The Clever Monkey

Dance Like Nobody's Debugging

So, A. and I completed the Wigl dancing robot, but the results were not that exciting. It appears to move, and the LEDs are lighting in a somewhat appropriate manner. I should report that A. successfully soldered the motors to the motor wires, and only once found out that you shouldn’t hold the iron like a pencil, near the hot end. (Don’t worry: no one was seriously harmed.) But, after some experimentation with various instruments, it appears it is about a half-step off from the expected.

The idea is that certain notes with cause the robot to turn, move forward, etc. And certain combination of notes will cause mode changes. However, after some head-scratching it looked like it was sampling such that a B♭ would turn left, instead of a natural B. Not only that, but the movements seemed, well, really jerky and weird. Almost like the sample rate was a little too fast. It wouldn’t respond at all, and then it would take off in some random manner. Something was not right.

This was, of course, a little disappointing.

Anyway, since we have some of the tools to figure this out, the first step was to confirm through the Crowd Supply page that the code published in the GitHub project repository is actually on the ATmega328P shipped with the kit. I also made sure the pin header pads on the board were typical sizes so I could order a 90-degree 6-pin header so I can start poking at the chip via a serial port.

It turns out that we can’t be sure that what Crowd Supply shipped was within spec, so Vivek said he’d order a project for himself and make sure what was shipped was what he designed. Meanwhile, I could experiment a bit with the software, which required reprogramming the ATmega chip. Since I didn’t have any programming headers around, I put the board down for a bit.

Fast-forward a few weeks and through the magic of global manufacturing and supply I have more 0.100″ (2.54 mm) Breakaway Male Headers than I will be able to use in my lifetime shipped right to my door for a few bucks.

Last night I soldered the header in and hooked up an FTDI USB-to-TTL-Serial cable to the board and started down the rabbit hole of talking to an ATmega chip via various AVR/Arduino tools. After talking with Vivek, the chip is supposed to be Arduino Uno-compatible with an on-chip bootloader but it sure isn’t behaving like it. It looks like the serial port circuit is sensing the DTR and invoking the auto-reset (or at least we see some LEDs flashing, which is the universal indicator for Arduino auto-reset) but otherwise I get the dreaded “avrdude:stk500_recv(): programmer is not responding” message.

Well, by then it’s 1:30AM and I actually have to work in the morning, so here we are.

The next step is to pull the ATmega chip and put it in the Arduino that I know works, and poke at that with the built-in FTDI and possibly the ISP. Luckily I am the proud owner of an AdaFruit USBtinyISP so I have the tools necessary if I need to burn a new bootloader on the chip. I should also check that my el cheapo FTDI cable is actually doing the right thing using my ‘scope. For all I know, the DTS is not being strobed and the Arduino bootloader on the chip has no clue I’m desperately trying to upload a new program image.

Hopefully by this evening I’ll have a better idea if we assembled this thing right, or if maybe the kit has some odd or out of spec parts. Perhaps the next Wigl update will be a video of actual robotic things.