Dropping a few tenths of a volt problem

Sorry, another misassumption on your part here. I haven't even gone over to the batteries yet. All the development so far at all supply voltages tested has been carried out using a 30A linear boat-anchor, which I think you'll conceed is well stiff enough for 4 to 5 volts at

Sorry for the late reply, John. Been out all day and only just got in a few minutes ago. Thanks very muchly for the circuit, but before I examine it in detail, and for the avoidance of false asumptions (of which there have been many on the part of some people who shall remain nameless) is this likely to function as advertised when powered by 3x1.5V in series alkaline cells whose voltage will amount to around 4.8v when new and head south from there as they age?

Sigh....

I'm not really sure what your requirements are, namely the input voltage and pulse width and the output requirement.

An RC type stretcher has a time stretch ratio that is limited by the ratio of the pulldown/discharge current to the pullup/charge current. The FDV301 can sink a ton of current, all the way down to ground, numbers like a quarter of an amp with 2.5 volts of gate drive. That will discharge a small cap pretty fast.

You can cascade two stretchers to get an extreme time-stretch ratio, but a 123-type one-shot is easier if edge triggering is OK.

That's what he simmed, yet he said he wasn't sure how narrow they were.

Read Cursitor's latest post. ...Jim Thompson

a) 74AC14PC b) Even HC is plenty fast enough.

There are plenty of problems with the original circuit--Fig. 12 of that 1975 Fairchild Schmitt-trigger app just doesn't work for this.

The problem is definitely the pulse-stretcher. It's flakey. It's not the supply voltage.

A CMOS 555 (John Fields) is a darn good notion, actually. Wide Vdd, low Iqq, reliable triggering on short pulses. The 2n3904 front-end looks a bit slow for 1-2uS triggers though.

Your circuit-- Vdd Vdd -+- -+- | | 20K 180K | | |\ +--. +---+---| >o--- | | | | |/ |/ | |\ R5 |/ --- C1 >--20k--| '--| >o--51K---| --- 80nF |>. |/ |>. | | Q1 2n2369 |Q2 | === === ===

(input protection omitted)

looks decent. Low Iqq, works over Vdd, dumps C1 quasi- quickly and almost to GND. Mr. Doom had better have

John Larkin's circuit does those, is faster, but needs input protection.

Vdd -+- | 1M | |\ +----+---| >o--- FDV301 | | |/ ||--' --- ||---'|--+ | | | === ===

That's nearly as good as a triggered monostable for catching short pulses, plus all the other attributes (Iqq, Vdd range, etc).

I wouldn't use the 74ACxxx for this; crossover current is beastly--I smoked some early samples just biasing them quasi-linear. 74HCxxx would be better, saves power.

It's hard to beat the above in this application. For shorter (faster) triggering I might prefer the 74AC123, or the 74HC4538.

Cheers, James Arthur

Groan... John, I know from old that you're a knowledgeable fellow and I always welcome your input, but your clearly having a bad thread here. I've already outlined the 'requirements' as such - and more particularly the difficulty in establishing them - to Jim and others and I'm about beyond any further re-interation at this point. Give up, mate. Please!

Many thanks for the clarification. Sorry for the further delay, but Infernal September had another major outage last night here. :(

Indeed they are. So short I can't measure them! Even at the fastest scan rate the peaks only appear as single points of light on the scope's screen. It's only after they've been passed through the second Schmitt gate that I can see any top on them at all. This is the main reason I didn't expect anyone to be able to re-design it for me: the critical info relating to input pulse characteristics is absent, so I thought, seeing as I'd got it running fine at up to

Thanks for your observations. I've carried out the measurements I needed to so the urgency is over now. I just used the GC's front panel meter; it took a bit longer but no matter. So, now I have a bit more time I'm going to build ALL the circuits suggested by the leading contenders here and see which one works best with these extremely short pulses. In fact I've already ordered those parts I don't have to hand, so by the end of next week I should have a winner!

Well...... I'm not at all convinced that simulation is of much help in this particular case, given I am unable to fully establish the characterists of the input signal; it's more a case of trial and error. ISTR there was a vigorous debate on here about 14 years ago about the advisability of modiying model parameters and that the consensus back then was that the practice should be heavily deprecated. Is that no longer the case? If so, what's changed?

If you want to continue with the sim of the inverter in Ltspice, you can adjust it's logic voltages, TC etc to behave much closer to the real deal.

In case you are not aware of this, Right click the inverter of choice and add a statement in a spiceline edit box.

Vhigh 5 ; this will generate 5 volts out for the on state Vlow .7 ; this is for the off state..

Rout 100; this should be done to make it look closer to a real deal.

Of course, you can adjust those closer Rhigh = 100 Rlow 10 in which case the Rout would be ignored.

Traise, Tfall should also be set to match the device you are using..

After doing all this, you can then work out the problems you currently have now.

Just a thought...

Jamie

If appears to me that your GC signal isn't broad enough to accommodate the Schmitt window when you raise your Vcc voltage.

Why not use a non-schmitt type and apply a little (+) feed back to the input for a marginal hysteresis, instead of the wide one you have now? Not only will it most likely work better in the long run, more sensitive, it'll work over a wider range..

I suppose you could even use some (-) feed back with what you have now to narrow the window..

Jamie

Well, whether it 'works better' in this particular instance or not is yet to be determined! Many thanks, gentlemen all, I shall be building the top four designs over the coming week and will pick the one that sounds sweetest. I've already decided to bin the one I've already built, since the consensus is that it's er, "sub optimal". ;-)

You have "outlined" (but not specified) the problem in half a dozen posts, here and there, a little at a time. Last I noticed, you had an oscilloscope that might characterize the "*very* thin and spikey" input pulses but you don't know how to use it.

I'm happy to give up.

Years ago I never used a simulator and I don't really use it much now.

I do most of my R&D at the bench with real test equipment and some old note books I keep around for handy little circuits I came across over the years or come up with myself. Years ago I looked at spice, but computers were so slow and spice being badly put together, not forcing proper argument passing to avoid confusing, I found it quicker to bench test and then let the customer be the next inspector, debugger :)

As for changing the parameters? what would a tool be good for if you couldn't override or set parameters? I wouldn't use a tool any other way.

Jamie

Thanks, John. BTW, the best scope I have for examining these pulses (which I've only just acquired) is this model:

But I've done nothing with it yet. Previously I've only used analog scopes of one kind or another so this is one is like totally unfamiliar to me. I would have liked to have measured the actual duration of the pulses and their source impedance, but this was not possible with any of my other scopes. Do you think I'd be able to resolve these very fast pulses with the above DSO and if so, can you give me a steer on the key optimal settings to use? Many thanks.