Pulse extender circuit

[scnoob]

This seems simple to me but I can't seem to find a proper circuit diagram to do what I need.  I've attached a timing diagram of my input waveform and the output waveform that I need to get.  Here are the basics:

1. I need to extend the width of an input pulse for a duration of x.  (I may need to adjust x later in the design process)
2. The output needs to restart the x timer no matter how many input pulses are given.
3. If the input stays high for a long time, the output must also stay high until the input goes low PLUS extend the pulse by x.
4. Simplicity is key, I need to use as few components as possible in a limited physical space.
5. Power utilization is a major consideration...This should be a minimal waste energy circuit.

Thanks in advance!

[CM800]

This seems simple to me but I can't seem to find a proper circuit diagram to do what I need.  I've attached a timing diagram of my input waveform and the output waveform that I need to get.  Here are the basics:

1. I need to extend the width of an input pulse for a duration of x.  (I may need to adjust x later in the design process)
2. The output needs to restart the x timer no matter how many input pulses are given.
3. If the input stays high for a long time, the output must also stay high until the input goes low PLUS extend the pulse by x.
4. Simplicity is key, I need to use as few components as possible in a limited physical space.
5. Power utilization is a major consideration...This should be a minimal waste energy circuit.

Thanks in advance!

sounds like you want something like this:


The Input buffer will charge the capacitor, keeping it charged while you supply an input. When the input goes low, the capacitor steadily discharges through the resistor. Once it drops below the output buffer's LOW threshold, the output would go low.

you can change the capacitance and dump resistor to a suitable time length.

[Ian.M]

Your requirements are incomplete.  Please specify the following:
* High and Low levels for input and output.
* Minimum input pulse width and maximum pulse rate.
* Minimum and maximum values of x, and desired resolution and/or accuracy
* Available supply voltage(s) and current (both average and max pulsed).

[jonovid]

good old 555    otherwise incomplete.

[T3sl4co1l]

How precise?  What's the minimum input pulse width?  Pulse repetition rate?

Should there be a minimum output-low pulse width too, or can that be interrupted by an incoming rising pulse at any time?

How much energy/power can be afforded, and in what states?  (Should the high or low state be lower energy?)  What are the statistics on the input signal?

The most basic circuit is an RCD network followed by a Schmitt trigger.  The R and C are in parallel, to GND.  The diode comes from the input, so that an active input always keeps the capacitor topped up.  If the capacitor is allowed to discharge, then it eventually crosses the Schmitt trigger threshold, and the output changes state.

Tim

[ataradov]

74AHC123 is a  retriggerable monostable multivibrator, exactly what you are looking for. I'm not sure about output staying high, you will have to investigate this part.

[scnoob]

Ah yes, here are some further details:

-In/out highs voltages are somewhat flexible, I am currently using a 20v high but I could bump that down to 12v if necessary.  Low is 0V ground (or close to it). 
-Input pulses are unfortunately rather random.
-In theory, the smallest pulse I could get is around 2 microseconds in duration, however my application is not so sensitive that missing one of those would be an issue.  In practice, I really only need to catch pulses longer than 50milliseconds.
-My proposed value of x would be very long....minimum 250ms up to 3 full seconds.
-I am unfortunately not privy to the detailed specs of my power supply, however I have a total power output of 20v, 1A to work with.  A good chunk of that 1A has already been used up by other components, and I will be having 5 of these circuits on my board.  I would hope for <50mA each draw at worst.
-Low state should be the most power efficient.
-The output low pulse width should have no minimum duration.  The idea is to drive output HIGH ASAP on a high input, then hold it there until the end of x.

[scnoob]

@CM800  I will investigate this circuit's characteristics further, it looks promising
@Jonovid  The 555 unfortunately doesn't retrigger on subsequent pulses.  I'll look into the pulse stretcher transistors some more.
@ataradov  Unfortunately the 74AHC123 is triggered on the trailing edge.  I would need a leading edge trigger.

Thanks all for your input so far

[ataradov]

Unfortunately the 74AHC123 is triggered on the trailing edge.

There are two inputs - direct and inverse.

[Benta]

For a 12 V circuit, just use the good old CD4538 or MC14538.

Works every time.

Benta.

[Zero999]

The 555, 74HC123 and CD4538 are no goers. The 555 goes high on negative edge and remains low until the end of the cycle. If the trigger pulse is low for longer than the cycle, then the output will remain high, until the trigger goes low again. The 74HC132 and CD5438 will trigger on negative/positive edge and will be unaffected by the input remaining high/low.

The 74HC14 is probably your best bet is not very accurate. Another option is the LM393, which will be more accurate, but it requires more components.

[StillTrying]

A 556 would do it.

[T3sl4co1l]

74HC7014 has very accurate thresholds, though it's only available in hex packages (74HC14 is available down to singles!).

50mA is basically unlimited power, as far as logic is concerned.  This is not "least power" territory!

A circuit with two or three transistors, or a single comparator (two comparator halves is kind of overkill, but they are pretty much for free, so who cares), can also be made to interface directly with the stated 12-20V logic level.

What's the logic source, anyway?  Is it a constant voltage source, open collector, high resistance, what?

Tim

[Gixy]

What about a tiny micro-controller? With internal oscillator, at the end probably less components than every discrete solution, and completelly adjustable by software.

[alanbc]

A 556 would do it.

I have a (septic) pump that's triggered by a switch inside a float.  Level rises, float rises, switch closes, pump turns on, level falls .. a wee bit, just enough to lower the float .. the pump turns off.

I want some hysteresis; switch closes, pump turns on, switch opens, pump STAYS on for another 30 seconds.

Timing is not that critical .. 25 seconds, 35 seconds, close enough.  A uController is certainly an option and as a long time uController programmer, is how I'd usually do it.

HOWEVER, this is a mission critical application, and so the simpler the better; simpler still, better still.  555s are about as simple an option as could be, and are pretty well understood.  Add a solid state relay, pretty much done!

But as a guy who built his first 555 circuit about 40 years and has built a whole lot over the years, I am embarrassed to admit I don't know how this circuit works!

I've simulated it and it does seem to work; I can reproduce the traces shown here.

But 555s are triggered on a negative going pulse.  These traces show the output of the second 555 going high when the input goes high ...

What am I missing?


Alan

[StillTrying]

But 555s are triggered on a negative going pulse.  These traces show the output of the second 555 going high when the input goes high ...

What am I missing?

With TRIG and THRES connected together a 555 is just an inverting schmitt trigger with switching levels of 1/3 and 2/3 of Vcc.
So there's 2 inversions.

When the incoming pulse reaches THRES the DIS charge transistor discharges the 100n cap to 0V which TRIGs the 2nd 555 and makes it's output go high.
When the input pulse drops below TRIG , DIS is released and it takes time for the cap to recharge up to THRES to make the 2nd 555's output go low.

The output pulse width is always (input pulse width + delay time), so a bit rough.

[alanbc]

But 555s are triggered on a negative going pulse.  These traces show the output of the second 555 going high when the input goes high ...

What am I missing?

With TRIG and THRES connected together a 555 is just an inverting schmitt trigger with switching levels of 1/3 and 2/3 of Vcc.
So there's 2 inversions.

Thanks for this; I get it now.

But oh my goodness, after all these years, I've found yet another use for a 555 and another way to wire one up.

If I ever chance to meet the guy who designed this thing, I'll get down on my knees and bow.

Thanks again.

[basinstreetdesign]

This seems simple to me but I can't seem to find a proper circuit diagram to do what I need.  I've attached a timing diagram of my input waveform and the output waveform that I need to get.  Here are the basics:

1. I need to extend the width of an input pulse for a duration of x.  (I may need to adjust x later in the design process)
2. The output needs to restart the x timer no matter how many input pulses are given.
3. If the input stays high for a long time, the output must also stay high until the input goes low PLUS extend the pulse by x.
4. Simplicity is key, I need to use as few components as possible in a limited physical space.
5. Power utilization is a major consideration...This should be a minimal waste energy circuit.

Thanks in advance!

Yup, as has been said the venerable 4538 CMOS one-shot will do it.  It is retriggerable and can be triggered on either rising or falling edge.  Since 4XXX logic (NOT 74HC4XXX) will work from any Vcc voltage up to 18V, is cheap and still sold (at least many types are) and since you are not fussy about propagation delays or precision timing it looks like it will fill the bill.  You do need two other gates, though, an OR gate and an output buffer.

Run the input into the negative-going trigger input of the 1-shot.  It will fire at the trailing edge of the input pulse.  Then combine the input with the 1-shot output to derive the output signal.  The first gate does the combining.  The little filter between the gates is there to filter out a small spike equal to the propagation delay of the 1-shot, about 200 nS or less.  The 1-shot timing is set with a R-C pair whose time constant is the same as the extension you want.

So 1/2 of a 4538, 1/2 of a 4001, two resistors and 2 caps, total.  Dirt cheap and you have 1 1-shot and 2 gates left over to use elsewhere.  Also draws squat power (< 1 mA).

[Zero999]

Just use the 74LVC2G07.
https://docs.rs-online.com/82a8/0900766b815f556b.pdf