Shadow signal

[abuhafss]

Hi

I need to generate a shadow signal of a (variable frequency) square wave such that it is an exact copy of the original signal and that the pulse is generated as soon as the original pulse falls to zero volts. The diagram below interprets the requirement:


The red pulses are the shadow of the original black pulse.

How can that be done?

 

[Arouse1973]

What about a mono-stable that works on the falling edge of the origional signal?
Adam

 

[abuhafss]

What about a mono-stable that works on the falling edge of the origional signal?
Adam

As mentioned in the first post, the frequency is variable. How will the monostable get the information about the width of the original pulse?

 

[KrisBlueNZ]

What is the range of pulse widths?

What is the minimum and maximum pulse repetition frequency?

 

[abuhafss]

What is the range of pulse widths?

What is the minimum and maximum pulse repetition frequency?

10 Hz to 250 Hz

 

[KrisBlueNZ]

Can you answer my other question as well?

Also, how accurate do you need the second pulse duration to be? What percentage of timing error is allowable?

 

[abuhafss]

Can you answer my other question as well?

Also, how accurate do you need the second pulse duration to be? What percentage of timing error is allowable?

The duty cycle remains 20% hence, the pulse width changes accordingly with the change in frequency.
An accurately timed pulse is preferable however, a minor timing error can be acceptable. To what extend, that has to be tested.

 

[abuhafss]

BTW, I have been working on this configuration. The width of the copy pulse is perfect but after three four pulses the copy pulse starts overlapping the original.................may be that is due to timing error. I can't solve that issue.

Moreover the height of the pulse is smaller but, that is tolerable.

 

[abuhafss]

An accurately timed pulse is preferable however, a minor timing error can be acceptable.

However, the copy pulse should start after the end of original pulse NOT before that. And I estimate that a 5% timing error would be acceptable.

 

[KrisBlueNZ]

OK well I have no idea how your circuit is supposed to work. I think the simplest approach would be to charge a capacitor with a constant current during the input pulse then discharge it with an equal current until it hits the initial voltage, to determine the width of the pulse. There are various ways to do it. I'll have a think and get back to you.

 

[abuhafss]

OK well I have no idea how your circuit is supposed to work.

The lower two boxes shows two identical oscillators with fixed duty cycle. The frequency is controlled by pot R2 in the left box and R7 in the right box (R2 and R7 are ganged). The right side oscillator is powered on as soon as the left oscillator's 1st pulse ends. Thus the right oscillator's output is delayed for the time equal to the width of the first pulse of the left oscillator.

 

[KrisBlueNZ]

Oh, OK. Sorry, "I have no idea how your circuit is supposed to work" was a bit insulting!

 

[abuhafss]

Oh, OK. Sorry, "I have no idea how your circuit is supposed to work" was a bit insulting!

Absolutely not!
I truly respect you..

 

[Arouse1973]

As mentioned in the first post, the frequency is variable. How will the monostable get the information about the width of the original pulse?

You could possibly use a ganged pot that adjusted both timings together.
Adam

 

[Anon_LG]

I could say use a micro controller but that may get me laughed at...

 

[(*steve*)]

I could say use a micro controller but that may get me laughed at...

Not by me.

Depending on the range of frequencies and pulse widths it may be the best answer. It also might be totally unsuitable for other combinations of frequency and pulse width.

Kris really nailed down the factors for deciding on a solution when he asked for the range of frequencies and pulse width.

 

[Laplace]

I could say use a micro controller

This is an ideal function for a micro-controller; however, the learning curve for using a micro-controller is quite steep. But if one is over that hump then the whole thing, including generation of the original pulse train and the delayed shadow pulse, could be simply implemented with a one-chip micro and not much else. Otherwise I would look at using an analog integrator where the first pulse charges the capacitor, and the absence of the pulse discharges the capacitor. The discharge cycle is equivalent to the shadow pulse. When the capacitor has been discharged the integrator will enter a quiescent state (no charging or discharge current) until the next pulse arrives. If the analog approach cannot maintain sufficient accuracy over a 25:1 dynamic range, then a digital approach using a binary UP/DOWN counter could be used (where the binary count is equivalent to charge in the analog integrator). That is actually what would be done with the micro-controller approach, and it is simpler to implement the function in code than in circuitry.

 

[KrisBlueNZ]

Absolutely not!
I truly respect you..

Thanks But I should have been more diplomatic.

Yes I agree that a microcontroller would be appropriate, but I'm working on the capacitor charge/discharge method that I mentioned in post #10 and Laplace described in more detail in post #17. When I finalise and post my circuit, perhaps Laplace could see if he can suggest any improvements.

 

[Rleo6965]

You can add 2 flip flop ic and an inverter to create that shadow waveform.

 

[(*steve*)]

This is an ideal function for a micro-controller;

Or a hybrid solution where the falling edge of the input waveform triggers logic to produce the rising edge of the output waveform. The duration of the input pulse pulse and the resetting of the output would be under the control of the microcontroller.

Of course, you need to define what happens if there is a long input pulse and then a shorter input pulse which ends *during* the shadow pulse.