Electronics > Metrology

Time interval measurement limits

(1/15) > >>

metebalci:
I wonder how time intervals and their distributions are measured and what the current state of art is in terms of resolution. Rather than a clock or a periodic signal, what I wonder is more general, measuring irregular events, either by start/stop signals or just by measuring the pulse width. Is it possible to have a resolution around 1ps or even better ?

I saw there was jitter/time interval analyzers (eg Yokogawa TA520) in the past having 25ps resolution using an hybrid approach, and Keysight at the moment have a freq. counter that can do 20ps I think using interpolation. TA520 resolves two reference clock period length of time by converting it to a voltage and then reading it with a 8-bit ADC. Just by using a 16-bit ADC and if needed a better clock (it has TCXO), it should be possible to improve this a lot, correct ?

jonpaul:
Bonjour as a digital audio researcher and transformer manufacturer since 1980s, we have used the old HP counters 5370A and 5372A but only 100 pS res.

In our papers for AES we were fortunate to have Yokogawa time analyzers: Several TA320 and a like new Yokogawa TA720

Those are unique as they use an interpolation with two different clocks. ~ 25 pS.

Simply digitizing the input  would need GHz FS rates to do better.

  Stanford Research has a nice counter  25 pS:  SR620  s Universal Time Interval Counter

The nucleonics groups have much faster single shot , using image intensifiers, MCP gigahertz scopes, etc.

If you can define the signals you measure and the aim of jitter, rise/fall, etc we can better respond.

Finally in all old digital instruments like the TA320/TA720 beware of old/dead backup Li batteries, that may hold CAL constants. The TA320/720 are not simple to open and reach the batteries.

Bon Chance,


Jon

2N3055:
It is tough to add something of value after Jon but just to repeat his question:
What is this for? Pure intellectual curiosity?

It certainly is possible. But those are specialized instruments.
Sub picosecond accuracy measurements are domain of CERN, femtosecond laser research, etc etc...

I cannot think of any "normal world" electronics application of top of my head that wouldn't be served with 25ps time resolution for instance.

Would you like to share what kind of exciting stuff you're been doing.  ;D

metebalci:
Hi Jon,


--- Quote from: jonpaul on January 03, 2022, 10:55:36 am ---In our papers for AES we were fortunate to have Yokogawa time analyzers: Several TA320 and a like new Yokogawa TA720

Those are unique as they use an interpolation with two different clocks. ~ 25 pS.

--- End quote ---

I find it very nice how they are designed. Very simple (at least it sounds simple) but very effective.


--- Quote from: jonpaul on January 03, 2022, 10:55:36 am ---Simply digitizing the input  would need GHz FS rates to do better.
  Stanford Research has a nice counter  25 pS:  SR620  s Universal Time Interval Counter

The nucleonics groups have much faster single shot , using image intensifiers, MCP gigahertz scopes, etc.

--- End quote ---

I think most of these are out of question without a method like TA-series have, I mean something more hybrid with analog circuitry. I dont know about RF/microwave freq. counters, if they can be used for this purpose, and also even if they are rated 100Ghz, that is still not enough I guess. Can you give an example of such a device if you have something in mind ? I dont even know how to search.


--- Quote from: jonpaul on January 03, 2022, 10:55:36 am ---If you can define the signals you measure and the aim of jitter, rise/fall, etc we can better respond.

--- End quote ---

The reason I ask this question now is because I was wondering if I can measure the SPDIF jitter and I was wondering how they are measuring the oscillators. However, these are not the point of my question, I am interested to know the state of the art in this type of measurement. As far as I understand the oscillators are measured in f-domain, I guess simply because there is no device that can measure the close-in phase noise in time domain (?). I dont know if I think and calculate this right, if I have a clock 10MHz so 50ns pw, and if I want to measure deviation at 1Hz level, this requires even a better resolution than fs, so I guess not possible to do in t-domain (?). and I guess there is no need to measure SPDIF or similar jitter below 25ps. It might be possible to do this with a sampling scope to some extent, but this requires a periodic signal and it will not be better than 25ps I think.

Actually my interest in time interval measurement started from somewhere else. I am interested in measuring interrupt response latency, how quick an embedded device -electronic+software- response to an input, so this is a time interval measurement with irregular events. I do this (as a hobby) using an FPGA based counter since there is no need to go below ~10ns there but it raised my interest on time interval measurement.


--- Quote from: jonpaul on January 03, 2022, 10:55:36 am ---Finally in all old digital instruments like the TA320/TA720 beware of old/dead backup Li batteries, that may hold CAL constants. The TA320/720 are not simple to open and reach the batteries.

--- End quote ---

That is very good to know as I am interested in getting one (most possibly TA520). When you say they hold CAL constants, I guess you dont mean they are dead when the battery is dead, but you mean it needs a working battery right ? Is there a way to check the status of the battery without opening up the unit, so it can be asked to the seller ?

metebalci:

--- Quote from: 2N3055 on January 03, 2022, 11:36:37 am ---It is tough to add something of value after Jon but just to repeat his question:
What is this for? Pure intellectual curiosity?

--- End quote ---

Yes intellectual curiosity


--- Quote from: 2N3055 on January 03, 2022, 11:36:37 am ---It certainly is possible. But those are specialized instruments.
Sub picosecond accuracy measurements are domain of CERN, femtosecond laser research, etc etc...

--- End quote ---

I am possibly saying this wrong but I think the lengths are measured with an interferometer for example, so I guess optical methods make sense, but I dont know much about it so cannot convert something in electrical domain to that easily in my mind.


--- Quote from: 2N3055 on January 03, 2022, 11:36:37 am ---I cannot think of any "normal world" electronics application of top of my head that wouldn't be served with 25ps time resolution for instance.

--- End quote ---

I guess it is natural for oscillators, but they are measured differently, so it might be unnecessary, but as an example, I can pose the question to what extent can I measure an oscillator jitter in t-domain ?

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version