My circuit's ringing

Weird. That isn't a reflection, because the ringing has much sharper edges than the excitation pulse. It looks like it's talking to the comparator, maybe via the grounds or maybe just via the big inductance you put in the circuit by making the ground currents go all over the place. Does the junk go away when you jack up the comparator threshold so that it never triggers?

Cheers

Phil Hobbs

"George Herold" kirjoitti viestissä: snipped-for-privacy@y2g2000prf.googlegroups.com...

Now that looks like my solderless breadboard designs :) Small resistor on hc14 output to limit slew rate maybe helps. Do you have good decoupling caps near that comparator?

-ek

"Does the junk go away when you jack up the comparator threshold so that it never triggers?"

Yes. No crud if the comaprator doesn't fire.

So perhaps it's a ground issue. (Again!) There's only a few mm's separating the two separate ground planes. I wonder if I can scratch down to the copper and stich the two sides together, maybe with some copper tape soldered along the edge?

This is someone else's board so I hate to 'trash' it on him, perhaps I'll let'm do it on Monday. At least that can be the first thing to try.

George H.

t- Hide quoted text -

m...

Yes there are decoupling caps. A bit of series resistance on the output you say. Got a value for me to try first? 100 ohms? 1 kohm?

Thanks,

George

Two separate ground planes? Why?

That's often a recipe for trouble.

What is on which plane?

John

ust

ld

The HC14 edges slightly precede the glitches on the input waveform, so Phil's right, the comparator output or the HC14's switching is coupling back.

Two ground planes? Oh.

A 100-ohm series resistor would tame the line-driving currents. That might keep the switching currents out of the grounds and stop the spikes from coupling, but it's just covering up the symptom. Better: fix the grounds _and_ tame the currents.

-- Cheers, James Arthur

R's work great to remove that problem..

But the PMT has lots of gain, why use a high-gain comparator after it? Just connect a monostable. The counter of a PIC isn't terribly fast anyhow, a retriggerable monostable will do fine, and filter out any false triggers.

Start with something like 22, 27, or 33 ohms and see what happens.

BTW - Is the PIC counter input able to pickup a 12nsec pulse?

on

it

to 50

the

just

er

hold

" Two ground planes? Oh."

Yeah, I believe there was some concern that 'stuff' would get back into the input. (Comapartor driving two inverters, one inverter is monitor out, 500 ohms in series and then 50 to ground. The other drives the pic. (I havn't looked at this 5 volt digital line.) The comparator and hex inverter define the 'input' section.

I got to see the layout for the first time after observing the 'ringing', (probably better called crud.) Hopefully this will be some sort of learning experience.

George H.

(I would have responded sooner, but our home computer's hard drive lost it's way.....)

The pmt pulses are ~100mV. The trigger level is set below this. Can monostables be triggered down that low?

ohm?

Thanks, after asking the question I realized I'd try and guesstimate the trace capacitance and then add an R for whatever RC time might work.

I don't know about the (timing resolution/minimum pulse width) of PIC's. I resurected a Data Pulse generator that has a two pulse output. The pots that set the pulse width and delay time were scrathcy, but at the minimum ~10ns pulse width and ~40 ns separation the counter still seemed to count just fine. (I'm not real sure about the exact value of those numbers.)

George H.

"My circuit's ringing"

So, did you answer it? ;-)

Cheers! Rich

Judging from the shape and timing of the glitche, I'd decouple the positive rail going into the inverter ~ 10 ohms in series, .01uf shunt at the supply pin.

RL

Thanks RL, there are already 10 ohm R's in the supply rails. I'm betting it's a long ground return path. (As Phil H. surmissed.) I'll find out on Monday.

George H.

The R has to be in the signal path.. To lower the Q in the circuit.

I've also seen series LCR tied across the circuit with approximation of the ringing. But that is very selective and problematic..

Might be easier fooling around with values of parts that are already present, then. Looking for effect.

Might be a scope ground issue in measurement.

You don't seem to be able to provoke any actual misbehaviour, and haven't indicated that there was a measurable radiation problem.

Is there a issue?

RL

PMTs are current sources with hundreds of volts of excitation. It's '100 mV' only because of the attached load resistance. A different load resistor, or a bit of extra gain at the dynodes (raise the bias voltage) will suffice. If that's inconvenient, a one-transistor amplifier, base to +1V, emitter to 1k ohms to GND, collector pullup 1k ohms to +5, makes a suitable pulse to TTL amplifier- just capacitor-couple the PMT pulse to the emitter.

Or, a NE555 trigger input really IS a comparator, internally; with a resistor bias string and capacitor coupling, it will operate nicely from the 100 mV (but beware power supply sensitivity).

ll

"> Is there a issue?"

Well, It's the uglies that I'd like to get rid off. Students will set the comparator trigger level. To do that they will look at the two channels that I displayed. It would be nice if all the crud wasn't there and there was just a 'clean' signal.

George H.