[time-nuts] How do I know my GPS stabilized oscillator is working?

Richard H McCorkle mccorkle at ptialaska.net
Fri Aug 11 14:51:58 EDT 2006


A very low cost way to test your GPSDO is to generate 1Hz from the GPSDO
with the TVB divider compare it to the GPS 1PPS with the Simple Time
Interval Counter with RS-232
Output(http://www.piclist.com/tehref/member/RHM-SSS-SC4/TIC232.htm), log the
difference readings over time on a computer, and feed the data into an Allan
deviation calculator. A true Time-Nut would use Stable32 for detailed data
analysis but at $395 this is not low cost. The free AlaVar calculator
(http://www.alamath.com/alavar.html) can remove any time offset, scale the
readings, has a built-in phase to frequency function to convert the time
data, and generates an Allan deviation chart (be sure to set Tau for the
sample period selected) that shows the stability of the GPSDO over time.
This provides a complete test system using 5-IC's for less than the cost of
the box it's mounted in.
     The Simple Time Interval Counter uses a Phillips HC4046 phase detector,
a 16F688 PIC, a 16M XO, and a MAX232 for RS-232 output. The low cost XO
"dithers" the TMR1 clock to provide good averaging, but a high stability
OCXO can be used for the TMR1 clock for high accuracy testing. Insure the
TMR1 clock and the DUT are from different sources for proper averaging.
Adding an offset by inverting the DUT 1Hz to the TIC improves operation
where the inputs are closely aligned. My test box uses a 14-pin 16F630 TVB
divider with a 5M/10M frequency select pin and the arm and frequency select
pins are driven from the TIC to allow arming and frequency selection by
serial port command. An inverted 1Hz output is provided for the TIC, and a
1PPS with 100us duration aligned with UTC provides the DUT 1PPS output.
(Thanks TVB for making this so easy!)

----- Original Message ----- 
From: "Didier Juges" <didier at cox.net>
To: "Discussion of precise time and frequency measurement"
<time-nuts at febo.com>
Sent: Sunday, July 30, 2006 11:04 AM
Subject: Re: [time-nuts] How do I know my GPS stabilized oscillator is
working?


> Along those lines, I have a question.
>
> I am planning to (finally) finish my GPS stabilized OCXO using a Jupiter
> receiver and an HP 10811-60102 OCXO recovered from a defunct HP 8672A.
>
> The question that comes up is: how do I know my GPSDSO is working and
> how do I evaluate its stability?
>
> I understand I could measure the control voltage from the phase detector
> to the OCXO and plot it on the computer easily enough, but that will
> only tell me the error between GPS and OCXO, which can be significant in
> the short term, and that won't tell me who is right (even though in
> theory, if the OCXO is any good, it should be the GPS, then within a few
> hours or half a day, the GPS should take over).
>
> I have a Takeda Riken (Advantest) counter TR5823H that has the high
> stability oscillator option (not as good as the HP OCXO) also and Time
> Difference capability but no GPIB port, so I cannot get the output to a
> computer. If anyone has the schematics for that counter, I may be able
> to jury rig something to get the display to a computer.
>
> I have a couple of HP generators that have their own OCXO, I believe
> they are also 10811 (a HP 8673M synthesizer and a HP 3586A with high
> stability option). I also have another HP 8672A (on loan for now, but I
> can get it back if needed) and I have an EIP 371 counter with an OVENAIR
> OCXO (part number unreadable until I remove the OCXO from the chassis,
> and I do not have the spec for it). Finally, I have a Tek 494P with the
> excellent OCXO Tek has put in it (I do not know who makes it). So any of
> these instruments could be used as a reference against which I could
> measure my GPSDSO, but how do I know who is right?
>
> Aside from the fact that some instruments specify the stability in days
> and others in year does not help determine the best oscillator, which
> should I use and how should I interpret the results? For instance, is
> the Tek 494P at 1x10-7/year better than the HP 8673M at 5x10-10/day?
> Since the HP 3586 uses a 10811 OCXO at 2x10-7/year vs. 1x10-7/year for
> the Tek, it seems the Tek has better long term spec, but what does it
> mean for short term stability?
>
> Should I go out and by an HP counter with time difference and GPIB?
>
> Is there another alternative that will not require another piece of HP
> gear in my lab?
>
> Here are the specs I found for the oscillators I have:
>
> Tek 494P       1x10-7/year (after initial 6 months)
> HP 8673M     5x10-10/day (after 10 day warmup)
> HP 8672A      5x10-10/day (after 30 day !!! warmup)
> HP 3586A      2x10-7/year
> TR 5823H      5x10-8/day
>
> Thanks in advance for any suggestion or comment.
>
> Didier KO4BB
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts




More information about the time-nuts mailing list