# [time-nuts] How can I measure time-delay of a cable with HP 5370B time-interval counter?

slawdabr@wp.pl slawdabr at wp.pl
Mon Oct 29 18:01:36 EDT 2018

```Hello,  Maybe this method will help:   1. connect splitter to generator ,any two coaxes to splitter and then connect them to start and stop channel of TIC (CAB I and CAB II)  2. Generate any time interval (period, pulse width etc)  3. Measured time interval is zero - level, reference level(TI0)  4. Connect coax under test to CAB II and then TIC channel  5. Generate the same time interval as in p.2  6. Measured TI is TI1  7. Replace CAB II and coax under test, generate the same time interval as in p.2  8. Measured TI is TI2  9. Calculate delay: T = TI0 - (TI1+TI2)/2   For best result, use the same timebase for generator and TIC

Dnia 29 października 2018 17:00 time-nuts-request at lists.febo.c napisał(a):

Send time-nuts mailing list submissions to
time-nuts at lists.febo.com

To subscribe or unsubscribe via the World Wide Web, visit
lists.febo.com lists.febo.com
or, via email, send a message with subject or body 'help' to
time-nuts-request at lists.febo.c

You can reach the person managing the list at
time-nuts-owner at lists.febo.com

than "Re: Contents of time-nuts digest..."

Today's Topics:

1. Re: How can I measure time-delay of a cable with HP 5370B
time-interval counter? (David C. Partridge)
2. Re: How can I measure time-delay of a cable with HP 5370B
time-interval counter? (Artek Manuals)
5. Re: How can I measure time-delay of a cable with HP 5370B
time-interval counter? (Chris Caudle)
6. Re: How can I measure time-delay of a cable with HP 5370B
time-interval counter? (Dr. David Kirkby)

------------------------------

Message: 1
Date: Mon, 29 Oct 2018 10:38:52 -0000
From: "David C. Partridge" <david.partridge at perdrix.co.uk
To: "'Discussion of precise time and frequency measurement'"
<time-nuts at lists.febo.com>
Subject: Re: [time-nuts] How can I measure time-delay of a cable with
HP 5370B time-interval counter?
Message-ID: <004801d46f73\$96e047b0\$c4a0d71
Content-Type: text/plain;	charset="us-ascii"

If you have a Tektronix 7000 series 'scope, then a 7S12 equipped with a S52
pulse generator and an S6 sampler head will talk all the pain out of
measuring cable length and will also show you any impedance mismatches.
This way you don't need to suspect a bad cable, you can prove it's bad.

If you need a cable checked, I can do it as I have 7S12 plugins.
Dave

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces at list On Behalf Of Dr.
David Kirkby
Sent: 29 October 2018 00:50
To:   time-nuts at lists.febo.com
Subject: [time-nuts] How can I measure time-delay of a cable with HP 5370B
time-interval counter?

I'm trying to do something which would seem conceptually easy, but I'm
getting results I can't understand. I wish to measure the delay (in
seconds) of a bit of length of coaxial cable.

I'm feeding a sine wave from a Stanford Research DS345 30 MHz function
generator via a coax to the START input of the counter, then with a BNC
T-piece, of 480 mm of 50 ohm cable to the STOP input of the counter. Here's
a photo of the complete setup.

www.kirkbymicrowave.co.uk www.kirkbymicrowave.co.uk
enerator-to-start-then-stop.jp

I've set the 5370B's START impedance to be 1 M ohm, and the STOP to be 50
ohms, so the function generator should see a 50 ohm load, as 1 M ohm in
parallel with 50 ohms is virtually 50 ohms.

The switch position on the counter are as shown here

www.kirkbymicrowave.co.uk www.kirkbymicrowave.co.uk
s.jpg

So the main settings are

* TI mode.
* +/- TI
* START. 1 M ohm, positive slope, level to preset position (0 V)
* STOP 50 ohm, positive slope, level to preset position (0 V)

With the cable 480 mm in length, the velocity factor of the cable being
approximately 0.7, I would have expected an electrical length of around 686
mm, and so a delay of

time =  distance / velocity = 0.686 / 3e8
= 2.29 ns.

I would not be surprised by small changes in delay with frequency, which is
what I wanted to investigate. But I'm getting the following readings, for
different frequencies of the function generator

1 kHz - unstable readings, around 100~300 us.
10 kHz  -> -21.3 us
50 kHz -> -4.27 us
100 kHz -> -1.90 us
250 kHz -> - 528 ns
500 kHz -> 1.837 us
1 MHz -> 956 ns
2 MHz -> 490 ns
3 MHz -> -2.6 ns
4 MHz -> -0.33 ns
5 MHz -> 0.90 ns
6 MHz -> 1.50 ns
7 MHz -> 1.93 ns
8 MHz -> 2.15 ns
9 MHz -> 2.38 ns
10 MHz -> 2.52 ns
11 MHz -> 2.60 ns
20 MHz -> 2.85 ns
30 MHz -> 2.80 ns

The numbers look believable  with a frequency input of 10 MHz or more. I
did not do the complete set again, but using a cable of 1.53 m in length,
where I would expect the delay to be around 7.29 ns, the results were

1 MHz  -> -26.51 ns
5 MHz -> 9.70 ns
10 MHz -> 9.70 ns
15 MHz -> -57.81 ns
20 MHz -> -41.64 ns
30 MHz -> 7.13 ns

Note, the function generator and counter do not share a common frequency
standard for this test. I have not tried it with them locked to the same 10
MHz reference, but I somewhat doubt that is the cause of these issues.

I must be missing something, but I'm not sure what it is.

--
Dr David Kirkby Ph.D C.Eng MIET
Kirkby Microwave Ltd
Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, CHELMSFORD,
Essex, CM3 6DT, United Kingdom.
Registered in England and Wales as company number 08914892
www.kirkbymicrowave.co.uk www.kirkbymicrowave.co.uk
Tel 01621-680100 / +44 1621-680100
______________________________
time-nuts mailing list --   time-nuts at lists.febo.com
To unsubscribe, go to
lists.febo.com lists.febo.com

------------------------------

Message: 2
Date: Mon, 29 Oct 2018 07:57:00 -0400
From: Artek Manuals <Manuals at ArtekManuals.com>
To:   time-nuts at lists.febo.com
Subject: Re: [time-nuts] How can I measure time-delay of a cable with
HP 5370B time-interval counter?
Message-ID: <b0ac7b03-c460-7e6d-1d5d-163fd
Content-Type: text/plain; charset=windows-1252; format=flowed

David

First let me say that I have never used a 5370B

Ignoring the lower frequency stuff for the moment, Can you measure (
accurately) the trigger levels of both the start and stop gates? Slight
differences in the trigger points at each end will will obviously add
error in the measurement.

The next flag for thought is your comment "assuming a velocity factor of
.7" What if the velocity factor is really .66 ? This would account for
almost half of the error.

Rerun you data with larger and smaller output levels from the function
generator. Increasing error" with lower signal levels (or vice versa
lower erros with increased voltage ) would implicate a trigger
differences as the source of error as well

Dave
manuals at artekmanuals.com

David on another note I tried replying to you directly and the email
bounced ?

On 10/29/2018 6:38 AM, David C. Partridge wrote:

<SNIP>

If

I'm trying to do something which would seem conceptually easy, but I'm
getting results I can't understand. I wish to measure the delay (in
seconds) of a bit of length of coaxial cable.

I'm feeding a sine wave from a Stanford Research DS345 30 MHz function
generator via a coax to the START input of the counter, then with a BNC
T-piece, of 480 mm of 50 ohm cable to the STOP input of the counter. Here's
a photo of the complete setup.

www.kirkbymicrowave.co.uk www.kirkbymicrowave.co.uk
enerator-to-start-then-stop.jp

I've set the 5370B's START impedance to be 1 M ohm, and the STOP to be 50
ohms, so the function generator should see a 50 ohm load, as 1 M ohm in
parallel with 50 ohms is virtually 50 ohms.

The switch position on the counter are as shown here

www.kirkbymicrowave.co.uk www.kirkbymicrowave.co.uk
s.jpg

So the main settings are

* TI mode.
* +/- TI
* START. 1 M ohm, positive slope, level to preset position (0 V)
* STOP 50 ohm, positive slope, level to preset position (0 V)

With the cable 480 mm in length, the velocity factor of the cable being
approximately 0.7, I would have expected an electrical length of around 686
mm, and so a delay of

time =  distance / velocity = 0.686 / 3e8
= 2.29 ns.

I would not be surprised by small changes in delay with frequency, which is
what I wanted to investigate. But I'm getting the following readings, for
different frequencies of the function generator

1 kHz - unstable readings, around 100~300 us.
10 kHz  -> -21.3 us
50 kHz -> -4.27 us
100 kHz -> -1.90 us
250 kHz -> - 528 ns
500 kHz -> 1.837 us
1 MHz -> 956 ns
2 MHz -> 490 ns
3 MHz -> -2.6 ns
4 MHz -> -0.33 ns
5 MHz -> 0.90 ns
6 MHz -> 1.50 ns
7 MHz -> 1.93 ns
8 MHz -> 2.15 ns
9 MHz -> 2.38 ns
10 MHz -> 2.52 ns
11 MHz -> 2.60 ns
20 MHz -> 2.85 ns
30 MHz -> 2.80 ns

The numbers look believable  with a frequency input of 10 MHz or more. I
did not do the complete set again, but using a cable of 1.53 m in length,
where I would expect the delay to be around 7.29 ns, the results were

1 MHz  -> -26.51 ns
5 MHz -> 9.70 ns
10 MHz -> 9.70 ns
15 MHz -> -57.81 ns
20 MHz -> -41.64 ns
30 MHz -> 7.13 ns

Note, the function generator and counter do not share a common frequency
standard for this test. I have not tried it with them locked to the same 10
MHz reference, but I somewhat doubt that is the cause of these issues.

I must be missing something, but I'm not sure what it is.

--
Dave
Manuals at ArtekManuals.com
www.ArtekManuals.com

---
This email has been checked for viruses by Avast antivirus software.
www.avast.com www.avast.com

------------------------------

Message: 3
Date: Mon, 29 Oct 2018 14:59:08 +0100
From: Attila Kinali <attila at kinali.ch>
To: Discussion of precise time and frequency measurement
<time-nuts at lists.febo.com>
Message-ID: <20181029145908.f7fde144af2bc1
Content-Type: text/plain; charset=ISO-8859-1

Moin,

I'm bunching a few mails together, to not clutter the mailinglist too much

On Sat, 27 Oct 2018 23:25:30 +0200
Magnus Danielson <magnus at rubidium.dyndns.org> wrote:

The integration is very important aspect, as a number of assumptions
becomes embedded into it, such as the f_H frequency which is the Nyquist
frequency for counters, so sampling interval is also a relevant
parameter for expected level.

An important thing to note here is that Gaussian white noise is,
as it is defined, non-continuous (by any continuity measure).
Ie if you take two samples, no matter how close they are time-wise,
their difference in value can be arbitrary large. If you are integrating
over (time-continuous) Gaussian white noise, you have to argue
carefully, why this integral is defined (meaning why calculating it
leads to a single, well defined value). In our case, it's usually
enough to assume that there is a finite cut-off frequency at which
point the signal falls off with at least 1/f^2 (or >=40dB/dec) to
ensure 1) continuity and 2) convergence of the integral.

For more details, see a textbook on Ito-calculus, e.g. [1]

On Sat, 27 Oct 2018 23:43:33 +0200
Magnus Danielson <magnus at rubidium.dyndns.org> wrote:

A simple trick to transform uniform distribution to normal distribution
like shape is to take 12 samples and add them together. A special trick
is to take them pair-wise and subtract them and then add 6 differences,
to avoid DC bias of typical uniform distribution generation (as typical
pseudo-noise generators does not have all 0 state in them). The result
of this subtract-add trick is a normal distribution like thing with the
standard distribution of 1. More or fewer sample-pairs can be added if
the product is scaled appropriately.

The Box-Mueller algorithm is another way to convert uniform distribution
to normal distribution.

as an approach for generating normal distributed values! Even if it will
get you something that looks like a normal distribution, it's quite far
from it. It is also a very slow method and uses up a lot of randomnes.

Box-M?ller is a usable alternative, though I would recommend using
the Ziggurat Method[2], which is very fast and leads to a very good
approximation. When I replaced the "take 30 samples and add them" of
Fran?ois Vernotte's Sigmatheta package[3] and used the Ziggurat Method,
combined with xorshift1024*[4] for random number generation, I got
a total speed up of a factor of more than 2 (including the FFT and
everything)[5] (yes, I know that xorshift1024* does have some problems
in the quality of random numbers generated, but they shouldn't be
relevant for the application at hand).

Attila Kinali

[1] "Stochastic Differential Equations", by Bernt ?ksendal, 2013 (6th ed)

[2] "The Ziggurat Method for Generating Random Variables",
by Marsaglia and Tsang, 2000
dx.doi.org dx.doi.org

[3]  theta.obs-besancon.fr theta.obs-besancon.fr

[4]  xoshiro.di.unimi.it xoshiro.di.unimi.it
or more specifically:  xoroshiro.di.unimi.it xoroshiro.di.unimi.it

[5]  git.kinali.ch git.kinali.ch
--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
Miss Matheson, The Diamond Age, Neal Stephenson

------------------------------

Message: 4
Date: Mon, 29 Oct 2018 15:28:28 +0100
From: Magnus Danielson <magnus at rubidium.dyndns.org>
To:   time-nuts at lists.febo.com
Cc:   magnus at rubidium.se
Message-ID: <2abfba15-f9f7-2ff6-bff8-0ec46
Content-Type: text/plain; charset=utf-8

Hi Attila,

On 10/29/18 2:59 PM, Attila Kinali wrote:

Moin,

I'm bunching a few mails together, to not clutter the mailinglist too much

On Sat, 27 Oct 2018 23:25:30 +0200
Magnus Danielson <magnus at rubidium.dyndns.org> wrote:

The integration is very important aspect, as a number of assumptions
becomes embedded into it, such as the f_H frequency which is the Nyquist
frequency for counters, so sampling interval is also a relevant
parameter for expected level.

An important thing to note here is that Gaussian white noise is,
as it is defined, non-continuous (by any continuity measure).
Ie if you take two samples, no matter how close they are time-wise,
their difference in value can be arbitrary large. If you are integrating
over (time-continuous) Gaussian white noise, you have to argue
carefully, why this integral is defined (meaning why calculating it
leads to a single, well defined value). In our case, it's usually
enough to assume that there is a finite cut-off frequency at which
point the signal falls off with at least 1/f^2 (or >=40dB/dec) to
ensure 1) continuity and 2) convergence of the integral.

There is aspects of noise which is more or less important depending on
what you do. As we leave WPM it is no longer gaussian anyway. For ADEV
and friends the shape of the PDF isn't as important as for other things.
The slope of the frequency range is however the important one. It is
only when we do the confidence intervals where the Gaussian shape
becomes relevant for the Chi-square bounds, but those are usually not
precise enough that even rough Gaussian shape is relevant. Even for
noises with none-Gaussian properties, the Chi-square seems to be valid
enough.

For other measures, like bit-error simulations, proper Gaussian shape is
much more important, but only to a certain point. For higher BER values,
the details of the outer part of the shape isn't all that important,
it's only as you push into lower BER numbers you need to care.

For more details, see a textbook on Ito-calculus, e.g. [1]

On Sat, 27 Oct 2018 23:43:33 +0200
Magnus Danielson <magnus at rubidium.dyndns.org> wrote:

A simple trick to transform uniform distribution to normal distribution
like shape is to take 12 samples and add them together. A special trick
is to take them pair-wise and subtract them and then add 6 differences,
to avoid DC bias of typical uniform distribution generation (as typical
pseudo-noise generators does not have all 0 state in them). The result
of this subtract-add trick is a normal distribution like thing with the
standard distribution of 1. More or fewer sample-pairs can be added if
the product is scaled appropriately.

The Box-Mueller algorithm is another way to convert uniform distribution
to normal distribution.

as an approach for generating normal distributed values! Even if it will
get you something that looks like a normal distribution, it's quite far
from it. It is also a very slow method and uses up a lot of randomnes.

Actually, for many simulations you do not need better "shape".
There is some simulations where shape comes in, but others where it has
little to no consequence.

Box-M?ller is a usable alternative, though I would recommend using
the Ziggurat Method[2], which is very fast and leads to a very good
approximation. When I replaced the "take 30 samples and add them" of
Fran?ois Vernotte's Sigmatheta package[3] and used the Ziggurat Method,
combined with xorshift1024*[4] for random number generation, I got
a total speed up of a factor of more than 2 (including the FFT and
everything)[5] (yes, I know that xorshift1024* does have some problems
in the quality of random numbers generated, but they shouldn't be
relevant for the application at hand).

Getting suitable PRNG polynomials isn't all that hard, if the length of
the "random" sequence is of concern compared to the length of the
sequence used. It's a solved problem.

Never the less, thanks for the many references. Will read up on them
eventually.

Cheers,
Magnus

Attila Kinali

[1] "Stochastic Differential Equations", by Bernt ?ksendal, 2013 (6th ed)

[2] "The Ziggurat Method for Generating Random Variables",
by Marsaglia and Tsang, 2000
dx.doi.org dx.doi.org

[3]  theta.obs-besancon.fr theta.obs-besancon.fr

[4]  xoshiro.di.unimi.it xoshiro.di.unimi.it
or more specifically:  xoroshiro.di.unimi.it xoroshiro.di.unimi.it

[5]  git.kinali.ch git.kinali.ch

------------------------------

Message: 5
Date: Mon, 29 Oct 2018 10:01:32 -0500
From: "Chris Caudle" <chris at chriscaudle.org>
To:   time-nuts at lists.febo.com
Subject: Re: [time-nuts] How can I measure time-delay of a cable with
HP 5370B time-interval counter?
Message-ID:
<4c942efd15520e6d47bdb068bf3b
Content-Type: text/plain;charset=iso-8859-1

On Mon, October 29, 2018 6:57 am, Artek Manuals wrote:

The next flag for thought is your comment "assuming a velocity factor of
.7" What if the velocity factor is really .66 ? This would account for
almost half of the error.

Propagation velocity has an inverse dependence on permittivity, and
permittivity changes with frequency.  Electrical delay time will not be
constant with frequency because of that.

In addition to the fundamental physics at play, there are instrumentation
difficulties.  The rise time at low frequencies is long enough that any
50Hz/60Hz interference from power line related current flow can modify the
trigger point and influence the measurement.  The fast rise time signals
proposed for evaluating the measurement setup get around that, but then of
course you are measuring a wideband signal, which rather misses the point
of the original goal of measuring vs. frequency, so at some point after
verifying the setup basics you will have to go back to narrow band
signals.

A 5370 is a somewhat coarse instrument for this type of measurement, a VNA
which has a suitable lower measurement frequency would probably be more
suitable.

--
Chris Caudle

------------------------------

Message: 6
Date: Mon, 29 Oct 2018 15:23:49 +0000
From: "Dr. David Kirkby" <drkirkby at kirkbymicrowave.co.u
To: Tom Van Baak <tvb at leapsecond.com>,   time-nuts at lists.febo.com
Subject: Re: [time-nuts] How can I measure time-delay of a cable with
HP 5370B time-interval counter?
Message-ID:
<CANX10hDi1yY82fXJZvYDVomnnAX
Content-Type: text/plain; charset="UTF-8"

On Mon, 29 Oct 2018 at 09:43, Tom Van Baak <tvb at leapsecond.com> wrote:

David,

Just to see if your setup is working:

1) Set the pulse generator to as fast a risetime as possible; ns or less.
Use a low pulse rate (100 Hz is fine).

Unfortunately, I don't have such a pulse generator, so I can't run that
test. But it is clear the system is sensitive to the trigger levels, so I
guess is the problem. I have done all the confidence checks in the manual
on this TI counter before and it was fine.

Also, I am interested in the delay of the cable at low frequencies, as I
suspect that might depart significantly from the usual figure based on the
"velocity factor". Certainly the impedance of coax  rise at low frequencies
because the normal formula

Zo = sqrt(L/C)
is not valid before a few MHz. A more accurate formula is

Zo = sqrt ( (R + j w L )/ ) / (G + j w C))

where R = Resistance per unit length
L = inductance per unit length
G = Conductance per unit length
C = Capacitance per unit length.

So feeding in short pulses brings the validity of such a test into
question.

2) Use a BNC tee at the generator, into two equal 2 meter cables, each one
into a 5370B input.
3) Set manual trigger, 50R, 1.0 V, DC
4) Now collect time interval data in block/stats mode. You should see a
mean of under +/-1 ns and a stdev in low ps.

/tvb

Thank you. I will look for a pulse generator. I would use the 1 pps from
the GPS receiver, but my HP 58503A GPS receiver has decided to pack up. I
need to have a look at that, but it is not the highest priority task just
now.

This possible trigger issue metioned by Hal Murray is probably a result of
the knobs not exactly lining up with the positions they are in. It is
fairly clear that the marker on one of them is vertical at 0  V, but the
other is not. I need to try to get the knob back on the shaft in a slightly
different position. But they were both set to preset, but it is clear that
the reading is sensitive to the trigger points.

I have a 100 MHz scope, and can borrow a 300 MHz scope, but I don't have
anything really fast.

I have a VNA which can make measurements of phase difference down to 300
kHz, but don't trust those because of the fact one calibrates with a 50 ohm
load, 50 ohm calibration standards, yet I know the impedance will rise well
above 50 ohms at low frequencies.

I was looking for a different approach than a VNA, to make comparisons with
a VNA.

Dave

------------------------------

Subject: Digest Footer

______________________________
time-nuts mailing list
time-nuts at lists.febo.com
lists.febo.com lists.febo.com

------------------------------

End of time-nuts Digest, Vol 171, Issue 38
******************************
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TIC_lenght.png
Type: image/png
Size: 12781 bytes
Desc: not available
URL: <http://lists.febo.com/pipermail/time-nuts_lists.febo.com/attachments/20181029/3e2e45fc/attachment.png>
```