I am working on a universal charge sensitive amplifier/pulse stretcher to use with either plastic or ZnS type scintillators. I'll post it and the companion discriminator circuit when all is said and done. However, using a charge storage capacitor appropritate for short pulse scintillators caused a problem that is driving me buggy. I'll digress a bit to explain what is going on...
A charge sensitive amplifier is nothing more than a current input integrator. This is nothing more or less than an inverting amplifier with a capacitor connected from output to input. A current pulse fed into the input gets converted to an output voltage proportional to the total amount of charge contained in the pulse. This is useful for two reasons: the integrator strains out a certain amount of noise and garbage, and it also stretches the pulse so that amplitude discrimination is much easier.
Commonly, a large value resistor is connected across the integrator capacitor and resets it to zero between pulses. This resistor also determines the "tail time" of the output pulse. One could assume that the output fall time constant is simply the RC time constant of the resistor and capacitor in question, but it ain't so in practice. One can (and does) get a fall time that can be anywhere from 2 times shorter than the expected value to 50 times shorter. I've been tearing out my hair trying to figure out the source(s) of the problem. I also found out that I wasn't alone in having this problem. I opened up an Ortec 109 preamp recently obtained through Ebay. The output fall time of this preamp is specified as 50 usec. The preamp has a 1pF charge storage cap, so that if everything was on the up and up, you would expect a reset resistor of around 50 Meg. Instead I counted no less than five
47 Meg resistors series connected around the capacitor. Hmmm....
So far, I've found out a couple of things: 1) leakage currents are important. I will probably end up mounting the input fet gate lead and its associated components on some push-in teflon standoffs, then hose the whole mess off with alcohol. 2) It helps if the input test pulse is derived from a real current source that goes essentially open circuit once the pulse is done, just like a PMT or silicon detector. The moral of the story is that simple circuits ain't necessarily simple....
Fall Time Deficit Disorder
-
Richard Hull
- Moderator
- Posts: 15037
- Joined: Fri Jun 15, 2001 9:44 am
- Real name: Richard Hull
Re: Fall Time Deficit Disorder
Right on there man..... I have studied the hell outta' my collection of about 10 charge sensitive amps and for a simple input stage, they sure cut and paste as needed to get at functionality. It is fascinating to see bootstrap engineering do its inevitable thing.
I have worked with PIPS detectors and Surface Barrier units for a couple of years and am glad I have had all these nice finished charge amps on hand, for they work as advertised. They make great front ends for BF3 detectors, too.
That little input FET has the full 3500 volts DC bias on its gate lead but for the sake of the 5kv .01uf disc coupling cap!
Ortec is famous for the warning sticker at the input BNC.
"Do not short the input connector!" "Wait at least 60 seconds before plugging or unpluging devices into input jack."
One wrong move on the part of the user and POOF....the gate/source/drain connections are eviscerated as the coupling cap dumps all its 3500 volt potential and seemingly limitless energy into the micron sized delicate FET die. Luckily, I have been rather paranoid and wait 5 minutes before connecting and disconnecting anything.
While Ortec is a favorite of mine, (NIM), and I do have about 4 Ortec charge amps; I really like the quality of the work done by Princeton Gamma-Tech in their charge amps.
I look forward to your "hacker amp" once completed. It may be tough for the average electronic tyro to get some of the higher value resistors that haunt the front ends of the these things, especially if a bias input is needed.
Richard Hull
I have worked with PIPS detectors and Surface Barrier units for a couple of years and am glad I have had all these nice finished charge amps on hand, for they work as advertised. They make great front ends for BF3 detectors, too.
That little input FET has the full 3500 volts DC bias on its gate lead but for the sake of the 5kv .01uf disc coupling cap!
Ortec is famous for the warning sticker at the input BNC.
"Do not short the input connector!" "Wait at least 60 seconds before plugging or unpluging devices into input jack."
One wrong move on the part of the user and POOF....the gate/source/drain connections are eviscerated as the coupling cap dumps all its 3500 volt potential and seemingly limitless energy into the micron sized delicate FET die. Luckily, I have been rather paranoid and wait 5 minutes before connecting and disconnecting anything.
While Ortec is a favorite of mine, (NIM), and I do have about 4 Ortec charge amps; I really like the quality of the work done by Princeton Gamma-Tech in their charge amps.
I look forward to your "hacker amp" once completed. It may be tough for the average electronic tyro to get some of the higher value resistors that haunt the front ends of the these things, especially if a bias input is needed.
Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
-
guest
Re: Fall Time Deficit Disorder
Funny thing about those high value resistors - right after reading Richard Hull's previous post in this thread, I popped on to Ebay, and managed to nail down a whole mess of 22 Meg metal glaze resistors. There were also some 51Meg, but I got outbid on those. The 22 Meg guys will serve well for time constant setting in my charge sensitive amp prototype.
-
Richard Hull
- Moderator
- Posts: 15037
- Joined: Fri Jun 15, 2001 9:44 am
- Real name: Richard Hull
Re: Fall Time Deficit Disorder
I guess everyone noted the high value resistors I have in my trading post section. I have a number of values. I work extensively with ion chambers and such. The lowest value I have for sale is 100 megohms (near dead short in this biz) and the upper end is 100 teraohms (precision, calibrated open). Most folks find that 10e11 ohms (10) gigaohms is more than adequate.
Check with me for values in the above range.
Richard Hull
Check with me for values in the above range.
Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
Re: Fall Time Deficit Disorder
You are right on target suspecting the capacitor's leakange resistance as a major factor in the actual time constant for the integrator. I believe the overall circuit gain also affects the effective resistance in parallel with the integrating capacitor.
If you have access to an electrometer, you can smoke out the really leaky caps. One bad actor is the solid tantalums - solid electrolytics..you need metallized polyester film or similar caps for the absolute lowest leakage conductivity. Also, low voltage leakage tests may not tell the whole story. Ideally a leakage current test at up to the bias voltage (couple kv ??) is needed to measure the actual resistance. Most DVM's will go to a microamp reading current directly, which converts to 1 G ohm per kV. You can get to nA sensitivities, however, by reading the voltage drop produced by the leakage current through a 1 - 10 megohm resistor. This will get you to the Tera Ohm leakage resistance level. (Just be sure to calculate the actual resistance of your meter in parallel with the current sense resistor.) Example: a 10 M input resistance meter across a 10 M resistor, makes an equivalent 5M current sense resistor. A 1 mV reading at 1kV potential, indicates a 0.2 nA current and an effective leakage resistance of 5 Tera ohms.
Dave Cooper
If you have access to an electrometer, you can smoke out the really leaky caps. One bad actor is the solid tantalums - solid electrolytics..you need metallized polyester film or similar caps for the absolute lowest leakage conductivity. Also, low voltage leakage tests may not tell the whole story. Ideally a leakage current test at up to the bias voltage (couple kv ??) is needed to measure the actual resistance. Most DVM's will go to a microamp reading current directly, which converts to 1 G ohm per kV. You can get to nA sensitivities, however, by reading the voltage drop produced by the leakage current through a 1 - 10 megohm resistor. This will get you to the Tera Ohm leakage resistance level. (Just be sure to calculate the actual resistance of your meter in parallel with the current sense resistor.) Example: a 10 M input resistance meter across a 10 M resistor, makes an equivalent 5M current sense resistor. A 1 mV reading at 1kV potential, indicates a 0.2 nA current and an effective leakage resistance of 5 Tera ohms.
Dave Cooper
-
guest
Re: Fall Time Deficit Disorder
I'm using NPO ceramics for the charge storage capacitor, so there is no question of leakage. There is no detector bias at present, as I'm driving the input directly from a test pulser. However, leakage current from the fet gate and PC board leakage both seem to be big factors. The time constants are much more in line with what's expected when I mount the timing components off the board and use a small signal mosfet instead of a jfet for the gain element. I hadn't considered open loop gain as a factor, but the point is well taken. I reduced the quiescent current in the circuit in a rather draconian fashion in order to get rid of a persistent 200-300 MHz oscillation. This reduces the open loop gain of the circuit. It sounds like what I need to do as well as the other expedients I have mentioned is to cautiously inch up the quiescent current again so that there is some reasonable open loop gain to play with.
BTW, the reason all of these factors become so important is that I'm using a 1-3 pF capacitor for charge storage in order to get a reasonable output signal amplitude (~1V) for short input pulses like what could be expected from a plastic scintillator. This is not unprecedented. The Ortec 109 preamp uses a 1pF charge storage capacitor. A larger storage capacitor (10-20 pF) is much easier to work with, and would be what one would use with NaI(Tl) or ZnS scintillators. With this value of storage cap, the output pulse amplitude is about 0.2V for a short input pulse. The output amplitude after pulse shaping is 50-100mV. This can be made up for with a post amplifier, and I have included one in my discriminator circuit. A fairly normal opamp can be used at this point, as the pulse is about 30usec wide. I plan to use 1/2 of a LF412. However, I think it is important to understand what is going on, so as not to be bitten in the future, and for my own edification. Also, there is an element of engineering pride here - if Ortec can do it, so can I....
BTW, the reason all of these factors become so important is that I'm using a 1-3 pF capacitor for charge storage in order to get a reasonable output signal amplitude (~1V) for short input pulses like what could be expected from a plastic scintillator. This is not unprecedented. The Ortec 109 preamp uses a 1pF charge storage capacitor. A larger storage capacitor (10-20 pF) is much easier to work with, and would be what one would use with NaI(Tl) or ZnS scintillators. With this value of storage cap, the output pulse amplitude is about 0.2V for a short input pulse. The output amplitude after pulse shaping is 50-100mV. This can be made up for with a post amplifier, and I have included one in my discriminator circuit. A fairly normal opamp can be used at this point, as the pulse is about 30usec wide. I plan to use 1/2 of a LF412. However, I think it is important to understand what is going on, so as not to be bitten in the future, and for my own edification. Also, there is an element of engineering pride here - if Ortec can do it, so can I....
Re: Fall Time Deficit Disorder
Whew... with such small input capacitance... you may need to think of the double shielded, scheme like Keithley uses to improve the response of their electrometers. You just use double shielded input lead and let the opamp or your input FET drive the intermiediate shield, close to the input voltage. You probably know the theory anyway.
At any rate, it sounds like a nice project. Pleas keep us up to date on the results.
Dave Cooper
At any rate, it sounds like a nice project. Pleas keep us up to date on the results.
Dave Cooper
-
guest
Re: Fall Time Deficit Disorder
Just last night, I was testing the newest version of the universal preamp. The timing components are mounted on teflon push-in standoffs to reduce board leakage. The pulser was soldered right to the input s of the preamp board. The fall time deficit with a 2.2pF charge storage cap and a 22M reset resistor is now about 2X, or about 20 usec time constant versus the expected value of 48 usec. The input pulse looks a little screwy, and I'm going to put together a pulser that has a better approximation of a current source output. The pulser I posted in the files section is good for driving current input transimpedance amplifiers with a low input impedance, like the Hamamatsu 3-transistor amplifier. A different type of pulser is necessary for getting accurate results with a high input impedance charge sensitive amplifier. You essentially need a pulser that puts out a current pulse, and then goes open circuit between pulses. To a good approximation, this is the way that PMTs and silicon diode detectors behave. I will post the results for everything (amp and pulser) when the results satisfy me.
We ship no wine before the truck arrives....
We ship no wine before the truck arrives....