[time-nuts] SRS SR620 External Source Issue -- Help Request

Dr Bruce Griffiths bruce.griffiths at xtra.co.nz
Thu Feb 22 15:54:53 EST 2007

TheInfamousFlavio at hotmail.com wrote:
> 1.  Same test with a Fluke 6680 yeilds 10MHz mean exactly.
> 2.Tried different length cables and swapping the cables.  Same result.
> 3. Autocal was done (a few times)
> ----- Original Message ----- 
> From: "Tom Van Baak" <tvb at LeapSecond.com>
> To: "Discussion of precise time and frequency measurement" 
> <time-nuts at febo.com>
> Sent: Thursday, February 22, 2007 00:12
> Subject: Re: [time-nuts] SRS SR620 External Source Issue -- Help Request
>>> I have an SR620 counter that I set up with a Z3801A as an external 
>>> reference.  If I put a bnc T connector at the output
>>> of the Z3801A and use two equal length bnc cables, one to the ext. ref 
>>> input on the back and the other to channel A
>>> then do a frequency measurement, I get a mean that is about .0015 Hz below 
>>> 10,000,000.0000Hz.
>>> Does any know why this might be happening? I would expect it to read 
>>> 10,000,000.0000 exactly give or take a couple on
>>> the last digit.
>> Three comments that may help.
>> (1)
>> 0.0015 Hz out of 10 MHz is 1.5e-10 which seems a little
>> high but not too bad. But do not expect exactly 10 MHz with
>> this sort of "test". What you are giving to the channel A
>> input is the most highly phase correlated signal you can
>> imagine relative to the internal clock and the interpolators.
>> This won't happen in real life with real input frequencies.
>> (2)
>> Try different lengths of the channel A cable and see if the
>> number changes. My prediction is it will.
>> (3)
>> Run the setup menu self-calibration if you haven't in a while.
>> /tvb
>> _______________________________________________
>> time-nuts mailing list
>> time-nuts at febo.com
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

The effect that Tom described (differential nonlinearity in the 
interpolators) has a period of about 11.11 ns, so changing the relative 
delay by around 2.7ns (or small odd multiple thereof) is most likely to 
uncover the effect. If your relative delay changes is a multiple of 
11,11ns or close to it you wont see much of an effect.
However the manual indicates that this is relatively small when the time 
interval being measured is longer than a few microseconds or so.

If the differential delay of the 2 synchronisers isn't corrected by the 
calibration procedure then a value of 150 ps would not be unexpected, 
the synchroniser flipflops are MC10H131's which have a clock to output 
delay of up to 1.5ns. As far as I can tell the autocal procedure will 
not correct such an offset.


More information about the time-nuts mailing list