# Easy ways to turn square/saw wave into narrow pulse?

#### devinw

Joined Dec 2, 2018
32
I know I can use an op amp comparator, but perhaps there is an easier way with a capacitor or other method? I would like the input to be a square or sawtooth wave centered about zero, about 10vPP and the output to be a very narrow positive-going pulse (say ~5% Duty) of the same frequency. I've found some cap/transistor circuits that do what I want but provide a negative-going pulse that rests at high V+, but I want the inverse.

Any help would be appreciated!

#### crutschow

Joined Mar 14, 2008
25,123
What is the signal frequency range?
Does it vary?

#### devinw

Joined Dec 2, 2018
32
This would be audio, probably even less. Say 10Hz to 10kHz.

#### wayneh

Joined Sep 9, 2010
16,398
I've used a simple capacitor coupler to light an LED for a brief pulse at the beginning of a longer square wave pulse. The pulse that's created is in the shape of an RC curve and is not digital, but maybe it's good enough? In other words, what do you need from the truncated pulse?

#### devinw

Joined Dec 2, 2018
32
How would that work? If you put in a 50% duty square through a capacitative coupler you get the same wave out, minus any DC offset right? It wouldn't do anything to shorten the pulse.

#### wayneh

Joined Sep 9, 2010
16,398
How would that work? If you put in a 50% duty square through a capacitative coupler you get the same wave out, minus any DC offset right? It wouldn't do anything to shorten the pulse.
Like this. The square wave enters upper left, ground at lower rail. The positive going edge of the square wave lights the LED as C4 charges. Max current set by R5 and time set by C4. R17 discharges the cap when the input wave goes low, to ready it for the next pulse. The component values are fixed, so this arrangement only works within a fairly narrow frequency range, maybe 10X or so. I haven't really tested that.

#### BobaMosfet

Joined Jul 1, 2009
1,118
I know I can use an op amp comparator, but perhaps there is an easier way with a capacitor or other method? I would like the input to be a square or sawtooth wave centered about zero, about 10vPP and the output to be a very narrow positive-going pulse (say ~5% Duty) of the same frequency. I've found some cap/transistor circuits that do what I want but provide a negative-going pulse that rests at high V+, but I want the inverse.

Any help would be appreciated!
schmidtt (sp?) trigger

#### ronv

Joined Nov 12, 2008
3,770
I know I can use an op amp comparator, but perhaps there is an easier way with a capacitor or other method? I would like the input to be a square or sawtooth wave centered about zero, about 10vPP and the output to be a very narrow positive-going pulse (say ~5% Duty) of the same frequency. I've found some cap/transistor circuits that do what I want but provide a negative-going pulse that rests at high V+, but I want the inverse.

Any help would be appreciated!
What direction does the fast edge of the sawtooth go?

#### devinw

Joined Dec 2, 2018
32
It would be like this:

#### crutschow

Joined Mar 14, 2008
25,123
Can the pulse-width be constant or do you want it to vary with the frequency?

#### devinw

Joined Dec 2, 2018
32
Constant is good crutchow

#### AnalogKid

Joined Aug 1, 2013
8,489
the output to be a very narrow positive-going pulse (say ~5% Duty) of the same frequency.
A constant output duty cycle over a 3 decade frequency range is not trivial, but can be done with two comparators configured as a window comparator. 5% of 10 V (the peak-to-peak input waveform amplitude) is 0.5 V. Two inverters set to trip 0.5 V apart will combine to form an output pulse that is 5% of whatever the input period is as long as the input amplitude is constant over the frequency range.

1 - LM339 running on +5 V and GND
1 - 2-resistor divider to create a trip point voltage of +0.5 V
1 - output pull up resistor
1 - input protection resistor
1 - decoupling capacitor

One comparator uses GND as it's inverting input reference, and one uses +0.5 V as its non-inverting input reference. The two remaining inputs are connected together to the input through a resistor to limit the below-GND input current into the comparator input stages.

ak

Edit: Just saw your response to Wally. Constant is ok if you're "that" kind of person ...

Or, with a capacitor and a PNP ...

#### devinw

Joined Dec 2, 2018
32
Awesome, AnalogKid! I am not talking “that” kind of person haha! Let’s see the PNP capacitor solution!

...close enough for government work in other words

#### ronv

Joined Nov 12, 2008
3,770
Awesome, AnalogKid! I am not talking “that” kind of person haha! Let’s see the PNP capacitor solution!

...close enough for government work in other words
Here is a poor mans 1 ic circuit.

Joined Feb 20, 2016
3,566
If you want a really short pulse...
First, scale the signal to 0 to 5V then feed the signal into a Schmitt trigger to convert it to a rectangular wave. This output goes to one input of a 2 input AND gate and to the other other input via an odd number of inverters. This will produce a very short pulse, depending on the delay on the string of inverters.
Using various logis IC families can give dirrerent delays. And adding an R/C network in line with the inverters will lengthen it.

#### MrChips

Joined Oct 2, 2009
21,341
You can use a monostable multivibrator IC such as 74LS123 or CD4528.