Low noise PMT conditionning

What sort of current do you expect from the PMT? What's the bandwidth requirement?

PMTs have a ton of gain and a lot of shot noise, so a really good amp may not be justified.

PMT's have a lot of umph, (current) I built a pmt pulse amp with an AD8001, I think it was your recommendation.

George H.

I think PMTs wear out eventually; they come with a limited supply of coulombs or something. That's not so much a problem for something like a scintillator that does narrow pulses.

AD8001 is a nice little amp, but current noise is high. Fine for 50 ohm systems.

For more CW stuff, I guess one would want to limit the toob gain and use as high a load transimpedance as practical. That's easy if the bandwidth requirement is low. A fet opamp would be good there.

As far as I know there are two fairly different ways to use a PMT: counting photons or measuring the average output current. Which are you intending to do? I still won't be able to give much advice but it might help someone else to help you.

I recall reading somewhere (perhaps a Hamamatsu app note) that in some situations it is better not to use an amplifier as it may not help and may worsen the performance of the system. So I guess another question is what are you trying to drive with the output of the PMT? - are you feeding an ADC? In this case, what is the input voltage range and current of the ADC, and what is the sample rate, and do you want to capture the waveform of individual pulses, to extract the peak, or measure the DC average?

Chris

few microamps and 80Hz (Photons flow is beeing chopped in the reacting cham ber)

Remember JFET inputs ... and low intrinsic noise ... is that your answer " so a really good amp may not be justified "

We are measuring photo current (not counting the photons)

At the moment we are feeding the photo current (modulated by a chopper) to the ADS1251. We are able to detect NOx concentration as low as 0.4ppb

The PMT is modelised by a current generator --> Cause of the JFET input op amp for the I/V converter.

If you are operating in the current regime rather than photon counting wouldn't a cooled CCD detector be worth considering?

On a sunny day (Mon, 02 Nov 2015 20:41:53 +0100) it happened Habib Bouaziz-Viallet wrote in :

Well, the issue of JFETs .. long time ago (sixties) when interfacing Vidicon image tubes, I accidently found that (same high output impedance) in case of current source I had better results (or just as good as the best) driving directly into a normal transistor base, THERE the impedance is very low (usually), and you have beta (current gain) as your friend. This results in decent bandwidth (was a requirement) and good sensitivity.

WTF go to a high impedance input, then make it low impedance with feedback, or lose your bandwidth? Anyways this is what I did:

This avi movie shows (here used to detect decay pulses) it in action:

One trick I used to speed things up is discharge the PMT (cable circuit) after every pulse. As you want current only, why bother. And YES beta varies, but a PMT varies a lot more not even counting external influences. And photons do not exist apart from the mathematic definition, Einstein should never have gotten a Nobel for that but so did some for a peace price, all politics.

This thing is still sitting here on the table and working fine after all them years (4).

By no means am I right and others[tm] may argue against this, but hey, it gets the work done.

;-)

or

rd,

Any reference ?

Best regards,

Something like this CCD spectrometer device (although this one isn't actively cooled)

Or even some of the CCD kit sold for amateur astronomy at lower prices.

are

the

for

me

uard,

le

you

ght

May be useful to characterize our reacting chambers for chemiluminescence, fluorescence, NDIR, ...etc. We'll see later, thanks.

You get about 1000 anode coulomb/cm**2 out of a normal tube, vs. only about 1 coulomb from a microchannel plate. The difference is that MCPs have a huge amount of internal area and can't be baked out very hot, so they're dirty.

A useful rule of thumb is that you're in the shot noise limit when your photocurrent drops more than 2kT/e (51 mV at room temperature) across its load resistor. (The amp has to be sufficiently quiet, obviously.) With a tube gain M, the amplified shot noise goes as M, so you can drop the load resistor by a factor of M**2, i.e. you're in the shot noise when the drop is at least 50mV / M. That gets easy pretty fast.

(I'm working on a high-SFDR PMT system for a customer today.)

Cheers

Phil Hobbs

• I have asked for this before, and guess i need to repeat: PLEASE write black on white,not on grey. At least (this time) the lighting was reasonably uniform, so that a simple conversion to black and white with selected transition level worked well.

• Again,poor contrast; pulses almost invisible. For starters, turn off the lights.

Have a look at an article in Linear Technology Journal of Analog Innovation (July 2015) by Glen Brisebois entitled ' Op Amp combines femtoamp bias cur rent with 4GHz gain bandwidth product, shines new light on photonics applic ations', dealing with the LTC6268 device. Search their web at

Scott, TIA Systems.