Maker Pro
Maker Pro

Delaying low voltage pulse

JPilot1

Nov 18, 2022
7
Joined
Nov 18, 2022
Messages
7
Hello all

I would like to delay a pulse waveform by some specific amount of time (a few us). I was looking at this delay line IC:


The issue is that most of these delay lines seem to use TTL logic as the input, however my input signals would instead be small voltage pulses, only ~10 mV amplitude and ~500 ns in width. Does anyone know if this would work? If not, are there any delay line ICs which can delay pulses of this size? Preferably something with an adjustable delay time?

Thanks for reading!
 
Last edited:

dragon

Oct 31, 2022
248
Joined
Oct 31, 2022
Messages
248
2 megahertz is in the range of sonic delay lines. if all u want is a delay (which isnt a huge delay), you dont need any circuitry for that except maybe a transistor on the other side, the delay probably helps if u want to feedback back into the delay from the output, if u dont need to do that just one of those tv delay lines that was on eev blog a while back is all you need. I bet u already know this video but ill put a link to it anyway JIC.
 

danadak

Feb 19, 2021
751
Joined
Feb 19, 2021
Messages
751
Note the bessel filter is not an allpass but exhibits approximate linear phase
over a range of frequencies.

1668856229342.png,

So you would not pass the wave shape unaltered over a wide range
of frequency, you tradeoff delay versus fidelity of pulse.



Regards, Dana.
 

JPilot1

Nov 18, 2022
7
Joined
Nov 18, 2022
Messages
7
Note the bessel filter is not an allpass but exhibits approximate linear phase
over a range of frequencies.

View attachment 57012,

So you would not pass the wave shape unaltered over a wide range
of frequency, you tradeoff delay versus fidelity of pulse.



Regards, Dana.
Thank you very much for the responses. So an allpass filter will delay a signal by some amount of time, which is based on its frequency components. But what is the "frequency" of a single pulse? Is it just 1/width? If I put a PMT pulse:
1669150998723.png
through an allpass filter, what sort of output would I expect? If I change the width by say 50%, do you know what kind of effect that would have on the delay time? And isn't the maximum phase shift 360 degrees meaning the max time delay is one pulse width? But I need a delay equal to 10+ pulse widths...

The other issue is that I have to make 64 of these. Are there any ICs that could simplify this?

Thanks!
 
Last edited:

JPilot1

Nov 18, 2022
7
Joined
Nov 18, 2022
Messages
7
2 megahertz is in the range of sonic delay lines. if all u want is a delay (which isnt a huge delay), you dont need any circuitry for that except maybe a transistor on the other side, the delay probably helps if u want to feedback back into the delay from the output, if u dont need to do that just one of those tv delay lines that was on eev blog a while back is all you need. I bet u already know this video but ill put a link to it anyway JIC.
Thank you for your response Dragon. Are these glass delay lines commercially available? I haven't been able to find them.
 

danadak

Feb 19, 2021
751
Joined
Feb 19, 2021
Messages
751
Thank you very much for the responses. So an allpass filter will delay a signal by some amount of time, which is based on its frequency components. But what is the "frequency" of a single pulse? Is it just 1/width? If I put a PMT pulse:
View attachment 57054
through an allpass filter, what sort of output would I expect? If I change the width by say 50%, do you know what kind of effect that would have on the delay time? And isn't the maximum phase shift 360 degrees meaning the max time delay is one pulse width? But I need a delay equal to 10+ pulse widths...

The other issue is that I have to make 64 of these. Are there any ICs that could simplify this?

Thanks!
A pulse as you know has many frequency components :

1669156557367.png

The freq of a p[use train is 1 / period. An allpass, ideal, will replicate the waveshape of the
pulse. Real allpass are approximations, so amplitude and waveshape are affected. Basically
you make a tradeoff between complexity and fidelity in designing them. The filters we are discussing
are LTI so changing the pulse width does not affect the delay time. But changing the width or duty
cycle does shift the edge in time within the period. But delay thru filter is unaltered. If you change
the frequency of the pulse, then delay thru filter does change.

Phase maximum is a f() of filter order, first order 90 degrees, 2'ond 180 degrees.......

There is more recent findings in filters using wavelet filters, scope of that is lengthy discussion.

Regards, Dana.
 
Last edited by a moderator:

JPilot1

Nov 18, 2022
7
Joined
Nov 18, 2022
Messages
7
A pulse as you know has many frequency components :

View attachment 57055

The freq of a p[use train is 1 / period. An allpass, ideal, will replicate the waveshape of the
pulse. Real allpass are approximations, so amplitude and waveshape are affected. Basically
you make a tradeoff between complexity and fidelity in designing them. The filters we are discussing
are LTI so changing the pulse width does not affect the delay time. But changing the width or duty
cycle does shift the edge in time within the period. But delay thru filter is unaltered. If you change
the frequency of the pulse, then delay thru filter does change.

Phase maximum is a f() of filter order, first order 90 degrees, 2'ond 180 degrees.......

There is more recent findings in filters using wavelet filters, scope of that is lengthy discussion.

Regards, Dana.
Sorry I'm still not understanding. In the graphic you shared the pulses are periodic. Would a single pulse be comprised of equal amounts of all frequencies?

I also tried using the rf-tools.com application you showed and changing the parameters - I see increasing the filter order increases the group delay like you said, but even at the maximum setting of order 20 it is still only ~400 ns. Maybe I'm doing something wrong, how would I get this into the usec range? It would also be prohibitive to build 64 filters, each 20 components.

In regard to distorting the waveform, that is not too important. The only information I need is whether the amplitude exceeded a certain voltage or not.
 
Last edited:

dragon

Oct 31, 2022
248
Joined
Oct 31, 2022
Messages
248
Thank you for your response Dragon. Are these glass delay lines commercially available? I haven't been able to find them.

I dont know where to buy them either, I plan on making mine from scratch using capacitors for the transducers, just needs high voltage, and doing it in compressed air, which has slower travel than in glass (300 m/s), and can still put short pulses through it. u get a 60hz (16ms) delay with just ~11 cm in length. (If the speed of sound in air comes through as valid.) (or im off by a bit maybe) - maybe a fair bit.
 
Last edited:

danadak

Feb 19, 2021
751
Joined
Feb 19, 2021
Messages
751
Sorry I'm still not understanding. In the graphic you shared the pulses are periodic. Would a single pulse be comprised of equal amounts of all frequencies?

I also tried using the rf-tools.com application you showed and changing the parameters - I see increasing the filter order increases the group delay like you said, but even at the maximum setting of order 20 it is still only ~400 ns. Maybe I'm doing something wrong, how would I get this into the usec range? It would also be prohibitive to build 64 filters, each 20 components.

In regard to distorting the waveform, that is not too important. The only information I need is whether the amplitude exceeded a certain voltage or not.
I concur, I had thought axis label was uS. I should have known better.
Tektronix in their old scopes used delay lines and they were high
order and only capable of nS delays.

My apologies. I think lumped element delay lines simply out of the realm
of possibilities.

Note if this is what you want why not a comparator ? Even a SAR could handle this.

The only information I need is whether the amplitude exceeded a certain voltage or not.


Regards, Dana.
 

JPilot1

Nov 18, 2022
7
Joined
Nov 18, 2022
Messages
7
I concur, I had thought axis label was uS. I should have known better.
Tektronix in their old scopes used delay lines and they were high
order and only capable of nS delays.

My apologies. I think lumped element delay lines simply out of the realm
of possibilities.

Note if this is what you want why not a comparator ? Even a SAR could handle this.




Regards, Dana.
So what I'm trying to do here is identify pulses from 8 different SiPMs (type of photodetector) with only 1 scope channel. So my plan was to delay the first SiPM signal by 1 usec, the second signal by 2 usec, etc. Then I can combine all those signals into the same channel (superimposing them) and identify which SiPMs pulsed based on time elapsed between a given pulse and some initial trigger. This is why I'm trying to do this analog rather than digital.

I guess I could convert all 8 signals to TTL using a comparator, then use a hardware delay line IC like:


to delay the TTL signals. The longest delay I can find is 500 ns, so I would need to cascade several of them to get some of the longer delays, but do you think that would work?

Also I don't really follow what you mean by SAR here.

Thanks again!
 

dragon

Oct 31, 2022
248
Joined
Oct 31, 2022
Messages
248
So what I'm trying to do here is identify pulses from 8 different SiPMs (type of photodetector) with only 1 scope channel. So my plan was to delay the first SiPM signal by 1 usec, the second signal by 2 usec, etc. Then I can combine all those signals into the same channel (superimposing them) and identify which SiPMs pulsed based on time elapsed between a given pulse and some initial trigger. This is why I'm trying to do this analog rather than digital.

you could also use a mux for that, but I guess that would be digital.
 

danadak

Feb 19, 2021
751
Joined
Feb 19, 2021
Messages
751
SAR = Successive Approximation Register A/D converter.

Basically use a SAR to digitize / sample pulses, feed to a memory delay line approach.

Analog delay possibly use a CCD approach :





Regards, Dana.
 
Top