Wednesday, September 17, 2008

Arduino Rebrain Project.

After getting an Arduino Decimila for my repstrap, Zach at RRRF developed the bigger-brained (RAM&ROM) & additionally-fingered (digital input/output) Sanguano.

Oh, and Zach has done a lot of clever work on the software side of things too.

Anyway, I assumed I could get one of these larger Atmega microcontrollers and stuff it into the Arduino so it uses all the PSU, USB etc. that are already there. This would need a new board with a header to go into the old brain's socket, and pins to carry the extra IO lines.

I use 'EasyPC' which produced the 3D. The 'old' brain is shown as a DIL package but the pins would be soldered in to project out of the bottom of the PCB. The grey square is a 44 pin surface mount version of the 644P microcontroller. The 3 sets of four pins are the extra 12 I/O.

Then it got a bit more complicated. The big hole in the board is there to clear the Decimmila's in-circuit-programming connector. Unfortunately, it is connected to different pins on the new brain, so I have added a new connector on my PCB.

Also, I am unsure of the performance of the crystal oscillator - the PCB track length between the original XTAL and up to the new brain may cause problems. I have added pads for a new XTAL and a pair of capacitors close to the '644P and a pair of vias in the tracks so they can be drilled to disconnect. This isn't shown above 'cos I only thought of it last night.

The schematic is a bit tricky to read - I don't like tags that go places that I cant find, but when I connected all the points by lines on this one it is just as tricky to trace. It was originally a quick 'n' dirty hack anyway. I also haven't put the effort into finding out how to make EasyPC schematics look good. Anyway, this is a 44 pin chip so there is a lot going on anyway.

The aim of course is to make a USB sanguano by reusing as much of what I have already got. I shouldn't really have bothered because of high parts cost in small quantities and all my time (I am better then most lawyers I have met, so that makes my hours for inventing very expensive indeed!) (I do seem to enjoy myself much more than many lawyers I have met though!!).

I could make these rebrain boards for under ten pounds if I get several tens of them. Anyone interested ?

Monday, September 8, 2008

Roll On !

I got the nice needle roller bearings, opened up the roller bores and trimmed them to my original design length. Beautiful, smooth.

In fact, even with my heavy springs fitted they roll well on an oval filament. Grip is good, but needs more so I will experiment with heavier springs, and maybe bigger ribs on the rollers. I also think the pitch angle needs revising. The photo shows the rollers in their 'nut' with radial mounted compression springs, all mounted in the housing. I am using red 'fibre washers' as thrust bearings mainly because I had them to hand. As the rollers are smaller than the tile cutters I was able to make up the brass tube spacers too so everything stays tight, but not tooooo tight and the rollers roll well.

On a different tack - I have been cutting toooooobs to make my repstrap bot structure:

I have some of the 22x22mm aluminium extrusions left over from an old project. It is nice stuff to work with with basic tools. The bearings (not skate bearings, but 70pence each from RS) are on M5 bolts, 3 washers between them and pull down onto special 'T' nuts in the aluminium slots. This allows pressure adjustment.

The bearings run on 6mm steel rods, simply sitting in the slots. The plastic cable tie each end help hold the rods in place whilst I fiddle/assemble. (the carriage is made from 2-slot and the rail is 3 slot extrusions)

This reveals an interesting thing with the bearings - they are designed for radial loads. In this arrangement, they are being loaded partially between radial and axial. Normally ball bearings can run with axial load, but in this case the axial component is ALL on one side which causes the outer race to cant over slightly - there is often a bit of axial slack.

This causes the bearings to 'cog' for some reason (like when you hand-turn a unenergised stepper motor) . It could be the balls passing the loading slot, or something to do with the cages. Anyway, it is very bad when I put 4 washers beetween the bearings and almost perfect with only 2 washers. 2 washers reduce the contact angle on the rods and the carriage shown popped off with only slight side loads when running along. Not good.

I have settled on 3 washers for my Mk.1 and I have skeched a layout with proper 45 degree shafts and radial loading for the Mk.1.5 if it needs to be smoother. 45 degree angles add considerably to the tooling requirements, but now my Dad has got his mill motor re-wound I could do it.

I assume this 'cogging' issue applies to Timothy Schmidt's chunky pipework design too.

Progress is a little slow this week - too much time spent in Jazz clubs and visiting friends for vegetarian curries. Nice.