[GPS_Standard] No joy here

[Dave Platt]

On 02/14/2012 03:28 AM, Chris Howard w0ep wrote:
> 
> 
> It is a new day today and things are looking better.

Excellent!

> I used my frequency counter to determine that direct
> input of the control voltage did change the output
> frequency.
> 
> I attempted to use a shortwave radio to hear both
> WWV and my box, but I had to go to the far end of the
> house to hear WWV at all.  That got me wondering whether
> the little GPS board sitting 2 inches from 5 unshielded
> wires carrying 10 mHz was being influenced. So
> I installed shielded wiring in those places.  (probably
> should also on the connection between the controller and
> the VCO). 

Probably a good idea.  Some oscillators have a low-pass noise
filter on this line;  others might not, and RF noise creeping
into the varicap might make the oscillator somewhat jittery?

>                       I wondered if I should also put the GPS
> in it's own sub-enclosure, haven't decided.

For what it's worth:  I have my system built into a relatively
small box (formerly a wireless access point controller):  Motorola
Oncore GPS board, oscillator/oven board, VE2ZAZ controller, and
a "schmoo" interface board.  No special sub-enclosures, no
extensive shielding, and the 10 Mhz output signals are just
going through short hookup wires to the BNC jacks.

I was careful to ground the GPS receiver to the chassis using
metal standoffs.

The GPS, VE2ZAZ, and oscillator board each have their own
+5 or +12 regulator(s).  I even added a second 7805 regulator
"piggyback" on the output-driver IC, to reduce the degree
to which connecting and disconnecting output loads would
make the board's main 5-volt rail drift around.

I haven't had any problem getting the Oncore to lock onto
the GPS signal, and present what seems to be a stable PPS
signal... the amount of jitter/error in it seems to be
consistent with what I would expect.  It locks and works
well even with a simple patch antenna mounted up under
the roof.  I do have a better outdoor-mount "bullet" antenna
to try but haven't gotten around to installing it yet.

> Let me know if you see anything else that I should
> attend to with the shielding. It may have just needed
> tweaking the whole time, but I'm pretty sure
> the shielding helped.

It could well... keeping the 10 MHz signal well away
from the GPS signal path can only help.  Whether you need
additional shielding, may depend a lot on the specific
GPS receiver you used.

I'd also suggest paying careful attention to power
supply regulation and decoupling... keep noise from
getting into the GPS receiver via its power supply
lines.

I've been playing around a fair bit with my own system
in the last couple of months, trying to figure out what
FLL tracking settings would be most effective.  I added a bit
of automation to the process, experimenting with automatically
changing the sampling duration when the system was tracking
stably, and reducing the duration to allow for faster oscillator
adjustments when the FLL was doing a bunch of "coarse" adjustments
in the same direction in rapid succession (sort of a poor man's
proportional-controller add-on).

You can see some of the results at

  http://snulbug.mtview.ca.us/dave/fll.png

Legend:

Blue (top) - unlocked/holdover/locked status.  It's spending
most of its time "locked" with a few "holdover";  the only time
it unlocks is if I power-cycle it.

DAC value (green, left-hand axis) - you can see it cycling up and
down each day (temperature variations in my garage), with an overall
drift downwards at first (oscillator setting down after having
been powered off for several months).

Sample-to-sample error (red, right-hand axis) - staying within +/- 5
counts most of the time (+/- 500 nanoseconds) which is probably
to be expected for an Oncore GPS receiver.

Long-term error average in parts per billion (magenta, right axis)...
recently the error average has been staying within +/- 1 PPB (10^-9)
almost all of the time.

Sampling duration (cyan, right-hand scale with a scaling/offset
factor I can't remember at the moment)... shows my "autotrack"
program's adjustments to the sampling duration based on recent
FLL controller adjustments to the DAC voltage.

At this point it looks as if the major perturbing factor in the
oscillator accuracy is still the temperature variations in my
garage.  I changed the oscillator coarse-adjust circuitry over
to a more stable +5 reference a few weeks ago;  this helped
somewhat, but I think I'll need to temperature-stabilize the
whole cabinet to improve matters further.  I've made up a
secondary-oven controller board to add to the chassis, when
I get a round tuit :-)