EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: lskop on July 09, 2015, 05:50:46 am
-
Hello all!
I'm tasked with building a gamma radiation detector. One component, which has become troublesome is designing and building a very fast sample and hold circuit. The input signal has a maximum value of around 1v, and is a short ~50ns pulse that may occur randomly anywhere between 10 to 100,000 times a second.
Basically I'm trying to design a circuit that will detect when the peak has occurred (to trigger an ADC), and of course a circuit to hold this peak value, so it can be sampled and processed (Building an energy spectrum).
If anybody could help in anyway, provide reference circuits our general references for me to read that'd be great.
Many, many thanks in advanced.
Kindest regards,
Leon.
-
I have no ideas myself however I would suggest try posting on sci.electronics.design
There are two guys in particular on that group that work in that area;
George Herald - develops electronics for physics
Dimiter Popoff - runs a company developing instruments for Nuclear science.
I don't know either of these guys, I've just seen them post on the group and they appear quite knowledgeable and helpful to others.
They might be good for some ideas.
EDIT - If you dont have one already it may be worth installing a usenet reader (Thunderbird works ok) as some usenet regulars filter and ignore people who post to usenet from GoogleGroups.
-
Sample and hold circuits of this speed and greater are readily available - inside fast ADC chips. In the current state of the art it might be a better approach to digitise continuously, say around the 100 to 200 Msps mark, and peak detect using an FPGA. Unless there are specific constraints to your application?
Sounds like an interesting project!
-
Have a look at the OPA615 (http://www.ti.com/product/OPA615/technicaldocuments). Peak detector and Sample/hold examples are shown in the datasheet.
-
An entirely different approach could be a digital counter (to count the pulses) and an integrator to accumulate total voltage-time integral. If the pulse width is near-constant you can easily derive pulse height from the two numbers.
-
AD8036 can become a 20MHz peak detector by wave of a precision full wave rectifier:
https://youtu.be/fs2MfTW4o_Q?t=6m37s
Not sure if that's suitable but thought I'd mention it.
-
What kind of detector is it you are using? If you're using a NaI(Tl) on a PMT then 20 MHz may be a bit overkill.
Do you require MCA features or are you more interesting in getting rid of the rubbish at the lower and upper end of the spectrum? There could be easier ways of doing this then using a peak detector that allow a bit more flexibility in the longer term - perhaps if you start looking at alternative scintillators and require alpha/beta discrimination
-
I have no ideas myself however I would suggest try posting on sci.electronics.design
There are two guys in particular on that group that work in that area;
George Herald - develops electronics for physics
Dimiter Popoff - runs a company developing instruments for Nuclear science.
I don't know either of these guys, I've just seen them post on the group and they appear quite knowledgeable and helpful to others.
They might be good for some ideas.
EDIT - If you dont have one already it may be worth installing a usenet reader (Thunderbird works ok) as some usenet regulars filter and ignore people who post to usenet from GoogleGroups.
Thank you, I will probably post there and hope they jump in! Also thanks for the tip with the usenet reader.
Sample and hold circuits of this speed and greater are readily available - inside fast ADC chips. In the current state of the art it might be a better approach to digitise continuously, say around the 100 to 200 Msps mark, and peak detect using an FPGA. Unless there are specific constraints to your application?
Sounds like an interesting project!
Definitely something I've considered. I started out using an awesome digitiser, and was making progress. However, it's very large and expensive and I need a smaller, more portable and affordable option. I'm going to look into modern fast ADC chips, I cannot imagine they'd be as expensive as the digitiser I was using, nor as large! Could even do peak detection in software after that, depending on capabilities.
Have a look at the OPA615 (http://www.ti.com/product/OPA615/technicaldocuments). Peak detector and Sample/hold examples are shown in the datasheet.
Thank you mikerj, I've already ordered a free sample of this! Looking forward to test it out. However, due to the nature of the signals. It's hard for me to reliably sample the system with my own signal, I'm looking for a peak detect-sample and hold circuit, basically. Peak detection to order a hold signal. Purely due to the random like nature of gamma radiation.
An entirely different approach could be a digital counter (to count the pulses) and an integrator to accumulate total voltage-time integral. If the pulse width is near-constant you can easily derive pulse height from the two numbers.
Thank you dom0, the pulse width has some fluctuation, but this may be something to look into. Thank you for the tip!
AD8036 can become a 20MHz peak detector by wave of a precision full wave rectifier:
https://youtu.be/fs2MfTW4o_Q?t=6m37s (https://youtu.be/fs2MfTW4o_Q?t=6m37s)
Not sure if that's suitable but thought I'd mention it.
Thanks EEVblog! I will definitely look into it, assuming it's you Dave, in your video you mention smoothing out the output from the AD8036? Did you achieve this with a cap? Or am I missing something basic!
What kind of detector is it you are using? If you're using a NaI(Tl) on a PMT then 20 MHz may be a bit overkill.
Do you require MCA features or are you more interesting in getting rid of the rubbish at the lower and upper end of the spectrum? There could be easier ways of doing this then using a peak detector that allow a bit more flexibility in the longer term - perhaps if you start looking at alternative scintillators and require alpha/beta discrimination
Thanks penfold. The detector is using a NaI(Ti) scintillator with a PMT, however and you're correct it seems a little over kill but I'm looking somewhere within that range. MCA features is something I'm hoping for, however removing some rubish in the lower end may not be a bad idea.
-
Then, what kind of maximum rate of detection are you expecting? NaI can be quite nasty for after pulses and general spectral rubbish so taking the integral would upset your energy spectrum (worse than NaI already does!), you are probably right to be wanting a peak detector. You can use a bit a pulse shaping to get the bandwidth of the pulse down to 1MHz whilst preserving much of the energy spectrum.
-
AD8036 can become a 20MHz peak detector by wave of a precision full wave rectifier:
https://youtu.be/fs2MfTW4o_Q?t=6m37s (https://youtu.be/fs2MfTW4o_Q?t=6m37s)
Not sure if that's suitable but thought I'd mention it.
Thanks EEVblog! I will definitely look into it, assuming it's you Dave, in your video you mention smoothing out the output from the AD8036? Did you achieve this with a cap? Or am I missing something basic!
Yes it's me. Administrator next to my name gives it away ;D
IIRC I had a sample cap on the output that was discharged after sampling.
-
Then, what kind of maximum rate of detection are you expecting? NaI can be quite nasty for after pulses and general spectral rubbish so taking the integral would upset your energy spectrum (worse than NaI already does!), you are probably right to be wanting a peak detector. You can use a bit a pulse shaping to get the bandwidth of the pulse down to 1MHz whilst preserving much of the energy spectrum.
Thank's again Penfold,
I've now swapped to an LSO Scintillator material. I am getting pulses that look like the picture included, sorry for the figure but the x-axis is time in seconds, and the y-axis is volts with no external gain(of course excluding the PMT itself)/pre-amp. For a Na22 sample. Now the current testing set up is not absolutely fantastic, which is why you can see a pretty substantial amount of noise in the signal. But it's just about understanding it at the moment.
Can you suggest any particular pulse shaping method/(s)?
Yeah the integral may or may not work.I'm still trying to work it out but the noise may be fairly symmetric around the x axis, and may cancel itself over the integral. Of course, that's still up for analysis.
AD8036 can become a 20MHz peak detector by wave of a precision full wave rectifier:
https://youtu.be/fs2MfTW4o_Q?t=6m37s (https://youtu.be/fs2MfTW4o_Q?t=6m37s)
Not sure if that's suitable but thought I'd mention it.
Thanks EEVblog! I will definitely look into it, assuming it's you Dave, in your video you mention smoothing out the output from the AD8036? Did you achieve this with a cap? Or am I missing something basic!
Yes it's me. Administrator next to my name gives it away ;D
IIRC I had a sample cap on the output that was discharged after sampling.
Cheers Dave! Fellow Aussie here. Makes sense, I am a subscriber of yours, and I learn a lot from your Youtube videos so I thank you. I'm impressed that you're actually here replying and talking with us all on the forums, means a lot.
-
Even if rather old, maybe those publications can be of some interest to you:
http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf)
http://www-pub.iaea.org/MTCD/publications/PDF/te_426_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf)
http://www-pub.iaea.org/MTCD/publications/PDF/te_530_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_530_web.pdf)
Using an FPGA:
http://www.slac.stanford.edu/econf/C0908211/pdf/028.pdf (http://www.slac.stanford.edu/econf/C0908211/pdf/028.pdf)
-
Even if rather old, maybe those publications can be of some interest to you:
http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf)
http://www-pub.iaea.org/MTCD/publications/PDF/te_426_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_363_web.pdf)
http://www-pub.iaea.org/MTCD/publications/PDF/te_530_web.pdf (http://www-pub.iaea.org/MTCD/publications/PDF/te_530_web.pdf)
Using an FPGA:
http://www.slac.stanford.edu/econf/C0908211/pdf/028.pdf (http://www.slac.stanford.edu/econf/C0908211/pdf/028.pdf)
Just briefly browsing through them and yes, they seem to be very useful. Thank you very much :) I'll get stuck into it tomorrow!
-
I can also vividly suggest to get the following handbooks:
Knoll: Radiation Detection and Measurement, Fourth Edition (http://eu.wiley.com/WileyCDA/WileyTitle/productCd-EHEP001606.html) (at least chapters 16-18, but the whole handbook is treated as a "Bible")
Leo: Techniques for Nuclear and Particle Physics Experiments - A How-to Approach (http://www.springer.com/us/book/9783540572800)
(they can be found at cheaper prices if you buy them from Asian/Indian sellers)
If I have some time I will search in my library for more documents/papers regarding your needs (but do not trust me too much: in this days I'm very busy!)
When I will found some more documents, I will add them at the end of this post...
http://www-physics.lbl.gov/~spieler/physics_198_notes/ (http://www-physics.lbl.gov/~spieler/physics_198_notes/)
http://www-physics.lbl.gov/~spieler/physics_198_notes_1999/PDF/Jan-19.pdf (http://www-physics.lbl.gov/~spieler/physics_198_notes_1999/PDF/Jan-19.pdf)
Physics and Engineering of Radiation Detection, 2nd Edition (http://store.elsevier.com/Physics-and-Engineering-of-Radiation-Detection/Syed-Ahmed/isbn-9780128013632/)
Experimental Techniques in Nuclear and Particle Physics (http://www.springer.com/us/book/9783642008283)
http://www.desy.de/~garutti/LECTURES/ParticleDetectorSS12/spieler.pdf (http://www.desy.de/~garutti/LECTURES/ParticleDetectorSS12/spieler.pdf)
-
I must have forgotten to reply Eliocor, thank you for your help. I'm still not quite at a good place with this circuit.
I've built pulse shaping circuitry now, and I have got this current situation.
Uni-Polar: The peaks last around 10us with approx 5us peaking time. The pulses are separated by at least 31us.
Bi-Polar: The peaks also last around 10us, with approx 1-2us peaking time (depending on the tuning). The pulses are separated by at least 31us.
I however, still need to design a sample and hold circuit with peak detection to feed an ADC.
I believe, there may be some issues using just a threshold detection (setting a trigger level), to interrupt an MCU (say on an Arduino), to use it's inbuilt sample and hold to maintain the peak whilst it digitises it. Due to timing etc, or if I built say a peak detection circuit, without a hold, I'll be at the mercy of the delay between the signal interrupting the MCU, and the ADC holding the value.
If anybody has any other suggestions, or systems they've used in the past this would be awesome. I'm running out of time, this is the last section of my capstone project, nor is it the most important. However, it is critical to get my system fully independent and running by itself.
Thank you for any assistance in advanced.