Monday, 7 October 2013

2-3-2, Dome Servos and Arduino Workshop

Building has been fun recently.  Decided to delay the magic panel and DBC installs, I want to look at the base of my code first.  Reduce a few signal lines and replace them with 2 way comms.

But I added the 2 relays to the dome, these control the power to the servos.  One for the 3 HP servos and one for the rest of the doors.  They both prevent any servo twitch as the main power is connected, so connection order is no longer an issue for the dome.  They also power off the door servos 2s after they are all closed.  This allows the doors to be pulled open when on, stops any servo buzz and saves power.  Really good, highly recommended.  These will be included in the BHD write-up.  Will also be adding these to the body servos, and maybe the dome motor for completeness.

I then moved back to the 2-3-2, James had offered to ask Roger to look at the ankles and leg rod fixings in the feet, so I took him up on the offer, hoping to get this finished once and for all.  The play in those areas was causing a lot of wobble in the 2-3-2.  So almost a full day of stripping the legs on the Saturday, then I set off to Roger’s last Sunday.  We remade the JAG push-in inserts for the ankles for a tight fit on the shoulder bolts, and made them a little longer too.  Then we redid the leg rod fixings in the feet, these now cannot move, and the rod more easily rotates on the brass bush and so it looked like it should solve the issues.

I also wanted to show you my ankle bracelet brackets. :) 

So then came 3 evenings of re-assembly!  So much work to get it all back together.  James came over Thursday evening to help as catcher.  So tested it, it looked amazing, so much more solid and reliable.  But all the positions were wrong.  So I uploaded the tweaked code…..then nothing.  Dead.  I spent about 5 hours debugging, thinking initially that I had knocked out a wire and then that I had shorted the Xbee.  Nope, I had accidentally deleted the line Serial.begin(9600);  not clever, obviously the serial wont work without that!  So once id found that I recalibrated all the positions, re did the 2-3-2 code so that the signal to drive the feet stops when the first actuator stops and not both (looks much better).  It is so nice now.  Really smooth and stable.  There is one little detail of the bolts that attach the legs now taking all the force so they undo themselves, but an easy fix with the right washers.

So here is my 2-3-2 as it should be, stable, balanced and reliable.  No face-plants, no wobbles, just 2 legs to 3 legs and back.  Maybe 60 cycles + now, and still running perfectly.  Long may it last.  But if you are doing 2-3-2, it is a nightmare, does require lots of work, it is different on a dressed droid, and just because it does it once, doesn't mean it will do it all year long.  Don’t give up, just keep finding more ways that don’t work till you find one that does.  My whole frame now needs a rebuild to allow more space for the CPU arm and claw, they fit, but not their mechs. :)

Anyone who says they have 2-3-2, unless you have done 100 cycles without issue, you haven't even started. :)

This Sunday we had another Builders day at my local hall.  We had a really nice Dalek, a beautiful B9, 3 complete droids and 2 part droids, oh and a really nice mouse droid.  There are a lot of talented folk in the UK.

The day included (or was taken over by) an Arduino workshop.  I wanted to get a few people working with arduinos and felt that the initial phase was the tough one, so doing an intro and making something fun would be a good start and maybe help others to pick up Arduino and run with it.  It went really well, they seemed to enjoy it, and were showing each other how to solve stuff by the end, which is usually a good sign.

I had planned to do a load of testing in an open space, but didn’t get chance. Next time.

I finally got the Relays for the feet, only a 5 week delivery!  Will be adding those ASAP.

No comments:

Post a Comment