The Unnamed (FW)

[Ellis]

Yes, as Mario says, that is likely the failure. It's a flaw with all Gimson/drill-based gearboxes. Surprising it happened so soon, maybe it was the additional load from the belts that accelerated the process. Pop it open, it's likely the output axle's carrier plate is toast and the second stage gears might be half eaten. If so there are a few fixes.

You'll likely to have to replace the shaft, typically the holes in the plate that these pins press into get badly stretched and won't serve again when they fail like this. I intend to try and solve the issue with some custom pins, that have an m4 threaded portion that taps into the carrier plate. With a bit of permanent loctite this should prevent the pins from ever loosening, I just haven't come round to testing :P. Another method may be to simply apply some of the same metal glue/loctite to the pin (popping it out carefully and putting it back in with gloop) but I don't know how effective that would be. There's a third solution which involves sourcing the right diameter rod, leaving it a little long (so protruding about 1mm on the output shaft side of the plate) and then ballpeening that little stub so it mushrooms out, again so the pin cannot pull through and fail. That method has been used with success by the RoboChallenge guys on their drive-all-day kits for years.

Other than that, the motor is likely deaded too, but the speedo should have taken it in its stride (they have for us, many times).

I'm sure a solution can be found, I imagine you're already cooking up your own!


edit: can be done too. v

[maddox10]

I think the solution is locking the axle to the bearings so it can't move longitutal. (in and out)

[Eventorizon]

How come you left slack in the belt? There are lots of calculators out there that will give you the exact distance between centres so they are perfectly positioned from the start. If you are using HTD belts, which it looks like you are, you may be able to buy slightly shorter belts to make your life easier. Looks good though, shame about the dud Gimson.

[Ellis]

I used those calculators, in fact I actually shrunk the dimensions a tad ahead of machining, and the belts on Rango were way tight, including the weapon belt. ^

Bearing material I used in the wheels played a large role but they were still a lot tighter than I'd have liked.

[Eventorizon]

http://www.technobotsonline.com/timi...alculator.html

I have used that to calculate every belt I have ever used and it has been perfect every time

[Pinski1]

One or more of the pins of the carrier on the outgoing axle came out and locked the gearbox.

I think the solution is locking the axle to the bearings so it can't move longitutal. (in and out)

I'll have a look at the gearbox closely then, seems a shame this is a known problem without a factory fitted fix. Unfortunately due to my setup I can't see a way of locking the axle to stop longitudinal movement, I guess a cir-clip on the shaft could help, but I don't know where I could fit it.

EDIT: Well, I must be on drugs. I checked the motor around lunchtime today and the gearbox was fine. The motor spins again, put it back together and drove it around some more. I'm utterly stumped, WTF it was catastrophically buggered Saturday night but now they work almost perfectly. I guess it was the drill motor fairy?

How come you left slack in the belt? There are lots of calculators out there that will give you the exact distance between centres so they are perfectly positioned from the start.

That's exactly what I did! I've got 300mm between drive axle and driven axle, I've got 22 tooth pulleys on both axles for HTD 5mm pitch 9mm wide belt. This works out to exactly to the standard 710 tooth belt which is what I bought and fitted. I've mesured the spacing and it's 300.00mm. *shrug* I knew I'd have to deal with some slack as the belt stretched and wore but I wasn't expected as much as I have right now. Of course it's not difficult to deal with, a delrin guide will reduce wear and keep the belt tight.

I'm sure a solution can be found, I imagine you're already cooking up your own!

Oh I've got you're solution right here!

Just kidding they're for messing around with, They're much too large for this bot, at the moment...

[Pinski1]

What a productive evening! I received a big bag of M6-2 clinch nuts in the post today and I decided to install them! Clinch nuts are a special type of nut that you press fit into a hole (8.75mm for M6) and the metal cold forms to lock the clinch nut in place so it doesn't drop out or rotate. It's ideal for putting threaded holes in sheet metal or in my case regularly used threaded holes in 0.125" aluminium channel.

Last night I used a CNC Router to cut out the bottom sheet of polycarbonate to ensure all the holes were in the correct spots. I also attempted to cut out the top sheet of polycarbonate but I made a mistake and while all the inner holes and cut outs came out correctly unfortunately the perimeter was not cut correctly. I've also fixed this tonight.

Put both completed tasks together and I have top and bottom armour that I can mount to the inner chassis pieces:


I cannot attach the front/side armour plates in that picture because I still have some work to do with the front wheels but soon that will be sorted.

My next tasks are to strip the outer chassis pieces, the wheel pods and add clinch nuts to them so I can fully bolt the polycarbonate down. I also need to finish mounting the battery and electronics.

[harry hills]

That's so cool man. Excellent robot

[Pinski1]

Its been a while since I last updated this. The biggest news is that I've been stalled working on the lifter. It was a very expensive decision point, if I got it right it should work with minimal work, if I get it wrong the lifter arm would be useless and the core the chassis would need to be redesigned. As not making a decision is much much cheaper in the short term I stalled. However I've reached my "oh fuck it" point and I've ordered the lifting arm.

Another reason I stalled is while conducting extensive driving testing (aka ragging it around to scare people) I killed the battery. My 18.5V 5AHr 5S1P battery now has a voltage of 13.92Volts, 4 of the cells are at 2.73V. It's dead, pushing up the daisies etc. It happened just as my bank account was due to be cleared out for 2+ successive months so I couldn't just snap buy a replacement. I'm aware that I could, very carefully, recover it but after seeing this I'm terrified of what could happen when it goes wrong.

I'm planning to show it off at the Manchester Mini-Makerfaire July 26-27th and at EMF Camp August 29-31st. I've also got a wild hope of taking it to Stevenage and *gasp* competing!!!

[Pinski1]

Mini update time! I finally figured out how to get feedback on the linear actuator position and how to write the code to drive it. I've also tweaked the polycarbonate mounting. Further detailed updates to come!

[Pinski1]

So here's that update I promised...

Its been a while and the robot has been somewhat dormant. Mainly because I just couldn't figure out a way to know where the lifter arm was so I could then control it without jamming it up or down. I had thought about using shaft encoders but they're a relative measurement and need an indexing signal so the ESC wouldn't know where the arm was if had just been turned on, say as a result of a momentary power loss.

However an off-hand comment from a friend at Cambridge Makespace got me thinking, could I attach a linkage which would then turn a variable resistor. After a bit of maths, a bit of head scratching at SketchUp and a laser cut tester I found a set up that would work!



After some careful machining and some painful filing (D shafts are really cool but an arse to make) I'd got my linkage attached and now had a variable resistance based on the position of the arm. YES!

The next big step is to get the Tz-85A working. Looking at the modified Tz-85A code here. It seemed really messy and completely ignored the 16bit timer1 in favour of an 8bit timer0 with an 8bit variable to extend it. So I rewrote it. This was a painful choice but I got it working pretty well last night. I just have to sort out the servo code to process the variable resistance and run a very basic control loop. Oh and write the calibration procedure.

This is my current set up:



Once I've got the calibration code somewhat sorted I'll post the GitHub repository. Unless you guys want to trial the current firmware?

Oh and I've entered it into the Burgess Hill competition in April, eep!

[harry hills]

So much legit robot going on there. Tempted to come to burgess hill just to see it.

[shakesc]

Nice work, looking forwards to seeing this

[Eventorizon]

Another reason I stalled is while conducting extensive driving testing (aka ragging it around to scare people) I killed the battery. My 18.5V 5AHr 5S1P battery now has a voltage of 13.92Volts, 4 of the cells are at 2.73V. It's dead, pushing up the daisies etc.

Low Voltage isn't a total killer for a Lipo. Due to a faulty charger we drained 2 of our 5S 2600mAh packs down to less than 2.2V per cell and were able to recover them using a Pro-Cell 10 charger (It can run recovery cycles and I think we even ran a NiHM cycle just to force in some power) They have worked faultlessly since in over 10 fights and countless demo's.

All that said, be careful, and you can't do this with a cheap charger either.

Lifting mech is looking great though!

[teamstorm]

That's exactly how the arm on Storm II works, it uses a 10k Pot mounted inside the gearbox block that drives the arm mechanism to relay end-stop (not position) information to the speed controllers. We avoided creating a 'giant servo' approach where the arm would fight to stay in whichever position we wanted it in for fear that in a fight that's just not practical, something jams the arm and you're looking at a stalled out motor with very little you can do about it.

Can't tell you enough how you do NOT want to run that as a giant positional servo