[Thermwood]

Hello everyone,

I'd appreciate some feedback on this matter since I cannot come up with a smart answer ;

we are considering building a cnc plasma from scratch with a work envelope of approximately 50" x 100" to accommodate 4x8 sheets.

We have two machines on the floor and are relatively familiar with router and machining center mechanics since we retrofitted both of them (Thermwood 5x10 c50 router and biesse 20 ptp machining center).

The plasma will see occasional use but we'd like to build it as if it were to run often since this may be a whole new line in our business ....

After looking at many grooved bearings home made rails, we are of the opinion that commercially made linear bearings are the way to go,either thompson supported rod type or thk slide type (we currently have both on our machines....)

It seems that considering the rail lengths needed, Automation overstock has the best deals with 30 mm Hiwin heavy duty rails and blocks X axis master and slave and y axis for maybe $ 1,200.00 total.

Please do not hesitate to comment on choice .... We could not located fully supported rod sufficiently long ...

Here is were the question arises ;

I can see mounting fully supported rod type slides on a mechanical tube frame assuming that the slide geometry grants it extreme rigidity that would counteract the poor tolerances of tube steel and span the variations, but in the case of the Hiwin rails, I am concerned about mounting the rail onto standard mechanical tube ( we were thinking 3x4x .1875 hot rolled) with .03 tolerances and basically transferring this poor tolerance to the slide assembly ....

On our biesse, I believe the frame is made out of hot rolled tube but a "flat" was clearly machined into the tube where the rail mounts . I assume to maintain straightness ... having no practical way to machine the tubes, wondering if the frame members supporting those rails x axis and gantry should be made out of cold rolled bar (not too happy about cost) or if there is a simpler solution such as shimming ?

omitted to mention that the basic design at the moment (conceptual only) is a rigid mig/tig welded steel tube frame with linear bearings mounted on the outside face of top tube rails in the X with a gear rack underneath thus my concern of steel tube tolerance in order to maintain parallelism and straightness between both X slides (also stand for the gantry Y...)

thanks for any input ....

Pierre

[louieatienza]

There are a few large builds that used self-leveling epoxy to create an accurate bearing surface for the rails. Thogh 30 series bearings seems pretty large for a machine with no cutting forces?

Check one process here:
MadVac CNC - home made 4'x8' cnc precision gantry router

[RCaffin]

Hi Pierre

To handle 4' x 8' sheets? I am tempted to suggest that 3" x 4" steel tubing might be a bit light for that size of frame. I would prefer to go up a bit in size, and add concrete blocks.

> Though 30 series bearings seems pretty large for a machine with no cutting forces?
No cutting forces maybe, but you have to handle the weight and inertia that huge gantry! I suspect that cutting forces might have been a small consideration in comparison.
EDIT: 20 mm would be adequate I think.

The use of epoxy for bedding the rail to the frame is well known. However, you would have to get the steel tube straight first, and given commercial tolerances on steel beams, thermal/welding distortions, delivery trucks, welding errors, etc etc, I would not assume that the bed will be straight enough at the start even with epoxy.

How do you build a big machine? You start with a bigger one. :-)
EDIT: or you hire Madvac to do it for you! Impressive stuff.

Cheers

[LeeWay]

A plasma cutter is not quite as critical with regard to perfectly straight rails. While you certainly want them parallel, the actual rail height or straightness can vary and you can still get great cuts with a THC. You can use shims under the rails in the low spots. Shims are adequate on a relatively light duty machine like a plasma cutter. Granted, it still has to be a solid frame, but not as much as a large router. You do still need high speed, but the gantry doesn't need to be as heavy as one on a router this size, simply because of the lack of cutting forces.

The initial setup for measuring and shimming your rail height will be the part where you have to get creative. You still want the rails fairly close to straight. That will increase bearing life, but how close is close enough? I would say within a 1/16" over the length would be very good. 1/8" is still usable. Once you start cutting on the machine, your slats will get slag on it and unless you grind that slag off between every sheet, it will throw off the height of the plate over the span of the machine. It will do that on a machine with perfectly flat milled surfaces for the rails too. This is where the THC comes into play. It corrects any unevenness on the plate.

There comes a point in a plasma cutter where being more accurate or closer to perfect will not produce better parts or cuts. It will only cost more money.

[RCaffin]

There comes a point in a plasma cutter where being more accurate or closer to perfect will not produce better parts or cuts. It will only cost more money.

Point taken. I learn something.

I wonder - could you get some 8' x 3"x3" square tubes ground/machined flat on one surface?
If so, weld up the frame first, let 'cure', then mount these 3x3 tubes on top with epoxy padding which you let cure first before torquing the bolts! That us, no welding for the mounting rails, just lots of bolts. Then mount the linear bearing rails on that, again with thin epoxy packing. Just check for any spiral along the length. Hum - yeah, possible.

Btw - a cute way to get the top surface flat is to first cast a flat floor, and then build the frame upside down on it. There are companies which will do the floor for you, with an epoxy top coat for flatness.

Cheers

[LeeWay]

There was a discussion on here a long time ago about these epoxy floors that are supposed to be flat.
The larger they are, the more pronounced it becomes that they are actually curved or convex when cured. Gravity pulls it to the curvature of the Earth.

[Thermwood]

Thank you all for the input,

Few design thoughts since my original post ;

Indeed plasma precision and forces are nowhere near those of a large mill or router ....
The 30 bearings may be overkill ; thought was that the cost of the overstock units between 20 and 30 was not much different and the fact that I preferred bigger bearing and blocks due to my intention to make the gantry out of steel tube as opposed to AL extrusions ....

My thoughts about rail parallelism is that since it is a master slave set up, it seems that the most important is to maintain decent parallel tolerance between both rails so the bearings do not end up fighting each other causing premature failure, and I am thinking of using 3x5x.25 mech tube carefully welded around an MDF jig that I would cut on our 5x10 router ... (bigger machine to build smaller machine) then set one linear rail as true as possible using either shims or epoxy behind. Then build a carriage off that rail and run a dial indicator to set the opposite linear bearing ... I think 1/32" accuracy could easily be attained this way ...
Also planning to use a taught piano wire to check straightness ...

As to matching rail height, planning to set up on top of our router bed which has a .5" aluminum plate 5x10 as bed substrate and indicate from that bed ....
I am planning the gantry so there would be mechanical adjustment in the way of slotted holes and machine screws at the horizontal / uprights junction to allow for squaring the y axis to the x axis .... Also definitely planning to use DTHC which should give me some leeway as to frame tolerances when it comes to slight height discrepancies between x master and x slave ...

The design has changed a bit, since I ran into a great deal ; I was considering a Dragon cut servo system with Mach 3, but ended up acquiring a full electronic package for near nothing (500 bucks) ; full licenced camsoft professional, paired with inboard 6 axis galil motion controller, a couple of interconnect boards with a ton of I/O's , 3 baldor servos BSM 90 7.8NM .... for the X, X slave and Y, as well as 2 small servos for Z (only one would be needed), matching AMC amplifiers for all axis ....
And the beauty if it all is that this was a laser retrofit made about 7 years ago by a CNC engineer who has consulted with us in the past ...

Next hurdle is to figure out gearing to bring down servos from 4 k RPM and a relatively large pinion (shafts are .94" with +/- 2.25" dia pinions at the moment) down to a targeted rapid of about 1k ipm , without breaking the bank by having to buy baldor matching gear boxes ......

Attaching a sketch of our design to date .....
Attachment 206944

[louieatienza]

No cutting forces maybe, but you have to handle the weight and inertia that huge gantry! I suspect that cutting forces might have been a small consideration in comparison.
EDIT: 20 mm would be adequate I think.

I visited a cabinet shop nearby that had a Biesse Rover machining center. From what I can tell, it had 30 or 35 series rails, and the gantry was CANTILEVERED from the back, at least 5 feet. I would think the spindle and carriage assembly alone has to way 500-700 lbs., and that does not include the weight of the gantry bridge, or upright, or ATC. So it's not just the size of the rail but how it's mounted.

[LeeWay]

I have THK HSR 25's on both my 80/20 mill and lathe. They are way overkill for that, but would work great in a 4'x4' machine up to about what the OP is making. Very nice linear ways. 30's would not be too much overkill especially if costs are similar to the 20's. Remember the rail is larger and that will help a lot if shimming has to be done.

[RCaffin]

Hi Pierre

Some detail comments for you.

I would replace the rack and pinion drive with a long length of GT2 or GT3 toothed belt (say 8 mm pitch and as wide as it comes), and put the drive motor on the gantry. A rack is sloppy and wears under load. The belt is far more accurate, does not wear, and is easily replaceable anyhow.

I am not entirely happy with the idea of running the bearings on their sides, but I can see lots of benfits to that config as well - like keeping the rail clean! In that case I would run the 30 mm units.

Reduction on the motor drives - I am not sure I would bother! You might find that the motors will be working hard handling the gantry anyhow. Yes, you could use a small reduction using a GT3 belt if you want. Much cheaper, less wear, quieter ...

> I think 1/32" accuracy could easily be attained this way
I would suggest aiming for a lot better than that. Use the carriage itself to get the alignment.

You will never regret building a slightly stronger and more accurate unit. :-)

Cheers

[RCaffin]

There was a discussion on here a long time ago about these epoxy floors that are supposed to be flat.
The larger they are, the more pronounced it becomes that they are actually curved or convex when cured. Gravity pulls it to the curvature of the Earth.

Indeed. For a shape the size of the proposed table, the middle may stick up 0.11 microns.
What a pity.

Cheers

[Thermwood]

Hi Pierre

Some detail comments for you.

I would replace the rack and pinion drive with a long length of GT2 or GT3 toothed belt (say 8 mm pitch and as wide as it comes), and put the drive motor on the gantry. A rack is sloppy and wears under load. The belt is far more accurate, does not wear, and is easily replaceable anyhow.

I am not entirely happy with the idea of running the bearings on their sides, but I can see lots of benfits to that config as well - like keeping the rail clean! In that case I would run the 30 mm units.

Reduction on the motor drives - I am not sure I would bother! You might find that the motors will be working hard handling the gantry anyhow. Yes, you could use a small reduction using a GT3 belt if you want. Much cheaper, less wear, quieter ...

> I think 1/32" accuracy could easily be attained this way
I would suggest aiming for a lot better than that. Use the carriage itself to get the alignment.

You will never regret building a slightly stronger and more accurate unit. :-)

Cheers

Thanks for the input,

not sure how the long belt set up works .... was going rack and pinion because both our biesse PTP and Thermwood retrofit router use rack and pinion with great accuracy ....
Servos are so large (7.8 NM) probably will have plenty of torque even at rapid and no cutting forces here so I am opting for a simple 3:1 timing pulley system just to reduce speed a tad ... Servos are rated 2000 RPM at 300V and have a 2" pinion so that's 12k RPM at the output ...)

Reason for tucking the servos parallel / behind the gantry are as follows ; no need for 90 degree box (do not like servos sticking out of the envelope, our Y axis servo on the thermwood does and I have nightmare thinking about someone hitting it with a cart full of stock ....) since gantry will not be very high, I figured it woud be acceptable and elegant ...

As to accuracy, certainly aiming much tighter than 1/32 but not sure that super tight tolerances and associated headache will lead to better results since plasma tolerances must be pretty loose ... still aiming for as much precision as possible .....

[steve123]

I did a lot of research on self-leveling epoxy for rails. Here is the scoop: http://www.cnczone.com/forums/vertic...ml#post1278466