PIC programmer, Stepper controller, CNC, other ideas..

Hi,

Just setting my equipment back up, after moving, and ran into this project, very close to my own way of thinking. Rather than post all over the place and bring a bunch of threads back up, seems easier just to make a single post.

First, some pictures..

Homebuilt ICD2
[www.superference.com]

CNC ($75 metal mechanics, ~$50-75 motors/controller)
[www.superference.com]

Xvid video of some of the lights I make..
[www.modellights.com]



For the ICD2, I have a build guide in pictures. No making a PC board, no having to read a schematic and figure out where to put the parts. Just follow the pictures and plug into holes as indicated and you have two ICD2s built in a couple of hours. Actually it starts with plugging the chips in, building the programming 13V supply, then build a simple LPT based programmer on board to program the ICD2 chips. Then tear it off and finish the ICD2. From no PIC programmer to ICD2 in about the shortest route possible, without having to have someone else program the chips for you. And simply dedicate two breadboards to it, since you can follow the guide and make the wires tight there's no real disadvantage to not building a PC board for it. I was going to lay one out in Eagle, it just never made sense to do it since the breadboard version was fine like it was.

On the CNC, it's dirt cheap, all metal where it counts, easy to assemble with only drill holes and straight cuts, easy to align. Only wanted it for PCB drilling, but easy enough to build it up into a heavier duty CNC. Actually it helped build itself, one slide and rotary tool were used to make the other straight cuts. Has been used for some light machining.
Note the controller. Old homemade PCB version, I have newer far better layouts on hand. MOSFET based, no need to source driver chips. Runs unipolar or bipolar 2 or 4 phase and 5 phase motors, what ever you have on hand or get cheap. Intelligent controller, runs 3 or 4 motors from a single card, and all you have to send is coordinate points, it handles the motion. The motion code is abstracted and only uses binary shift divides, it could easily coordinate 10 or 20 axes at once with other driver boards if needed. With a little more software you could probably just send it the 3D file and it would do everything without intervention. Easily done, I had it doing HPGL for drawings..



Ideas:

First, why the move to a flexible cable drive on the extruder? Just wondering what prompted the move from a straight shaft.. Hurst makes feed through linear actuators, where the motor drives a hollow nut. Seems that or some other connector nut design where the material feeds through a nut could be simpler to build, not as tolerant of different diameters though. If the material is light and wants to spin with a nut then feed between glass cutter wheels right before the nut for anti-torque.

I like the overall design of Darwin, but not sure on the bar stock for frame. Only place you might want it is the slides, if you don't change them. Aluminum angle with drill holes and screwed into the blocks should be a lot simpler, steel if you don't care about weight. No need for the blocks on the diagonals at all just screw them into the corner block, and no need for such complex corner blocks to accommodate all the bar stock.

I like the gear making and belt drive, but I'd direct drive the four bed posts. Surplus motors are cheap. Done with discrete FETs and a single controller, driving the extra 3 motors is also cheap. It could then easily allow self leveling the bed. That would allow the next part.

A cube notoriously hogs space, so split the frame. Most of the motors and electronics are on one plane. With direct drive you can make the outer two bed drives shorter, and hinge the inner bed mounts against that plane. Raise the bed and fold it down against that same plane. From there it won't take too much design to split the main plane from the opposite plane, and store the machine in 1/2 or 1/3 the space. Just starting it and letting the machine level the bed itself will keep the setup time short, and easy storage would let many build a larger machine if desired. If a 2'x2'x2' machine has to be left up, it's unwieldy. If it can be leaned against a wall, and only come out 6 or 8 inches, it's much less so. Actually I'd hang it up in a closet, to get it off the floor and less likely to get hit etc.

Four small contact points embedded in the bed near the corners can make it easy to level. And get them exactly square and it lets you reference the squareness of the frame. Doesn't matter if your frame is 18 steps out of square from one side to the other, as long as you know it and can feed in the 18 steps, and make the tool path straight. Frame being exactly square and precision corner blocks etc aren't necessary.

This is fairly light duty. Using locating pins and NIB magnets and you could easily pop the extruder head on and off. A similar mount with rotary tool head would give access to two worlds.. Replicating parts slightly over then machining important surfaces could give the best of both.
[www.modellights.com]
If you put a bag around the work and tool it can be about as clean to cut. A small fan blowing in to slightly pressurize the bag will tend to keep it away from the work and tool. In general it's easier to deal with than a vacuum and not nearly as much noise.


Not like I'm trying to take over or anything, but I have some fairly mature designs that'll be easy to tack on the needed additions for doing RepRap.. Just for an idea, I have on hand probably 2 or 3 K of many PICs, 500-1000 stepper motors with many of 50+ per type, thousands of logic level 13A MOSFETs, etc etc. Probably $10-20K in parts at surplus prices, I switched from buying singles to buying by the hundred or more about 6 years back, got tired of running out.

Ah well time to get back to other things, but I'll be looking at this a lot shortly..

Alan

hey alan, welcome to the project.

thanks for your input on darwin. its definitely an experimental design at this point. the problem with developing on darwin is that not everyone has access to a stratasys, or even a CNC machine, so its hard to do work on a 2nd gen machine, when you dont even have a 1st gen machine.

that is what i've set out to do, and it appears you have a solid version 1 design as well. once we get extruders back in stock, you should pick one up and play around with printing stuff. the sooner we perfect the process of printing things in plastic, the sooner we can really get down to designing an amazing, printable machine.

Ah, well, CNC is against my religion, I only use it where an absolute miracle is needed. First thing to do is go to Lowes and find already engineered parts to make an extruder.. Way too many bulk manufactured items out there to be making your own, better to re-purpose mass produced items to what you want. For sure somehow one of the many glue guns and some other items can be made to operate in an acceptable manner.
Actually now that I'm looking at it, my CNC works great, I may split the Z and do the adjustable bed like Darwin so it'll fold up a little easier. Extrusion is a really light load compared to what I'm used to, the accuracy and 12"x12"x8" or 12" is way more than enough for most uses, so I'll adapt it. Won't take too long to get it flying, then it'll be a 'go to Lowes, buy these parts, and make it all' design..