Delaying a periodic pulse signal while preserving the voltage levels (no microcontrollers)

Thread Starter

Jeffrey Chen

Joined Jul 6, 2019
3
Hi there,

The input signal is a periodic pulse signal with an unknown DC offset and changing pulse amplitudes (f=10Hz). I want to delay the pulses by ~tens of microseconds without using a microcontroller. Common delay circuits e.g. with 555 timer IC won't work because while it does preserve the correct logic, it rescales LOW to ground and HIGH to Vcc.

Any ideas? Would an all-pass filter work?

Cheers
 

ericgibbs

Joined Jan 29, 2010
18,766
hi JC,
Welcome to AAC,
Does the F=10Hz have a 50/50 high low period.?
What amplitude input range and do you want the delayed pulses to have the same amplitude.?
E
 

Thread Starter

Jeffrey Chen

Joined Jul 6, 2019
3
hi JC,
Welcome to AAC,
Does the F=10Hz have a 50/50 high low period.?
What amplitude input range and do you want the delayed pulses to have the same amplitude.?
E
The duty cycle is quite low ~0.1-0.3%. Amplitude input range 0-9 V and yes the delayed pulses should have the same amplitude.
 

Analog Ground

Joined Apr 24, 2019
460
You have probably already considered using passive delay lines with multiple stages. Data Delay Devices has parts to get you 2 microseconds per stage.

For example, http://www.datadelay.com/datasheets/2214.pdf

I think a lot depends on the pulse characteristics and bandwidth you need. Pulse rise and fall times, etc. Could you post the pulse characteristics? Required timing and amplitude accuracy? Do you need the delay to be adjustable? If so, range?
 

Analog Ground

Joined Apr 24, 2019
460
In answer to your question about all pass filters, I recall their function when used with pulses is to provide equalization of a transmission path. So, they will probably distort the pulse quite a bit.
 

MisterBill2

Joined Jan 23, 2018
18,167
A transmission line could delay the pulses but there are losses, since the same energy gets stretched over a longer time interval. A sample-and-hold circuit triggered by the pulse arrival could do it, but to very accurately preserve the amplitude and timing gets complicated. One possibility is a string of unity gain amplifiers that have fairly good frequency response, but not fantastic frequency response. What property of the pulse must be preserved most accurately? Amplitude, width, rise time, or fall time? That affects which options may work most easily.
 

Analog Ground

Joined Apr 24, 2019
460
My crazy idea of the day. A/D converter->FIFO memory->D/A converter all controlled by an FPGA. Potentially very accurate and adjustable but very, very complicated!
 

Thread Starter

Jeffrey Chen

Joined Jul 6, 2019
3
Can you describe the purpose for the delay, why its needed ?



Regards, Dana.
Hi there
The signal is the output of a phototransistor sensor. I want to create a delayed copy to explore what an instrumentation amplifier can do in terms of rejecting common mode interference (ambient light).
Cheers
 

MrAl

Joined Jun 17, 2014
11,388
Hi there
The signal is the output of a phototransistor sensor. I want to create a delayed copy to explore what an instrumentation amplifier can do in terms of rejecting common mode interference (ambient light).
Cheers
Hi,

Just a quck idea.
A digital delay (couple of inverters and cap and resistor) followed by an open collector output stage, where the Vcc of the collector is given by the pulse itself but stretched, so you would also need a pulse stretcher that could stretch the pulse by the same amount (or a little more) as the delay.
The offset would be hard to duplicate though. Maybe a comparator with open collector and emitter, where you would have to stretch the low part of the pulse also maybe with a sample and hold to feed the emitter.
 

absf

Joined Dec 29, 2010
1,968
Analog bucket brigade aka CCD, Charge-Coupled Device.
Those are actually called BBD. It is still available on eBay but very costly.

They were usually used in pairs. 3009 id BBD and 3010 is the clock generator for the BBD.

s-l400 (3).jpg
 
Last edited:

absf

Joined Dec 29, 2010
1,968
My crazy idea of the day. A/D converter->FIFO memory->D/A converter all controlled by an FPGA. Potentially very accurate and adjustable but very, very complicated!
No so crazy idea....:D

This technique was already employed in the PT2399 chip. It was used as an echo and reverb effect in guitar and voice systems.

See P1 and P4 of datasheet attached
 

Attachments

MrAl

Joined Jun 17, 2014
11,388
Hello,

Here is the result of a square wave being passed through an All Pass filter that has a linear phase vs frequency response.
Note the phase shift in this case is negative, and a delay is achieved, however the amplitude is slightly tilted.

The blue plot is a square wave the way it would appear with no filter with harmonics from 1 to 55.
The red plot is a square wave passed though the filter, with harmonics 1 to 55. The top is tilted.

The delay is dependent on the phase shift, and so is the tilt. However, small phase shifts do not affect it too much. The rise time is affected slightly.

Also note this is for a continuous input. There may be exponentials not present in the plot that would be due to a changing amplitude and/or pulse width and/or pulse position.

PhaseShiftedSquareWaveComponents_neg-PhaseShift-1.gif
 
Top