I have a need for a circuit which will convert a 1 pps 200millisecond pulse to a 1pp 20 microsecond pulse. I wanted to use an op amp with an integrator but couldn't find a circuit. Voltage is TTL (5vdc) and I plan on using a 9 volt wall wort for power.
\
Any help will be greatly appreciated.
PM

 

Keeping the same PRF of 1Hz makes it rather simple ...use a one-shot. A 555 timer can be configured as such (see apps in datasheet), or use a design specific IC such as the 74LS123.

Using this type of IC the 200ms wide pulse can easily be reduced to 20us, but still occur at a 1Hz rate.

The 555 can operate from 9V, and then divide the output down to TTL levels. Otherwise, the 9V will need to be regulated down to 5V first.

 

Here's a somewhat dodgy circuit, but it's simple enough to build and test with inexpensive components.

Note that the 74LS132 can be just an "S" type, but it must have the Schottkey inputs. You could use a different Schottkey-input gate if you like, this was just the first one I came across.

C1/R1/R2 form a differentiator circuit, it's output is a series of relatively narrow spikes. The negative spike is clamped to ground via D1. The Schottkey input detects the slow-falling low level, and squares up the waveform. U1A's output is inverted, U1B gives a positive 20uS pulse.

I chose to use one fixed and one variable resistor for R1/R2 to allow for fine adjustments in the Spice simulator. You may choose to use a 10K pot for R1, and a 500 Ohm pot for R2 to allow very fine adjustments.

Be aware that you'll wind up with propagation delays by running through a couple of gates. Consult your datasheets for the numbers.

 

Hmm - I don't think a 74xS122/123 will work, because it'll be continuously retriggered by the 200mS pulse. Same problem with the 555 timer. But, a 555 timer could be controlled using a differentiator circuit similar to the one I posted; however it would need a level translator circuit as well.

 

Sir,

You're definitely on the right track with the integrator. You should research both the integrator and comparator. The integrator shouldn't be the standard 'ideal' circuit, but one that considers input offset voltage. You will recognize this circuit as an opamp with a resistor to the negative terminal, and a capacitor in parallel with a resistor in the feedback loop. You should tie a resistor equal to the input resistance to ground from the + terminal.

So, you can get a triangular waveform from your integrator if you set it up properly. Just about any opamp will meet your slew rate and small signal bandwidth requirements. Then, just feed this into a comparator with the reference set up the slope to where you want it. Use a potentiometer with a bypass capacitor for a clean setpoint, then measure with the scope to ensure you have the right pulse width.

You might have a bit of inaccuracy via temperature drift, but it should be okay for most stuff.

Steve

 

SgtWookie
Thanks for the practical approach; the LS132 might just be what I was looking for. It's been a while since my digital logic 101 days but I do remember the retriggerable constraints on the 555. I think the schottkey is the answer. I also have another box with a low voltage that can be stepped up to TTL level using the same type LS132 chip.

Also thanks for introducing me to 5SPICE application. It looks like a winner. Ed

 

The LS123 is not going to do it for you. Its a retriggerable mono so all it will do is stretch your pulse by a further 20uSec.

What you need to use is an edge triggered non-retriggerable mono e.g. the 74121 should do the trick.

 

It will retrigger and increase the pulse width ONLY if the retrigger pulse occurs within the time-constant of the output.

The attachment LS123_sym displays the sym schematic using a 470pF and 60.4K timing resistor. The input is a 200ms pulse at 1 Hz (with 100ns rise/fall times).

The attachment LS123_full is the output file displaying the results, where:
blue = input, 200ms pulse @ 1Hz
red/green = outputs, 20us pulse @ 1HZ

The attachment LS123_ex expands the output displaying a 20us pulse width between cursors.

 

Here is a variation on the approach used by sgtwookie. This circuit uses an RC integrator to determine the new narrower pulse width.

The diode serves to rapidly discharge the capacitor and ready things for the next occurrence of a positive going pulse. This is not likely to be a problem with your 1Hz pulse repetition rate so you could forego the diode if you so chose.

hgmjr

 

I was looking at sgtwookies approach and I could see the integrator charging a couple of TC's before the first gate turned on but I couldn't see how it would turn off until the input pulse decayed to baseline. I wasn't sure if I was missing something there. Your circuit ANDS the integrated pulse with the original and will output my desired result. Width is adjustable with the R or C. This is what I was looking for. I didn't want any delay. Looking at sgtwookies ckt, I can see a gate reaching transition level but I didn't see the 20 microsecond pulse starting at Tzero. I will breadboard and tweak as soon as I can get to it. Thanks to the forum; it's only as good as it's contributors! Ed

 

Hello again Ed,
Here's the circuit with more "virtual O-scope" data points taken (see attached)

The effects of the leading edge of the pulse are passed through to the 74LS132 gate U1A via C1, giving a very sharp edge. The R1/R2 network drain the voltage level towards zero volts. As mentioned before, D1 is there to clamp the negative transition edge of your input pulse to ground, to avoid damaging the input to the gate.

Note that there will be a cumulative propagation delay due to the shaped pulse passing through two NAND gates (U1A, U1B). You could likely reduce this considerably by using a single Schottkey-input AND gate; check the spec sheets of the gates you're considering.

The input signal toggles between 3.3v and 0.8v, as those are typical for TTL.

I used a program called CircuitMaker 6 Student, you can find it available for downloading on the Web by doing a search.

hgmjr,
I tried modeling your proposed modification, but it didn't work. What program were you using to model your circuit?