[time-nuts] 9390 GPS RX

Ed Palmer ed_palmer at sasktel.net
Thu May 30 01:04:19 EDT 2013

On 5/29/2013 10:14 PM, Mark C. Stephens wrote:
> Ah, The old Section 2 part (in Don Adams voice)
> Okay, I have been watching the timer out of the corner of my eye, and there is a trend.
> The TI difference gradually falls down to about 3ns (good) then shoots up to 103ns (bad) and then slowly falls down to 3ns in just over an hour, it then shoots up to 103ns again and repeats the process.
> This is why I get good readings for short term.
> The frequency can't be out that much, surely?

No, that's okay.  Nothing wrong at all.  The frequency of the FRK is 
slightly different from the 10 MHz out of the GPSDO.  Over a period of 
just over an hour, it drifts 1 cycle which, at 10 MHz, is 100 ns.  So 
the time interval goes from being 103 ns (think of it as being 360 
degrees out of phase), gradually falls to 3 ns (in phase) and then jumps 
to 103 ns (360 degrees out of phase).  This is called 'phase wrapping'  
If you connect the FRK and the GPSDO to the channels of a dual-trace 
scope in parallel with the counter you can watch the drift.  (Talk about 
boring!)  If you manage to stay awake, you'll see that when the counter 
is reporting a 50 ns time interval, the two waveforms are 180 degrees 
out of phase.  Why the extra 3 ns?  Because the 5370B has trouble 
dealing with signals that are that close together and gets a little 
confused.  If I did the math right, you're looking at a frequency offset 
in the range of 3e-11 which is better than the original spec for the FRK 
frequency error. Well done!

Timelab or Plotter or Stable32 (I assume) takes care of these phase 
wraps.  In Timelab, if you click on Measurement -> Phase Difference 
(Original) you'll see the sawtooth pattern.  If you click on Phase 
Difference (Unwrapped) you'll see how Timelab takes the angled parts of 
the sawtooth and sticks them together in a straight line by 'unwrapping' 
the phase wraps.  It then analyzes the straight line.

If you were to nudge the frequency of the FRK a bit, you'd see that the 
time between phase wraps would increase as you came closer to 10 MHz.  
Actually, it would be a pain to watch for that so you'd just watch for 
the counter to increment or decrement more slowly.  But there's no point 
in doing that because you're just going to put it back in the 9390.

> Anyway, as you say, Put it in the 9390 and see if 9390 can phase-lock it.
> That certainly will be interesting if it does!

I can almost guarantee that it will.  It will use the EFC to nudge the 
frequency of the FRK so that it is EXACTLY 10.000 000 000 000 000 MHz.  
Well, maybe not quite that many zeros. :)  And it will keep nudging the 
EFC to keep the frequency there as the FRK slowly drifts.  So there 
won't be a frequency difference.  That's what a GPSDO does.


> -marki
> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On Behalf Of Ed Palmer
> Sent: Thursday, 30 May 2013 3:31 AM
> To: time-nuts at febo.com
> Subject: Re: [time-nuts] 9390 GPS RX
> On 5/29/2013 9:58 AM, Mark C. Stephens wrote:
>> Uh, well. No one told me about the heatsink thing :)
> Section 2 in the manual talks about mounting and heatsinking.  You mean you didn't read the flippin' manual??? ;)  Anyway, it's good that you were able to get it on frequency.
>> I am using a HP5370B.
>> I am comparing to my best GPSDO.
>> For instance my 5x Z3805A's get down to the low -12's after 1 hour.
> Remember what these measurements mean.  Result = sqrt(<measurement
> system>^2 + DUT^2 + Reference^2).  Your measurement system and reference
> have to be significantly better than the DUT or they will affect the result.  The 5370B should be low enough to be ignored at these Tau values.  Your Z3805A should be okay, I think, but not by much.  My Z3801A happens to have a 'disciplining hump' that peaks at around 1000 sec. and was affecting the measurement at 800 sec.  I was also using a 5370B.
>> 4.7e-11 at 1 hour 12 minutes.
> Yeah, that's high considering that the monthly drift is supposed to be less than that!  Let it soak for a month and see what happens.
>> Do you think the FRK is ready to go back in the 9390 :) I hope so, I
>> like that 9390 and I want it in my rack :)
> Yes, I think so.  Like I said, see what monitoring capability the 9390 has.  I suspect that the FRK will improve over the next while. Only time will tell how much.
> Congratulations!  It looks like your FRK is alive!  After a little time to work out the last few kinks it should be in good shape. (Knock on
> wood!)  :)
> Ed

More information about the time-nuts mailing list