Monday, 29 October 2018

Tears, Cheers and Motor Gears

October has been a tough month for us. After rebuilding the robot in plywood (much easier to work with than acrylic) and putting in some motors with more torque, we were ready to test again.

The motors we tried were, again, micro metal motors from pimoroni - but 298:1 gear ratio instead of 50:1. It went forward - and backwards - but again, even with the extra torque refused to turn well. It was better - we could get it to turn if we put on some less grippy wheels - or were moving at the time. But not great. The large, soft, wide wheels we want to use (to give us some natural suspension over obstacles) just have too much grip. 

So - back to the internet and we bit the bullet and ordered some bigger - more expensive- Pololu 20mm 6V motors that were fast and also had much more torque. However - when we received the shipment, they only had one in stock - so we had to wait - just over a week as it turned out.

In the meantime I mounted the sensors and brought some crimping tools for joining wires together - since our efforts with twisting wires together and wrapping them in electrical tape were not the most reliable.

Then - that joyful day arrived - the new motors arrived. I had already drilled the holes to mount the brackets, so we quickly fitted the new motors - and powered everything up. 
It was slower than expected - and again, refused to turn! 

This was the low point in the whole experience - and one of our team (OK me) was ready to give up and admit defeat. 

However - testing again we measured the voltage coming out of the RoboHat motor controller - it was only around 4V per channel - from an input voltage of 8V (from 2 x 18650 Lipo batteries). There are not many specs for the RoboHat controller - so we didn't know the maximum continuous current it supported. Quickly wiring up a spare L298N motor controller - we tried again. But then the Pi refused to boot. 

Hmm - another low point - we had no backup of the SD card (although the code was all in a remote git repository). Fortunately we had documented the setup and so half an hour later, we were good to go (Actually - the Pi now boots up faster!). We tried again, and this time it worked. There was 6V going to the motors from the L298N controlller and this time - on a smooth surface at least - the robot turned. Relief - but further questions - one of the reasons for getting the RoboHat was that it supplied the 5V power to the Pi and we wanted to keep this kind of setup. 

In the end we decided to follow the lead of other robot builders and get a ThunderBorg motor controller. Whilst waiting for it to arrive, the new motor controller code was written. The new controller arrived. We popped it on top and the new code worked fine. And what a difference. The robot is shooting around the house - turning on any surface we throw at it. It has speed - and power in abundance. The batteries may not last very long - but that's not a big issue for Pi Wars. At last we had a robot that can move the way we want it to. The wine was poured - and poured some more. 

Sputnik 0.4 - proudly sporting its new ThunderBorg hat.
Next step - rewrite the code to allow for different operating modes so we can start implementing the automation. This also requires getting the Raspberry Pi and the Arduino to talk to each other. But that's coding and something that one of us at least (the quitter of the family) is much more comfortable with.

No comments:

Post a Comment