[time-nuts] Re: Composite sky GPS - with an added digression

[Lux, Jim]

On 6/30/23 4:52 PM, glen english LIST via time-nuts wrote:
> Thanks for the replies so far. and very Interesting Jim !
>
> application is improve the GPS fix for field portable installed 
> TABS/conspicuity in experimental aircraft with metal skins or carbon 
> fibre skins. In a nutshell they broadcast a 1 sec intervals on ADSB 
> 1090 MHz the posiiton, and info) . they are usually suction cup fitted.
>
> There are conspicuity units that suction cup to the inside rear window 
> and they are terrible for fix, +/- 300m in H, +/- 200m in V.  even 
> worse if a high wing and a wing in the way that is full of fuel .
>
> It's passable on the front of the instrument panel at the forward end 
> where it can see a fair bit of the sky through the windshield, but not 
> great for vis, and also affects compass swing . The units are very 
> popular with recreational experimental, and assist to avoid traffic.
>
> Real TSO GPS require GPS antenna on the top of the skin to get an all 
> sky view.  (like my PA28)
>
> Possible solution is 2 or 3Â  antennas looking at different parts of 
> the sky but my feeling is most likely just one antenna on either side 
> window left and right would eb a substantial improvement. I am yet to 
> QUANTIITIVELY see exactly why it is so poor- I need to put one of my 
> UBLOX radios there in the device's place and get some data.
>
> sidenote - As an RF guy, I would have thought combining the signal 
> from both GPS antenna (downstream of their LNAs)Â  would lead to a fair 
> bit of infighting, and be highly sub optimal  . maybe not as bad as I 
> imagine.
>
> -


Assuming the feedlines are equal length, and you combine them. You'll 
get classic grating lobes where one is exactly in or out of phase with 
the other.  BUT.. since the pattern is already rolling off (a typical 
patch has a 90-100 degree HPBW), both of them are fairly far down at the 
"dividing line" equi distant, so it may not be as bad as you think.

This is easy to model in a spreadsheet - Just assume the antenna has a 
cos^2(theta) kind of pattern and then calculate the sum of the two 
signals (including the phase difference for the angle off vertical, 
which depends on the spacing between the antennas).

There's enough other lumps and bumps in the pattern from surroundings, 
the grating lobes might be in the same general magnitude.   Or just try it.

One thing to watch out for: if you're using antennas with built in LNAs 
- it's easy to build an oscillator - the amplified output of antenna 1 
goes through the splitter and comes out the port to antenna 2 and if the 
reverse isolation of the LNA isn't good enough, it radiates back to 
antenna 1.  Fortunately, the antennas are narrow band, so it might not 
happen to oscillate.  But if you get weird results, try covering one of 
the antennas with aluminum foil (or absorber, if you have some).

And these days, you can get a spectrum analyzer that goes up to 2GHz for 
about $100Â  - it's not awesome, but it sure would see an oscillation at 
GPS frequencies, and would prevent all the GPS users around you from 
hunting you down like a rogue wolf.