Needing more than a spark test?

[rwm]

In an xray tube you would prefer a nice bell shape curve. But using a W anode you still have large peaks at the characteristic energies of W.
(Emission curve shown below)



This complicates calculation of filtration and mean energy. You can adjust the kVP (peek energy) which shifts the curve. You will not get xray photons with energies greater than the voltage you apply. Most machines max out at 150 KV. If you are using IV contrast you should adjust the kVP to match the K edge of the contrast. For iodine, (K edge 33 keV) you shoot for 70 kVP so the average energy will maximize photoelectric interaction. For Gadolinium, you reduce the kVP somewhat.

 

[graham-xrf]

@rwm : Thanks Robert
I have already seen that the spectra that come out can be a bewildering mess from all sorts of interactions, spin states, and some from shells delivering energy into other shells as consequence of returning to their usual stable state.

For us, there is no problem about the detected X-rays including the main excitation. Being Americium 241 instead of a tungsten target, and set so that none of it goes to the detector, we don't have the unwanted thing in the plot. That is, unless we want to. Showing it a spare smoke detector should show a 59.54KeV peak, and a couple of lower energy peaks, 14.6KeV, and 18.8KeV.

I know that we expect the "peaks" in the spectrum for the lower energy returns also normally happen with lower counts, so the plot also looks "lower". This is because the probability of exciting the L-shell is lower than for the K-shell, even if the energy needed to make it happen in L-shell is less. In theory, one could be plotting a huge high count at a bucket representing quite low energy, but this should not happen

It's OK not to know it all right now. Once we get it working, I expect to have some fun showing it some elements, which I am going to have to get some of for myself.

@homebrewed Mark once posted a picture of a super set of pure elements he scored from someplace. It seemed to come in a sort of long string of little bubble packs. I think I should chase down something like that.

The "simplified version" is, I think, this one..


The wholehearted bonkers one is here ..

 

[graham-xrf]

I'm not seeing any evidence of 60Hz (or 120Hz) noise on the output of my signal conditioning board.

I get it that one is pleased not to see the interference - but if the board is in the open, do you not think you should (be able to)?

 

[WobblyHand]

I get it that one is pleased not to see the interference - but if the board is in the open, do you not think you should (be able to)?

What does the frequency response of the amplifier chain look like? If there's a lot of attenuation at these low frequencies then not much gets through. Of course if one has 60 dB + DC gain then a lot of power line frequencies will get through.

 

[graham-xrf]

What does the frequency response of the amplifier chain look like? If there's a lot of attenuation at these low frequencies then not much gets through. Of course if one has 60 dB + DC gain then a lot of power line frequencies will get through.

Honestly, I don't know how it can even respond to 60Hz. It looks like if given a "sudden pulse", it will wander up slowly to (some) level
The gain of the circuit is not at all 60dB. more like a couple of hundred dB, but we have to take care about what is gain in a TIA. It's dV/dI, which is expressed as dBΩ. By calculating backwards, starting with a nice full height 2V pulse into the ADC using 2.048 for it's maximum count 65536, and dividing by the gain, I tried for for how much of a current pulse started out.

Even though I drastically reduced the gain of the TIA by setting Rf = 1MΩ instead of 66MΩ, and adjusting the next stage to have gain x220, the total is still about 2.2 billion, (including the x10 gain on the signal conditioning board). The pulse that started out would be about 0.9pA. One of my previous attempts at counting electrons into 230pF in 13uS estimated 45pA. Certainly less than the 2fA to 4fA bias current, but I can't really reconcile that with the rest of the datasheet, hence the desire to try LTSpice.

Simulate the LMC662 ?
I wanted to do a (careful) restricted simulation of the LMC662, and I have downloaded a PSpice model from TI.com. It looks as if it will be usable for LTSpice, possibly after some compatibility edits, and a read through. The text warns that "asymmetric gain is not modeled" which may have something to do with there being two opamps in the package A and B. It also warns that noise is not modeled, though this might be added from using the datasheet and stealing from the LTSpice model.

I know the Pocket Geiger was always a threshold driven Geiger pulse maker, and I agree wholeheartedly that purchasing one just to get hold of the PIN photodiode is well worth it. I never did think the electronics would do, and the tracking itself was awkward for the pin-outs I wanted to use, which is why I did not try and use it.

Then Mark powered up, and started measuring stuff. I am envious at the progress, but I am concerned that the little circuit may be giving him confusing symptoms. Real known pulses is something one cannot argue with, and I can only admit my lack of understanding. Yes, if I have a chance, I may try and verify with LTSpice, but not now. I have enough else to catch up on, so I am seriously considering working up something on that evaluation board of post #1180.

KiCAD
I have been getting this added to. I have made a symbol for the Raspberry Pi header interface. I planned to add in the connections for ADC and all that to the Teensy. I got the library yesterday, but it's not yet integrated into KiCad user library path.

LTSpice verifications
My two basic circuits have been simulated, including .OP state to see if there is offset buildup, and noise simulations from 1Hz to 1MHz. The "simpler" version is almost college electronics simplistic, using LTC6268 so I can have guard rings. I have simulated with current pulse of about 45pA, and another of 2.85nA. My calculations of how many electrons result from a 59.54KeV direct pulse from Am241 are not consistent. I won't embarrass myself by putting them here (yet).

I also added an (optional) mains noise notch filter. I trawled several types, and I discovered that some intially disappointing notches would magically reach to spectacular dB depth just by changing the frequency sweep to 20Hz thru 80Hz, and setting it to "linear" with 3200 values. One of the state-variable types gave great results, but it may not be best if one wants to be sure of not saturating the op-amps.

The mechanical stuff
Not forgotten! You will soon be wanting to go for some 3D-printing. I am trying to figure how to drill the tilted holes. Mount the source carrier at an angle on a rotary table vise I think. Maybe just make them face straight ahead? Anyway, playing with FreeCAD must also be done.

Oops - OK, it's that time of year. I just got the PM to renew the donation.

 

[homebrewed]

Hi Mark
Recall your integrator from the past?
On the last page of the TI datasheet for LMC662, I see this trick in use to drive the offset voltage out of the op-amp's output.
Is that something like what you used to do?

View attachment 433999

Sort of, but the offset correction was applied in a sneaky way. The feedback path is via the dual amplifier's power supply current, which is common to both devices, thusly:



The capacitor connected between U1's output and the collector node of the PNP/NPN transistors is a compensation capacitor.

I built an audio power amplifier using this scheme and it worked OK but didn't sound all that great. I have to add that I did NOT come up with the original idea, I found it in an electronics magazine.

 

[homebrewed]

I get it that one is pleased not to see the interference - but if the board is in the open, do you not think you should (be able to)?

Both boards are inside the aluminum enclosure I made. If I open the enclosure up, I DO see line interference on the output.

I'm pretty sure I have some interaction between the two boards though. If they're too close together the output goes into oscillation!

 

[homebrewed]

I am trying to figure how to drill the tilted holes. Mount the source carrier at an angle on a rotary table vise I think. Maybe just make them face straight ahead?

I skipped the holes. I shaped my mounting ring on my lathe, basically turned a shallow cone but I didn't go all the way through -- that left a lip for the sources to rest against. I glued the little Am241 source disks in place with a dab of epoxy to hold them in place. I'll take a photo of my source assembly and post it when I have a chance.

I also found it necessary to fabricate a lead ring to place between that cone-shaped piece and aperture hole in the lead sheet. It's needed to keep primary xrays away from the detector. It totally was a consequence of my effort to get both the sources AND detector as close as possible to the sample, while keeping the detector out of the direct "sight" of the sources. As I subsequently discovered, it wasn't entirely successful, either.

This approach seems to be more sensitive to small misalignments and/or fabrication/assembly tolerances. I think a flat ring of sources likely would be more forgiving, and probably not exhibit much of a reduction of primary xray fluence onto the sample.

 

[homebrewed]

The current discussions regarding the pocket geiger circuit are all good. A different implementation of the front end would be very instructive. Or perhaps "illuminating" is a more appropriate description .

 

[homebrewed]

Sort of, but the offset correction was applied in a sneaky way. The feedback path is via the dual amplifier's power supply current, which is common to both devices, thusly:

View attachment 434031

The capacitor connected between U1's output and the collector node of the PNP/NPN transistors is a compensation capacitor.

I built an audio power amplifier using this scheme and it worked OK but didn't sound all that great. I have to add that I did NOT come up with the original idea, I found it in an electronics magazine.

BTW, the version I made used NPN and PNP transistors in a sort of cascode configuration to supply a more constant power supply voltage to the opamps. That reduced the PSRR requirements on the opamps, and also allowed me to use higher supply voltages to obtain higher power output to the speaker load.