Microwave Imager Updates.

When I built a DIY microwave imager earlier this year, I left some of my code unfinished. The high-resolution option seemed a little tricky at the time, since it used an unreliable and little-documented feature of the Dish Tailgater known as a “nudge”. This command, sent over USB serial connection or from a set-top box, would run the brushed motors in the antenna for just a second, pushing the antenna slightly closer (hopefully) to the best signal. Each azimuth nudge is approximately 0.2 of a compass degree, although as I found out later, this wasn’t the case for elevation.

Obviously, I wanted to improve my original low-resolution scan (seen above), that showed geostationary TV satellites in the Clarke Belt. In the above image, each pixel represents one degree of azimuth and one degree of elevation. Panning the dish back and forth through 180 degrees of Southern sky took a whopping 3+ hours to complete. Due to a quirk of the antenna programming, that was the absolute fastest I could make it run. And even that took some fancy handling of the signal data being returned by the serial terminal.

Below is a close-up of the inset box seen above. This is still using the low-res code, where each colored square is one degree wide and tall:

Originally I had the dish scanning back and forth in alternating directions. This was (slightly) faster, and looked cooler, than having it return to the origin azimuth for each elevation. However, I had ongoing issues with gear meshing (switching from clockwise to counterclockwise had some slack or play in the motor). I also had issues with my indexing that never quite went away no matter how I massaged the python data array or bitmap. Making things worse, the “nudge” motor runs aren’t consistent in each direction. Clockwise nudges are a different amount of antenna travel than counter-clockwise, so the image slowly drifted off at an angle. Commenters on Youtube and Github kept suggesting I ditch the alternating scan and just go in one direction each time. At the expense of my cool looking dish motion, I finally gave in and did that. The result is that high-resolution now works!

I wasn’t certain this would work at all, even after getting the motor movements to cooperate. For one thing, the beamwidth of this little 14″ dish is more than a degree, so I was worried that 0.2-degree movements would just give me a mess of noise and artifacts. It turns out that the smaller movements do get a better picture, although you can still see some fuzz and reflections around each satellite transponder source.

Another issue is that (as mentioned before), this Tailgater satellite dish isn’t designed to do any of this. I’m running the motors nearly constantly for hours at a time, when the typical TV-watcher-on-the-go would only run them for a few minutes and then leave it alone until they moved their RV / campsite / fish house. I noticed the more I ran the dish, the more horizontal bands and artifacts showed up in the high res scan. I also started hearing squeaking and grinding noises from the antenna as the poor overworked motor struggled to act as a radio telescope. I took the antenna apart and shot some silicone spray into the worst areas, but eventually I’ll probably ruin the thing!

If have or can find a Dish Tailgater and you’d like to try this yourself, my code is at https://github.com/saveitforparts/Tailgater-Microwave-Imaging.

Keep in mind that the high-res code is 5x more detailed in the x direction and 3x more in the y direction. so it will run 15x slower than the low-res version!

There are also some pull requests from people with suggestions to improve my code, which I have been shamefully ignoring since I don’t understand them and haven’t had time to test them out. If you’re better at Python coding than I am, feel free to poke around and make this better!