MCA Project for Scintillator

Go To Last Post
8 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi guys. I have spent hours researching scintillation and how multi channel analyzers work.

I basically want to read the height of pulses and sort them into bins amd count each pulse heights occurance. The pulses however are to fast for an adc on the AVR. I have though of other approaches that would hold the voltage around just long enough for an ADC such as a capacitor but I beleive there must be a better more accurate way than that. Any tips on where to look would be a great help. I don't want designs just ideas and guidance into learning a better approach.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To hold the value with a cap, you need a sample/hold, and that requires a trigger detector that determines when to close (and open) the sample switch. You might have problems with pulses that are too close together (for a single sample/hold). I am guessing that is what "multi-channel" is all about. Multiple (maybe 4, maybe more) S/H. The trigger is set up to activate the S/H switches sequentially. The amplitude measurements, however, are combined into a single set of statistics. That's my guess.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Jim,

That makes sense what your saying about a capcitor and something to hold the value temporarily. I was also considering some lol of switch like a FET across the cap so after the sample is held I can discharge it ready for the next one.

I regards to the samples that are close together a few missed samples here and there is ok as the purpose of the excercise is sample over a standard period. Pulse height is related to energy of the light entering a photo multiplier tube. And count these into virtual bins so after the six hours a spectrum can be plotted. But multichannel from what I have found that refer to the multiples of these virtual bins. The method you mentioned is what I beleive the multichannel scaler method. Sample a fixed intensity or channel over a set time then move on to the next until yoube cover the spectrum.

I have a look into ways to hold the signal temporarily and update once I have thought of the best way.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ecw23able wrote:
The pulses however are to fast for an adc on the AVR.
XMEGA AU have 2Msps ADC and an event system; each pulse would generate an event to one of a few or several XMEGA where the event is sunk by an XMEGA's ADC.

Would have sequencers and multiplexers (analog, digital).

Akin to the multiple ADC in some digital real-time oscilloscopes.

XMEGA AU have fairly fast analog comparators (30ns typical for propagation delay).

http://www.atmel.com/devices/ATXMEGA16A4U.aspx?tab=documents

http://www.atmel.com/Images/Atmel-8331-8-and-16-bit-AVR-Microcontroller-XMEGA-AU_Manual.pdf (page 71 for Figure 6-1. Event system overview and connected peripherals.)

 

"Dare to be naïve." - Buckminster Fuller

Last Edited: Thu. Jun 23, 2016 - 08:05 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I will look into this AVR it may be a good choice to use. It would certainly make it more precise which is a plus.

I have been thinking about the switch and hold method but can't quite figure out how to make the pulse automatically trigger the hold. Could one use the pulse itself as a trigger input so the pulse presents itself and latches its value into a hold then read and reset with micro. I am hunting suitable external components for a switch and hold circuit

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How many photo detector tubes are you using?

Multi-channel sometimes refers to multiple photo detector tubes to collect scatter.

 

That said, what is the emitter, and hence how fast do the signals typically appear?

Do you have a spec for a current device you are using to act as a baseline against which to design your new and improved device?

 

The Xmega128A1U for example, has a 2 MegaSample/Sec ADC, (12-bit resolution).

I've used one of the earlier Xmega128's, but not the new and improved versions with the built-in USB module.

That is a pretty fast sampling system, (all relative, of course...).

That said, the ADC is a complex module with lots of features, but many of the features are mutually exclusive.

You have to study the data sheet very carefully to see whether it makes sense to use single ended inputs or differential inputs, with which ones can you use the internal gain, what, exactly, is "ground level", etc.

It is worth getting a development board with the chip and testing both the sampling rate, and the ADC setup, before developing your own PCB, if you eventually go that route.

 

I would have a look at Wiki Precision Rectifier and scroll down to Peak Detector to see how to actually drive a cap to hold a peak voltage.

Likewise, just Google Peak Detector for more info, and schematics.

The point is this, even sampling at 2MS/Sec may be too slow to capture the peak of the photo detector output, especially if you are after amplitude info.

The op-amp based Peak and Hold will capture the peak for you, and then the ADC can take its time reading the peak, and then resetting it, (NFet to short out the cap).

 

With the right front end you could sample at 2 MS/Sec.

If the ADC reading is too low, below a given threshold, then no photon was detected during that last 0.5 uSec's.

If the ADC is above the threshold, quantitize it, and increment the appropriate bin, (e.g. 100-200, 201-300, 301-400... or whatever).

 

With this setup you use a Timer/Counter to fire an ISR and read the ADC every 0.5 uSec's, (2 MS/Sec).

 

The systematic error with this approach is that the Sample & Hold will always capture the higher of multiple pulses occurring within the sampling interval.

So you might miss some lower level samples, as they were masked by closely timed higher amplitude samples.

 

You haven't mentioned how closely spaced the samples that you wish to distinguish are occurring.

People often frown on multiple processor "solutions" to a problem, but obviously one could use two Xmegas, and have the secondary processor obtain a sample 0.25 uSec after the Master processor.  Each processor would have to gate its own sample and hold to its half of the 0.5 uSce sampling rate window.  One can carry this to the extreme, with additional processors, each watching their time slice of the overall 0.5 uSec window, essentially a time division multiplexing system.

 

I'd start with one Xmega see how it compares to your present measurements.

 

The next topic to address is data storage.

If you were to sample at 12-bits, say two bytes without any data packing, at 2MS/Sec, that is 86 GBytes of data over your six hour sampling period.  That is a lot if you needed to store the individual samples for measuring exponential decay, signal contamination, or other processing.  The USB link to a PC might be very handy in this case.

 

If you are just incrementing the bin counters then the data storage requirements are obviously much less.

 

Consider whether or not you need to periodically measure the photo detector's supply voltage or not during your sampling period.

If you have a "good" power supply, and you are operating in the middle of the tube's "plateau", then you don't need to bother with this.

 

Also consider if you need to measure a background measurement tube's data, or not.

This totally depends upon the measurement environment, and the variation in background counts that you expect to see over your 6 hour window.

If you need to accurately quantify the source emissions you might need to simultaneously measure the background counts to subtract them from your total counts.

(Storing the data on a PC for post processing starts to look better and better!)

 

It all depends upon exactly what you are trying to measure, and in what environment, and to what degree of statistical accuracy, as to whether or not any of this other stuff is needed.

 

When I was still very new to micros I made a 5 microcontroller project with four micros doing sequential ADC measurements and a fifth one as the Master controller and USB to PC driver.

I would do the project much differently now, (with one Xmega!), but that is what a learning curve is for.

 

Good luck with your project!

 

JC

 

 

Edit:  Opps, forgot the link.

 

 

 

 

 

 

 

 

Last Edited: Fri. Jun 24, 2016 - 02:48 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am intending to use one PMT with a NaI scintillation crystal in front, From my research signals can appear any where from a few nS apart to hundreds of nS apart, missing a signal isn't so much of a big deal as gamma spectroscopy is at its most accurate when averaged over a period of time.

 

The baseline readings will be taken in a lead shielded enclosure and subtracted from the end signal. 

 

I intend to use single ended inputs and only star with 512 bins for count events, This should give a respectable resolution, I plan to include a discriminator to eliminate weak or overly strong pulses.

 

In regards to how frequently the samples occur this is completely random as this varies with samples and radioactive sources. 

 

For data i intend to stream the raw data to a PC were software will interpret it and plot it. 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey guys I have hunted down a sample and hold ampifier of reasonably high speed from Analogue devices. The AD783. This coud be a good chip to build into the front end.