MPPC amplifier

It's not uncommon when using a slowish amplifier (60 MHz @ min stable gain, 625 V/us) with a much faster transducer. The op amp can't control its summing junction adequately in the inverting configuration in that case, leading to measurement nonlinearity.

Also the MPPC's output is far from shot noise limited (due to the high gain and fixed charge per segment), so you don't care much about the Johnson noise of the load resistor (probably 50 ohms).

If you want a fixed, finite integration time, I recommend using ping-pong gated integrators. Each one integrates maybe 60% of the time, and you overlap the two so that at least one is integrating at all times. You can make them closer to 50% duty cycle if you like, but you run the risk of missing stuff if you have both switches transitioning simultaneously, i.e. at exactly 50% duty cycle each.

Cheers

Phil Hobbs

Thanks! I was thinking about integrating short intervals (10-40 ns) and then summing in software if I needed longer integration times.

I like the idea of a ping-pong integrator, but I don't understand why I should overlap their integration periods. I would assume that this would lead to blurring of consecutive samples. Wouldn't it be better to have the integrator be high-Z on its input when it is not integrating? Any charge being generated will just accumulate during the dead-time between the ping-pong when both integrators are off.

Slowish?

George H. The op amp can't control

Ah, okay, assumed that the 400 MHz number was the GBW. Still way too slow to be safe as an inverting TIA with a MPPC.

Cheers

Phil Hobbs

Perhaps the devices take a finite time to move from integrating to non-integrating, and you want to put half the signal that lies on a transition into the earlier interval, and the other half into the later interval.

A finite transition time from integrating to not integrating will have that effect too.

Perhaps, if it works that way. Summing junctions that move away from the default voltage can do odd things.