Separate pulses from edges?

But meanwhile does anyone see any disadvantage in the method I came up with, as linked earlier?

Apart from simplicity, it has the advantage for me that the trailing edge signal is already present in my Schmitt:

No, you don't. You'd have to use an EX NOR for the first stage, not an EX OR.

Agreed.

After making the XNOR correction, the finished circuit looks like this

Although giving nice clean outputs it has the disadvantages of extra complexity, components and space, the latter making it a non-starter in the present project.

Still hope to hear views on my simpler approach.

Use a PIC.

Not what I am saying. You replace the EXOR gate with a single NOR, and you are done with the negative edge detector.

Again, I am saying replace the EXOR with a NAND. You can reuse the signal from the first RC. So you are down to a resistor, a capacitor, a NOR and a NAND, and you need one more inversion for the NAND (unless you have an AND available).

It is too complicated.

And you are missing my suggestion, also.

I don't think that works either!

Yet less so than yours, after adding either two extra differentiators or some further logic gates, to get the brief pulses specified?

I did misinterpret your original description, yes. A drawing would have helped! And I tacitly assumed you must be proposing XORs/XNORs to get the required brief outputs. But I believe my revision gets it right?

Terry Pinnell wrote: (snip)

(snip)

Sorry for my stupidity. I meant to say that you need to invert the delayed (RC filtered) signal before combining it with the original signal with the NOR and AND gates (but that the inverted, delayed signal can be used for both gates).

I dislike pulse circuits that include diode drops. But you are free to like them.

{trimmed}

Again, unless I'm still

Yep, I do.

-- john

Bingo - that works fine, thanks!

I'm not mad about diode drops either. But which particular diodes did you mean please? The two here?

- Lightness detected at dawn (i.e. 'Trailing Pulse' above)

- Power up pulse, for predictability So I have to use diodes there. With similar past circuits, it *seems* to work OK, but...

Thanks. Actually, that second one looked awful in the 'scope when I got around to trying it. Probably not a good idea to feed off the 'innards' of a very high impedance Schmitt?

And, although I haven't yet had enough time to work on it, so far even the first 'basic' approach (which I've definitely used successfully in similar circuits before) failed to deliver reliable outputs this time. I was getting spurious (µs) pulses just before the Schmitt triggered. That seems just like the behaviour I reported in the separate thread '4001 as Schmitt', attributable to slow switching because of the smoothing cap. I thought I'd sorted that by modifying the Schmitt's input to a T-filter, as discussed. But looks like it needs more study. I'll post again if I can't sort it.

(snip)

Yes, those diodes. In these cases the diode drops allow the signal to exceed the supply rails, which tends to latch up 4000 series CMOS. I would use small signal Schottkys if I built these circuits.

Terry Pinnell wrote: (snip)

Why not use a gate to combine those two signals?

CMOS is pretty tolerant of a signal not quite reaching a supply rail, but it does tend to increase the supply current. But the drop from diode gating is not so dangerous as diode drops involved in outside the supply rails clamping.

(snip)

You still have one extra inverter in there. The NOR and NOND inverters have similar inputs and outputs. Use one for both pulse formers.

I'm missing something here! I don't follow what you mean. As I see it, that circuit's now fished, yes? Where's this 'one extra inverter'?

Good point. I suppose the only reason is that, building on stripboard, I'm in the habit (maybe a bad one) of minimising space. A few vertically-mounted diodes versus a 14-pin IC holder for a 4001/4011. And presumably, to use my 4001 stocks, I'd need 2 gates for each OR? So that's 2 diodes versus 1/2 x 4001, yes? 4 solder joints versus

Thanks for the tip, hadn't realised that. See it in so many published circuits.

You show a NOR inverting the output of the RC delay filter (outputting the inverted signal at pin 10 of the NOR), and also a NAND inverting the same signal (outputting that inversion at pin 10 of the NAND). Those two gates produce the same result. Save one of them and use the other to drive both the NOR AND the NAND pulse former circuits.

Duh - of course. Can't think why I did that, especially as you reminded me earlier!

In article , Terry Pinnell wrote: [....]

This sounds like a job for a dual oneshot, or if you don't mind being a bit mickey-mouse:

Ascii art

----- ----! ! ----- ! ! NOR !---------------! ! +----! ! ! NOR !----- Out1 ! ----- ----------! ! ! ! ----- ! ! in--+-+--/\\/\\--+----+ ----- ! ! +---! ! ! --- ! ! NOR !--- ! --- ---! ! ! ! ! ----- ! ----- ! GND ---! ! ! ! NOR !-------- Out2 -------------------------------! ! -----

Thanks Ken, very neat. JP beat you to it with very similar circuit. See discussion up-thread .

However, yours has great merit of only a single new IC.