[reiyuki]

I have a TAIG mill and was looking at building something bigger, more like a router, with perhaps a 3x6ft work area.

I saw the thread elsewhere talking about the massive stregnth and rigidity coming from homemade epoxy granite. I was thinking of building a frame out of 8020 rails, and then capping the ends and filling 1-2 channels with epoxy granite for rigidity.

Shouldn't hold too bad for weight, and I figure it'd give enough added strengh to handle deep cuts.


Thoughts?comments?

[Stepper Monkey]

I seriously doubt that filling the t-slot channels would help at all. Some 80/20 extrusions have large enough open center sections that might work out for you to fill, but the channels themselves would not help at all with that small flat cross-section. Even filling the center segments would only really help torsional rigidity. Deflection at center span would likely not be much affected.

Aluminum has plenty of strength already, what it lacks is rigidity.
I don't use the stuff for long unsupported free spans the way you might need to for a gantry router, but I have always thought that perhaps simply using a steel truss rod through the length of it to pre-tension and put the member under compression might work very well to solve exactly the problem you are worried about. Something I'll have to try eventually when I come to it. Cheap, easy to do, and very light.
It would be a necessary precursor to filling it with epoxy concrete in any case, as to get any real effect on deflection you would need it in there anyway.

[ger21]

If you read the whole thread, the general consensus is that filling the extrusion won't add much strength or rigidity at all. The epoxy granite's main benefit is damping vibration.

[jpelosi2002]

Putting a steel rod in tension through the center of the extrusion is a good idea on paper. I haven't run any numbers but my gut is telling me that the amount of tension needed to have a noticable improvement in rigidy would cause alot of unwanted side effects on the extrusion, most notably a compressive deflection. If I get some extra time I'd like to get the actual numbers or hear of anybody's experiments.
Jim

[bearwen]

Why would you want to run rods and fill with concrete etc.. any aluminum extrusions just choose the right aluminum for the job look at the new shopbots they have an extruded aluminum gantry its just of large size before you go screwing up your expensive aluminum I would suggest you download the deflection calculators for what ever aluminum you have 8020 tslots etc... and crunch the numbers you will find as I am currently building that you can get the right size aluminum to do a 66" to 72" gantry that will deflect only .001" or less with over a 100lbs load in one solid piece I dont know about you guys but I really dont think I put that much force on the cross bar of my gantry most likely you are going to burn wood or break a bit if you have that kind of load well not trying to preach just my 2 cents.

Bearwen

[Zumba]

Filling an extrusion with E/G is certainly a good first step, assuming the extrusion is of adequate size. Fill the main cavities, not the slots. For a 6' span, I would recommend going for the 3" x 6" extrusion size.

You can add a significant amount of rigidity by bolting angle iron or aluminum angle to the unused faces of the t-slot extrusion.

[whitis]

Why would you want to run rods and fill with concrete etc.. any aluminum extrusions just choose the right aluminum for the job look at the new shopbots they have an extruded aluminum gantry its just of large size before you go screwing up your expensive aluminum I would suggest you download the deflection calculators for what ever aluminum you have 8020 tslots etc... and crunch the numbers you will find as I am currently building that you can get the right size aluminum to do a 66" to 72" gantry that will deflect only .001" or less with over a 100lbs load in one solid piece I dont know about you guys but I really dont think I put that much force on the cross bar of my gantry most likely you are going to burn wood or break a bit if you have that kind of load well not trying to preach just my 2 cents.

Bearwen

Check your calculations. At 72" and 100lb load using 1 3060 extrusion ($212) over two end supports, I get 3.461mils in one direction and 12.393 mils in the other. The numbers are only 0.865mils and 3.098mils if the ends are rigidly supported. Most gantry's would hardly qualify as rigidly supported - they get their rigidity from the crossbeam rather than provide rigidity to it. If you stand on the cross beam, it will bow down like a 2x4 (only less so) and the sides will bow outward (pivoting around the rails). If you are using round rails, your bearings will forgive you; if you are using profile rail, you are looking at the possibility of rapid bearing failure.
And this is only static load. Under dynamic loads (cutting forces), your beam may resonate and buck like the tacoma narrows bridge with excursions an order of magnitude higher than suggested above. It also isn't counting torsion. And this one axis is only about a tenth of your error budget. Table, Y rails, Y bearings, vertical part of gantry, gantry beam, gantry rails, gantry bearings, z-axis rails, z-axis bearings, spindle, vise, and tool, and workpiece. Plus backlash. In order to make a tool that is accurate to 1 mil, the beam itself may need to be stiff to 0.1mil.

There is a reason most of these machines are called routers and not mills. Because they can only handle soft materials to woodworking tolerances, not hard materials to metalworking tolerances.

Now, if you use 6 of those extrusions, ($1274), and connect them together properly (which is a royal pain), you end up with a beam which weighs 270lbs and improves things considerably. Back of the envelope approximation is that the deflection would be 0.02 mils, static. You could drive your car across it and only move half a mil. But it still may need damping, depending on the application.

One machine design used a 2 foot diameter steel tube for the beam and still had to damp it for an ultimate accuracy of just under 0.1mil (one "tenth") and a working area of only a couple cubic feet, even though the entire machine was the bigger than a big router - it was all structure and very little travel. Of course, the machining forces were insanely high. The machine was for production grinding of end mills.

So, the original poster may have been more or less on the right track for the size machine being contemplated. Over a 36" span, one 3060 extrusion deflects 0.2 and 0.8mils with 50lbs of machining forces. More than one would be needed.

[Jkountz]

Hhmm I just ran almost the same numbers with the 8020 deflection calculator and got totally different numbers. I put in 3060 at 60" long and a 50lb load. The load centered number was only .001 on the X and only .005 on the Y. I went with load centered where as the load is the Y and Z car of a typical 3 axis router. The deflection was slightly higher when speaking of end load (ie weight at the end of a diving board) but that wouldnt apply with a gantry style cnc router. You would want to go on centered load since that would be the most force at one time. As the Y car moves closer to each end the numbers would only get better as the Y car approached the side supports. Does this make sense or am I missing something?