开云体育

----- Original Message -----
From: GEOelectronics@...
To: [email protected]
Sent: Sat, 12 Dec 2020 01:44:50 -0800 (PST)
Subject: Re: [XRF] Proportional Counters and Preamps

"which brings me to burnout

Randall"

Not even close, Randall, once the electronics are as good as they can be (did you mention coaxial cable capacitance?), the tube itself, or rather the gas inside it opens a whole new con-of-worms. Imagine the excess contribution of heavy noble gasses like Argon or Krypton or Xenon muddying the XRF waters at the atomic level. Internal XRF peaks, selective X-Ray absorption etc.

Beryllium windowed Proportional Counters were the only game in town at one time, kudos to those patient enough to work with them.

PS they do make for excellent XRF targets, for element XRF right through the Be window. Probably the most practical way to add Krypton and Xenon to your element collection.

Geo

----- Original Message -----
From: "Randall Buck" <rbuck@...>
To: [email protected]
Cc: [email protected]
Sent: Friday, December 11, 2020 10:44:30 PM
Subject: Re: [XRF] Proportional Counters and Preamps

Hi,

There seem to be quite a few interrelated electrical parameters, physical properties (of the x-ray/PC chamber) and design compromises involved.

I don't have it all organized into one simple view.

1) Because the PC chamber has an intrinsically low capacitance, the output pulse should have a very fast rise time, hence the need for an op amp
with a high but not excessive slew rate.

2) Since we are interested in the accurate reproduction of the pulse from the chamber (but at a higher amplitude for further processing - ultimately
by the MCA), the settling time spec is also critical. Settling time for the op amp output peak to reach the amplified version of
the peak input pulse from the PC chamber. Settling time is usually specified to within some percent of the final value, which is only obtained after
a relatively long delay; typically, to 0.1% of the ultimate output peak. A more accurate pulse amplification process would result from the same settling
time to say 0.01%. For us DIY'rs, the actual settling time is not too critical since we can always slow down the count rate via x-ray attenuation and just
wait longer for the result. So we want a precision op amp with a slew rate that exceeds the rise time of the PC output, settles to an accurate amplified
pulse magnitude then back to zero within a reasonable amount of time to make room for the next pulse.

3) For MCA applications, the repeatability of the PC chamber output pulse height for a given x-ray energy input is important since it controls the fundamental
energy resolution of the of the final result. I.e., begin by assuming a monoenergenic x-ray input to the PC. If all the output pulse heights from the PC are identical
then a perfect condition exists and we only need to amplify those PC pulses, and route them to an MCA. The result would be an ideal vertical line of zero FWHM,
i.e., perfect energy resolution. Of course, that happy world is only a fantasy degraded by noise and nonlinearities throughout the system,beginning with the PC,
all the way to the MCA itself.

4) Noise: since we are connecting the op amp as a charge amp, its transfer function will be measured in say, 2 nC/V (two nanocoulombs of input charge will result in an
output pulse with a peak magnitude of one volt). So 550 electrons, FWHM means that a charge equivalent to 550 electrons (= 550 x 1.6x10E-19 Coul per electron =
8.8 x 10E-8 nano coul = 0.4 nV uncertainty in the amplified pulse height) -pretty good. That translates to a 0.4 nV uncertainty in the left to right position on the MCA display
(low pulse heights on the left,high pulse heights on the right.) Too good to waste on a PC. The 0.5 keV noise spec must be at some particular amplification in order to
compare with this example.

5) PC capacitance: A higher terminal capacitance will result in more integration of the output pulse (squished down in peak value and spread out in time) This makes
accurate peak pulse height measurement more difficult, which results in more peak height uncertainty and that translates to a larger FWHM.

which brings me to burnout

Randall





----- Original Message -----
From: Soren <justinhuber@...>
To: [email protected]
Sent: Fri, 11 Dec 2020 17:07:54 -0800 (PST)
Subject: Re: [XRF] Proportional Counters and Preamps

Dear Randall,

Yes. You're right about the Hamamatsu - it is for a SiPIN. I have a vague memory that I saw some articles about the A-250 being deployed on satellites and I seem to remember the detectors being proportional counters - but I could be wrong. Looking at the data sheet for the A-250, it mentions proportional counters as a possible detector.

To a certain degree this is more of an academic exercise - to understand the underlying circuit theory of these "state of the art' preamps. I have a Amptek SiPIN on its way in the mail right now. But I figure that since I have the proportional counter, why not try to get it working as I think it will be more sensitive in the higher energy regions than the SiPIN.

It's a little hard to compare noise specs as the measures seem to be different. The Hamamatsu is listed as having a "550 Electron/FWHM noise characteristic." The A-150 instead provides a chart indicating that with certain front end FETs, such as the 2SK152 (of which I just received 5 in the mail), at 2pF capacitance the noise is about .5kev FWHM. I'm not sure if those are just different terms for the same thing. The patent claims to have measured "equivalent noise charge of less than 20 electrons r.m.s." But as I'm a bit new to this, I'm not sure if these measures are all the same.

Earlier you mentioned

Now to compare the op amp specs with the Ortec 142PC

There may also be other, higher performance alternatives.
With such a low capacitance in the Porp Chamber, stability, slew rate and
settling time (at a good accuracy level) will be important characteristics
if the goal
is energy discrimination, not just detection.

Comparing the gain bandwidth product of a couple options, we get;

* AD823:??3 dB bandwidth of 16 MHz, G = +1
* Amptek A-250:?300 MHz with 2N4416 FET

Having just learned what slew rate and gain bandwidth product are about 4 days ago, I'm not sure that I'm equipped (read: too lazy to do the math) to compare other characteristics.

But I know that the are much better op amps out there such as the AD829, which appears to be what Amptek is using in their PA-230 preamp. This has a 230V/?s slew rate and a much better settling time than the AD823. The AD829 and the AD823 are about the same price ≈$7 but you'd need two of the AD829s since its only a single op amp chip.

A quick DigiKey search yields some other options;

* LTC6253 - $7 - 280V/?s - 720MHz
* OPA2690 - $6 - 1800V/?s - 300MHz
* LMH6626 - $6.5 - 360V/?s - 1.3GHz
* THS3062 - $13.5 - 7000V/?S - 2.2GHz

I imagine that slew rate is more important as the count-rate goes up. And gain bandwidth product is more important and one needs to amplify smaller and smaller signals? Is that right or have I gotten my wires crossed?

Then there is the ADA4530-1 which has an input bias of 20fA. But that might be for different applications altogether.

I know with SiPIN low capacitance is a good thing (hence the use of high reverse breakdown voltage PINs). What are the issues with a low capacitance proportional counter?

All the best,
Soren