Repstrap project

Hi, my name is Corwin. Some of you may remember me from such brilliant posts as "whoop, please ignore the above, I'm a moron"... Anyway, I'm starting work on a repstrap for my highschool senior project, and I thought it might be a good idea to run some of my plans by the people on the forums here to see if they can catch any obvious gaping errors.

My basic design is a CNC-gantry style cartesian bot, with the X stage moving the build platform, and the Y and Z stages manipulating the tool head; hopefully giving me enough stability to do isolation pcb routing with a dremel. Based on discussions with my grandpa, who used to build XYZ stages for microscopes and machining, I've decided (or more precisely been browbeaten into deciding) to put a good chunk of my budget ($500) into the positioning, which means using some 8" long linear slides for guide/support rails. I'm planning on getting ACME thread 1/4"-16 threaded rod from Mcmaster and anti-backlash leadnuts from www.dumpstercnc.com

On the electronics end, I'm thinking the Sanguino looks like an extremely good choice (I've run out of Arduino pins for things before) but I may hold off on that, at least temporarily, and use one of the Arduino boards I already have on my workbench in the basement to save money. At the same time I'm looking at getting an EZ CNC driver board from hobbycnc.com (http://www.hobbycnc.com/products/hobbycnc-ez-driver-board-kit/) instead of the driverboards at RRRF. It's not opensource, which pains me, and it isn't strictly speaking meant for 12vdc (it wants 24, but is rated to run at 12 minimum), but is is $20 less expensive than buying 3 of the standard boards - plus it supports microstepping.

At the moment I'm not sure about how I'm going to get a thermoplastic extruder. I do have some limited access to a machine shop but I'm not exactly a great machinist. If anyone has suggestions on that end (besides ordering from bitsfrombytes, transatlantic shipping costs being what they are), I'd love to hear them.

I also may be working on a chocolate extruder, my dad thinks I should try and make this pay for my college tuition by printing custom chocolates XD

Hi

What motors are you using? EZ CNC drive is unipolar & will not run 4 wire motors according to their web site. Also for Reprap microstepping is no advantage. The Arduino registers currently used run out of resolution. I have 20 TPI screws and I can only use half step. This will be altered one day but I doubt if it will be that soon.

---

Regards

Ian

Hi Corwin,
If you build a machine sturdy enough for milling you can make an extruder by milling plastic. That is how I make my first one.

If it is big enough you could also use a mill to cut the parts for the laser cut design. That wasn't available when I made mine, but should be easier as the parts are just 2D slices. Acrylic is quite easy to mill but you have to do fast shallow cuts to prevent it melting.

Ian,
Microstepping makes motion smoother and quieter even if the additional resolution is of no use. I think the gen 3 electrics will include it so the firmware will get updated then, worst case.

---

[www.hydraraptor.blogspot.com]

Hmm.. I hadn't realized the memory problem with the gen 2 arduino setup reached to the point of effecting the positioning resolution that way, it sounds like I should definitely be aiming for a Sanguino based system, then I could expand into the full microstepping mode when it starts to get put into the updated firmware.

Ian, the standard motors sold by the RRRF can be wired in unipolar drive mode, as can a fair selection of others, and the EZ CNC board can drive up to 3 amps which should be enough to make up for the lower torque. I also might be able to go to 24 volts for the drive power, my dad's work is donating a bunch of surplus odds and ends to my robotics team and he said he got them to toss in a couple variable powersupplies that I can pick through first XD. The other thing is that using a 16 tpi rod makes microstepping more useful - one rotation is going to send the stages about 20% further (I think), so the added resolution could wind up making a bit of a difference (probably not though - by my calculations one standard 1.8 degree step should mean 0.0003125" linear distance on 16tpi, so 1/16 microstepping would be rather excessive - something like 2*10^-5).

Having the machine mill it's own extruder sounds like an awesome idea, Nophead. Thanks!

Corwin Wrote:
-------------------------------------------------------
>
> Having the machine mill it's own extruder sounds
> like an awesome idea, Nophead. Thanks!
>

That's pretty brilliant, all right. With a flexible drive shaft off of a Dremel and very shallow cuts you ought to be able to do that kind of milling even on a light frame repstrap machine.

I've got to do that. That idea solves lots of problems.

Unfortunately my dremel is the wrong model for mounting to flex shafts, or at least I haven't found one for it. I'm going to actually have to mount the whole thing on there..

Actually, even if you have a light milling head and take shallow cuts you still need a machine with a high degree of stiffness otherwise the cutter snatches and chatters.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> Actually, even if you have a light milling head
> and take shallow cuts you still need a machine
> with a high degree of stiffness otherwise the
> cutter snatches and chatters.

Aw shit!

Nophead is right, you shouldn't underestimate the stiffness needs for a good milling setup. That being said, _very_ shallow cuts at high RPMs in relatively soft material may be feasible on a RepRap (witness the PCB milling in another thread here). It will just take someone to try it!

Dremel can get 30,000 rpm!

well hopefully thanks to the linear slides when mine is built it should be pretty rigid... plus that I'm building it like a gantry style CNC machine instead of like the McWire should help.

Corwin,
Yes your proposed machine sounds like it will be stiff enough for milling.

Forrest,

A very rough rule of thumb I found was that if you grab the frame of the machine with one hand and push the head with the other then if you can see it move it is not stiff enough. That is why I ended up with an 18mm MDF frame and 6mm AL plate on the z-axis braced by 10 x 6mm AL bars.

Non intuitively, it seems that milling something hard like copper on a PCB requires less stiffness than soft plastic. That seems to be because soft plastic snatches and pulls at the head regardless of the feed rate.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
>
> A very rough rule of thumb I found was that if you
> grab the frame of the machine with one hand and
> push the head with the other then if you can see
> it move it is not stiff enough. That is why I
> ended up with an 18mm MDF frame and 6mm AL plate
> on the z-axis braced by 10 x 6mm AL bars.
>

Mine doesn't move more than about a mm, but it does move.

> Non intuitively, it seems that milling something
> hard like copper on a PCB requires less stiffness
> than soft plastic. That seems to be because soft
> plastic snatches and pulls at the head regardless
> of the feed rate.

I'll rig mine in any case because I really want to be able to mill my own PCB's.

As usual, you're a font of good information.

Just got the gantry and z-axis mounted on Tommelise 2.0. Seems pretty stiff.

hmm. can we get some stiffness data? newton milimeter numbers? I suppose a dial gauge would be required and a bit of string, pulley (to check horizontal loadings)and bag of flour/beans/2l bottle of coke to apply the force.

What is the SG of coke? Probably better to dump the coke and fill with water at 20 deg C.

Could make a gauge using a ruler & an 'amplifier' arm say 10:1 ratio. I assume we are looking for movement in the region of 0.5mm but at what loading?

At the load applied by the mounted steppers I would guess

The steppers are the ones that force the movement, well and the weight of the machine itself of course (milling head)
but I cannot imagine anything else that is forcing the machine to bend.

So if you can fix the Toolhead somehow and try to run the steppers, they shouldn't move, If they are moving, you require some more stifness
(I'd guess the steppers wont be happy with this procedure though)
Or am I wrong?


Or, you can calculate the force applied by the steppers, and use that as a hint what load you should use to test.

'sid

I like the reverse thinking idea - but you would still need to record the motion on the stepper. Also the force would be an indirect derived measurement which with steppers would be tricky = change in supply current.

Stepper deflection - use a long sticky stick stuck to the pulley. Measure angle. This would include total system stiffness IF the tool tip is clamped to the bed.

... one of the biggest problem beside bending of the frames with steppers is blocking/stalling when the motor-frequency hits some resonance-points of the frame. This is the main source for limitating maximum frequencies ...

This could be reduced by smaller steps (half-stepping with 0,9

Got home from school today and the 3-axis EZ CNC kit had arrived. It's a very nice looking kit, low part count, no SMD soldering, on-board 5 and 24 volt regulated supplies (for logic and a cooling fan respectively). My bench supply can only give 5 amps, so even when I finish building it I won't be able to run steppers at full power until I get a dedicated powersupply for the repstrap, but I will be able to test.

One thing I'm a little worried about is frying the SLA7078MR driver chips, since I don't know of a source for buying a replacement, but I guess I'll just have to be careful.

I'll let you know how building t goes.

Corwin, et al,

FYI, a quick google search turned up at least one supplier for this stepper driver:

[www.bdent.com]


To avoid the need to replace, a generous heatsink would help, as will a fan and thermal compound.

-- Larry

Thanks Larry,
I definitely intend to use a fan, like I said the board even supplies 24vdc for one so it's clearly a good idea... unfortunately the kit did not come with a heat sink so I'm probably going to have to a) make one or b) buy 3.

Good news and bad news..

The order of parts I made from the RRRF shop came in today (awesome turnaround btw). The steppers are absolutely beautiful, and the biggest and shiniest I own. The kits I ordered look good, and once I finish the soldering on the EZ CNC board (just 69 pins left to solder) it'll be replaced in my soldering station by the pwm driver kit. Looks like I'm well on my way to getting this thing really started now...

Did have one problem though with the order. When I finished taking things out of the box, I was fiddling with the GM3 and realized that when I turned the output shaft, there was very little resistance - and no movement on the motor shaft. So I went to test the motor real quick when I finished taking everything out of the box, and it sounds like it's working fine - drawing something like .07 of an amp at 12vdc. It turns out motor arrived without the gear that connects the motor shaft to the gear train... I've already sent off an email about it, we'll see if we can't figure out what happened.

Thought I'd update this since my plans have changed considerably since my last post and I had an idea that I wanted to get some opinions on.

First, the changes in design: For a few reasons, I'm no longer planning to use the 16tpi acme. The new design will use a setup more closely resembling the way the Darwin is organized. The X and Y stages are going to be MXL belt driven, I've ordered open-ended belt, pulleys, and clamps from sdp-si.com. The X stage drives the build platform at the base, and the Y stage drives the Z on top. The Z stage is going to use 1/4"-20, at least to start. I'm building each stage on a section of 8020 aluminum extrusion because I was able to get it very cheaply, and the YZ assembly will be supported by a wooden frame. I'm currently working on the stepper motor brackets, which I'm making out of 2" aluminum L.

The idea that I wanted to check on is about Z axis homing. Recently the Metalab guys started using a copper PCB blank heated by nichrome wire for a build platform. What I was wondering is if it seems to anyone else like it would be feasible to use the copper plate as a touchplate for Z axis homing (ie, an alligator clip attached to the plate for one terminal and a clip to the extruder barrel for the other). This seems like it would make homing easier, since the location of the switch can't get out of adjustment - if you go too low, you bump into the plate. If it's raised on springs like I recall someone suggesting for leveling, it won't be damaged by a very slight impact, either. In my head this all makes sense, but my head occasionally loses its connection to reality, so...

Just to touch on something from the very first post, as a FYI...

When you start experimenting with chocolate extruding, I'd suggest making custom molds using a miller setup with a rounded dremel head for awhile--you can get a jumpstart on your cash cow this way by doing it the older way first.

Quick update with photos... Mechanically I'm at the XY plotter stage now. I expected to be at this point a week ago but got held up because of parts not arriving due to snow. I'm going to throw together an arduino sketch to let me plot XY with some pushbutton switches for fun in a couple days - I'd do it tonight, but the FIRST kickoff is tomorrow and I have to be up early.

The X and Y stages both move nicely by hand (unless the leads get shorted and back emf jams it, that freaked me out the first time it happened, thought I'd ruined my bearings!). They're each 16 inches long, the wooden frame is 16 by 16 by 13.5 not counting the actual Y stage. Once I have it going in plotter mode I'll get the Z built, it should go pretty quickly at this point - the stepper mount is already done and I've made jigs for the trickier machining parts.

I have a lot more photos, I documented the X-axis build so I could go step by step for senior boards and I've added a quick litte mdf bed to the X axis since the XYFrame photo was taken.

That's pretty sweet! It looks REALLY good.

What did you use for slides on that 80/20?

Demented

---

[www.urbansurvivalists.com]

The slides are NSK linear guides. here's an (overpriced) example: [cgi.ebay.com] They're nice because they provide rigidity in all directions and are very straight, so my own work doesn't have to be as precise or well designed - my x and y axis rigidity comes from those slides, not any precise alignment on my part. Getting the 3mm bolts for mounting the stages was expensive, though, and having to drill, tap, and cut my own aluminum t-nuts to mount the slides to the 8020 was quite tedious.

I'm currently doing some rework on my Y axis. I mounted up the Z axis and found that the aluminum L channel on the Y had too much flex in it (I may need to come up with something lighter than 8020 for the backing of the Z axis), so I'm taking my spare linear slide (I ordered 4 when I got them originally but had hoped to use 3) to put a second guide rail over the one being driven by the Y axis stepper.

The Z axis is running on 1/4-20 rod. I found a flanged nut with two mounting holes at the local hardware store for the stage, and drilled out a 1/4" threaded shaft coupling and added a set screw to mate the stepper shaft with the threaded rod. I only used one bearing at the top to support the shaft to help deal with alignment issues - the rod hangs free at the far end. It seems to work fine, though running it along side the belt driven stages makes me very glad I didn't use threaded rod on all of them - the threaded stage can't even get close to the speed of the belt-driven.