Hi, I'm trying to generate a signal having different pulse widths but same OFF time. For example like this one:

I know such signal can be easily obtained by a microcontroller but my aim is to use discrete components.

First idea I came up with is using 555 timer and a multiplexer and counter. It seem to be working in theory but I couldn't obtain such a waveform. Also my intention is to have a long signal such this so I don't know what to use for that. Could you give any idea about obtaining such a waveform.
Thank you

 

Is the pattern always the same?
You can use a Johnson counter such as CD4017.

 

Does the pattern need to be random?
If not, how many different pulse widths do you need?

 

Pseudorandom sequence generator?
https://www.maximintegrated.com/en/design/technical-documents/app-notes/1/1743.html
You can use a single CD4040 instead of the all the 74LS74s - it looks like this might be an old application note!

 

Is the pattern always the same?
You can use a Johnson counter such as CD4017.

No pattern is not the same. Thanks for your comment

 

Does the pattern need to be random?
If not, how many different pulse widths do you need?

Yes pattern has to be random and I guess 5 different pulse widths enough

 

Something like the LTspice simulation below of a pseudo-random generator, such as Ian0 suggested, with an added two flip-flop one-shot to give a low output that's never more than one clock-pulse period in width?
It requires 5 IC packages plus whatever may be needed for a clock generator.

 

It requires 5 IC packages plus whatever may be needed for a clock generator.

You can probably use a 4060, but it might not have the outputs that you need pinned out.
There's a table in Horowitz and Hill showing where the feedback points need to be for various different lengths. I'll scan it for you this evening.

 

You can probably use a 4060

My circuit is a shift-register whereas the 4060 is a counter, so it's not clear how that would work.

 

My circuit is a shift-register whereas the 4060 is a counter, so it's not clear how that would work.

You’re absolutely right. I was thinking about multiple-stage shift registers and looked up multiple stage counters.
And the part I was thinking of is a 4006.
(Damn! I could have explained that away as a typo)

 

Since you want one phase of the output to be a constant pulse width while the other phase varies, two thoughts:

1. Rework the circuit in post #1. If the output of the 555 clocks the counter, you will get a four pulse repeating cycle. Bigger mux, longer cycle, but always a fixed pattern based on the resistor values. A variation of this eliminates the mux; the counter outputs drive the resistors.

2. Use any of a zillion PR circuits to generate a bit pattern where both phases vary in width (such as post #7), and have that signal fire a true monostable. The monostable will guarantee a fixed pulse width whenever triggered, and the time between triggers will be as random as you have money to throw at the PR source. Sunspots are nice.

BTW, the #7 circuit is a linear feedback shift register (LFSR), the cornerstone of many pseudo-random circuits and systems. National Semiconductor used to make an 8-pin IC that did this.

https://en.wikipedia.org/wiki/Linear-feedback_shift_register

I should have asked sooner - is this school work?

ak

 

Is this not a PWM stream ?
or do you need a set of unique patterns to detect,

If its a number of patterns look at the Walsh transform,
For any given length of sequence ( power of two is easiest )
it provides a bunch of orthogonal sequences.

 

Since you want one phase of the output to be a constant pulse width while the other phase varies, two thoughts:

1. Rework the circuit in post #1. If the output of the 555 clocks the counter, you will get a four pulse repeating cycle. Bigger mux, longer cycle, but always a fixed pattern based on the resistor values. A variation of this eliminates the mux; the counter outputs drive the resistors.

2. Use any of a zillion PR circuits to generate a bit pattern where both phases vary in width (such as post #7), and have that signal fire a true monostable. The monostable will guarantee a fixed pulse width whenever triggered, and the time between triggers will be as random as you have money to throw at the PR source. Sunspots are nice.

BTW, the #7 circuit is a linear feedback shift register (LFSR), the cornerstone of many pseudo-random circuits and systems. National Semiconductor used to make an 8-pin IC that did this.

https://en.wikipedia.org/wiki/Linear-feedback_shift_register

I should have asked sooner - is this school work?

ak

Thank you all for the responds and interests. (sorry for the drawings I edited the screenshot in Paint) :


In fact I'm planning to use such a 'weird' signal to a generate a signal like above. The ON times of the square waves at a specific frequencies are determined by signal and the OFF times are constant. And the there may be, I don't know, 5 different pulse widths for timing and signal should be much longer.

Again I know it's super easy to generate it with a microcontroller, some code, pwm signal and constant delay but my aim is to use discrete components if it's doable.

 

Hi, I'm trying to generate a signal having different pulse widths but same OFF time. For example like this one:View attachment 238583

I know such signal can be easily obtained by a microcontroller but my aim is to use discrete components.

First idea I came up with is using 555 timer and a multiplexer and counter. It seem to be working in theory but I couldn't obtain such a waveform. Also my intention is to have a long signal such this so I don't know what to use for that. Could you give any idea about obtaining such a waveform.
Thank you
View attachment 238584

You need to create a controllable counting circuit, first.

 

Use any of a zillion PR circuits to generate a bit pattern where both phases vary in width (such as post #7), and have that signal fire a true monostable.

U8 and U9 perform the monostable task as a digital one-shot.

 

At the end of the day, an MCU solution is a zillion times easier to implement.

 

At the end of the day, an MCU solution is a zillion times easier to implement.

Hello sir I need some help in my studies. If you can help kindly let me know where I can contact you.

 

Hello sir I need some help in my studies. If you can help kindly let me know where I can contact you.

We don't do private tutoring. Post your questions in Homework Help and someone will try to guide you with your work.

 

U8 and U9 perform the monostable task as a digital one-shot.

Missed that. Nice.

Any reason the 6-bit shifter couldn't be replaced by a single-chip 8-bit SIPO?

ak

 

Maybe use a CD40174B HEX-D to reduce the DFF chip count in post #7?