[amsat-bb] variable phase delay / phased arrays

[DC <dc@xxxxxxxxx>]

I throughly enjoyed Tom Clark W3IWI's "C-C Rider" 5cm satellite concept 
paper (can't find the link right now).  He discusses using phased arrays 
as an alternative to mechanically steerable dishes.  I think that phased 
arrays could be a really valueable tool for Amsat microwave links (both 
the terrestrial and space ends).  Some nifty tricks are possible: 
satellites could form multiple autotracking beams centered on current 
uplinks and could also actively null out interference sources, the 
redundancy and scalability aspects are also very cool.  I especially 
like the ability to dynamically control "effective dish gain" and tx 
power by activating and deactivating elements.  We could have satellites 
power a number of elements proportional to current activity (allowing 
formation of multiple autotracking beams) or proportional to current 
available power.  There are also some interesting options for 
terrestrial moble applications using these---autotracking at 60mph with 
no moving parts.  There was a recent slashdot article (day before 
yesterday) showing a flat phased array based mobile DTV reciever.

Down to the nitty gritty, lets build a recieve phased array for say S 
band (or C---doesn't much matter)...
Disclaimer: I've been researching these today rather heavily but I'm 
still rather "in the dark"

It seems that if we want to build a patch based array we'd need 25 
2.4Ghz patches for a 5x5 array.  I'm curious what the advantages of 
different array shapes/spacings are.  Would a conventional 2.4Ghz 
circularly polarized patch like we use for our feeds work well?  (say 
120 degree effective beamwidth)
Each patch would need its own LNA with a decent noise figure say 0.7db.  
Buying 25 DEM 13cm ULNAs would get expensive in a hurry but the actual 
active parts these preamps are based on cost $2-$5.  Parts I've found 
thus far (need help here) are Aligent ATF-33143: NF=.6db, $1.89 large 
quantity and there must be others.  The actual preamp reference designs 
I've seen are simple, one active part and a dozen passives + a voltage 
regulator.  We could make the reflector of the patches actually a double 
sided PCB (one side serves as the reflector and the other contains the 
preamp circuitry) and share common components like power regulation and 
minimize the number of large and expensive ($$$ and loss) connectors 
required. We might be able to get cost per preamp/patch module to 
$15/each or $375 total which isn't so bad.  I've seen some pretty 
expensive crossed yagis.

All of this has a catch:
How do we implement the variable phase delay for each of the 25 
patches?  I haven't found any, but do discrete phase delay modules 
exist?  One idea for an easy/cheap digitally controlled variable phase 
delay would consist of a section of 50ohm stripline on a board 1/2 
wavelength long.  The signal would be fed in one end of the stripline 
and PIN diode taps could be placed at intervals along the length of the 
stripline to provide access to the signal at various phases.  The ouput 
sides of all the PINs would be connected to a single output connector.  
By biasing one diode at a time, we could create a fast switching and 
variable phase delay board.  This could be implemented on the same board 
as the LNAs.  Note: my understanding of pin diode design is shakey, 
please let me know if there is a conceptual problem with this idea.

This idea only yields a discrete number of available phase shifts, some 
experimentation would have to be done to determine the smallest 
effiecient number of phase shift steps needed.  One other idea for the 
phase delay is to downconvert the the output of each patch/LNA to some 
reasonable IF and then digitize that IF and do the delay in software.  
The hardware required for this implementation is available now but would 
cost $$$, however in the near future this may be a better option because 
silicon only seems to get cheaper with time (its only sand after all).

What other ways can variable phase delay be implemented?

The output of all the phase shifters would be summed together and fed 
into your reciever of choice.   Some fairly simple geometry and bit of 
code running on your favorite computer are all that are necessary to 
calculate the necessary delay for each element to point the pattern 
towards a given sky location. 

Some things I'd still like to learn are:
-How many elements do we need for reasonable gain say 25db?
-Will conventional dish patch feeds work?
-How reasonable are my patch/LNA modules? Will the second board layer in 
close proximity to the patch reflector cause problems?
-Any comments on variable phase delay.
-Anything else that fits in well here...

Thanks for reading this monster,
-David Carr
KD5QGR

----
Sent via amsat-bb@amsat.org. Opinions expressed are those of the author.
Not an AMSAT member? Join now to support the amateur satellite program!
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org