Let's Make Robots!

Towards a auto-aiming servo controled directional wifi antenna

I'm planning a project to get an auto-aiming directional wifi antenna working to provide a long-distance repeater for a boat at anchor connecting to access points we don't control.I think this is a robot, but a simple one.  I think it's a robot because it senses its environment and responds to it.  Sadly it won't be very pretty but it might be useful.  It will definitely be fun ;)

The reason it's worthwhile thinking about this is that the boat rotates through 360° on the anchor line.  This means most people use an omni antenna, but the range is not always sufficient and there is a ton of interference; worse the omni is pumping out even more interference to the world and that makes me sad .  Lots of interesting background at Wikia and best description of the issues here.

Since this is a repeater situation (ie multiple client wifi devices onboard) this is a 2 radio situation. Currently that means a commercial router making a local network onboard, linked over ethernet to the picostation. I looked hard at platforms with single radio/two antenna but they are almost always diversity antenna (and therefore won't work well as repeaters) or you get one antenna at a time and I can't figure out how to bridge and switching antennas at high speeds!

Inspired by the UAV homebrew antennas I think it's possible to have a high-gain directional antenna (most likely a panel) on a rotating platform, driven by a (single) servo motor (and some simple re-aiming algorithm based on monitoring the signal strength in software). I'm a little concerned about side-to-side chop so maybe one day this will be a pan-and tilt setup, but currently I'm planning just a panning setup.

As a basic platform I considered commercial routers (e.g. ASUS 520GU) connected via USB to a servo control unit (e.g. Pololu micro) but since we need 2 radios anyway (ie one long distance and one for low-power local network) I'm attracted the the Ubiquiti RouterStation platform. It has three miniPCI slots, USB, serial and a few GPIO connectors.  It's also powerful enough to run one end of a VPN tunnel (which is a nice to have requirement) and not too expensive at ~$62.

I'd be very interested to hear if anyone has an opinion on whether this can be used to directly drive a servo; RouterStation full specs here.  There's a good description of doing this from the Linksys WRT GL that gives me hope (lost the link right at the moment).

Currently I'm thinking of sticking everything into a StationBox which can have an integrated 14dbi antenna (for $23 box and antenna) and hanging it from the top of a pole, using the servo to "creep" around the pole.  Lots to think about there, might also do it just with a rotating platform and the antenna on top.

One challenge is how to cope with the rotation for the power supply; I found this site through that actually, searching for slip-rings and seeing the suggestion about the headphone plug.  I might use that or I might use a cable-wrap and just handle the unwrapping in software (when network activity is low).

Any thoughts welcome!

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

If horizontal is the required alignment then a simple balance gimbal(?) should fit your needs. Slip rings are a usual way to transfer power from stationary to rotating devices, search for POV clocks, I would guess many of them use some form of slip ring.

In my mind, maintaining an aim would require knowing which way the antenna needed to rotate, which means the system must know which way it is rotating. On that note, I envision what you are building as a balancing bot that is always trying to keep itself aimed at a distant antenna.

You are going to need a wired connection to the AP that you put in the box, right? Maybe you could use Power over Ethernet PoE to cut down on the extraneous cable runs.

My 2 cents anyway.

Thanks birdmun.

Yeah, the gimbal is an excellent idea, especially since it's unlikely to need to move more than like 20° on the horizontal. Or maybe just an electric gyro and an extending arm to maintain horizontal. It's kinda of an open question as to whether fluctuating signal strength is going to be sufficient to guide the antenna back to the right spot, or whether that in combination with an accelerometer/gyro (so you know which way you moved and can move back without starting in the wrong direction) will be needed.

I considered using an electric compass and just locking into a heading, but since the platform (ie the boat) swings on a 50'+ rope the needed angle actually changes, so locking onto the signal seems better (especially as it can find the best signal itself that way).

The POV clocks are awesome, lots of good ideas there.

PoE is definitely handy and the board does have it. The cool thing about the RouterStation has two radios (actually potentially 3 :) I'm actually not going to need an external wired connection, will just need to run in 12VDC, which is great because that's only 2 wires to get through the slip-ring (if I go that route).  Not sure yet if there's a way to get enough voltage off the board for the servo, as a last resort there's always the 5VDC from the USB, but I think there's probably another way (serial headers?). Also not sure if there are enough input/ouputs to handle the servo and an accelerometer.


From what I have seen mentioned on this site more than once, servos are power hungry. 1A has been mentioned often. USB doesn't typically offer more than .5A.