RE: [CAD_CAM_EDM_DRO] Chain drive question????

>I'm building a cnc gantry table to use with a plasma torch and I'm
>considering using a stepper motor with a chain drive, however I was
>wondering if it was accurate enough???

The biggest problem is the weight of the chain causing sag (catenary
curve, y = a.cosh(x/a) ).

You minimize this by putting tension on the chain. When the table
velocity
changes the tension changes and you get some non linearity. It is also
possible to get bad oscillation, but that should be easy to damp out.

So what you need to consider is:

Maximum load on chain due to table acceleration.
Pre-stress on chain.
Maximum tensile strength of chain.
Ratio of static/dynamic load on chain.

Obviously you want the change in load on the chain to be fairly small,
which
means high pre-stress. This requires a very stiff mounting base for the
gears
you run the chain on.

It is probably not a good idea to mount the chain directly on the
stepper motor
shaft, it won't survive the high bearing load. You'll need to gear down
a lot to get adequate torque anyway.

Cheap gears like you get on bicycles don't give you linear motion on the
chain,
but the relationship between angular motion of the gear and linear
motion of the
chain is predictable, so you can compensate for the error in software.
It is
probably possible to get gears that minimize the error. There are
mechanical
tricks around it as well.

To combat oscillation you want a heavy chain, but that is a trade off
between required
acceleration, size of the motor etc...

The first CNC machine I ever played with was a 5m x 15m plasma cutter.
It
used a fixed chains both sides of the gantry with single motor driving
gears
both sides. It was fast and apparently very reliable. The chain snaked
through
three gears:

/O\
O| |O
-------/ \------

Top is driven, other two are idlers. Chain is tensioned at one end. The
whole
thing was mounted on a huge slab of concrete. Gantry ran on railway
tracks that was
solidly mounted one side only, other side floated since the difference
in thermal
expansion between the concrete and steel was huge.

When using a chain there is no guarantee that the links are all equal
length. So your absolute
accuracy is a bit of a gamble. If you use a recirculating chain any
error will be all over the
place, which is why you typically want a fixed chain with a moving gear.
If you keep the chain fixed
all the techniques to compensate for a non-linear screw applies to
compensating for the error in the
chain. Measuring and sorting the links helps, but it's a tedious job and
you need a good
measuring jig and a temperature controlled environment to work in.

Chain drive should work much better than acme screws, unlikely to as
good as zero-backlash
ball screws. Whether it ends up costing less than ball screws depends on
what your time is worth.

Nic.