I uploaded some photos of my new minisumo project (tentatively called The Thin Man) here, and a photo of my latest circuit board project here. I haven’t had much time to get the software written for the circuit board yet, but it appears to be working as designed with a couple of quick tests.

Ever since my trip to Calgary for the Western Canadian Robot Games I’ve been thinking about a new mini sumo robot that would be competitive with the best robots I’ve seen. My current top-contender Velox really hasn’t been very competitive for quite a while now but I wasn’t sure I really wanted to build another mini sumo. After all, I’ve also got a few other projects on the go already…

Last weekend I finally decided to start building. My new design puts emphasis on a low profile and speed. My goal is to make a thin wedge that is less than 3/4″ tall and moves at 5 inches per second. It’s going to be a very challenging project. I’m starting with a block of aluminum that is a little over 10cm square and 3/4″ thick, and I’ll cut and mill away most of it, incorporating the motor mounts, front scoop, PCB mounts, etc. into one frame. I’m also planning to design my own PCB that puts all the electronics (including sensors) onto a single board.

The machining should be complete in two weeks or so (mostly evenings and some time on the weekends) then I’ll start on the PCB design. I plan to compete at Robothon in Seattle in the first week of October so there is no time to spare – especially since I need time to prepare my other robots too!

I spent last Saturday at the Western Canadian Robot Games in Calgary, Alberta. This was the 16th annual event but I had never been to it before. My friend Dave Hylands has been going every year for I think the last four or five years and always suggested I attend. He was right, it was a lot of fun.

I took some pictures (not great quality, but whatever). Find them here along with a short description for each one. The crowd was reportedly smaller this year than last but it was still well attended. The setup was very professional and the hosts were well organized.

I entered three robots: a linefollower, a minisumo and my walking metal hexapod. Eileen says it looks like a bug so I call it Metal Insect – got any better suggestions, leave me a comment. My linefollower did ok but not great. My minisumo didn’t do much better, its really starting to show its age against the tiny fast wedge robots. But the Metal Insect did catch quite a few eyes and was awarded Coolest Robot by the judges. I really was very proud as I spent quite a bit of time giving this robot some personality.

An annual tradition at the WCRG is the Solarbotic’s Barbaque afterwards. Dave Hrynkiw and the rest of the Solarbotics crew host a big dinner every year. It was great to spend another few hours talking about robots while quaffing beers and chowing down. I got to meet Dan Gates in person- he joined Solarbotics recently and is already busy creating new designs for them.

I’m definitely going back next year, I had a great time. I’m already planning a new minisumo robot, and I hope to have my Robomagellan robot working.

A few weeks ago I wrote about getting some printed circuit boards made for my robot projects. They arrived in the mail today and they look pretty good.

I also received my Digikey order of parts last Friday. That company is excellent to to work with – a pretty good website with a good search engine, links to data sheets, good prices and easy shipping options (basically a flat fee using Purolator, Canadian duty already included). My only complaint is their search engine doesn’t use permanent links – the link expires after a few minutes so its hard to send a link for a part to someone else.

I’ll probably start assembly tomorrow or Wednesday. Probably will have one of the boards done by the time I leave for the WCRG Robot Games on Friday night.

I’ve just finished my second and third printed circuit board designs and sent them off for production. The Spark Fun guys have a really cheap deal for hobbyists as long as you (1) aren’t in a huge hurry and (2) are willing to accept back what you send them. User beware, you have to proof you designs carefully as they aren’t going to do it for you. Which is why its cheap.

My first board was made for my small six-legged walking robot built by following the mechanical construction described in Karl Williams’ Insectronic book. In addition to the mechanics the book shows you how to make a circuit board using a PIC controller but I wanted to use an Atmel AVR. Plus I added an ultrasonic sensor with a pivoting head which required more pins than what the original design could support. Hence I made my own. It’s a very simple circuit (before laying out the board I built the whole thing on a breadboard in about 30 minutes). It uses an Atmel processor, a crystal oscillator, a bunch of headers and an LED or two. As I said, it’s simple. I really should find some time to take pictures + movies and post them, its a cool little robot.

This time I’m using surface mount packages for most everything on both boards. And these boards are more complicated. In addition to the Atmel processor, both are using a Microchip 2515 CAN controller + a transceiver. The first one is using a chip from FTDI that allows the processor to be accessible from USB. It’s basically going to be a bridge between my desktop or laptop and a CAN network. The second board doesn’t use the USB interface but instead plugs into a nifty little GPS module; this module requires a voltage level shifter to translate to the 2.85v used for serial communications rather than the 5v used by everything else. The idea is to create a board that reads GPS data and writes it back to the CAN network.

The last time I used the the Spark Fun service it worked out very well but I noticed the layout was shifted on the actual board a small amount from what I expected it to be. Fortunately I was very conservative with my traces so the board still works even if its not “perfect”. I used Eagle CAD again for these latest designs and carefully marked out the board size, thinking I must have made an error last time. But the Gerber files that got saved were offset again (and both boards were offset by a different amount). Seems like its a bug in Eagle.

I manually applied offsets and like magic the Gerbers lined up perfectly. The Spark Fun support forums talk about this issue but more about trying to fix it mechanically (and nobody else concludes its an Eagle bug so my conclusion might be hasty). My approach was to place “corner marks” at the four corners of the board directly onto the silk screen layer so it was included in the resulting Gerber files. Then I previewed the files in Pentalogix ViewMate and tweaked by hand until I could see the marks lining up correctly.

As soon as I get them back I’ll take some before and after assembly photos and post them.