Delay measurement help...

[Sankha]

Hi, I'm sankha from University of Colombo, Sri Lanka . Currently we are doing a projects to measure time delay between two high speed pulses. approximate time delay between the two pulses varies between 2 to 10 ns.  Any idea of how to move forward?

[Ice-Tea]

Get a good scope? 

[coromonadalix]

https://forum.arduino.cc/t/measuring-delay-between-two-pulses-using-millis-and-micros/341893/3

[Marco]

Hi, I'm sankha from University of Colombo, Sri Lanka . Currently we are doing a projects to measure time delay between two high speed pulses. approximate time delay between the two pulses varies between 2 to 10 ns.  Any idea of how to move forward?

Repetitive or single shot?

[Henrik_V]

Repetitve : part of PLL?

Or a fast XOR that loads a cap?

Classical instrument is a counter with can measure the A to B interval,   KS 53230A ?
Depending on the needed  resolution / accuracy
Just keep in mind that every meter cable is good for ~5ns
(A sharp edged pulse source and) a power splitter come in handy to check your setup...

Or try to get your hands on a HP 7371A ??

[Marco]

Some PIC microcontrollers have the CTMU, which can be used for time to digital conversion and a whole lot cheaper (though a bit noisier) than fixed function ICs. It also requires a bit more DIY which seems a good thing if you're doing it for a school project.

AFAIR Microchip has some documents on the topic, but it might be easier to understand with this project from EDN :
https://www.edn.com/use-time-domain-reflectometry-tdr-for-low-cost-liquid-level-measurement-part-iii/

[Sankha]

Thx,  but i'm working on nanosecond range.

[Sankha]

It is single shot.

[Henrik_V]

Whst timing resolution do need  and/or would like to have ?

Using a counter:  fast with known uncertainty (spec) but $$
using the PIC CTMU (see EDN article):  a lot cheaper but need time and some knowledge

So it seems to be a problem to simply staple coins on, or have some students and a lot of time

[jmelson]

It is single shot.

Oh, that makes it harder.  Well the classic approach is a FF that is set by the first signal and reset by the second.  The output of the FF starts charging a capacitor when true and stops charging when false.  Then, you look at the voltage on the capacitor.
Jon

[Marco]

That's exactly what the CTMU does.

[hgl]

It depends on which resolution and which repeat accuracy you need.

If you want to build it yourself take an AS6501 otherwise buy a time interval analyzer.

[Picuino]

Some PIC microcontrollers have the CTMU, which can be used for time to digital conversion and a whole lot cheaper (though a bit noisier) than fixed function ICs. It also requires a bit more DIY which seems a good thing if you're doing it for a school project.

AFAIR Microchip has some documents on the topic, but it might be easier to understand with this project from EDN :
https://www.edn.com/use-time-domain-reflectometry-tdr-for-low-cost-liquid-level-measurement-part-iii/

https://www.microchip.com/stellent/groups/SiteComm_sg/documents/DeviceDoc/en542792.pdf
https://ww1.microchip.com/downloads/en/DeviceDoc/61167B.pdf

You can reach 0.1ns 200ps resolution in several models of PIC MCUs.

https://ww1.microchip.com/downloads/en/DeviceDoc/30009995e.pdf

[hgl]

0.2 ns resolution are 10 ADC counts at 2ns measuring time.

From PIC24FV32KA304 Datasheet :   

[jonpaul]

Bonjour,

One shot TD in nS..Ps require a very fast analog comparator and counter or multi gHz digital sampling.
Cables, connectors and layout are all critical. I would be skeptical of any microP based or home made solutions,  results may be impossible to quantify or verify.
Such applications are in the nuclear, defense, military, weapons research and testing. EG nuclear devices, one shot laser weapons, EMP, etc.

In our research, We used\ dedicated  time interval analyzers : the ancient HP 5370B, 5372A, and Yokogawa TA 720 are specifically designed for these measurements but may not have the resolution you require. On ebay.

Modern state of the art at Stanford Research (SRS) 25 pS res

https://www.thinksrs.com/products/sr620.html

Perhaps Zurich Instruments, 
https://www.zhinst.com/

Hope this is of use to you!

Bon chance,

Jon
 

[Picuino]

Also 0.2ns is equivalent to 6cm at the speed of light. Any length of cable a little longer can alter the measurements.

One way to eliminate systematic error is to insert a longer wire for the second input cable. Inject the same signal through the two cables and measure the time shift. That is the time that will have to be subtracted to find the actual time shift between the two input signals.

[Sankha]

thx a lot for the information

[FriedLogic]

  An AMS/Acam GP21 should do this as long as you know when the pulses are coming to within a 4ms or so as that is its maximum timing range. 60ps RMS noise, if that's OK. Cheap little boards with them on ebay, so no QFN soldering required.
  A TI TDC7201 should do too, but in a BGA package. It has an 8ms timing range, and a little less noise than the GP21. A fair bit easier to set up too - or at least the TDC7200 is, so I would guess this one too.
  The AMS AS6500 (or AS6501) already mentioned would give better resolution, and a 16s timing range, but you would need to make your own board, or get one of the official eval boards from Mouser etc. at $65 or so.