[time-nuts] T.I. experimenting - newbie question

[Magnus Danielson]

On 04/03/2011 04:03 AM, Bruce Griffiths wrote:
> Bruce Griffiths wrote:
>> Joseph Gray wrote:
>>> I have a borrowed HP 5328A with the Universal Module. I'm using
>>> averaging mode and auto triggering. Counter does not have the OCXO
>>> option. I'm using the internal oscillator, not locking to my GPSDO.
>>>
>>> I was just messing around with a length of Belden 9273 and the 100 KHz
>>> square wave output from my Spectracom distribution amp. Start signal
>>> is the 100 KHz signal. Using a tee, the stop signal is the same 100
>>> KHz after passing through the length of coax. I'm reading 18.9 ns. The
>>> coax specs give a VOP of 66%. This should calculate out to a length of
>>> 3.74 M. When I measure the coax, I get 3.8 M.
>>>
>>> I am including the BNC connectors in my length measurement. If I
>>> subtract them, I'm much closer to my calculated length. Should the BNC
>>> connector lengths be included or not? Or, is the measurement I'm
>>> getting as good as it gets with this HP counter?
>>>
>>> Joe Gray
>>> W5JG
>>>
>> The measured delay includes the time offset between the start and stop
>> inputs.
>> This is unlikely to be zero, you need to measure it and subtract the
>> offset from the measured result.
>> Since the 5328A can only measure positive time intervals measuring
>> this offset is somewhat tricky to do.
>>
>> Bruce
>>
> The simplest way of measuring the cable delay with the 5328A is to
> measure the delay with a short length of cable between the start and
> stop inputs and then increase the length of cable by connecting the
> cable to be measured in series with the first cable measure the delay
> again and subtract the first measurement from the second.
> This procedure corrects for the start to stop offset as well as that of
> the first cable.
> Then the effects of finite rise time, trigger thresholds and reflections
> should be considered.
> One then has to consider the effects of the various connectors,

You would also need to consider the delay effect of the BNC 
sex-converter to hook the two cables up.

You can do that by doing the following measurements:

Measure the short cable.
Measure the short and DUT cable.
Measure the DUT cable.

t1 = ts
t2 = ts + tc + tDUT
t3 = tDUT

Solving for the connector delay tc is simple.

You won't resolve the delay offsets of the measurement itself.

Cheers,
Magnus