I'm not sure how to succinctly phrase my question, so bear with me. I'm trying to make a very simple tail-pulse generator for a specific application. In the old days, this was done with a mercury-wetted relay. Imagine a SPDT relay. The common terminal would be attached to a capacitor, which was also in series with a resistor. The time constant of the tail is then just the RC time constant. When the relay is in the NC position, the capacitor is charged and a constant current is flowing through the resistor. Then the relay switches and the NO contact connects the output to the capacitor, the voltage on which decreases as it discharges through R.

So, that's the basic principle. In my case, this has to go somewhere where a mercury-wetted relay is not possible. I thought I could replace the relay with MOSFETs, but my simulation is giving me strange behavior. It works with a generic/ideal MOSFET, but as soon as I pick a real MOSFET, things go wonky. The trace on the left below is what I get in LTSpice with a generic MOSFET. The trace on the right is what I get with any randomly selected MOSFET. I've tried different Rdson values, etc.

Any advice? Where should I go next for troubleshooting? Maybe there is a better way to do this?

Thanks!

 

hi tm,
Please post your LTS asc file.
E

 

I don't think it's relevant to the problem I'm having, but the amp in the circuit is not in the standard LTS library, so I also attached the model I downloaded from Analog Devices.

 

I suspect this is because C2 never gets to discharge fully. When M2 is off there is no current path for C2.

 

I suspect this is because C2 never gets to discharge fully. When M2 is off there is no current path for C2.

I tried adding a resistor between M2 and C2. That definitely changed the shape of the trace but didn't improve things. See attached image. It's also strange to me that, when the real MOSFETs are used, the voltage level is much higher than expected. With a 10 pF cap discharging into the 100 pF cap, I expect a voltage less than 500 mV. But I'm also a physicist, not an EE. Sometimes things work better on paper than they do in the real world!

 

hi,
As AH explains there is no current path for the cap to discharge.

If I understand correctly you want a low going pulse with a fast leading edge and a slow trailing edge.??
Your circuit looks over complex for that requirement,

What is the specification of the source of the pulse and the timing periods, in and out.?
E

 

It works with a generic/ideal MOSFET, but as soon as I pick a real MOSFET, things go wonky.

That's because the inter-electrode capacitances of a real-world FET are much greater than those of the generic FET and also much greater than C3 and C2.

 

This has to model a scintillation pulse, so a rising edge on the order of a few nanoseconds and an exponential tail on the order of 1-5 microseconds. Ideally, it is a current pulse and not a voltage pulse, meaning you set the amount of charge by selecting C3 and then discharge C3 into the rest of the electronics.

The circuit is exactly how it's done in every "old school" tail pulse generator used, except that they used relays instead of MOSFETs. (E.g. one of these: https://www.berkeleynucleonics.com/model-bh-1)

I'm completely open to other ideas, though. It just needs to be small and work in any orientation (no mercury relays!).

 

That's because the inter-electrode capacitances of a real-world FET are much greater than those of the generic FET and also much greater than C3 and C2.

Ah! The light bulb went on. Thanks, Alec_t. Any suggestions on how to better approach the problem?

 

hi tm,
Read the BH-1 datasheet, it explains the pulse OK.
I guess you do not want to duplicate the full specification as that d/s.??

What is your trimmed down specification.?
E

 

Do you really need a blitz fast op-amp? If so, then why are you using MSOS-FET's that are snail slow and have massive amounts of capacitance?
How fast a tail pulse do you need?

Look here for some very fast tail pulsers:
https://forum.allaboutcircuits.com/threads/fast-risetime-falltime-pulsers.124518/#post-816114


edit: Just saw your post. I see your pulse does have to be fast.

 

After seeing your link to to tail pulse generators I have a big question. How many of the features in the link do you need?

 

Try this. Pulse rise time is ~15ns, decay time ~1.2us.

 

I guess you do not want to duplicate the full specification as that d/s.??
What is your trimmed down specification.?
E

After seeing your link to to tail pulse generators I have a big question. How many of the features in the link do you need?

I do not need all of the features of a real tail pulse generator. The goal here is to provide a known pulse that can be injected into the system at regular intervals for baseline restoration in the DPP algorithm. I need something with a fast rise time, on the order of 10 to 100 nS, with a decay time on the order of 1-10 microseconds but these values do not need to be adjustable. I need to be able to activate to provide a burst of pulses with the pulses coming in at something like 500 Hz. The duty cycle on these bursts will be on the order of 10s of minutes. I'd like some kind of arrangement where I can easily choose between 2 or 3 different pulse heights but it can be very coarse. The exact pulse height (i.e. amount of charge) has not yet been determined for this system but we're talking about anywhere from 10 to 400 picocoulombs of charge. The charge is "negative", i.e. it's a negative voltage the appears on the input cap, C2.

The fast amp *is* needed and is actually the preamp for the rest of the system. It has already been successfully implemented, I'm just looking for a circuit I can tack on. I should point out that the actual MOSFETs shown in the image were random. I literally chose them by accident and noticed that it changed my output, and then wondered why.