Spad games III

The local broadline distributor

sells 2N7002 for $A0.085 in small quantities, going down to $0.065 in hundreds.

One hundred 2N7002 would cost $A6.50 and I could certainly wind a transmission line transformer for less than that - a couple turns of twisted pair around a ferrite core.

John Larkin's irrational fear of transformers has warped his judgement again.

Twisted pair can work just as well, and is cheaper and more compact.

Not to John Larkin.

5GHz transistors might work better. I built a three stage Percival distributed amplifier once, and got an 800psec wide pulse out of it. A cruder approach got us a 500psec pulse (but it cost a bit more).

--
Bill Sloman, Sydney

It better than stripping and soldering coax and winding it through a toroid. That connectorized chunk of cable is stock from Digikey, and plugs into mating connectors that are pick-and-place.

I have a pic of the production PCB around somewhere...

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

.

more

or

ep up

r up

ntage

what

Coax connectors are always expensive. We found three- an five-pin coax term inators that you could crimp onto a coax cable and solder directly into a b oard. They were cheaper than either the socket you had to solder onto the b oard or the plug you had to crimp onto the coax cable, and you only needed one per cable end.

Twisted pair can be soldered into or onto a board even more cheaply.

--
Bill Sloman, Sydney

Granted you don't want t make your own cables, but for the record, I'm curious why it's hard to find Ipex (or U.FL) cable end pieces. I can only find the SMD sockets, and pre-made cables with various things on the other end.

Does anyone know where to buy Ipex cable ends cheaply? For tiny co-ax, like 1mm, 0.7mm, etc.

Also, is there anything like this that works like ribbon cable, say from 2 through 6 ganged co-ax cables?

Clifford Heath.

Like this ? ....

piglet

Yes, but something better for the gate input coupling. That gets interesting.

Who is Mario?

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Actually as drawn it won't work. You're right my input coupling is wrong. But might work with inverter gates. Back to the little grey cells.

Mario is the guy behind the SPAD active quench paper George quoted some time back. Mario devised a very nifty comparator with hystersis shift to reduce after pulsing "twilighting" spurious events.

piglet

Inverter gates give positive feedback as the Marx chain elevates. That could be all the gate drive that you need (caps to ground) except for the prop delay skew.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Huh, I was powering it from a resistor/ capacitor string. How much delay is there toggling on the power pin?

George H.

Mario is the author/ designer of the spad circuits I've been using/ copying.... I haven't found anything else useful on the web. In Mario 1 he uses capacitive coupling... inverters at

7 V supply. and then in Mario 2 he dumps current from an inductor, through an npn into the spad node.

George H.

piglet

Have you done that with, say, 74hc14's? and if so, what's the delay? Three in 'series' might be enough for me. Or just a transistor and 24V.

It's just a goofy idea; I've never done it.

You should be fine with a transistor or a small mosfet.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

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

Very little delay.

This version might work ...

and could beat a 2N7002 with passive pullup to 25V?

piglet

I haven't done that with HC14's - I guess they'd be too slow for you where every nano-second matters. Schmitts can be very slightly slower than vanilla logic. Decades ago I had great fun building things like that out of CD4007 but very slow and with little current drive. The delay that matters most is the first inverter at the bottom of the chain that has to lift the rest up, then a staircase ripple.

piglet

Huh, that's cute enough that I'll have to try it. Thanks GH

Right, well HC14's is what I've been using. I guess once it's working I can work on faster. You know I don't see as much propigation delay as is quoted on the spec sheets. (All TI IC's) Is the delay mostly a function of the load capacitance?

The first thing to do today is remove the diode between spad and comparator and confirm (for myself) that this is shielding the comparator input C.

Later,

George h.

Well shorting the diode means I have to redo the front end... but hanging an extra 10 pF on the spad node made things ~twice as slow.

Anyway I nailed the source of a 2n7000 to ground, hooked the gate to the hc14 output, drain pulled the normally high PS (100 ohm series R) to ground. max V ~20V It's a sloppy layout, but it worked, mostly.

As the reverse biased increased, there were more and more of these events, where the pulse was not enough to quench the discharge.. 'more' is judged by the intensity of ghost images on new fav DSO. Single shot they are rare, but reducing the quench voltage give more, like this from before.

It takes ~1-3us to reset. I was thinking it would be great, if my new 'love' of a 'scope had a pulse width trigger. And it does! I want a keysight for xmas.

I'll post pics tomorrow.

I need a circuit that only triggers on long pulses, so I can count them. say more than 0.1 to 1 us. (That seems easy enough...as long as the rate is low. low pass into hc14.)

George H.

I can see 100ohm pullup would get fast risetime but isn't 4W a lot of heat near a spad you're trying to keep cool?

piglet

Sorry bad description on my part.

20V---R(100)---+---C(~500pF)--spad node | D hc14------G (2n7000) S | GND

Hey I fixed the long pulse problem.. I'm going to call it after pulsing or something. My one-shot wasn't resetting fast enough and so a fast double pulse hung it up. (I'm just guessing the fast double pulse was after pulsing.) One shot needed a diode.

V+--+--+

----+ | _ |