[time-nuts] Re: Lady Heather v6 for Raspberry Pi......?

[Lux, Jim]

On 6/4/21 4:02 PM, mpb45 at clanbaker.org wrote:
> Hello, Time-Nutters--
>
> One of several things keeping me off the streets and out of
> trouble these days is a home-brew system for downlinking and
> displaying imagery from the NOAA low-earth polar-orbit
> environmental satellites.
>
> A satellite tracking software package runs/controls the AZ
> and EL gear-motors that steer the 10-ft diameter dish.
> This dish beamwidth at 1700 MHz is only a couple of degrees.
>
> If the tracking program does not have pretty good time
> information the dish AZ & EL tracking is not close enough
> to provide a solid signal from horizon to horizon.  The earth
> orbital periods of these NOAA birds is very close 100 minutes.
> Seems like they pop up over the horizon and disappear in just
> a couple of minutes. 

I've done quite a bit of tracking of LEO satellites with 3 meter dishes 
- there's a few key things:

1) the movement is on the order of "1 degree/second"

2) It's not like pointing beams with 10-20 degree beamwidth, where a 1 
second error makes no difference - this is the fatal flaw on adapting 
amateur radio VHF/UHF tracking software.

3) Many tracking systems are "dead on" when sending the command to the 
positioner, but don't allow for the time delay in sending the message 
and then the actual control loop doing  positioning, so you wind up 
always a bit behind.   This is really annoying for the modem, since 
you've got a 1 (TBD) Hz sawtooth amplitude variation as the set point 
jumps every second.

4)Â  Then, there's two aspects to motion if you go in 1 second steps.

4a) If the motors have enough torque, the dish actually "steps", with 
some shaking on each step (not good for high rate data).

4b) It's not just the amplitude jump, it's the phase jump, because the 
phase center (usually at the feed) is moving a significant fraction of a 
wavelength, and the phase locked loop in the receiver doesn't like 
transients. The feed is often about a couple meters from the axis of 
rotation, so a 1 degree move of the dish is a 4-5 cm move of the phase 
center.  At S band (15 cm lambda) that's 120 degrees of phase!

See the attached plot - several dB variation in received power - two 
factors involved - one is that it's a second (or more) behind, so it's 
"off the nose" of the pattern and over on the steeper part of the 
pattern where it varies much faster with respect to angle, the other is 
that it's going in steps.


A more subtle problem is if there's an interfering source, it's phase 
and amplitude also move in discrete jumps.

Take home - You want "smooth" motion - The best way is to have a fairly 
fine time step on the commanding of the loop, and a loop bandwidth that 
is chosen appropriately to get a nice continuous motion in both az and el.

Some controllers have a feature where you can load the track in, and it 
smoothly interpolates the motion (cubic spline or something like that).

Then it's a matter of precomputing the axis motion before the pass, 
loading it in, and making it start on time.

Interestingly, you don't need super accurate time-nuts timing - 1/2 
second is a fraction of the HPBW so the gain loss from pointing error 
isn't all that big - the key is the smooth continuous motion.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: rcvpwr.png
Type: image/png
Size: 24688 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210604/25811f34/attachment.png>