PMT pulse amp

On a sunny day (Fri, 29 Oct 2010 07:49:58 -0700 (PDT)) it happened George Herold wrote in :

Jus tan idea. If you have 1mV in 50 Ohm, then you have a 20 uA pulse from the PMT How about sticking that right into the base of a NPN? Zi of the NPN is low, That way you use beta to multiply that current, use a low collector resistor, So for a beta of 100 and a Rc of 50 Ohm, you would get 2 mA pulses in 50 Ohm,

100 mV. no opamp needed.

Yep. I did that with a vidicon pre amp stage, worked perfectly.

ns

it

usually you should just pick the Rf recommended in datasheet and set gain with the other resistor.

too low and Rf goes toward unstable, too high lowers BW

I believe you can:

-Lasse

ons

tor,

Ohm,

ext -

Thanks Jan, I'm not much of a transistor level design guy. (Mostly I stick them on the output when I need more poop.) If I'm getting a current gain of 100 at a frequency of ~100MHz. (~10nS) What transistor would you use? How does the current gain change at high frequency?

George H.

ions

t

bit

y

Thanks Lasse, I think I mis-spoke when I said I hadn't used a CFA before. I've been using the CLC450 for years. (I just never knew.)

George H.

PMTs have a habit of blowing out amps. Put a back-to-back schottky pair in the front end.

You don't need a TIA, as PMTs have lots of signal and fast TIAs are a PITA.

If you need speed, an inductor in series with R1 can help a little.

Look at THS4303. Fixed g=10, and the tricky inverting node is handled internally for you.

John

On a sunny day (Fri, 29 Oct 2010 09:06:55 -0700 (PDT)) it happened George Herold wrote in :

Well, I would look at what I had in the box :-) In the old days it would likely a BFY90 or similar. Maybe today a BFR96. As for the gain - changes - it depends on what you want to do. If you need to measure amplitude very precisely then perhaps some feedback system could be applied. If it is just to detect the pulse, it would not be so critical. In my case with the vidicon I could control system sensitivity by changing the target voltage. It all depends. You will probably need some protection circuits too, PMTs can make huge spikes.

ons

bit

Excellent... schottkies added!

Yes! I like it simple.

Oh, add a bit more gain at high frequency. But I really don't care about speed. It wouldn't matter if the pulse got streched out to

100nS. We're replacing a current preamp sold by Amptek (A111) that was used in the original design.

That's a screamer. I'm a bit afraid of having more frequency response than I need. (With no access to the inverting node, can I still roll the gain off?)

Thanks, George H.

k system could be applied.

g the target voltage.

pikes.

I'll look at the BFR96. We are just pulse counting so I don't care if the pulse becomes a bit distorted.

George H.

Gain of 100 at 100MHz with a BFY90? Err ... that's like trying to win the Rallye Monte Carlo on a moped :-)

ft is 5GHz. You need north of 10Ghz for this job or use two in a row. If you never did transistor stuff I would not recommend doing this unless you know an RF guy in the neighborhood.

My favorite amp is the THS4021, not a CFA, very behaved. Heed Johns advice about protection but make sure the diodes never come on unless needed.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

No stock at Digikey? Gasp!--what have you done with the real Joerg?!

-- Cheers, James Arthur

a BFR96. As for the

Thanks Joerg. I wasn't sure if the same sort of Gain * BW =3D~ ft, equations could be used for transistors. I would much rather stick with opamps though! (I'm afraid a 5GHz transistor will start singing above 1GHz, and I'll never know it...)

Excellent, I'll add a few to my order. (Hey these babies aren't cheap! ~$10 for onesies)

George H.

You could get 10x gain at 100 MHz from a $2.50 CFA. Voltage-mode opamps tend to have low slwe rates, too, so be careful there.

John

Or a 99 cent MMIC.

John

Huh?

Over 2000. How many do you want? Mouser has some as well. If you need oodles of them order at Arrow. 6 weeks leadtimes, which is next to nothing these days.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

In George's case probably even a 40 cent MMIC. But MMIC are in essence glorified RF transistors. You pay a 200% premium for having the bias resistors in there, and if not careful they can sing the blues up there in the gigeehoitzes. "Hey, why do all the bars on my cell go away when you turn that there thang on?"

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

.

If you search for "THS4021" all results save one are listed as "Not stocked"--that's what I was referring to. So, they have them, but no guarantees.

You must admit, I gave you a scare :-).

-- Cheers, James Arthur

Well, I only look at the qty on hand, it's all that matters at Digikey. Larger qties are bought at mainstream distributors anyhow.

:-)

I wasn't too concerned because the client where I designed it in never called me about it. And they would have if they couldn't get it. 6 weeks leadtime for reels ain't bad, I have seen north of 40 weeks.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

I've recently seen >150. ...and then there is Maxim, where 150 weeks is good.

Most of the InGaAs darlington mmics are claimed to be unconditionally stable. I've used lots of them as wideband pulse amps, and so far they all seem to work fine. They are about as much GBW per buck as you can get.

I have a bunch of data on using the ERA5 as a pulse amp, around here somewhere...

John