favorite circuit for 8 - 20us pulses

Standard pulses: tr 8us, td 20us to 50% down. What's your favorite circuit for making device test pulses, from a square wave I suppose.

--
 Thanks, 
    - Win
Reply to
Winfield Hill
Loading thread data ...

Is this logic level or HV? If logic then I am unclear on the terminology, do you mean 8us rise time, 20us delay? Do you want output crisp logic edges or slow exponential rise/fall? My starting point would be RC either w-w/o schmitt HC14

If HV then cap discharge I guess?

piglet

Reply to
piglet

74121 or 74221 (dual) monostable.

About 0.1% stable with a good quality capacitor, and a 19mm potentiometer trimmer lets you set up exactly the pulse width (or delay) you want.

If you want to count edges from a square wave

formatting link

is nice. You can clock it at up to 10 MHz (if you run it off a 5v rail).

This is obvious stuff. What kind of part do you actually want?

--
Bill Sloman, Sydney
Reply to
Bill Sloman

It sounds like Win wants 8 and 20 us rise and fall times.. but I'm not sure.

George H.

Reply to
George Herold

Y'mean an IEC 61000-4-5 surge?

You need a network like this:

formatting link

Note that the impedance is not resistive, it's impedance abuse because the waveform depends on load impedance as well. Namely it's a 1.5-50 open circuit and 8-20 short circuit, and the ratio of peaks is the "impedance", which is supposed to be 2 ohms.

There's something like 10 or 20% loss in this network IIRC, so set C1 IC={desired peak voltage * 1.1} or so.

I've also made an automotive load dump model which is square-wave-based, for SPICE purposes only (I wouldn't recommend building it that way ;) ).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

This is standard terminology. First, these are current pulses, with high compliance-voltage capability, higher than the breakdown voltages of the D.U.T. The 8us tr refers to the time to reach the peak current, and 20us td refers to the delay time, from the beginning, for the pulse to drop to 50% of the peak value. You will recognize this pulse's waveform drawing.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

What is the required peak current ? 100 kA ? Positive or negative peak ?6

Reply to
upsidedown

10A is a common value. Need both polarities.
--
 Thanks, 
    - Win
Reply to
Winfield Hill

Only incidentally related but a 2D21 thyratron sawtooth oscillator has ripping fast edges at the end of the ramp (de-ionization time?)

suitably divided down and buffered a 10 MHz GBW 20v/uSec op amp can barely keep up with it.

Reply to
bitrex

Oh, the 8/20 sounds like ESD testing. I'm pretty sure the ESD testing "guns" use a spark gap to trigger the pulse, and a network of R's and C's to shape it and control current. This must all be public domain stuff by now. So, all you need for these is a HV power supply to charge the first cap.

Jon

Reply to
Jon Elson

That is pretty low, it doesn't even fulfill IEC 61000-4-5 Class 0 (25 V / 12.5 A into 2 ohm), intended for well protected environment.

More approximate for long signal cables would be Class 3 (2 kV / 1kA into 2 ohm or 2 kV / 48 A into 42 ohm) or for mains connected devices Class 4 (4 kV / 2kA into 2 ohm).

Reply to
upsidedown

A Moog synthesizer's solution to the pulse problem is (basically) a three-transistor multiplier. You capacitor-load it and the rise and fall times depend on the current programmed into the emitter. Set the multiplier input positive for rise, negative for fall, and program the current differently in the two phases...

Another is the gas-filled tube; conductance rises quickly when it avalanches, drops slowly as the gas cools. It'd be a hard problem to match risetime/falltime to a model, though, because there's pressure/geometry/quench-gas variables to consider. Still, if you need to make a CERN deector that's the size of a three-story building, fill gas is the economic solution at that scale. Georges Charpak earned his Nobel prize doing that.

Reply to
whit3rd

Thanks, I realised that some after posting when I connected it to your other posts on using a BJT e-b for ESD protection!

I think that while the 8/20us wave may have a use representing line borne transients it is way too slow for simulating ESD events which can have very steep rising edge.

piglet

Reply to
piglet

Yes, but if our part can pass 8/20, it can also pass faster risetimes, provided there's a little capacitance at hand. On my sensor stick, each I2C line got a 100pF cap, along with the CD143A. Blast away! (BTW, for anybody who remembers that discussion, my final choice for a data cable was a miniDIN with gold pins, CS-DNPDM6MMX2-006 from Cable-On-Demand, $5 each. Surprise! It came with a braid shield, so we're double safe.)

--
 Thanks, 
    - Win
Reply to
Winfield Hill

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.