CNC router (stepper motors in parallel)

Hello folks,
I'm slowly amassing the parts to buiild an 8' x 4' CNC router and most of the designs I've seen have dual leadscrews on X.
What are the mechanics of driving/synchronising these?
If we construct a crude 'flow diagram', it would go:
software -> parallel port -> breakout board -> stepper driver board -> stepper motor
Where should the signal be split? Can TurboCNC or Mach 3 send simultaneous and identical signals to two sets of parallel port pins?
If I split at the breakout board and send to two driver boards, is the signal in any way weakened by doing this?
If I split before the driver board, are there any synchronisation issues?
If I split after the driver board, do I need to consider, if you like, 'power issues' i.e. if my motors are 2.5A, does the board need to deliver 5A?
How important would things like keeping the wires the same length be?
Thanks very much.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

If it was me Id split the feed into the driver, and use 2 (identical) driver boards. The input impedance to the driver boards *should* be high enough that the drive from the parallel port will be sufficient. A quick look at the data sheets should confirm this, and if not then a buffer amp to split the signal is an easy fix. I wouldn't trust 2 pins to drive 2 drivers and rely on them being synchronized, Thats hard enough to achieve in properly dedicated hardware with properly dedicated software (my day job) that anything lesser is not likely to work 100%, and its going to fail just when you dont need it to (assuming it works in the first place).
Keeping the wires the same length is theoretically very important, however unless your very unlucky (or have some expensive test equipment) differences in the order of a couple of feet will be hard (impossible) to detect.
Have you considered a mechanical split? ie use 1 stepper (approx 2X the size?) and a timing belt to drive a pair of leadscrews. that simplifies the wiring, and I suspect that if you do the calculations you probably dont need twice the size of what you are currently planning on.
hth
Dave
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Sun, 10 Jun 2007 13:01:42 -0700, dave sanderson

...
Dave,
Thanks very much for the suggestions. As a comment only, and not wishing to start an argument, your advice given on the basis of your 'day job' (i.e. not possible electronically) does seem diametrically opposed to that of later replies.
If I do use a (long, flappy) timing belt, how do I cope with backlash?
Thanks.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
John Montrose wrote:

Nope, you misunderstand, not impossible, just likely. In widoze, iirc you cannot guarentee anything to less than 10mS accuracy, adn given the scheduler and other programs running it is likely that you will get pins waggling at different times. whether this *actually* affects you in the real world I dont know, but given that you would get in a mechanical lockup if it happened and the other simpler alternatives which cant give this problem. Dayjob wise errors in the >1mS range can be a problem, I once spend a week chasing the differenct between 2 inputs on a chip...<Tears hair out>

Simple, dont use a flappy timing belt ;) Timing belts are designed to give acurate relative rotation, its there purpose. In a car they only rotate 1 way, so you can get away with a simple tensioner, there is no backlash, so a simple spring will take out the flap and the tension side is always the same.... on a mill you have more space,and are less constrained by packaging / robustness etc, so design a tensioner that keeps the belt tight in both directions, soemthing like a fixed tensioner pulley on the backside of the belt, or even the correct length belt. The Yanks seem to use timing belts a lot on their CNC conversions so they can mount the motor out of the way.
Dave
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Sun, 10 Jun 2007 19:50:26 +0100, John Montrose
wrote:

Two ways of doing this, mechanical or software. Mechanical requires two screws to be driven by timing belts or similar and geared together. Software approach below.

Correct but after the breakout board we have N x stepper driver boards to go to N x motors where N = the number of driven motors, not always the number of axis, again more later.

Can't comment on Turbo CNC only Mach3. In Mach 3 you have 6 axis X, Y, Z, A, B and C X, Y and Z can be slaved to A, B or C quite simply by selecting from a menu which ones need to slave to which. So in your case on a router assuming no rotary axis you select X to slave with A and wire the two identical motors up to Z and A step and direction pins from the breakout board to 2 of your 4 drivers, then onto the motors. You only need to setup the X axis as the slaved axis [A] mimics the X axis. All driver settings like microstepping has to be the same for both axis.

No it treats the supply as being 4 motors just that two are slaved.

Doesn't apply.

You need to add all your motors together to get a total amperage and aim for 2/3 of that.

Under normal build constraints this shouldn't cause a problem but if you feed the two slaved motors centrally and then go different directions the cable lengths will be more or less equal.

-- Regards,
John Stevenson Nottingham, England.
Visit the new Model Engineering adverts page at:- http://www.homeworkshop.org.uk /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Sun, 10 Jun 2007 21:04:16 GMT, John Stevenson

John,
Thanks for the reply. Please read this with the implied huge smilie: wiring identical motors to Z & A will surely have, shall we say, interesting results!
I'd prefer to do it this way as it makes the build simpler, if more expensive. Directly connecting each motor to its leadscrew means I don't need to worry about coping with backlash in a (minimum 8') long timing belt.
Thanks again.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 11 Jun 2007 21:49:59 +0100, John Montrose
wrote:

No no, TWO identical motors, not two THE SAME motors <bg>

From your other post to Tony on about lost steps if you are thinking about using the Big Bertha's from Arc and the large drives then just direct couple them via a coupling and don't worry about reductions, belt drives etc. The loads on a router are quite small unless you are building it from RSJ's and these motors will be more than enough.
As an idea I have motors slightly less powerful than these of Ketan's direct driving 0.200" pitch ball screws with similar drives to Ketan's on 2 axis and one of Ketan's on the other. This is on a 3 tonne machine that does some serious work.
If you was going the belt drive route the best arrangement would be to mount the stepper central and using two separate pulley and belt sets, drive each screw. This way you will have approx 2' centre's and the belts would be less. Backlash is dealt with by using an idler and getting the belt tight. Timing belts are designed to be run tight, and I mean tight.
Other methods that may bear looking at is to hang the gantry off high level rails as opposed to the normal method of bed level rails that everyone follows. Upside of this is two fold, one the rails are out of the crap and secondly you can have the X motor mounted centrally on the gantry and drive both side via two jack shafts and a spring loaded rack and pinion.
Some of the high end laser cutters run this method as it's faster than balls screws due to whip over an 8' length.
Other alternatives is to replace the rack with a long length of timing belt stretched and glued to a rail and a spring loaded timing pulley as a pinion. The same has even been done with lengths of chain and sprockets but that smacks of being designed by Harley Davidson and I'm sure anyone can do better than that.
-- Regards,
John Stevenson Nottingham, England.
Visit the new Model Engineering adverts page at:- http://www.homeworkshop.org.uk /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
John S has given you the SP on doing "slave" configurations - 2 motors driven from the same axis signals. This can work OK under normal conditions; however, it is worth considering what happens when (as will inevitably happen) you drive the slaved axes too hard. One of them will lose steps first, at which point, the axis will be driven slightly "crab-wise" - one motor leading the other - till it locks up good & solid. At that point you will need to unwind one of the motors to get that axis back in operation. In contrast, if you take the mechanical approach and drive both leadscrews from a single motor via toothed belts or whatever, this particular problem goes away.
Regards, Tony
On Sun, 10 Jun 2007 19:50:26 +0100, John Montrose

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

...
Tony,
Thanks for the reply. As you say, once steps are lost, there is a problem. I'm hoping to minimise the possibility of this by overbuilding using Arc's bigger motors and drivers, with a belt or possibly a planetary reduction box.
The thing I don't quite understand is how one deals with backlash in a long, toothed belt set-up. If the router can cut 4' boards, the belt needs to be minimum 8'. It won't be like my car where the belt always runs in the same direction; there will be continual reversals.
Thanks for your time..
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 11 Jun 2007 22:03:31 +0100, John Montrose

The answer is simple.. If you have, say, 3mm of backlash and 50mm pulleys that's about 7degrees. With a 5mm pitch leadscrew that's about 0.1mm of error. That's bugger all on an 8x4 sheet of wood. It's equivalent to a 1 deg C change in temperature of the steel frame of the router.
From experience with tape libraries using 16 foot toothed belts direct driving the picker arm, I'd expect 3mm to be on the outside of the possible backlash. 1mm would be closer to normal.
Mark Rand RTFM
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 11 Jun 2007 22:45:37 +0100, Mark Rand

Is bugger all an Imperial unit or Metric? <G>
Regards, Tony
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Bugger all is metric. The correct Imperial quantity would be three fifths of five eights of F*** all.
Mark Rand RTFM
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

AND in Fahrenheit !!

-- Regards,
John Stevenson Nottingham, England.
Visit the new Model Engineering adverts page at:- http://www.homeworkshop.org.uk /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.