Tracking a radio signal

[Mike]

Thanks for all the help guys. I've been busy writing proposals/preliminary design reviews/requirement documents so that we can secure some equipment and funding.

We're moving forward with the Doppler idea. What kind of precision can I expect from say, an 8 antenna array? I realize there may be more factors than the amount of antennas, just throwing the question out there.

Adding GPS to the beacons has been ruled out for two reasons
a) Financial scalability. Much easier to secure government funding to put these beacons in every house in the < $10 per range.
b) The more science experiment aspect of this project with the Doppler technique.


Thanks for the help guys, this is why I came back to CD after a few years when I had a problem that required the brightest.

[DonRotolo]

If you are moving and accurately keep track of your motion, it's possible to use math to bring the accuracy well below 22 degrees. That's just the precision of the Ramsey kit. There are alternatives.

I'm doubtful that adding antennas will help accuracy or precision, instead it may just add complexity for no performance gain.

As for the math, think about watching the display with your eyes and noting where it points when the dots change, do that enough times and you can narrow down the most probable direction. Or, to put it another way: Triangulate from ~50 points and you can narrow down the 'zone'.

[EricVanWyk]

Originally Posted by Don Rotolo

I'm doubtful that adding antennas will help accuracy or precision, instead it may just add complexity for no performance gain.

It took me quite a while to take the mental leap to where adding antennae help. I was stuck on the mechanical analogy of having an over constrained system: Having more constraints just seemed redundant. The way I got there was imagining a probability beam coming from each antenna and over-lapping their fields, and comparing the process to over-sampling. Each antenna gives a slightly different piece of the puzzle.

It helped that my intern cube-neighbor / lunch partner really knew what he was talking about

[rsisk]

So, speaking of radio waves, check out this video using an LED probe to visualize the radio transmission from an RFID device, nice lobes.

http://www.popsci.com/gadgets/articl...using-led-wand

[DonRotolo]

Eric:

Yes, it is absolutely true that adding data to the system will often provide for a more complete solution. However, for this application I'm not convinced that, say, 8 antennas will improve the pointing direction accuracy by even 10 degrees.

Doppler isn't working on signal strength, but the doppler frequency shift caused by the (electronic) rotation of the antennas. In order to have a detectable frequency shift, the antennas need to be "rotated" very quickly. More antennas means more switch 'changes' per second, less time to capture and process the doppler-shifted signal, and the design and performance burdens associated with having to do all this faster and with less signal.

We often use 4 antennas because it is easier to do the math, however 3 would also work, as would 8. Some Doppler units DO use 8, but I have yet to find claims that they point to the transmitter any 'better' than a 4-antenna unit. All that is gained is capture area, which is only useful if the signal is close to the receiver's lower limit (i.e., weak) - and even then, the effect is marginal.

Thus my statement.

[EricVanWyk]

Originally Posted by Don Rotolo

Doppler isn't working on signal strength, but the doppler frequency shift caused by the (electronic) rotation of the antennas. In order to have a detectable frequency shift, the antennas need to be "rotated" very quickly. More antennas means more switch 'changes' per second, less time to capture and process the doppler-shifted signal, and the design and performance burdens associated with having to do all this faster and with less signal.

That is really cool.

The system I looked at worked a little differently. I know I'll mangle it, so please be patient . As I understood it, the system had N omni-directional antennae, and measured the relative phase between any two antennae. This created N^2-N virtual pseudo-directional antennae. Each of these virtual elements could tell you what direction it was relative to the line that passed between its corresponding real antennae, but had a couple of faults: Its accuracy was a function of direction, and it couldn't tell front from back from top to bottom - it carved out a fuzzy hollow cone. With enough of these fuzzy hollow cones, the original direction could be teased out.

With that type of setup (aka money is no object), more does equal better. I can see why with the other set-up (wicked sweet), more equals more complicated. Science is awesome.

This virtually spinning doppler thing looks really neat, I'm going to have to check it out! Thanks for the info

[DonRotolo]

Essentially, doppler "spins" the antenna in a circle very fast - fast enough to create a measurable doppler shift at several hundred Megahertz. You could just mount a single antenna onto a turntable (think phonograph) and spin it really fast (think CIM at full speed), but mechanical issues favor the electronic switching method, since the difficulties created by switching are easier to manage than the mechanical ones of spinning an antenna really fast.

Your system is phase-differential time of arrival (ToA), which has the potential to be more accurate but at a two or three order-of-magnitude uptick in complexity. You need N receivers in most implementations, fer instance.

[Mike]

Ahh busy busy busy. Proposal is done, will be sent back to me soon for editing.

Seeing as we have deadlines (little longer than six weeks) I can't really go from teaching the younger IEEE members what a resistor/capacitor/etc. is to designing a doppler DF. Plus I don't really have the time myself to go through that prototype/troubleshoot/etc. process. I've looked around and I think we're going to go ahead and use the PicoDopp system, http://www.silcom.com/~pelican2/PicoDopp/PICODOPP.htm and then integrate it into our flight hardware.

Again, thanks for the help everyone. I'll try and keep the thread updated.

[DonRotolo]

Good choice on the PicoDopp. Didn't know it had GPS input, that'll help.