Open RSSI Antenna Tracker

[kaos]

the 2 axis and 'planned features' are exactly what is needed/on my mind, especially the auto calibration.
Keep busy.

[Rob Thomson]

Just sorted auto calibration out.

Got the calibration mode working.

Very simple to use.

- Toggle calibration switch turns off servo, and starts a running average of the rssi diff.
- Take model 50 meters from you.
- aim center antenna at model
- wait for 10 seconds for stable signal
- Turn off calibration switch

Seems to work quite well

[kaos]

Looking good. This is only for horizontal plane right? I assume the vertical plane can be done at the same time?
Are you making that little girl to walk out 50 meter and hold the plane for you again?

[tilmanb]

You should see her calibrating the vertical axis.

[kaos]

You should see her calibrating the vertical axis.

yes, Like to see a vid that she is calibrating while Rob is running back and fourth the 50 meter and jumping up and down - gasping.

[Rob Thomson]

I just added a kalman filter to my rssi readings.

The whole thing feels a little smoother now.

Now.. Time to shelve this one, and start work on the pan tilt version!

Sent from my GT-I9300 using Tapatalk

[Bill]

kalman filter

Wouldn't a simple weighted average work just as well or better?

[Rob Thomson]

I tried it. It 'felt' more inaccurate

[Bill]

What happens when you are zipping around in front of you won't the prediction algorithm introduce a delay as you change direction?

[Rob Thomson]

Not an issue.

The only point this could be an issue is close in, and in that scenario it would fall back to the clover

And of course a prediction is re-sampled every clock cycle

Rob

[aadamson]

Not an issue.

The only point this could be an issue is close in, and in that scenario it would fall back to the clover

And of course a prediction is re-sampled every clock cycle

Rob

That would be *every loop cycle*... I doubt you are sampling at 16mhz ... Just saying...

I noticed your most recent code only samples 2 RSSI values, is your diversity setup such that the center antenna and the clover are on the diversity ports? And the other 2 are on their own RX so you have 3 RSSI values in effect (unless the diversity RX has 2, then you'd have 4), but you only need the two on the left/right?

Am I getting the gist of your process?

Alan

[Rob Thomson]

My diversity is already happening on my receiver. It is a dual receiver, so I had no need to keep the code in for it.

So basically, I am only sampling to left and right for tracking.

Sent from my GT-I9300 using Tapatalk

[aadamson]

Very cool... I designed a little STM board that I'm looking for a project to play with...

This little guy...


Basically an STM32F3 with 256k flash, 40kram, and runs at 72mhz... it's a M4 ARM core with hardware floating point ... Timers are already structured to support motor control etc.

I use STM's on lots of other projects as so this one has been in the works for some time, it started as an F1 design, but when the F3 came out I quickly moved it to it.

At some point, it's completely Arduino pin compatible... feature, function, etc.... An other friend is working on the maplelib for the F3 so you'd just be able to build arduino code and it would run on this hardware...

But that's not the purpose of this other than I just like STM's ...

I think I'm going to do something similar to what you've done but with Steppers and maybe an IMU and maybe some advanced features.

Couple of other comments...

if the diversity RX has RSSI, wouldn't you want to somehow know when you are on the clover so that the tracking doesn't try endlessly to pick up a better signal when high overhead. in fact, I wonder if you put this in a matrix and did some matrix calculations on it if you could *predict* when you were over head and heading in what direction (hmm... doppler would be good if you could figure out a way to do that) such that you could swing and be ready for the correct orientation when you were back on the beams... Just thinking outloud, and IMU could detect acceleration and rate and you might be able to *forward* predict the next step. Obviously, you could get fooled, but I'd like to see if it's possible to have smooth motion even though you never really leave the sweet spot in the beam width.... Getting smart about this with the right beamwidth *might* alleviate the need for a second axis I would think...

Lot to think about, but great job in putting the brain cells in motion!
Alan

[Rob Thomson]

You should check out this unit.

http://www.ebay.co.uk/itm/Pan-Tilt-asse ... 0851514226

I am planning on using it for my next version.

Stepper control etc..

For the receiver modules, I am going to use actual modules - not full built receivers.

http://pages.ebay.com/link/?nav=item.vi ... 0764762871

This works out much cheaper.

My plan is a full pan tilt system.

In practice, I have found that there is no teal need to check, and stop if running off the clover. You find that regardless, the seek happens. When you do go overhead, and loose lock, my tracker pauses for 5 seconds before starting a seek. The end result works quite well - with very little actual seeking randomly

Sent from my GT-I9300 using Tapatalk

[aadamson]

Very cool RF module... that's a great idea as opposed to a separate full RX.

[Rob Thomson]

If you open up the standard boscam receiver - you will find inside the same module.

The receivers simply add the dip switch and vreg.

Sent from my GT-I9300 using Tapatalk

[aadamson]

Yeah time for a board design anywhere to get the layout of that module and pinout?

[Rob Thomson]

This help?

http://iw.suntekstore.com/images_cache/ ... nengtu.jpg



Sent from my GT-I9300 using Tapatalk

[aadamson]

what I really need is dimensional data... What I'd like to do is create a simple board, think of it as a carrier board. It will have it's own power supply so that it can power what it's carrying.

2 modules
processor
interface for stepper driver (it's not work re-creating those when you can buy them all day long on ebay for around $2.00 us- built and they control 2 motors (look at the L298n).

In fact, maybe a step backward here... *if* we designed a simple carrier board... *what* would it include?

- footprint for a mini (could be my 32bit one or the pro mini)
- 12v in 5v/3v3 output for oh say 2amps
- stepper driver interface, this should include both low power, IO and high power (problem here is that steppers are all over the map in what they want to drive power)
- external interface - serial, usb???
- IMU interface just because I'd like to play with one... that's just a simple 2 wire i2c with power and ground


What else without getting too complicated?

Alan

[tilmanb]

Multiple modules on one board and a video switcher for diversity.
Plus a CPU. Voltage regulators and a few servo ports.
Just add a big row of headers for additional exotic stuff that not everyone needs.

[tilmanb]

I mean stackable boards

[Rob Thomson]

Guess it really depends on how complex you want to get - and if it is worth the effort.

My intention is to just 'bread board' and direct solder up to the modules. For the Arduini/Stepper drivers - I am simply using 'plug-able' modules for everything.

One of the difficult things with this sort of system is making the parts small, and light.

In my next version, chances are my receivers will be stacked on their own PCB, up on the rotating head of the antenna. The power & rssi will be fed down through the slip rings to a secondary board containing the stepper drivers, voltage regulators etc. This way the actual moving antenna head can be kept light enough to avoid over taxing the motors!

[Rob Thomson]

I had a box of receivers arrive yesterday.

So bar a few minor bits I am nearly ready to start building tracker ii

Sent from my GT-I9300 using Tapatalk

[kaos]

this thing is getting better and better.

Rob: I do want to see a vid when you calibrate the tilt function.

[Rob Thomson]

Nothing to it.

Just point at model, and calibrate.

It should not matter about vertical height.

Sent from my GT-I9300 using Tapatalk

[Rob Thomson]

Just had a change of plans.

Ordered one of these pan tilts.



Its an ATPRO-1000

[kaos]

did you order the kit with servo? or just the kit and use stepper motor?

[Rob Thomson]

With the servo. No reason other than that they will be garuanteed to fit.

No issue doing then code with a stepper. If you wanted to build one like that I could probably sort a test rig.

Sent from my GT-I9300 using Tapatalk

[kaos]

I may be ignorant. The reason I think stepper motor works better than servo is when the rotation gets to the end of the servo turn and pass it, it has to rewind 360 degree, but with a stepper motor you don't need to. Am I thinking it wrong?

[aadamson]

I may be ignorant. The reason I think stepper motor works better than servo is when the rotation gets to the end of the servo turn and pass it, it has to rewind 360 degree, but with a stepper motor you don't need to. Am I thinking it wrong?

yeah... cables.....