[time-nuts] Comparing the frequency of two gpsdo's.

[Magnus Danielson]

On 11/13/2010 08:12 PM, Bob Camp wrote:
> Hi
>
> Ok, so right now you are looking at about 10 degrees out of 360 where one cycle is 100 ns. More or less you are in the 3 ns range.
>
> Some cheap stuff that will do better:
>
> HP 5334
> HP 5335
> HP 5345
> HP 5370
> HP 5371
>
> All are in the "sub $300" range on the normal sites. Some are sub $100. All are available with GPIB for logging.
>
> For a bit more money
>
> HP 53131
> SR 620
>
> You may find one for sub $1000. Often you see them listed for nutty prices. I certainly would not pay anywhere near $1,000 for either one.
>
> The one I'd go for is the 5334. It's smaller than the rest. They likely are the cheapest of the group. The ones I have *seem* to be more reliable than some of the rest of the stuff listed.
>
> All of them can run in any one of three modes:
>
> 1) PPS to PPS timing
> 2) PPS to 10 MHz edge timing
> 3) 10 MHz to 10 MHz timing

Technicality - you want to have a stable rate of measurements. One way 
to achieve this is to use a PPS or similar trigger. The TADD-2 
divide-down kit can be handy to proivde triggers from a 5 MHz or 10 MHz 
source.

In particular you would enjoy triggering the measurement from a PPS (or 
divided down 10 MHz to say 10 Hz or 100 Hz) and then do Time Interval 
measurements with one 10 MHz rising edge (say reference) to the other 10 
MHz rising edge. I have had great success with this method myself.

> Number 2 on the list seems to have the fewest issues.

Actually, depending on the GPSDO style the PPS may be the "raw" GPS PPS 
or the smoothed variant being a divide down from the 10 MHz. Method 2 
would not be as clean as you would wish for some GPSDOs. Using the 
trigger variant described above is preferred since it would would have 
less issues with the noise of the PPS relative to the 10 MHz.

> A completely different approach:
>
> Pump both signals into a phase detector (RPD-1 or X-OR or what ever)
> and use a DVM to log the voltage. You can get some super  overkill
> DVM's for less than you can get any of the counters. They will
> easily get you into the sub ns range on resolution. Weather the
> setup will be accurate at this or that  level is in the "that depends"
> category.

Hmm. Gotta try that one.

Cheers,
Magnus