Sensl/On Signal-driven multiplexing for SiPMs

IME the fast output is a disappointment in real life. It's quicker, but it's really just an RC differentiator, so the resulting amplitude depends on the slope of the light pulse. For a general signal (one whose pulse edges are not necessarily super fast), doing nonlinear stuff like that will leave you with a signal that's very hard to interpret.

In niche cases it might be a win. Of course it only works if you're tiling SiPM chips, since each one has only a single fast output.

I'd be much more likely to cascode the fast outputs with a biased SiGe:C transistor and condition them separately with linear circuitry. The chips cost $50 or more, so if you're using 4 or 9 or 16 of them, there ought to be space in the BOM budget for that. Since they're AC-coupled anyway, the DC bias is not a big worry.

The other thing is that the fast output is only super fast on the leading edge--its RC decay time sets the trailing edge speed.

We make a nice MicroFC-based PMT replacement gizmo that produces nice Gaussian-looking pulses about 4 ns wide. It ignores the fast output entirely.

Cheers

Phil Hobbs

I used BGA427 for that.

You also see the RC recovery of the slow path in the fast path waveform.

You like to share the schematic?:-)

It'll be okay for a lot of things, but its Zin is 10x higher than a BFP640 running at 5 mA. (The BJT will need a base stopper, but it's not a big deal below 500 MHz or so.)

It's not as pronounced because the fast path recovers relatively slowly. The fast axis waveform isn't very S-shaped at all, it turns out.

Super happy to license the design--that's what it's intended for. We charge considerably less in royalties than what folks save in BOM costs, besides being able to manage their own supply chains.

Cheers

Phil Hobbs

Phil Hobbs wrote: ...

I guess you mean the BFP640 in common base circuit, with low Zin to keep RC small, the collector loaded with 50 Ohm and a ferrite base in the base as "base stopper"?

...

Well, experiment people normaly only asks for few devices and what request come next I do not know. But I will keep in mind.

B.t.w: Any opinion on the original question about SDM?

I thought I answered that: doing a nonlinear thing like that with the fast output will give you a signal that is very difficult to interpret, in general. If you have nice big pulses, it might be OK, but I haven't found a good use for the fast output yet, despite trying. (It's useless for speeding up the leading edge of the slow output, for instance--way too much bandwidth is missing in the middle.)

Cheers

Phil Hobbs

Right. A BLM15BA050 or BA100 will keep it stable up to several milliamps I_C, with a good layout. I'd probably put an 0402 footprint in series with the emitter as well, just for emergencies. Gotta decouple the collector bias carefully, too.

Cheers

Phil Hobbs