Circuit for Photomultiplier Tubes to detect Terrestrial Gamma-ray Flashes

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
Hello,
I am currently working on a design for a system that will convert analog currents from a photomultiplier tube to digital voltages that will be processed in an FPGA. The signals I want to detect are Terrestrial Gamma-ray Flashes (TGFs). What I have done in order to convert analog currents to digital voltages is to use a transimpedance amplifier to convert the currents to voltages. It is also at the transimpedance amplifier I placed all the gain in order to have the least noise, and it is easy to change the gain if necessary. After making the signals to voltages, I have placed a 4th order Butterworth filter; A 2nd order KRC filter followed by a 2nd order MFB filter that is placed externally on the ADC driver (ADC driver is used as a single-ended to differential converter). After the 4th order filter, the signals are sent to the ADC.
The components that I have used are the ADS5404 (500MSPS ADC), OPA855 (for the transimpedance amplifier), THS4303 (for the KRC filter) and THS4505 (for the ADC driver).

What I am having problems with is that I don't know the signal strength from the PMTs. It says on the OPA855 that it can withstand up to +-10mA of continuous currents, should I have a current limiter in front so that it does not go above this maximum? In that case, what can I use? I also require a voltage limiter so that the differential output signal on the ADC driver does not go above 1-1.25 Vpp. Any suggestions as to what I can do here? I am thinking of using a diode clamp to limit the signal to +-0.6V at the single-ended input of the ADC driver.

If there is anything that you think should be changed otherwise, do not hesitate to say! I am a Bachelor student in electronics, so I am quite inexperienced, but I want to learn as much as possible!


Hope you can help,
Best regards,
Sindre
 

pmd34

Joined Feb 22, 2014
503
Hi Sindre, For PM tube counting applications you should be looking for individual pulses from the PM tube rather than an continuous current signal.
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
Do you have a datasheet or part number for the PMT?
Sadly no, we have not been given any information about the PMTs... We've asked the physics department, but the answer is rather vague and doesn't tell anything about the PMT that will be used.

Hi Sindre, For PM tube counting applications you should be looking for individual pulses from the PM tube rather than an continuous current signal.
The problem is that the OPA855 only tells how much it can handle continuous current signals. Will the PMTs give out currents that far exceeds the limitation for the OPA855 in spontaneous currents? In that case, is there a way to limit the current so the component doesn't get fried? Or is there a better way to design the current to voltage stage for a PMT?
 

Yaakov

Joined Jan 27, 2019
1,612
Sadly no, we have not been given any information about the PMTs... We've asked the physics department, but the answer is rather vague and doesn't tell anything about the PMT that will be used.
Frankly, designing something for an unknown and critical part is... not very good. You might want to insist that you need critical design parameters and can’t consider the PMT an interchangeable or theoretical component.

By the way, being a physics department, make sure they don’t give you an order of magnitude estimate of the part number. You’ll need full precision. :)
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
Frankly, designing something for an unknown and critical part is... not very good. You might want to insist that you need critical design parameters and can’t consider the PMT an interchangeable or theoretical component.

By the way, being a physics department, make sure they don’t give you an order of magnitude estimate of the part number. You’ll need full precision. :)
Just asked the Senior Engineer that is helping me with the project, and he has no information about the PMT that will be used. The project I'm working on is a continuation of another project, where certain components had to be upgraded (ADC with a greater sampling rate, larger FPGA, etc.). The engineer helping me out was the one designing the previous project's circuit board. He has information on the PMT previously used (H10720/H10721 series), which might hopefully be the one used in this project.
Otherwise, I have no information on the PMT and I probably won't be receiving any, so the best I can do is to design the system with the previous PMT in mind.
 

Sensacell

Joined Jun 19, 2012
2,567
Sounds like they are setting you up to fail.

This is where you buck-up and insist that things get done properly, rather than accepting a fools errand.
 

Yaakov

Joined Jan 27, 2019
1,612
Just asked the Senior Engineer that is helping me with the project, and he has no information about the PMT that will be used. The project I'm working on is a continuation of another project, where certain components had to be upgraded (ADC with a greater sampling rate, larger FPGA, etc.). The engineer helping me out was the one designing the previous project's circuit board. He has information on the PMT previously used (H10720/H10721 series), which might hopefully be the one used in this project.
Otherwise, I have no information on the PMT and I probably won't be receiving any, so the best I can do is to design the system with the previous PMT in mind.
The data sheet for those devices seems to say the output is in the μA range.
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
Sounds like they are setting you up to fail.

This is where you buck-up and insist that things get done properly, rather than accepting a fools errand.
You are right. I have asked the project manager multiple times but he hasn't received any information from the physics department. So I have contacted the physicists in charge of the project now, and will hopefully get some information from him.

The data sheet for those devices seems to say the output is in the μA range.
Yes, noticed that the current from this PMT wouldn't be of any problem for the OPA855. But I have contacted the lead physicist to get a proper answer, just in case this isn't the PMT that will be used.
I really appreciate all the help, thank you so much!
 

MrChips

Joined Oct 2, 2009
21,047
A scintillator (such as NaI and certain types of plastics) are commonly used to detect high energy X-rays and gamma-rays using PMTs.

The pulse output from the PMT is high enough to detect without the need for preamps.
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
Hello,

Not knowing the type of PMT you are using, we can only give general information.
Here are some PDF's about PMT's.

Bertus
Thanks for the files! Will use the documentation wisely! :)
I am pretty certain now that the H10720/H10721 is the PMT that will be used, just waiting for the physics department to verify this.

A scintillator (such as NaI and certain types of plastics) are commonly used to detect high energy X-rays and gamma-rays using PMTs.

The pulse output from the PMT is high enough to detect without the need for preamps.
The scintillators are LYSO crystals, and it seems like we are using the same detectors as is used in the ASIM BGO project (the predecessor to this project). Even though the pulses are high enough to be detected without a preamp, wouldn't this limit the usefulness of the 12-bit resolution of the ADC? And both the Butterworth filter and the ADC requires voltages, so wouldn't it be quite simple to just place a transimpedance amplifier first that converts the currents to voltages as well as amplifies the signals?
 

MrChips

Joined Oct 2, 2009
21,047
The purpose is to create a stopband around 200 to 250 MHz in order to support Nyquist's theorem. So since the ADC is operating at 500MSPS, we need a stopband that is less than or equal to 250 MHz.
Don't believe everything you read or hear. You don't need anti-aliasing filter.
 

MrChips

Joined Oct 2, 2009
21,047
Is it because the detectors won't give out continuous signals, but shorter bursts?
It is because your PMT response time is not fast enough to give you 250MHz signals, unless you are using super fast PMT.

This is why you need to get the spec sheet of your PMT.

If your pulse widths are shorter than 5ns, you are going to miss your pulses, filter or no filter.
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
It is because your PMT response time is not fast enough to give you 250MHz signals, unless you are using super fast PMT.

This is why you need to get the spec sheet of your PMT.

If your pulse widths are shorter than 5ns, you are going to miss your pulses, filter or no filter.
Which timing variable in the spec sheet is it that tells the pulse width? In the spec sheet for H10721, the only variable remotely close to the ns area is the rise time, which I read in a document is approximately equal to the response pulse width.

So the filter won't do any good in the circuit at all? Cause when I got the project for my bachelor thesis, I was told one of the design requirements was a filter... But then again, as you said: "don't believe everything you read or hear". I'd rather be taught the right things and make a good design than follow what they think is required.
 

MrChips

Joined Oct 2, 2009
21,047
The purpose of anti-aliasing filter is to remove frequencies above the Nyquist limit when performing frequency analysis. You are not doing frequency analysis. You are doing pulse shape analysis, pulse counting and pulse height analysis.

A LP-filter will simply smooth out the signal pulses.

If your input is a delta-function you will end up with a pulse-width approx. equal to the rise time + fall time. In other words, a delta-function gets convolved with the response function of the system.

The pulse width will be determined by the decay time in the scintillator and the collection time in the PMT.

I am doing exactly what you are doing. I am detecting gamma rays using a PMT with NaI scintillator, digitizing straight into an ADC with no preamp. Then I collect energy spectrum after processing the pulse height.
 

Thread Starter

sindre jacobsen

Joined Apr 1, 2019
40
The purpose of anti-aliasing filter is to remove frequencies above the Nyquist limit when performing frequency analysis. You are not doing frequency analysis. You are doing pulse shape analysis, pulse counting and pulse height analysis.

A LP-filter will simply smooth out the signal pulses.

If your input is a delta-function you will end up with a pulse-width approx. equal to the rise time + fall time. In other words, a delta-function gets convolved with the response function of the system.

The pulse width will be determined by the decay time in the scintillator and the collection time in the PMT.

I am doing exactly what you are doing. I am detecting gamma rays using a PMT with NaI scintillator, digitizing straight into an ADC with no preamp. Then I collect energy spectrum after processing the pulse height.
I see, so the only thing required before the ADC is to convert the currents to voltages and amplify the signals? Do you think a basic transimpedance amplifier using an OPA855 is good enough for this, or should I rather go for something like a CSP?
 
Top