Sanity check- 6'x10' table

[Zephrant]

I did some tests this weekend. An error of 0.010" or more in the wrong place is easily discernable, and would cause additional steps in the assembly process to fix. Picture two glass panels mating at 90 degrees. Even a tiny difference is easily visible, and in my case, creates a weak point in the chemically welded joint.

If I cut out a 2'x8' piece (front panel), then a 2'x2' piece that mates up to it (end panel), a difference of more than ten thousands in width means the joint needs time consuming hand work, one on the main things I am striving to eliminate. With experience, I might be able to tell where the table is "off" at and adjust my code to fix it, but since I build full-custom equipment, it is rare that two tanks are the same size.

Can a rack and pinion system be made ridged enough such that it can't get jerked around by a typical 1/2" bit in acrylic?

You know, if people would just accept fish tanks made out of solid wood this would be a whole lot easier.

Thanks all-

Zeph

[MikeA]

Zeph, if you made them out of wood, would'nt be a barrel?
Can a rack and pinion system be made rigid enough? This is some what of a loaded question. The rigidity of the machine will be a by product of how it is consructed. The afore mentioned Thermwood weighed in at 12,000 lbs., used 1.5" and 2" Thompson linear bearings. The Multi Cam I have now weighs in at somewhere around 6000 lbs. The rack and pinion I feel are only a small part of the equation. At 5x10 like you want, the frame and guides are what will most likely determine the rigidity and accuracy. Where that threshold lies of what's needed to cut 1/2" acrylic smoothly, and day to day I am not 100% sure?
I sense your eagerness to attempt your own machine, but if these tanks are a profit earning thing for you, could you look to perhaps outsourcing your parts for now and then look to buy or build the right machine later? Mike.

[Zephrant]

Mike- Good points- It is not cost effective to out-source the custom parts. I'd pay too much to someone to generate the code on each unique part to make any money at it. If I found another hobbyist that was local, it might be do-able, but the big CNC houses in town are too expensive.

There are some standard parts that I could out-source, but the quote I got back was $10 each for the 8" flange shown above. Considering I use 13 in one product that sells for $250 (due to marketing constraints), it is not cost effective either.

I agree that the rack and pinion are a part of the equation, but all the slop does add up. I am working on the assumption that if I do the best I can reasonably do at each step, the over all slop will be adequate when I get done. That's a lot of assumption, but I know of no other way to proceed.


My goal is to make a "better" machine than a shop-bot, for similar or less money. Their total accuracy is rated at 0.015", which is larger than I desire for my machine, so I'm looking at where I can get the most improvement for the least money. At this point, I'm thinking of a linear shaft/ball bearing setup to carry the weight, and am waffling on the Rack or ball-screw for the drive.


Thanks for the help-

Zeph

[balsaman]

What happens is machines loose the accuracy when they change directions. If you make a rectangular part, I doubt the straight lines will vary by .015, but if you make two rectangles that are 10" x 15", the size may vary by .015"

My machine built for low dollars only varies by .0055", and I took that out in the software. It's only 12x24" mind you. Measuring several identical parts shows more like +/- .002".

I would go with acme rod and antibacklash nuts. I would turn the nut and not the screw on a table that size. I would probably go with 3/4" OD rod. There will be no backlash when the machine is new. Changing the nut will be easy if some develops that you can't live with. The plastic/delrin antibacklash nuts are pretty durable.

Just my 2 cents.

Eric

[HuFlungDung]

Zephrant, another possibility, although it could be expensive, is to mount scales on the machine for feedback, instead of spending the money on encoders for your servos.

You might only need a scale with .001" accuracy which is not a premium scale in today's technology. There is a Newall brand scale, based on a sensor which reads position from a tube filled with ball bearings. You would need to contact them to find out if you can get a suitable digital output from the scale unit to interface into your motion controller.

I did this with a Sony magnescale unit on a lathe, but the scale was only 8" long, not 10 feet. The conversion box cost $400, plus the $1500 (Canadian) for the scale. The trouble is the Magnescale is likely overkill for your application, as it is capable if incredible precision. Anyways, using a scale for feedback eliminates the positioning errors due to backlash, although a loose fitting screw makes a very poor "clamp" when the servo on that axis is supposed to hold a given position while the other axis is busy.

[Zephrant]

Thanks guys- I've been out of town for a few weeks, just got back and am in the scheming stage again.

BM- Thanks for the info- One of my requirements is to cut out four pieces of acrylic that are 4x7", and have them all come out the same size +/- 0.005" or better. If one pieces is 0.010" larger than the other, the glue joint gets marginal. Much more than that and the box is not water tight without additional work.

HFD- I have seen scales before I think, magnetic ones, but was scared off by the prices. I like the absolute feedback they present, I'm just not sure I can afford them.


Thanks for the ideas-

Zeph