Variable threshold on sawtooth thru comparator with fixed output-pulse at 1μs

Thread Starter

rawbot

Joined Sep 20, 2016
22
Hard to come up with a fitting topic, but it do explain it well. :)

Hi everybody,
First time post on this amazing forum.

So here is my issue -
I have a sawtooth going thru a comparator (made with a TL072) with variable threshold with a potentiometer. Just to be able to change the timing/delay of the output-pulse regarding to the saw. Simple stuff.

The following part is what im not sure of yet.

As the threshold changes, so does the width of the pulse obviously. I need the width to be of a fixed value of 1μs however, so I've tried using a 555 timer for this, but after reading the datasheet i discovered it isn't recommended using a lower value than 11μs, so i ditched that idea and made it all with discretes instead.
Thats a big part cost though, and i want it as simple and with made with as few parts as possible.
The CD4098 is my next option to try, as it can produce smaller widths.

Anyway, it feels like im overdoing it by using two IC's and a bunch of external parts just for this.
It feels there must be a easier way...?
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
Forgot to mention im using a +/-12V supply, so certain IC's can't be used.
Another word for what im trying to do could be PPM/Pulse Position Modulation. I'd love to hear about people and their experience with this.
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
What is the required output pulse amplitude? Unipolar or biolar?
What is the allowable tolerance on the 1uS pulse width?
What is the sawtooth frequency/amplitude?
It would be great if it was -4V to 0V but nevermind that. I can shape whatever voltages later. Current is not important as it will be used for a FET.
Tolerance +/- 10% preferably.
Sawtooth is 0-7.5V, and freq is not fixed as it comes from a oscillator, as a part of a synthesizer. Hence the decision on using the saw as a scanned guidline for the pulse.
 

AnalogKid

Joined Aug 1, 2013
10,986
A single CD4093 quad Schmitt trigger NAND gate can perform both the variable oscillator and true monostable output functions in one 14 pin package. Frequency stability over temperature will not be as good as a 555, but maybe good enough for your application.

One gate is a hysteretic oscillator (with an enable input), two gates form the monostable (with both Q and -Q outputs), and one gate is a spare for something else, like inverting the oscillator enable signal. I am without schematic software temporarily, but I'll see what I can whip up.

ak
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
Thanks for the interest.
Havent had time to try the CD4093 yet, but this is what im playing with at the moment.

Note -
The active highpass is there as a try to even out the amplitude changes caused by the variable PWM.
Not sure yet if im going to go that way or another, so here is 2 images done with the 555 timer.
 

Attachments

Thread Starter

rawbot

Joined Sep 20, 2016
22
The more i think of this, the more of a problem it seems to be, regarding the frequency-window this thing should work on.

I mean, the CR loading times will vary too much between low freq and high, and not only between 1% - 99% PWM.

Perhaps it's possible doing this in another way without using capacitors for the pulse-out width?

I have a vague memory than I've seen a circuit using only transistors, but can't find it now.

Anyway, any kind of help would be really appreciated!
 

AnalogKid

Joined Aug 1, 2013
10,986
10 Hz to 10 kHz is a 1000:1 frequency span. If you want to cover that span in a single range, as opposed to a range switch, you will have poor adjustability and repeatability no matter how you do it. Is the goal to cover the entire span with a single turn of one knob?

ak
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
10 Hz to 10 kHz is a 1000:1 frequency span. If you want to cover that span in a single range, as opposed to a range switch, you will have poor adjustability and repeatability no matter how you do it. Is the goal to cover the entire span with a single turn of one knob?

ak
I guess using a octave-switch set to different CR-values would be better, but i would love to be able to create a 1μs pulse out of the PWM'ed sawtooth no matter what frequency i set. Must be a way..
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
It's for syncing a saw-core oscillator.
I could have made it easy and use its own pulse from its schmitt trigger section, but i wanted to change it up a bit.
 

AnalogKid

Joined Aug 1, 2013
10,986
I guess using a octave-switch set to different CR-values would be better, but i would love to be able to create a 1μs pulse out of the PWM'ed sawtooth no matter what frequency i set. Must be a way..
The constant 1 us pulse regardless of repetition rate is the easy part. The frequency span vs adjustability is the decision that affects usability.

ak
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
The constant 1 us pulse regardless of repetition rate is the easy part. The frequency span vs adjustability is the decision that affects usability.

ak
Just to make sure i follow you.
Are you talking about making an oscillator as well as making the pulse?
I actually already have the oscillator made, so the only thing needed is to make this narrow pulse working properly.

Feel free to let me know if you have any ideas about that part. As said, i would highly appreciate it as i feel out of ideas at this point. :)
 

Alec_t

Joined Sep 17, 2013
14,280
It's for syncing a saw-core oscillator.
Never heard of one of those, but many sync arrangements respond to the leading edge of a sync pulse and aren't too fussy over sync pulse width. If that's the case, a simple RC circuit could provide a sync pulse.
 

AnalogKid

Joined Aug 1, 2013
10,986
Here is a classic monostable circuit. Output pulse width is approx. 0.7 x R2 x C1.

Because it has positive feedback, the output pulse width is independent of the input pulse width. So the input can be 100 ns or 100 s, and the output will remain 1 us. I prefer using a CD4093 for RC timing circuits because of the Schmitt trigger input stages, but that is more important for long pulse widths like 0.1 s or more.

This is not my schematic, but it covers the basics. If you need a negative output pulse, you can get it at the left gate (U1A) pin 3. If you have a negative-going trigger edge, you can use one of the spare 4001 gates as an input inverter. Note that they used the same pinout for both gates, while a real part has 4 independent gates (U1A, U1B, etc.).

If you do use a CD4093, then the input and output polarities change and R2 goes to GND. Also, one gate in a 4093 can be the oscillator.

ak
Monostable-1.jpg
 

Thread Starter

rawbot

Joined Sep 20, 2016
22
Never heard of one of those, but many sync arrangements respond to the leading edge of a sync pulse and aren't too fussy over sync pulse width. If that's the case, a simple RC circuit could provide a sync pulse.
Yes, thats usually the case, but not this time. :)

Here is a classic monostable circuit. Output pulse width is approx. 0.7 x R2 x C1.

Because it has positive feedback, the output pulse width is independent of the input pulse width. So the input can be 100 ns or 100 s, and the output will remain 1 us. I prefer using a CD4093 for RC timing circuits because of the Schmitt trigger input stages, but that is more important for long pulse widths like 0.1 s or more.

This is not my schematic, but it covers the basics. If you need a negative output pulse, you can get it at the left gate (U1A) pin 3. If you have a negative-going trigger edge, you can use one of the spare 4001 gates as an input inverter. Note that they used the same pinout for both gates, while a real part has 4 independent gates (U1A, U1B, etc.).

If you do use a CD4093, then the input and output polarities change and R2 goes to GND. Also, one gate in a 4093 can be the oscillator.

ak
Thanks, that worked great!
Now to decide on the offset voltage part. Either an opamp, or something as simple as this could work.
If i could use the last part of the CD4093 for the PWM-section or the offset voltage that would be great, but i cant come up with a solution that really works for that. I guess its not meant to be used like that.
 

Attachments

Top