply to give the electron a speed (kinetic energy). What happens in a reverse biased PIN is much different, as all the liberated electrons, now called "carriers" get collected up, and drift around the circuit.

Here is where we maybe need @RJSakowski I thought that a single photon could only liberate one electron, and whatever was left after overcoming the work function only gave the electron speed kinetic energy. A zippy fast electron can be as energetic as you like, it's only going to ping around in the metal and silicon at speed. It's not going to deliver extra current. So at this point, I no longer understand how this works.

I put the point aside for now. I don't need it to get on with the design. I can still make a TIA amp.

- - - - - -
One thing that is always depressing, is having to pay my taxes. Somehow, however much it is the civic responsible and good honest thing to be doing, I can never work up any sort of cheerfulness over it. (It happened today)!

 

[graham-xrf]

FR4 Microphonic Noise?
I have been searching to find any information, or even anecdotal mention, that FR4 can generate noise currents in traces printed on the PCBs, except in the context of it being a hard material able to carry vibrations to other components, like ceramic capacitors of the "wrong" type. FR4 itself seems to be regarded as very good to use in the most demanding (audio) applications.

Yes - I have had to navigate a mountain of audiophile tosh. If we are getting mechanical piezo noise, then I am finding it hard to see the mechanism.
Yet - the SparkFun counter was designed specifically to detect and flag this as something to be dealt with. A circuit with comparators to detect and invalidate pulses which we don't know how came to be there in the first place!

 

[homebrewed]

High-K ceramic capacitors are reported to be microphonic. But it's hard to believe that the 1pF cap in the input amplifier's feedback circuit would contribute much. Like you say, we just don't know where the microphonics are coming from (yet). The .1uF coupling cap between the first and second stages would likely generate some microphonic noise, but that part of the circuit doesn't have really high gain.

At one point I'd thought about using X7R-based capacitors as piezoelectric actuators but never got around to testing the idea. It's unlikely they would produce much motion, but at that time I was thinking about making a home-made nanoprober, which wouldn't need much range of motion anyway.

It sounds like it's time to fire the board up again and start tapping some of the components to see which one(s) is/are most sensitive to vibration.

Regarding concerns regarding how many hole-electron pairs an x-ray photon generates, all I can say is that silicon drift detectors (basically huge PIN diodes) make pretty good detectors for commercial MCA's. Ketek makes some whose energy resolution goes down to less than 200eV, but they're 'way out of our price range. At least they don't need liquid nitrogen to cool them so they're known as "dry" detectors". They have built-in thermoelectric coolers so they still are operated well below room temperature.

 

[graham-xrf]

What goes on capacitors far outweighs anything that happens in FR4 boards.
The pF value of the capacitor need not be related to the amount of piezo voltage it delivers.
A thick chunk of ceramic, with capacitor plates well apart, might be 20pF, but deliver volts!

Re: "tapping" -a quote [Not my original]
"I submit that if you open up your amp and start tapping parts with a
pencil, the loudest noise you'll hear is the sound of parts being
tapped with a pencil. To solve this particular problem, quit hitting
parts with pencils".

But you are right. If we can get at this mechanical noise, we take out a huge issue. Particularly I wonder about it being specifically always making a pulse in the opposite direction as the wanted photon pulse. It must do it that way, to have provoked the designers to set up the go/no-go pulse comparator as they have. This is why I speculated it was maybe bias related. In one direction, it can overshoot noisy. In the other, it would have to get over the bias - but I just don't know.

 

[homebrewed]

It must be from positrons .