[time-nuts] Sub mm measurements with gps timing antennas?

[Magnus Danielson]

On 04/25/2012 08:42 PM, Attila Kinali wrote:
> Moin,
>
> On Wed, 25 Apr 2012 10:05:00 -0700
> "Tom Van Baak"<tvb at LeapSecond.com>  wrote:
>
>> Are you sure the customer said sub-mm and not sub-meter? I know
>> post-processing is really helpful, but the LEA-6 is a single frequency
>> receiver so all the advantage of L2 is lost for this customer. The
>> bullet antenna's don't even have an arrow for North ;-)
>
> Yes, it is sub-mm. They are already doing sub-cm with the current
> setup, but it isn't precise enough yet. IIRC during the first tests
> last year they got<4mm precision. And yes, using a LEA6-T with a Trimble
> Bullet antenna. I don't know what they actually do with the raw phase
> data, beside that they average over several hours and use a 100m baseline
> reference consisting of two stations mounted at a fixed position.
>
>> One thought -- seeing how this is a research project. It might be
>> possible to cross-correlate the post-processed data against the
>> Az-El of each SV along with ambient temperature over days or
>> weeks and thus actually measure the phase center profile as well
>> as tempco of the system. This would be no small effort, depending
>> on the math and programming skills of the researcher(s), but the
>> advantage for them is that is costs time instead of money. Then
>> armed with this "calibration" data (possibly unique to each unit),
>> it would be possible to reduce these effects, improving precision.
>> I have no idea how much. Still, an interesting project.
>
> Hmm.. That would be an idea, but i don't know whether it is feasible.
>
> Maybe i should here expand a little bit what the project actually is.
> The effort is part of the Permasense[1] Project, which does measurements
> of different parameters of permafrost soil/rock in high alpine regions.
> The idea of this sub-project is to measure the exact movement of rocks
> and other "solid" and "unmovable" things with regard to weather conditions.
>
> Beside the harsh environment, this also means that the devices cannot be
> reached for most of the year (there is usually a 3 months window when
> the devices can be accessed)..and some years they cannot access them at all
> (adverse weather conditions preventing ascend). For installation, a helicopter
> has to fly everything up into the mountains, which means that pre-assembled
> stuff cannot be transported, because it's too bulky (prevents whole
> system calibration in a climate chamber). After installation the system
> runs on solar power with a backup battery. But this doesn't guarrantee
> power at all. The solar panel could be below a meter or two of snow.
> Hence the whole system has to cope with periodic power los and has to
> be as low power as possible (ie no OCXO, no Rb, no temperature stabilization).
>
> The snow also prevents the use of choke rings, because they would accumulate
> a lot of snow and ice, which would then cover the whole antenna.

If you buy the right choke rings, they come with a plastic cover with 
fairly steep slopes, so snow-buildup would not be dramatic. Just bring 
it up from the ground with good support.

> But at least, there is hardly any electronic interference, a good sky
> view (no trees), unless the system is mounted near a wall. And the
> antenna cable is quite short (it was 15cm in the previous version
> of the device and should be<50cm in the next)

You get a well defined phase-center, and also known. You can then bring 
in calibration files and reduce the phase-center in the post-processing. 
This is standard stuff, and there is plenty of information online on it.

Consider that the C/A code has a rate of 1,023 Mchip/s, so anything 
within 150m will not be de-correlated by the C/A code. Here is another 
benefit of the P(Y) recievers, they bring this into 15m radius.

>> A simple test that could be done locally (refrigerator, sauna, etc.)
>> would be to measure the tempco of the entire system (antenna,
>> cables, LEA-6T) before they deploy it to a mountain. It may also
>> be the case that the system has both a temperature coefficient
>> and a temperature change coefficient so it's not a simple 2-point
>> test. You can probably ignore humidity and barometric pressure.
>
> I think the ETH has climate chambers that could run such tests.
> But i'm not sure how you'd test the antenna in such a chamber.

1) Network analyzer measuring phase delay and group delay in the range 
of interest.

2) GPS simulator times from a rubidium, and then compare how timing 
deviates.

>> Another test would be to rotate the antenna at 1 RPH (revolution
>> per hour) and then look for modulation in the post-processed
>> solution.
>
> Hmm.. that's actually a quite nice test. Cool idea, thanks!
>
>> This would give a hint of the quality of the antenna. As
>> a baseline, try the same test using a precision gps antenna. I
>> have spare pin-wheel, choke-ring, and ground-plane antennas
>> that I could loan, but surely these are available where you are,
>> and probably cheaper than postage from here.
>
> Yes. The problem is, antennas that perform well in a city environment,
> where temperature swings are quite limited, fail in high alpine environment.
>
> But that's what my question originally aimed at. How much better can
> we get using a better antenna? Is it worth doing? Or is it just a
> waste of time and money?

Anther benefit of using a better antenna, is that it is intended for 
20,46 MHz bandwidth around L1 and not 2,046 MHz bandwidth. This 
translates into lower Q and lower group delay to have temperature 
dependency on.

> BTW: what's a pin-wheel antenna? Google tells me contradictory things.

Look at Novatel
http://www.novatel.com/products/gnss-antennas/high-performance-gnss-antennas/

It is a "new" class of antennas which has a bit different buildup 
compared to the patch antennas and the helix antennas. It's a 
combination of a few techniques and they end up giving the choke rings a 
run for the money at the performance they give. Phase center is well 
defined, surpression of multipath is good. Size is decent. You can put 
one in a back-pack as you go up the mountain.

>> It seems that everyone else that does sub-ns precision timing or
>> mm positioning uses a large combination of tricks: dual-frequency
>> antenna and receiver, geodetic-quality antenna, passive or
>> active temperature control, phase-stabilized cables, GPS and
>> Glonass, external frequency reference, and post-processing.
>> Your customer is only using one from this long, expensive list.
>> So there may be a lesson there.
>
> The problem with most of those techniques is, that they are not available
> for the price the customer can afford. A dual frequency receiver costs
> a lot more than an of the shelf LEA6-T. Also these modules are usually
> build with larger power budgets in mind, e.g. the Trimble BD920 uses
> 1.3W typical, while the 0.3W max(!) of the LEA6-T already hurt us a lot.
> Using an external frequency reference is not possible with the LEA6-T.
> It would be possible to do that when using one of the GPS chipsets from
> u-blox, but therefor we would need to take at least a full reel (iirc 2000
> pieces), which isnt exactly cost efficient. Beside, we would still need
> to use a TCXO, because there is not enough power available for an OCXO
> or even an MCXO.

The CSAC was actually intended for this very purpose. Good frequency 
stability for the size and power.

Cheers,
Magnus