adding functoinality and parts

before i begin i thought i would say that tinkit.it.com (uk)have some adrinio boards, and sorry for the spelling etc , im very tierd.

ive been trying to explore milling for the reprap, but no matter how i tryed to work it out for good and accurate 3d modals we'll probably need multiple heads other than the extruder, and add anouther degree of freedom ( eg we can mill some layers as there put down but some may be obstructed by support material or be at a angle where you cant reach in a downward directoin,so they will need to be milled when the modals compleated)

So long story short im going to try to attempt to get a robotic arm to use a extruder etc, to see how it works out, unfortantly im having a REALLY hard time finding one thats not A)useless for this or hidiously expensive.

So i hoped if anyone knew any good sites etc for robotic arms ???

k

Hi Noz,

... there were some kits with plastic-robot-arms for educating and hobby in the last twenty years, so a search at ebay could help ...

Another way is the Lego-NXT/Mindstorms-set with three servos and a controller, so you could by on your own.

Or dive in the robot-comunity - for hexapod-robots see Matts blog: [builders.reprap.org] - here they use 4 to six 'simple robot-arms' as legs of a poly-pod.

With the same method you can assemble three or more RC-servos to build a leightweight arm or two parallel arms for higher loads and forces ...

Viktor

Been going through alot of stuff,premade stuff is either beyond budget or useless for the task.

The current electonic setup, i think uses all the pins, most of the memory etc, So would be diffacult to add anouther degree of freedom.

ALTHOUGH at the moment im trying to find a company with motors with CAN interface, ive heard people talking about them but i cant find them.

Means if people wanted to they wont have to use a motor control and the motor drivers also much easer to add more motors etc into the setup as the limit on a can network is 255. Only thing would be the solunoid and the heating element but if worse comes to worse i could use a small PIC to attach them to the network as well.

I dont surpose anyone knows where i can find CAN motors?

Sounds like it might be either awkward or expensive or both. But there are quite a few CAN-capable microcontrollers out there, so you could always rig up your own controller board...

You might want to have a quick think about bandwidth and lag if you're having lots of things all of which will need to be talked to often or be polled often on the same bus.

Ahh the catch 22!

Im trying to avoid the pitfalls of a microcontroller!

dont get me wroung, there great, so are the drives, my job is to make new software in embeddid C. I work with simple PIC's, ARM's (7 and 9 mainly),drives,lcd drivers etc all day long. Im not a expert but i got a good idea of strengths and weakness's.
If reprap was going to never be addid to then i would do that route, but i want to have an idea , chuck another motor on and only need to modify the code on the PC or/and physcial stuff.

like the controller now , has no spare pins! so we could make anouther controller untill we need something else etc. Sorry im going against the grain here but just for a scratchpad i thought it would be easer. less need to modify hardware/software to modify behaviour.

also if there cheep enough (cheeper buying 3-4 CAN drives than all the electronics at the moment) it would also be much simpler for people to get involved in.

Would make is more assesable to some people, I have spoken to people from my work about it and they like the idea but theyve been overwhelmed by it all! dammit ive got to assume im not too stupied about this stuff scince i did 3 years of cybernetics at university(got a first) and im starting a phd about robotic minds and i had to look over everything a couple of times before it all fell into place!

Any comments on anything ive said would be appriated

I thought there was a way to join two Arduino's together to give us more pins/memory to play with. I'm pretty sure that Zach designed the firmware/hardware with this in mind.

-Mark

that i did not know, guessing hes using the serial or spi,emi type of config. Zachs got a picture in his head but i dont see it yet .

I Think i'll still try to find some CAN motors. Who knows it might have some benifits that people would want, anouther mutatoin to the cause.

Hi,

The original PIC electronics were based on a ring network where further boards could be added/addressed. I seem to remember that the Arduino based system can also be networked to include further boards to expand I/O capabilities. I think the single Arduino system is mostly the result of the higher cost and higher processing power of the Arduino making it sensible to control as much as possible with as few as possible.

If you are associated with a University or the IEEE I found this paper that sounds interesting. I'm not associated with either and so couldn't tell you if the content is any good, but the summary seems very appropriate.
[ieeexplore.ieee.org]

A couple of challenges I can think of are:
- minimizing resources dedicated to network communications and synchronization
- possibly limited availability of combined motor/driver/CAN interface devices
- I suspect the devices you are thinking of are servo motors not stepper motors

Very intreasting , i could only read the abstract as well, but ive heard of this kind of stuff before.

A CAN network might work well, worth exploring anyways

servo or stepper motors are fine , i can work with either but i cant see for any reason if i find one i cant find the other. There are hundreds of CAN shaft encoders out there so im guessing i'll find motors soon enough, main problem is that Google brings up too much useless data when searching!

synchronisation shouldnt be a problem at the distances were talking about , the smaller the physical network the faster it can be pushed!

Unfortantly i need to find the motors before i do the physical design as they'll have a big push, but my general directoin is basicly wood blocks with steel rods and a couple of screwthreads. I think i could build the acutual physical shell in a day with only needing to goto homebase or wicks.

anyways back on track,before anything need to find CAN motors.

I could have sworn I'd posted a detailed reply to this earlier today. Argh. I wonder where it went.

Quote

Im trying to avoid the pitfalls of a microcontroller!

And, as generally happens with this sort of manoever, I rather suspect you're heading into a different sort of pit. You already seem to have stablished than CAN controlled motors are rare... assuming you find any at all, it seems sadly inevitable that they will be expensive and probably only available from a wholesaler.

Making your own driver board that can read stepper instructions from a CAN packet and pass them on to a motor driver board isn't exactly going to be a trivial task, but then again it isn't the most difficult challenge ever.

Find the cheapest CAN capable microcontroller, and make the simplest possible firmware. This will make it cheap, and significantly easier to debug. Moreover, if you start something, the beauty of open source is that you've got a good chance of there being someone else who also thinks it is a good idea, and happens to be a uC assembly language wizard or similar... a simple circuit and simple firmware make a project open to everyone. Expensive and hard to find proprietary motor drivers are pretty much the opposite of this.

Quote

I thought there was a way to join two Arduino's together to give us more pins/memory to play with

This was indeed the case. The initial arduino electronics page had details for a dual arduino setup. The basic idea was that the host, and both arduinos were part of a token-ring network linked using a powercomms board (used for the older PIC electronics, I believe) connected to the UARTs of the arduinos and the serial port of the host. This presumably only worked using the SNAP firmware. No additional uC support or network protocols are required, so no SPI or anything like that.

This was presumably taken down because of the effort required to build the more expensive hardware and maintain two sets of firmware. You can still find the details on line... thats the beauty of wikis; version control!

[farm3.static.flickr.com]
[farm3.static.flickr.com]
[farm3.static.flickr.com]

If you're feeling suitably keen, I'm sure you can dredge the rest out for yourself.

Ive just happened to write a microcan open protocol in the past month

CAN started off in cars,breaks,engine management,and ive read in a couple of places electric windows So i would of thought there would be someone willing to make steppers.

If we wanted much much cheeper controller we could chuck a arm7 on a board ( about

Quote

If we wanted much much cheeper controller we could chuck a arm7 on a board ( about

been working with a arm7tdmi chip lately, actually ive been using a CAN chip with it. Not a bad sized amount of onboard memory. If it goes this way i can help sorting out the memmory map, possible even migrate the code from the current board onto the new board.

Still liking the CAN idea better, get the PC to do more of the workload, PC is a heck of alot more powerfull, and easer to work with (in some respects).

Im more curious to look into the PC side of things, I want to make a C,C++,C# port of it. Try to change it so we can overlay files etc so 3d modal with a PCB layout as well. etc etc.

We could always switch to a robotic arm (an idea im playing with). If we all didnt tinker with established idea's it wouldnt be any fun

Quote

Still liking the CAN idea better, get the PC to do more of the workload

I'm not disagreeing with you here... but you're going to need some hardware that can understand the CAN protocol, and it doesn't seem to be just leaping off the shelves at you.

Quote

Im more curious to look into the PC side of things, I want to make a C,C++,C# port of it

I'm still patiently waiting for the documentation of the current Java host to be completed. I'm a great fan of C[++]... fast, lightweight, and remarkably platform independant, all things considered

Quote

We could always switch to a robotic arm

Someone, possibly Viktor, linked an interesting SCARA design, which effectively used two SCARA arms linked at the business end. With appropriately clever parallel kinematics, it gave a particularly good level of precision compared to the normal serial kinematics of a standard SCARA.

Course, I don't have anything like the right sorts of parts or machine tooling to start assembling complex things, but these are the sort of projects which could perhaps be partially or largely constructed from reprapped components...

Noz Wrote:
-------------------------------------------------------
> ive been trying to explore milling for the reprap,
> but no matter how i tryed to work it out for good
> and accurate 3d modals we'll probably need
> multiple heads other than the extruder, and add
> anouther degree of freedom ( eg we can mill some
> layers as there put down but some may be
> obstructed by support material or be at a angle
> where you cant reach in a downward directoin,so
> they will need to be milled when the modals
> compleated)
>
> So long story short im going to try to attempt to
> get a robotic arm to use a extruder etc, to see
> how it works out, unfortantly im having a REALLY
> hard time finding one thats not A)useless for this
> or hidiously expensive.

Any robot arm is going to add the extra imprecission of going from cartesian to angular coordinates. And to replicate even the imperfect the precission of RepRap's XYZ platform, you're going to need a very precise, very steady robot, which will probably be a very expensive commercial unit.

A robot arm might be usefull for adding mechanical assembly capability to the RepRap, but for the time being it probably won't be too useful for extrudung parts.

Maybe someone could put some work into designing a robot arm using RepRaped parts, and then refinements can be made until the arm becomes usefull for assembling simple items, turning screws, etc. Eventually the RepRap might be able to extrude special-purpose expansion parts for itself, and self-upgrade.

And then the RepRaps will march over the hill and kill all the humans.

Robots arms have at the moment proven unsucsessfull, either too much money or to build it cheep enough be a heck of alot of work for me. Thinking maybe ajoint on the z axis, at the moment the z axis moves the board up and down, im going to try and move the arm up and down, if i get sucsess with that i'll try to add a joint thats capable to attack the object at 90 degrees. in theory we could go further and add more degrees as 90 wont be able to do complicated modals but it'll be a start.

Still trying to find CAN motors. Wish i could find any at all! then if there too expensive at least i can knock them off my possibilitys list!

A cheap arm7tdmi platform is the GBA, adding extra IO isn't trivial but has been done by others and so is probably documented somewhere.

>Im more curious to look into the PC side of things, I want to make a C,C++,C# port of it

Without wanting to start a debate on which programming language is the best, as we all know ML would win ;-}>, I don't see much to be gained by changing the platform. If the host software were being moved to an embedded or resource reduced environment then C might be called for but even then there are Java virtual machines designed for low resource environments that wouldn't require large changes to use.

oh the port really isn't for any reason other than i know C better, was thinking of trying out modificatoins to the code, specially scince im going to try and shift more processing to the pc.At the moment i beleave we give it a directoin etc and the board works it out, where i'll be attempting to control the pins directly.

Ive settled on an idea today , the whole CAN thing is just too expensive. Even though its wastefull i think i'll try the multiple head aproach, eg have one milling head pointing to ground, have anouther head at 90 degrees. Which may be more expensive as need two motors etc when we might only need one, turned at an angle. but it means i can drasticly cut costs.

I think i can get around using optos, probably by using physical switch's and i'll attempt to put relays directly onto the board , controled by multiplexing.

not soo sure weather to try and make my own extruder though.

cheers for all the help
and if anyone else trys to follow this path of insanity , distrubuted controllers might be your best bet!

If you're going the PC control route, have you considered using EMC? It would be a shame to reinvent another set of wheels, when I'm sure you can put your talents to better use!

hopefully not reinventing the wheel too much going to use the Arduino Diecimila with a extendid breadboard. have pins in pairs of two for the motors, which will use a multiplexing chip to produce four outputs that will go through relays directly into the motors, cutting out the motor drives. more of a slash and bang job but good enough for me as im planning to use screwdrives on all axis's. greater accuracy but lower speed. Although it should enable me to make a much strounger reprap that could take more tourque, which will be usefull for milling or drilling etc.

then everything else will be the same,except for the code. Hopefully i'll be using the usb to 232 to direcly control the pins. Im hoping to find some C code out there for the Arduino. It will make life much easer.

Think ive got most of it planned out now. Only thing thats a bit grey is the C code, and weather the motors etc will output too much noise and blow something up.

timing might be a issue but im hoping not, and theres only one way to find out

also thinking about making my own extruder, make it more as just a "head".

need to take a better look at it though .

Quote

four outputs that will go through relays directly into the motors

The driver boards make up a significant amount of the price of the electronics; if it were possible to replace them with something much cheaper I'm sure they'd have been the first to go.

Given that they do their job pretty well, what would make you want to avoid them? Maybe I've misunderstood what a screw drive is. I'm rather assuming it is a linear stepper of some kind.

the current driver boards do clever things like chopping voltage etc, which i might be able to do but im not too sure, like i said only one way to find out
It might all end in tears,but with a educated guess im thinking it is possible. Delays and noise etc might prove too much to get accuracy.

the z axis on the darwin is a screwdrive.

I'll give it a try soon. Already started messing with the board.

Quote

the current driver boards do clever things like chopping voltage etc

Yeah, thats what I was thinking. There don't seem t obe a whole lot of obvious alternatives, and given there's already a cheap and convenient supply of the bits, I certainly wouldn't bother

Quote

the z axis on the darwin is a screwdrive

Ahh, so something like a motor housing that slides along a leadscrew with the aid of a captive nut?

I've been peering at other motor things since Forrest blogged about linear steppers, which looks like very nifty little toys. But seeing as how I'll probably be assembling a slightly more rigid metal reprap that I can do a little milling on, I've yet to work out if they're strong enough.

I think I'll just follow the path more travelled, and do the usual rotary stepper/leadscrew thing.

Ru Wrote:
-------------------------------------------------------

> I've been peering at other motor things since
> Forrest blogged about linear steppers, which looks
> like very nifty little toys. But seeing as how
> I'll probably be assembling a slightly more rigid
> metal reprap that I can do a little milling on,
> I've yet to work out if they're strong enough.
>


[www.hsi-inc.com]

They cost about $85. The lead screw is about $1.50/inch if you cut it yourself and $4.50/inch if they cut it for you.

its a balancing act, although a theme of reprap is ment to be mutatoin,and mutatoin doesn't always end well.

I think ive finalised my physcial design though,except for the extruder.
maybe i'll use the motor drivers as well, cant decide yet , need to look more into there specificatoins.

Sort of a update, Been playing with some electronics, going to buy the set from the online shop soon. but i have been coding a new interface, i think it may only work on windows though scince im using c# with the .net framework. I'll try to get the stl files to load in tommorow. Im going to try and get around using gcode and alike and possibly give some more optoins, unfortantly i dont think it'll work with the current pic code.

I think i may need to fit a extra motor in somehow though, to allow exchanging of toolheads.

Quote

im using c# with the .net framework

There's an outside chance that it will work with mono under linux. I've never played with mono, however. I imagine it is also fairly likely that .net assemblies can be loaded into a mono application, so it wouldn't be too hard to use th bulk of any code you write with a small compatibility layer which will probably be needed for the hardware interface.