[partsman]

I've stolen so many ideas from this forum over the years for different builds that I figured it was time for me to pay back and post one of my own builds. This is my second CNC plasma build. The first was a 4'x4' which has been used in a production environment since 2009. It has worked flawlessly but has a lot of limitations - size being the biggest.

This new build will be a 5' x 11'. One of the things I found with the 4x4 was to leave plenty of room on the sides of the sheet you are cutting so that the water in your water table stays in the water table. The 4x4 had exactly 4' in each direction so cutting down a 4' sheet made a big mess. Our new needs are to cut a 5x10 sheet so we are going about 5.5' x 11'.

We are building the frame out of 2"x6" 3/16" wall tubing.
The legs are 4"x4" 1/8" wall.
The gantry is 2"x4" 1/4 wall aluminum that I ordered from onlinemetals.com.
The electronics are CandCnc.com We used these electronics on our last build and they are bulletproof. Love them.



Here is the table framed up. We used the sheet of 5x10 that we are going to use for the water table floor as a big square. Measurements from corner to corner were very good (+/- .06" over 11'). You will never notice that with a plasma cutter.

[partsman]

We welded the rack (from McMaster Carr) directly to a 2x2 welded to the main frame. I did this with my last build and it really works nice. It provides sort of an umbrella for the gear and for the rail that will go on later. It also protects it from being bumped from the side while loading & unloading material.

[partsman]

Here is the sump in place. We will drain the water table every day into barrels under the table. This sump should help keep some of the gunk from collecting in the barrels. The table is ready for some primer and paint.

[partsman]

I'm using an aluminum gantry this time. It will save weight and looks pretty nice also. I'm going to weld the gear rack to a piece of angle and then bolt that to the aluminum. I thought about trying the 3m VHB tap[e but I just could not convince myself that it would stay. I'd worry about it every time I made a big cut in an expensive piece of material.

I decided to use a sealed gearbox on this build. My last build was direct drive and I didn't really like the resolution that I got. I also just don't like the idea of an exposed belt drive system. I made a little gearbox using off the shelf gears. I will fill it with grease and it should last forever. It ends up being about a 3:1 ratio which is perfect for this machine. I will need to build three of these.

[partsman]

Here is the gantry mounted to the table. I used rails that I purchased from ebay. I got two 12' long shipped from China for around $350/pr. They are the same rails from the same place as I used on my last build and they work really nice. I have my limit switches installed and everything seems to work well so far.


This is where I am so for on the table.

You can see in the last picture that there is a rail mounted to the right side of the floating Z assembly. This is where I will be adding a plate marker later on.

[partsman]

One of the things that I hated on my last table was the keyboard & mouse. We had them on a bench and they always got burried in junk, the mouse didn't like the dirt & grime, and the keyboard keys always stuck. This new machine will not have a bench (for that exact reason) and I really wanted a control panel like my Hass machine.... so that is what I'm going to build. There are going to be switches for all of the common functions plus a MPG with a 19" monitor on top. I can't wait!

I have holes for two fans in the sides plus a mount on one side for the power supply. The back will hinge open to access the insides.

I welded a plate to the table where the control panel will mount. Then I cut out and welded up a new control panel, primed it, and caulked it. Ready for paint.

I thought about having a PC under the table somewhere and then decided why not just put it in the control panel. Well.. a standard PC is too big and I could not find a compact PC with a parallel port so I decided to build a PC inside the control panel. I'm going to use a Asus Micro ATX motherboard and make the control panel a stand alone unit.


This is where I am so far. Right now I am wiring up the switches for the control panel at my desk and answering phone calls

More to come soon!

[partsman]

I stuck the switches and MPG into the switch panel to see how it looks. I think it will look pretty good. I'm going to organize the clusters into movement, setup, and plasma controller sections. I'm going to try to print the graphics on some of the weatherproof laser labels. Not sure how well they will hold up but the reviews on Amazon are really good.

The other picture is how I'm going to organize the buttons... I think.

[partsman]

Not sure what happened to that last picture...

Here it is again. Sort of my "artist" rendering of the final switch layout. These are the buttons that we seem to use the most here.

[partsman]

I made some more progress today. I built a mounting plate for the CandCnc box under the table. The face of the candcnc box will be flush with the sheet metal skirt that will go all around the bottom of the table. Should give it a nice clean look. I also got the controller arm & swivel built and the controller temporarily in place. We will be able to swivel the controller out of the way when not in use. There is about 100% humidity today so the paint is drying veeeeerrrrrrrryyyyyy slow. I cant do much else until the paint dries a little more.

[partsman]

Miserable FAIL!

The weatherproof laser labels are amazing! I scrubbed on them with soap and water and the print & color held up perfectly. I thought I had something.... until it came time to actually apply them. You get ONE shot at those and then you are left with a big mess. The adhesive on the labels sticks to the panel and has to be removed with acetone... which also eats off the paint. Back to square one. sanded the panel down to bare metal and will try again tomorrow.

I might just cut out the individual labels and stick them on there.

[BTA PLASMA]

You may want to think twice about a gear drive, gear-to-gear-to gear system. Your backlash will be obvious in Y right away and in X and the slaved axis it only works well enough when the backlash is the same (matching within less than .001 or face vibration). You may be adding more frusteration than you want.

[Kerryfisher]

Coming along well, I like those gearboxes!

[partsman]

You may want to think twice about a gear drive, gear-to-gear-to gear system. Your backlash will be obvious in Y right away and in X and the slaved axis it only works well enough when the backlash is the same (matching within less than .001 or face vibration). You may be adding more frusteration than you want.

It looks to me like I should be able to take out all of the backlash by sliding the motor tighter against the other gear - same as with a belt drive... only in reverse I guess. I played around with it on my bench and could not find any measurable backlash. What am I missing?

[BTA PLASMA]

Here is what pops out at me

While aluminum seems like a great idea for a gearbox the 6061 you most likely used will see temps of 140 degrees due to the operating temp of the stepper motors. The aluminum will move and the backlash will get worse. One thing you should do is use a indicator to see if you have any measureable backlash on the involute of the drive gear. Typically aluminum is decent material and used when you have a timing belt where a little thermal expansion wont cause backlash. On opposing axises backlash is tolerated because one is fighting the other. That is the only way a gearbox works in a precision machine like a laser or plasma or even router. But the single axis will show the backlash. Mach's backlash comp isnt ideal for plasma operation. At least not that I have seen. Then the other thing that pops out is the amount of power loss you will have turning a large steel gear. You will loose accel and deccel or worse steps. You can definately get that to work but there are much better options for movement.

Just an observation.

[partsman]

Very valid observations. Thanks. I'm pretty well committed to the gearbox idea at this point so Ill let you know what I find. I guess I might just be shooting for a different level of accuracy than some people. When I see plasma cutters advertised at .000X" accuracy I always think "why?". Unless there is something wrong with my Hypertherm (there isn't), the taper on a cut will always be several (4-5) degrees- accounting for MUCH more accuracy issues than I predict my expanding aluminum gearbox will have... but I might be wrong. I can totally see on a milling machine or even a router but plasma, to me, is just not that critical. Why stress out over .000X" when the plasma cutter itself is making "errors" that can be measured with a tape measure. Just my opinion.

[BTA PLASMA]

It is the accumulated error that causes bad geometry. Imagine being 1/4-1/2 off at the end of a complicated piece. If you were just cutting squares you may not notice but like any CNC machine that doesnt use a resolver and an encoder or use backlash compensation you will see that a change in direction will start the accumulation of errors. Not just so I cover the bases but so you know if you have any issues later you will know why.

[partsman]

It is the accumulated error that causes bad geometry. Imagine being 1/4-1/2 off at the end of a complicated piece. If you were just cutting squares you may not notice but like any CNC machine that doesnt use a resolver and an encoder or use backlash compensation you will see that a change in direction will start the accumulation of errors. Not just so I cover the bases but so you know if you have any issues later you will know why.

I don't understand how backlash would be cumulative. I can understand lost steps or belts slipping being cumulative but not backlash. If you start a complicated 5 hour cutting project on a machine with 0.01" of X axis backlash it seems to me like the most you would ever end up off at the end is 0.01" +/- on the X.

[BTA PLASMA]

bad example!

[BTA PLASMA]

Bad example

[partsman]

BTA PLASMA

Here is an example with .010 backlash
G0 X0
X1.
X0 (BACKLASH WOULD ONLY BRING IT TO X.010)
X1. (ACTUAL TRAVEL TO 1.010)
X0 (.020)
X1. (1.020)
X0 (.030)
X1. (1.030)
And on and on

So the more changes you make in direction the more the error adds up.

That is incorrect. If you gain 0.01" in one direction you will lose it in the other. Backlash is not cumulative.