Needing more than a spark test?

[graham-xrf]

@homebrewed
Mark - do you have a fave suggested switched gain range amplifier?
I was looking at various LT6370, but not sure..

 

[WobblyHand]

@WobblyHand
I screwed up with the FreeCAD colours.
I need some way to set them back to standard, or something, so I can see the sketches I "revolved"

No worries. Take your time. Save early and often! I have been messed up with colours before, so I can appreciate what just happened.
Have you tried undo?

 

[homebrewed]

This is, I think, my point. The wideband noise racket (currents) is associated with the DC dark current, and gets bigger with bigger dark currents, which get bigger with temperature, but that wideband noise is not the same thing as the DC dark current. Knowing that we can see a noise floor lower that what comes out of a reverse biased diode, I think we can have this gadget work as well as any at the kind of workshop temperatures most of us would tolerate on a hot day.
BUT..
I do have one of those "flat" heat pipes about15mm wide, and 3mm thick. If simply put in there, as if it was part of the EM shielding, one might put the end of the heatpipe sticking out of the machine into a glass of ice.

I don't think we will need to go for cooling, but there is easily the potential to accommodate it.
Mark - thanks much for the Id equation. I know I have it somewhere, but that was handy.

The X100-7 data sheet specifies the typical noise current at Vr = 12V and T = 23C (300K) to be 6.1E-14 amp/sqrt(hz). 61fA/sqrt(hz).

 

[homebrewed]

@homebrewed
Mark - do you have a fave suggested switched gain range amplifier?
I was looking at various LT6370, but not sure..

Nope. I really haven't looked at that type of amplifier. Something like that might be a good choice for a simplified signal conditioning circuit.

 

[homebrewed]

@homebrewed
Mark - do you have a fave suggested switched gain range amplifier?
I was looking at various LT6370, but not sure..

Actually, I just recalled that I bought an instrumentation amplifier for an ionization detector, aimed at possibly characterizing Am241 sources. It is an MCP6N11 with somewhat better specs than the LT6370, and a MUCH better price ($1.77 in unit quantities). The former: Vos = 3mV, Ib = 10pA. It can operate with supply voltages ranging from 1.85V to 5V.

 

[graham-xrf]

The X100-7 data sheet specifies the typical noise current at Vr = 12V and T = 23C (300K) to be 6.1E-14 amp/sqrt(hz). 61fA/sqrt(hz).

Yes - exactly as I said in #1101, which I multiplied by 300kHz to come up with 20nA.
As I understand it, that is a DC current. In theory, it can even be a totally noise free pure DC, though we know it is not.
Not the same thing as the random fluctuations waveform with frequency components over a wide band, ranging from the troublesome 1/f shot noise through to microwave random hiss.

Getting a value to put on it might be quite hard, but one starts with √(4*k*T*B*R)

We know the racket depends on the dark current, which depends on temperature, and there is also some contribution which depends on temperature alone, and is there even if we don't connect it to allow a dark current to flow.

Since you have those (great) scope traces of real pulses, you can tell they are there, with only a little wiggle on the line thickness. When I saw that, I thought OK, so long as we make the front end clean enough, we have a good chance of measuring even much smaller pulses.

 

[graham-xrf]

Actually, I just recalled that I bought an instrumentation amplifier for an ionization detector, aimed at possibly characterizing Am241 sources. It is an MCP6N11 with somewhat better specs than the LT6370, and a MUCH better price ($1.77 in unit quantities). The former: Vos = 3mV, Ib = 10pA. It can operate with supply voltages ranging from 1.85V to 5V.

Thanks for that - I will be checking it out for design.
I fully expect that one may want to "zoom in", and be able to have the ADC resolve better for low energy arrivals, knowing that the buckets that would have been overloaded are not even being looked at.

I thought perhaps 3 gain ranges. Lowest gain, then perhaps 10dB higher, and the most sensitive state set at 20dB higher.
One might choose something else, like a x2 doubling and then a x5

 

[homebrewed]

I've been doing some experiments to get a handle on the various noise sources I'm seeing in the MCA results. One of the photos I posted earlier showed a couple of significant peaks with my x-ray sources in place but no sample, and another with the x-ray sources removed, which showed one lower-energy peak. One experiment I did was to slide in a lead shield in place of the x-ray source setup, and I was very surprised to find that the overall count rate dropped by a significant amount compared to the no-source (but no shield) result. By "significant" I mean that the peak counts dropped from hundreds to less than one hundred -- for an overall accumulation time of about 2-1/2 hours.

This result suggested that I might have some kind of noise problem due to interference of some kind, rather than stray xrays, so I did some additional tests. I made an electrical shield out of some thin copper sheet and inserted it between my pocketgeiger and signal conditioning board -- and the count rate dropped by a significant amount. Like 5X. I replaced the lead shield I had on the other side of the pocketgeiger with another piece of the same copper sheet -- only a small improvement, perhaps down in the noise. This suggests that at least some of the counts are the result of feedback between my two boards. Much earlier I noticed oscillations when I opened up my enclosure enough to permit access to the offset trimmer on my signal conditioning board so I have concluded that at least SOME of the low-energy noise (as presented by my MCA plotting code) is due to undesired feedback between the pocketgeiger and signal conditioning boards. The xray detector is a 100mm^2 piece of silicon connected to a high-gain circuit so it's not totally surprising that some feedback is going on. 100mm^2 is a large antenna in a high-gain circuit like this one. More partitions in my enclosure, with some beefy aluminum plates separating them seem to be in the near future...

I also noticed large variations in the amplitude of that higher-energy peak due to relatively minor changes in the distance between the xray detector and xray source plate. Moving that source plate further away from the detector resulted in a dramatic drop in the stray count rate. So I conclude that the high background count rate due to the sources may due to two problems: Poor control over the distance between the source assembly and detector; and poor design of the source assembly, which allows primary xrays to get to the detector.

To facilitate these experiments I modified my Teensy XRF code so it keeps track of the average count rate of qualified pulses. The count rate actually was a better diagnostic than the MCA spectrum, since it's auto-scaled so the accumulated data fits on the screen no matter what. The count rate is more of an "absolute" indicator, although as written it _does_ depend on the chosen trigger level. For all my experiments I chose a 100mV trigger level. That's with a 10X gain selected on my signal conditioning board, stock gain-setting components on the pocketgeiger.

My code converts the binary 12-bit ADC result to floating point, using the Aref voltage I measured on my particular Teensy4.0 -- 3.286V. Hopefully the characterization info will help other folks adjust their measurement systems accordingly. NOTE: The ADC's maximum resolution is 12 bits but the SOC system noise is reported to reduce its ENOB to closer to 10 bits. While my pulse integration approach should greatly improve the effective SNR, starting out at 10 effective bits isn't a great starting point....

 

[graham-xrf]

I think, start with no sources, and short-circuit diode, and look to drive down the residual noise signal as far as possible.

If the plot axis has a high count of lower energy buckets to the left, then the crowd of unwanted stuff is probably noise. If the crowd distribution is not very wide, then it might indicate a single amplitude signal leak or oscillation.

With the diode active, think that some x-rays might go right through the diode package. It might benefit from a 2mm lead shield stuck on the back. We have a scene where there is radioactive stuff right up close to metal, splattering X-rays all over the place. X-rays mess with electronic stuff. Everything in there with gain needs not to see any x-rays, nor any electric fields, nor any magnetic fields. You can see why the pocket-geiger came wrapped in foil.

More subtle, but neither should the electronics in there see any conducted noise, hum, and other energy that any wires to and from the kit might make it prone to. Our gadget is likely as sensitive to fields as purpose built EMC equipment designed to sniff them out.

 

[WobblyHand]

My interest has been piqued enough to take the plunge. What specifically do I need to start with? Is it the Sparkfun Pocket Geiger Radiation Sensor Type 5? This is to access the large PIN diode? Is any of the rest worthwhile, or is this just a way to get the detector? Or is it better just to buy the diode and make a TIA?

Speaking of analog hackery anyone know where to get little glass standoffs like old analog guys used to use 30+ years ago? You would drill a hole in the PCB and solder them in and they had a terminal to solder a wire. They had low capacitance. I have a circuit with a few of these. Don't know what the proper name is for these terminals. Are they made any more? Used to do galvanometer like work with them, low leakage and capacitance.