[mmcp42]

I'm starting to work on a DIY CNC table
Main use will be PCBs but may move on to 3D printing

Target parameters:

Outside dimensions 300mm cube (that's 12" cube in old money)
working volume 200mm by 200mm by 10mm
accuracy - I want to be able to make tracks 0.5mm wide
£100 for mechanical parts (incl spindle)
£100 for electronics (incl motors)

My background is electronics and software

I already have a prototype contoller and motor drivers running
Also software to take gcode and squirt to the controller.
The controller interprets gcode directly

The design will be heavily influnce by microcarve's A4 machine
I've been following the various threads and it looks like a superb machine


I'm trying to work out the drive train arrangements
I think it's something like:

Motor :: coupling :: threaded drive rod

Drive rod has a bearing at each end to stop it falling to the ground
It also has two thrust bearings to stop it moving side to side
So the drive end looks like

collar :: thrust bearing :: radial bearing :: thrust bearing :: collar

The radial bearing is mounted in a static block of some sort
the collars are attached to the drive rod

The floating end just has a radial bearing in a mounting block


I'm looking at 10mm threaded rod with anti-backlash nuts
rails will be 25mm chromed steel (curtain poles to you) available at local DIY store; oilite bearings

frame will be 15mm MDF


I'm trying to use google sketchup to draw the beast, but I find it quite klunky
Is there something better I should try?
(penil and paper are looking very appealing!)

Am I making sense or talking out of body part usually reserved for sitting upon?

feedback most welcome

[harryn]

Hi, your budget goal is very - thin, but we all start out there. I would have normally guessed about 3-5x that, but perhaps you are a good scavenger.

There are some useful discussions in this section of the forum which might be helpful for pcb machine building:

PCB milling - CNCzone.com-The Largest Machinist Community on the net!

0.5 mm traces sort of implies similar size spaces, which is the main challenge for pcb milling. That sort of work is pretty delicate.

Rather than fully milling out the copper, it might be worth considering your router as more of a "printer" to open up / scrape away the resist on a board, then do the etching as is common. I know that the goal is to skip the etch step, but accurately milling the copper layer can be challenging.

For me, I gave up on doing the boards myself and just have Zig do them for me. He posts down in that section and his prices are pretty reasonable.

[mmcp42]

Hello harryn
thanks for taking the time...

I want to get away from etching, been there done that, got the stained Tshirt to prove it
I'd use laser/hot iron if I went that route btw

Speed isn't an issue
Budget is a target (no good once you've hit it!)

Thanks for the link - good stuff there

My worry with belts/gears is the backlash

I'll only ever be producing very low volumes of one-off boards, so sending them out probably won't work

For me it's an challenge to build it as much as to actually save time or money!

What did you think of the drive train setup I propoesed?

cheers
Mike

[harryn]

Hi, I am glad the links were useful.

The oilite bearings are probably fine, in fact there is a long thread in that section about a guy who builds up small MDF machines for precison parts and pcbs.

Personally, I can't stand MDF, but as noted in that long thread in the PCB area, there is a lot of variability in the quality. MDF does have its place, but I have not had significant success with it. If you really want to make it from a wood like material, I would instead glue up your own laminations from thin wood stock and epoxy glue, then dry it in a vacuum bag. This process is actually not as bad as it sounds.

I was surprised to learn that most Al can be worked with using more or less the same methods and tools that I am used to using with wood. Face and eye protection are even more important as the chips really get thrown, but 5mm thick walls of Al can be tapped, which is really handy. Al tube is not very expensive and certainly will not tear out like MDF.

It is hard to beat a fixed gantry design for precision work.

Threaded rod in theory is ok, in practice you need some that has the larger, flatter (not sharp) threads - I think it is called ACME. If you use conventional threaded rod, it will wear out fast, especially if it is zinc plated.

For my own wood router project, I spent a LONG time looking at different drive system approaches. In the end, it appears that for similar results, the price is similar, regardless of approach, especially for hobby size machines. I think a lot of people tend to use what they can get used.

As far as jobbing out the boards, in some ways, I would really like to do them myself, but I also could never justify it for my own work. Ziggy is actually quite reasonably priced, and his shop specializes in 1 - 2 off + low volume. For my project, he was competitive with any shop out there, including the chinese. There also are some secondary operations like plating and solder mask that he does for people - as needed.

Anyway, its always interesting to have options.

[RomanLini]

Here are a few low-budget suggestions;

1. Use brass threaded rod, you only need less than 2 foot of it and brass thread is milled not roughly rolled like steel thread so brass threaded rod is quite accurate.

2. Use a solid coupling and no thrust bearings, it saves size and complexity and actually increases accuracy. It uses your stepper motors for the thrust bearing, this is quite ok with light loads like lightweight machines and fine cuts (ie PCB use). For the far end of the leadscrew you just need a plastic slip bearing to stop it whipping around.

3. Use stainless steel curtain rods, or better still stainless steel polished bar you can get from most metal suppliers. I woudn't mess with anything plated or flimsy.

4. Instead of oilite bronze bearings that cost money and may have slop, use plastic bearings. All you need to do is cut plastic sliders that are snug on your stainless shafts. These are good for low speed use and more accurate as the plastic allows snug sliding and metal bearings can't do "snug".

5. Likewise make plastic leadnuts, you can tap them to suit the brass leadscrew thread and make them snug by not tapping all the way or by heating and deforming the nut afterwards. Again, cheap and more accuracy.

6. There's nothing wrong with pen and paper designs! You will probably have to change a lot of things when you actually start building so spending a lot of time in CAD work may be wasteful.

Good luck on the build and I would enjoy reading your build log! A tiny cube style PCB milling machine is one of the rare CNC projects you can actually make (and perform well!) even on a tight budget.

[mmcp42]

many thanks for a very full answer
especially as I like a lot of the content too

funnily enough I'm already looking at chromed curtain poles and chromed water pipe
both seem to be very reasonable (cheap)!
I've managed to find some really cheap oilite beraings so that's not a problem
I hadn't considered brass for the drive - hmm that's got to be worth a few minutes in the store!

appreciate you taking the time

[mmcp42]

I just re-read your reply
I do like the plastic lead nut suggestion, very clever stopping just before the end to make it snug :cheers:
watch this space!

[RomanLini]

I'm not sure you can find oilite bearings AND curtain rods/water pipe that just happens to be a precision sliding fit. PCB's need precision.

If you use rod that is nice and round (and a uniform diameter) like polished stainless rod you can easily make some plastic bearings that are snug by drilling holes and even filing the holes out a little to suit, or deforming the holes inward a little to make it snug.

Last time I bought 5/8" polished solid stainless steel rod it cost about $4 a foot, it is not that expensive for a tiny machine. I wouldn't even mess with curtain tube which will be flimsy or water pipe that won't be round.

[mmcp42]

fair point!

what plastic do you suggest?

[RomanLini]

Delrin (polyacetal) is not too expensive and is quite slippery on most metals, provided you don't need big speeds (which you don't for PCBs). It can be drilled, tapped, sawed and filed easily and is a dream to machine.

Also there are plenty of engineering plastics that are priced quite reasonably. My local bearing shops both have plastic rods and blocks on the front counter, they have a whole range of special plastics that contain lubricating materials etc. If you buy 40mm square plastic rod and drill holes in it then saw it into blocks you can easily make some plastic bearing blocks that can also be bolted to your flat surfaces like the Z axis plate or Y axis moving table etc. Be aware some of these engineering plastics don't like to be wet, or exposed to sun, and some are so darn slippery you can't file them and they are quite annoying to work with!

Have a look around your industrial bearing shops.

[mmcp42]

finally made some progress
started building a prototype to check out my abilities and the soundness of the design (posh word for pile of scraps of paper with squiggles on)

here's the base and x-axis (well it's the first one so it must be x)
first design parameter has gone out the window already
I'd hoped to make it a 12" cube, but that gives me bugrall travel
so it's now a 400mm cube (ok the motors might stick out a bit)

base is 400mm by 400mm by 11mm marine ply
ends are 50mm high

drive rod is M8 steel
guide rails are 7mm stainless steel
bearings are skateboard (10 for £3)

[mmcp42]

been away for a while
(damned day job keeps getting in the way, but it does pay the rent and feeds my habit)

I got a prototype up and running using scavenged scanner parts
not man enough for any serious grunt work
but gave me a toy to test the electronics on

The Arduino is interpreting Gcodes received over USB from the softly "wot I rote" running on a laptop
the Gcode is generated from Eagle using pcb-gcode
seems to do a reasonable job
Didn't need to implement arcs as the pcb-gcode convertor generates lots of X and Y movements to give circular pads

I have implemented this sub-set:

G0 rapid move
G1 controlled move
G4 dwell
G20 set inches
G21 set mm
G28 move to origin
G90 set absolute
G91 set relative
G92 set home

together with:

M0 stop
M2 program end
M3 spindle clockwise
M4 spindle anti-clockwise
M5 spindle stop
M115 get firmware version
M226 pause
M254 program start

I also implemented some M code extensions for my own evil purposes:

M110 set line number
M111 set debug level (handy when you can't work out quite what's going on!
M114 get current position
M255 get last line executed

communication is serial over USB
I have added a checksum to command sent
there's a "heartbeat" to show that it's still alive
full support for 6 limit switches (1 on each axis masquerades as "home")
E-Stop (uses one of the buttons on the scanner body )

yet to add
most important will be acceleration (but that needs dark room and intravenous coffee)

first pic shows the electronics
Arduino as controller; Sparkfun EasyDriver talking to the X stepper
Y driver is hidden under the Y stepper

second picture shows a disastrous attempt at Z
the whole thing struggles and creaks
then the gear on the stepper came loose

kept going long enought to prove the concept
and confirm that a "proper" machine is called for

and finally, here's a video of the kit performing a square dance - just to prove it works!

[RomanLini]

That's some pretty cool stuff you wired up there, and programmed it yourself too!

[mmcp42]

ok plan B!
I've been away working in Kuwait, but now back in harness

Squandered $10 on Solsylva's 10 by 9 plans

finally got around to doing something with them

1 to 4 - the tools I'll be using
5 - work in progress
6 - a nice pile of parts, or so I thought
7 - you can see the wood is quite warped, wonder if it'll be any good?!?
8 - complete set of wooden parts

I think the warped parts might be ok as they go in pairs, glued together, so I'm hoping I can have the warps opposite ways, glue and screw and force them flat.

next up is some sanding and installing lots of holes
then on the the bearing trucks

[mmcp42]

a little more progress
cut the chromed rods to length
(ok cut one of them 1/8" too short so back to the store tomorrow - dammit)
so 5 out of 6 on that score

cut the aluminium angle ready to start building trucks

got the first truck started, but then supper was calling
so that's it for today

[mmcp42]

2 1/2 trucks done; #4 drilled ready
ran out of M8 25mm bolts

[RomanLini]

It's good to see the bigger machine coming to life!

[mmcp42]

you have no idea how satisfying it is
my first serious build :grins:

popped to the store and cleaned them out of M8 25mm bolts (all 10 of them!!)
but I now have 4 trucks ready!

[mmcp42]

I'm researching steppers now
Leaning towards these..

opinions, please?

[mmcp42]

managed to get another couple of hours in the garage

trucks are breeding nicely!
started gluing/screwing the Y axis supports, too