[Craigie]

I'm considering building something along the lines of the attached sketch- is it nuts ? Has anyone experience of making a large format machine ?

Craigie

[neilw20]

If you are using long cable runs for data signals, encoders, limits etc, any run above 3m requires differential line drivers or fibre optic cables so that noise does not corrupt the signals.
For limit switches slugged with a filter you will get away with long cables, but for encoder wires used with servos you need RS422 line drivers or fiber optics.
5m long will need 8m cable runs.

[Craigie]

Thanks for that - This is probably why I should stick to things I can fix with a hammer !
I'm guessing fibre-optics is a stage too far- I'll need to look into what a differerential line driver is/ does.

The electronics were always going to be the bit that bit me - if it means I can use out-the-box parts I could alway reconfigure the set-up with the drivers etc centrally-mounted? (means moving long rack and rails to underside ) Otherwise I'm going to have rely on a lot clever people having a lot of patience.

Any links to clear explanations of differerential line driver would be appreciated.

Craigie

[neilw20]

RS-422 - Wikipedia, the free encyclopedia
RS422, specifications and in depth tutorial

[Craigie]

Neil
re differential line drivers - I'd got as far as cost first - and it's a relief to find it won't be these bits that'll kill the project.
I've very quickly scanned your link and I'll have a proper study over the weekend. i think the principle has filtered through !
I'd be interested to know what kind of difference "noise" will make....? Does this translate as loss of a MU of accuracy or full-on leaps and stops ?
If it's the former I can live with a pencil line thickness for most cutting: otherwise I'd better get reading .
Thanks
Craigie

[neilw20]

Using steppers. Then just make limit switch wiring a problem, easily solved.
Using servos and encoders, the signals will be corrupted and the drives will fault and you will get positional errors, more than likely.
At low feed speeds it would still work OK, most times, if you are lucky.
Invariably you have power wiring to run a spindle, and spindle power transients will muck it up.
If you have spindle index signals, they can be corrupted.

Bottom line.. Long cables - use differential drivers. $3 per 4 channels + wiring and extra bits.
http://au.element14.com/jsp/search/p...sp?sku=1564965

[Craigie]

Neil
thanks - it's good to know there's cheap solutions to at least some of this.

i've got some prices through for some Rolls Royce track and Roller pinion from Nexen.
Ouch ! I was working on the principle of over-specing to paper over any lack of accuracy in the bits I make.
As a short term option I may have to go to roller chain for the long axis: I've followed threads on various sites that suggest this can lead to "bumpy" cuts. Support the chain in some kind of channel arrangement ? increase cog size. ( apologies if I don't know the correct terms )
Better stil can anyone recommend effective and economical rack ?

[neilw20]

Back to back cog bets nested in each other.
Belt acts as a rack.

[Craigie]

Neil
I'm not sure if it's what you mean but I can see fixing down a cog belt then making a "caterpillar" track meshing with this over a short section could be interesting- although how do you avoid slip of the inside out belt ?.
Any thoughts on flexible rack .
Craigie

[neilw20]

http://www.cnczone.com/forums/linear...tml#post618586

Nice picture shows how it works here. Scale it up or down to suit your needs.
Be aware that the steps are much coarser than using a screw, but you could gear box drive it, then you need to consider backlash in the gearbox.

Direct microstepping drive may have sufficient resolution for your needs.

[asuratman]

hi neil,
Is there any info to use TEXAS INSTRUMENTS - AM26LS31CD. - IC? Somebody sellingkits? I have ask Marcus at Gecko, that we can use cui amt 102 w/o that TEXAS INSTRUMENTS - AM26LS31CD. - IC with no problem as long as 100 m. No data corrupt.

[Craigie]

Neil - That's very elegant. I see also a version with reduction gearing. Is that needed ?

If this works my scheme may take a sideways leap - more to follow...
thanks again.

[alheira68]

If you have all cables with shield and well connected for sure you will not have problems with communication for encoders.

For your drawings looks will be slowly and light machine. Tae care about the forces and problems you will get during operation at high speeds.

Adira SA >> sheet metal working machinery: laser cutting, press brakes and shears -> they made an 6m x 2m high performance laser cut machine...

Some aspects to take care in the project time

[neilw20]

Sorry, alheira68,
But you probably will have problems with long runs, even if shielded.
Shielding protects against electrostatic interference - capacitively coupled interference, but not electromagnetic interference. That needs balanced lines, so the induced interference cancels.
Sure if the machine, has no motors, and no high currents nearby it can be made to work, but exceeding 3-4m is asking for problems,
and this is MACHINE SAFETY - - humans can be injured, so use balanced lines or optic fiber.

[alheira68]

Ok, I understand... probably our numerical control already make it!!

But for homemade project what can be done to avoid any change of normal operation (voltage control) is apply signal converter between the numerical control and the encoder... so, for the numerical control this is transparent.