Pulse Shaping

[guest]

I have talked in earlier posts about using a charge sensitive amplifer to stretch the output of a PMT or other detector to make life easier for the following discriminator or SCA. There is a way to take this even further by using cascaded RC and CR circuits to spread out the pulse. The rationale is this most preamps and CSAs put out an exponential pulse with a fast attack and slow decay time. Not much time is spent at the upper voltage levels where one would most like to set a discriminator. Putting the pulse through a RC integrator and buffer spreads it out and makes it easier for the discriminator. Spectroscopy amps and MCAs often cascade several of these stages, so that the resulting wavefom is well-nigh gaussian in shape. This is discussed at length in Helmuth Spieler's notes on detector electronics (see the old intranets forum for the URL) and also in the latest edition of Knoll's book on radiation detection.
Practical applications? At present, my discriminator circuit uses a LM306 comparator, which is fast (60nsec), but power hungry. It also has a high input bias current, is limited to +/- 12V rails, and is hard to find to boot. Using pulse shaping, I can settle for an LM311, which is slower (200nsec), but can run off +/-15V, has lower input bias current, and is much easier to find. I won't use these techniques on my current model discriminator (I just finished taking the waveforms tonight), but I will certainly apply them to the next model. Of course, you can expect some circuits somewhere down the line...

[Richard Hester]

In the files section is a simulation showing the effects of RC shaping on the average preamp output waveform. This will stay up a week or so, vanish, and then perhaps reappear as part of a tested circuit.

[DaveC]

Just a thought or two to add to the discussion....
To measure the peak value of a signal, why not simply route it through a fast diode in series with a small capacitor to a good HF FET (JFet is probably a better choice than a MOSFET) with or without a few extra pF betwen the gate and ground. With the FET connected as a Source Follower and you will have a good low impedance driver whose decay time will be about as long as you want, given the high gate resistance. Put a few meg pot for gate to source resistance adjustment and I think you will have a simple "pulse stretcher". If you want to go the opamp route, check out the Burr-Brown data.. they have a few very fast devices. More pricey than LM3xx ICs but still affordable.

To make an easy discriminator, add an adjustable DC voltage source to the anode side of the input diode ahead of the input capacitor. Only those voltages that exceed the discriminator bias plus the junction voltage will forward bias the input diode and get through. Depending on the source resistance you select, you should be able to have the normal forward diode drop taken care of by the JFet's quiescent current.

Just some more grist for the circuit design mill.

Dave Cooper