It's the capacitance, stupid

Well, you could use something like a 6CW4 Nuvistor in parallel with the SPAD to turn it off. Small HV transistors such as the 2N6517 run 5 pF or thereabouts.

Tubes still own some corners of the design space.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

You only need to pull the spad voltage below its latched zener-type voltage to turn it off. That could be done from a voltage source through a schottky diode, or better yet a few diodes in series.

I still don't understand where the current spike comes from.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

The passive resistor quench circuit lets the SPAD do that by itself.

Problem is that you need like 5 time constants (5 x 3 pF * 500k ~ 7.5 us) for the bias to come up again, leading to a nasty dead-time correction. To improve this, you need to reapply bias to the diode ideally in time for the next event. A SPAD is a tens-of-picoseconds sort of device intrinsically, so that time constant limits your measurement speed pretty badly. Even small SPADs have a high false count rate, so fast reset is key to getting decent dynamic range.

As you say, another approach is to use a much smaller bias resistor and reset the SPAD like an SCR. I like the diode string idea, as long as the string doesn't mess up the time domain response of the SPAD.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

It's something like 400K. It looks like there is a pretty big and fast spike into the sense resistor. Is that purely the current through the

400K? Or is there some not-shown capacitance to ground?

Ideally the top of the spad would be driven by a fast-switching voltage source, with maybe a low value series resistor to limit the peak current.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Thanks guys, I really like this self quenching circuit. First off, I can stick a switch down the bottom and change the sense resistor to ~1Meg. (not sure the exact value) And then measure the 'zero' bias voltage and higher APD response. And sorta measure the QE, at DC.. having never done the measurement, I'm not sure how well this works. Secondly, students then can find the timing deficiencies, (about 1 us... you can decrease R_quench.. but then can't turn the bias voltage up so high.) of the circuit... And then I need the fix. How about I just stick a ~1-2pf cap into the node I want to charge and pulse it positive for C*V/time. Maybe the lower voltage, bias power supply can soak up any over voltage.

Phil, The paper above is worth a glance. They talk about twilight events, that happen when the diode is recharged, but compartor still off. Paul Kwiat, is the grand old man of entanglement, (though younger than us. :^)

George H.

The diode has about 3pF of C. The 400k is mostly irrelevant, (except with 200k and 10V bias it doesn't always quench right away)

George H.

I downloaded it yesterday and will give it a squint. I should know more about SPADs than I do.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Why not go to the literature for a circuit that works.

Sergio Cova has published a lot on the subject. This is one of a number of papers from his group. There are plenty of other publications.

I liked the one that used a 685-style ECL comparator and exploited the input-latch feature of that style of (fast) comparator.

There's not a lot of sense in tryong to re-invent that particular wheel.

--
Bill Sloman, Sydney

Right, I've been reading a little of the literature. I linked to this above,

Mario seems like a nice guy.. He'd like to sell me some pcb's which is also an option.

Mostly I have this dream of running the spad in different modes as a learning/ teaching experience. Linear gain =1, APD gain ~100, spad, spad with reduced dead time.

Did I post this pic?

It shows the dead/ reset time. 'scope is on infinite persistence, triggered on the ~10ns pulse in the center... and then filled in with other pulses. The lower pulse height right after the trigger is because the spad hasn't fully recharged.

George H.

Yeah, me too! One new thing I gleaned from the paper is that there may be a sweet spot in temperature space. I always thought the colder the better. (lower dark count.) But the above says you get more after-pulsing if you cool too much.

George H.

Oh, I should add that I don't really understand where the current spike comes from either. I see ~400 mV pulses for 10V over bias and yet for some simple model the pulse should be the ratio of R_sense/ R_quench, ~1/1000, 10 mV. Maybe I need to add some parallel C to my quench resistor?

George H.

Seems like capacitance from the high side of the spad to ground (not many pF) would make a big fast spike into a low value sense resistor. That would keep things fast. The spad substrate capacitance, or circuit strays, may be doing that already.

I'd scribble out a wonderful circuit, but I have to take the cinnamon buns out of the oven.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

"Which the margin of this post is to narrow to contain." ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

they taste cin-ful

NT

I wonder what device he used for Q1? Is that 1N6263 schottky used as a low capacitance reverse breakdown device?

Whenever I hear talk of fast HV pulses I am reminded of 1940s transmission line PFNs but that paper's resonant LC idea seems very neat. Fascinating stuff to me who knows nothing of SPADs!

piglet

Yeah that works, a few 100 fF (0.1pF) from the spad to ground or across the 400k quench resistor gives me ~volt level spikes, of ~1ns or less.

One thing that doesn't explain, is why it get's spikier when I raise the bias voltage higher above the breakdown. Hang on I'll take some 'scope shots.

Here's with 3V over the breakdown voltage.

and here's at 10 V

Anyway my capacitive coupling idea was silly. I guess I'll have to reduce the resistance a bunch and actively quench... grumble.

George H.

I wonder if an inductor somewhere would help. Change the recharge from exponential to critically damped.

Of course, active quenching would be best.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I can ask him (about Q1). I was a little confused by the 1N6263, but it's biased on, (about

0.5V on one end and whatever Vce is on the other. When the spad fires node A goes high and turns it off, and that starts the whole chain. (at least that's my understanding.) I guess after that it's there to prevent the ~25V quench voltage from frying the comparator.

Re: LC resonance.. TBH I was going to skip that part on my first try.. just use a bigger voltage, that might be slower.(?)

George H.

Oh I should add that this is in my LTspice simulation.