[Motivator-1]

Hi,
I have followed hundreds of threads about gantry router construction, and am now designing one to be made of aluminum. I am planning on a 50" x 50" X & Y with about 6" of Z travel, and plan on using THK components. I will be machining my Router's parts on a Haas mill with 40" of X travel, so accuracy of the parts should not be a problem.

The question I have is how do you get the Gantry (Y axis) to run a perfect 90 degrees to the table (X axis). I hear how some machines claim to have accuracy within .001 or .002, but they might be machining in a parallelagram? Even a mild twist in the Gantry from end to end would compound the error at the cutter. I can build the machine to make adjustments for this, but how do you measure to this .002 or so accuracy. If this has been explained somewhere else, please lead me to that thread.

Regards,
John

[wizard]

A lot depends on the design of the machine. There is a big difference in ease of adjustment on a fix gantry machine VS a moving gantry machine in my estimation.

There are likely to be a number of ways to square the machine but back when I ddi machine tools professionally we used a high precision Starret square and a dial indicator that could resolve well below 1000 of an inch. Install the square parallel to one axis of motion and adjust the other to run square to that axis.

Dave

[vger]

If adjustment can be made after the machine is completed, you could cut or edge 2 stacked pieces at the same time in a square pattern, then flip one piece upside down aligning the X edges and one corner. Then check at an adjacent corner for the error.

[mcarvey]

You could always cut out a square and measure the diagonals. If they are equal then it will be square. If you're like me and don't have 24" calipers, you could always take a bar a aluminum 24" or so long, put a pin in one of the ends and put a lip on the other end to effectively lengthen your calipers. Since only the relative measurements of the diagonals matter, you could compare the diagonals of a 20" square using 6" calipers.

-Matt

[JerryFlyGuy]

I went through this issue a few weeks ago, my machine is a bit bigger than yours and what I did was added elevation bolts at either end of the gantry, but cranking these I can lift either end as well as tilt the gantry forward or back. Then using the rail's I'd installed I leveled them two ways. Next the Z axis is made using THK type rails. I installed gib plates of sorts on each carriage, this allows me to again, tilt the Z axis two ways and gives me fine adjustment for the final alignment. These gib plates will have a small gap between themselves and the Z axis frame which will be filled w/ potting epoxy.

To get your gantry sqr w/ the length of the machine, use a small v-bit in your spindle and mill [ I've a wood spoil board] traverse [along the x ] your mill while cutting a slight groove in the table top, you can then use a large compass to create arc's which center on the groove. Where the arc's intersect is 90Deg to the milled groove. You have to tram your mill to this line. Use a large machinist straight edge and a dial indicator to get it sqr all the way along. Then you can test some parts, measure w/ a caliper and tweak as needed.


If you find a better way, let me know. I' haven't done the sqring thing yet, just the leveling part

Jerry

[Motivator-1]

Thanks Guys,
You have come up with some good ideas. I figured anyone making a machine would have to deal with this. After hearing from you guys, I thought of maybe having a shop with a CNC mill with a relatively long X & Y travel machine 2 edges square on a plate, and then indicate the y out to that distance. That would make my machine at least as good as the CNC machine that it came from. Any more thoughts?
John

[Eson]

Why not just take a face mill and cut the table with the machine itself? The only problem is that the Z-axle must be perpendicual to the gantry.

[JerryFlyGuy]

Eson, I don't think your understanding the question. When you have a long Y axis [gantry], how do you know/ quantify the perpendicularity of the Y to the X axis?

It's a fairly in depth problem.

John, your sqr plate would work for sure, it might be a bit pricy to do however, depending on who you know in the industry, it might be cheaper.. all depending..


Jerry

[One of Many]

Using the router, you can also cut 2 slots in the table 4' between centers with the X axis(same as a slot as Jerry posted) off to one side of the table surface.

Using the slots with keys dropped into them, and butt up an edge of 4' square material against the 2 keys. Now you should be able to cut along the Y axis edge, flip the square over maintaining the original edge against the keys and then cut the opposite Y axis edge. Measuring across the 2 Y cut edges should reveal an out of parallel or bow if the X axis reference edge is repeatable.

It should become evident which way the gantry must be adjusted to make the Y axis parallel in this test. Although this does not take into account any twist in the Y axis rails. So confirm the Y axis does retain Z tram throughout its travel first.

The 1/32 diagonal of a square spec on one of our smaller routers(dual servo rack and pinion) describes their process to adjust the homing switches since it homes each servo independently. You can watch it rack the whole assembly during homing about 3 times until it agrees each side is synced up. What a goofy POS!

+/- .002 repeatability, my A$$! We are lucky if we can keep the parts within .015. Even that changes drastically in different areas of the table surface.

DC

[JerryFlyGuy]

DC, do you think I could convince you to do a sketch of your method for me? I'm missing something here. The way I read your description you'd end up w/ two parrallel grooves long the Y axis but they may not nesc be sqr to the X, just parralel between themselves? I'm sure thats not what you ment so I just wanted to get my head around this. I've still got to sqr my mill so any help I get here will be much appreciated!

Thnx
Jerry

[Motivator-1]

Hi Guys,
From what I have been reading here, it would seem that some sort of strict procedure should be determined to complete this square-up process. Perhaps something like the following:
1) Completely level the 4 corners of the machine.
2) Determine whether your Gantry is free of twist (However this can be done, possibly a precision machinists level?).
3) Adjust either the Table or Gantry Y axis for constant depth of cut (Assuming you already have a constant X depth).
4) Follow the "yet to be determined method" to square the Y axis.
5) Once squared, bore 2 holes for dowel pins into the table at each end of the Y travel so you can re-verify the squareness with an indicator in the future.

I designed my Router to have the THK rails facing up on top so that I could have slots for tweaking and shims for twist correction of the moving Gantry.

If there are any other Professional Router builders out there who have a viable method, and would not mind sharing it, please post here.
Regards,
John

[One of Many]

Here is a pic to show some detail.

Cut the slot or key slots in the X axis spoil board first. Place a good left edge (perfectly straight) of the test piece against the keys. Then cut the lower edge with the Y axis. The sheet must be then be flipped over to cut the second edge of the test piece, with the same original edge butted against the keys. The first Y axis cut will now be at the top of the test piece since it was flipped over. Taking the second cut on the lower uncut edge with the Y axis on the test piece.

Measuring the parallelism between the 2 Y axis cuts will show up in the test piece. The perpendicularity and straightness of the 2 Y axis cuts are amplified from flipping the part over in reference to the X axis keys.

Flipping the sheet over and reseating it against the keys must be done in a manner that keeps the material against the keys. Any glitch that might shift the test piece(during fastening or cutting) into or away from the keys can throw off the test results. So the keys must be tight and be aware of any chips between the reference edge and the keys.

DC

[JerryFlyGuy]

DC your one smart fella! Thats a great way to check the sqr-ness of a setup! The thing I like about it is that its measurable. My [plagerized] idea is a TLAR way of doing it [that looks about right]. There isn't really a way to measure the error. Being able to measure is very important to quantify how to correct the error.

Thanks Chief!

Jerry

[One of Many]

Forgot to mention one thing. Remember in this procedure, is that the compensation adjustment will be 1/2 the error!

Continue recutting and flipping and again recutting the test piece after each compensation of the Y axis alignment until it does come out parallel.

DC

[JerryFlyGuy]

Yup that makes sense..

I can see I'm gonna have to buy a 60inch++ caliper..

The wife's gonna love THIS... (nuts)

Jerry

[Motivator-1]

This thread is developing nicely in my opinion.
Thanks, DC for your contribution here. Please reference my last post in order to understand the whole in-depth process I'm speaking of. Assuming your machine is bolted down firmly, you should not see a variation from the adjusted Y axis, once you have it set unless something moves unexpectantly. You will need to check this squareness from time to time. Your method, DC, (In principal) would certainly keep that alignment in check very well.
My point is that we need to make minute adjustments in order to narrow down the error, and then try to machine the squares again and check for the squareness again, and then again, and again. Do we use the old "Loosen the bolts & hammer the gantry assembly over", tighten, and check again. We are talking about fine adjustments to be made here, so we can check it again and achieve the .001+- tolerance we all want. Clearly there is more to consider in order to achieve these tolerances.

As I mentioned in my above post, I have some adjustment available in my design, but it would also be crude at best (Perhaps, except for the shimming ability for twist). Some micro adjustability would be useful (Like two opposing set-screws), and I stress the importance of occasional checks if your are trying to hold tolerances.

I think this thread may be a good oppurtunity to bring minds together to develope a set of rules and procedures that we can both design machines around, and also address this issue that all of those building machines need to deal with eventually when they finally get to this point in their project, and for those with existing machines that are wondering about their tolerances. (Does anybody really know how accurate their machine really is?)

When I got to this point in my Gantry design, I began to think about these issues seriously. Let's keep the thought process moving on this one.
Regards,
John

[One of Many]

This thread is developing nicely in my opinion.
Thanks, DC for your contribution here. Please reference my last post in order to understand the whole in-depth process I'm speaking of.

I will try and write up my experiences in response to that post.

Assuming your machine is bolted down firmly, you should not see a variation from the adjusted Y axis, once you have it set unless something moves unexpectantly. You will need to check this squareness from time to time. Your method, DC, (In principal) would certainly keep that alignment in check very well.

Your are right, you shouldn't see variations, but most likely you will. Unfortunately, I suspect the design should have some fail safe measures that will inevitably happen in a crash. Moreso for large gantry routers capable of high traverse speeds, moving large amounts of mass and high torque servos to achieve it. The least of which is realigning rather than replacing parts or worse being forced into living with it tweaked due to major damage. Making it rigid enough for cutting loads might be under estimating, but the weakest link shouldn't be the most expensive or hardest part to replace.

My point is that we need to make minute adjustments in order to narrow down the error, and then try to machine the squares again and check for the squareness again, and then again, and again. Do we use the old "Loosen the bolts & hammer the gantry assembly over", tighten, and check again. We are talking about fine adjustments to be made here, so we can check it again and achieve the .001+- tolerance we all want. Clearly there is more to consider in order to achieve these tolerances.

As I mentioned in my above post, I have some adjustment available in my design, but it would also be crude at best (Perhaps, except for the shimming ability for twist). Some micro adjustability would be useful (Like two opposing set-screws), and I stress the importance of occasional checks if your are trying to hold tolerances.

Again, as you can imagine, this is very design dependant. Planning ahead helps. As does avoiding designs that limit post assembly adjustment, or just plain make it a hassle. I have no doubt that micro adjusters on a moving gantry could be made to work and assessed per the configuration. These micro adjuster might needed backed off once all is tuned in, so as not to lock things up too rigid.


DC

[One of Many]

Hi Guys,
From what I have been reading here, it would seem that some sort of strict procedure should be determined to complete this square-up process. Perhaps something like the following:

From my experience, you can align all the sub-assemblies you have to the nth degree, then still find that bringing them all together, there is no way to really escape fine tuning for dynamic running alignment verses static alignment.

1) Completely level the 4 corners of the machine.

As a first step the base and long rails to which everything is aligned to influences the rest of the machine. I see this in terms of 4 conjoined right angle planes that I must bring into one common plane. In other words, each rail end to end must be straight and level, then from rail to rail must be level over their entire length. A good machinist level of .0005/ft works well, but only after the machine is completely assembled. Minor gantry motion can shift weight, making you chase your tail trying to get things level. Starting with a .005/ft level will put the machine in the ball park quicker.

2) Determine whether your Gantry is free of twist (However this can be done, possibly a precision machinists level?).

The gantry Y axis rails should be first aligned in its free state. Again, a precision level could be used, but if the gantry were laid over on its back, the physical loads of its own weight could be quite different standing. My guess is that once the gantry was on the machine the Y axis rails should be indicated and set parallel in Z to X axis rails, directly under the Y axis rails.

3) Adjust either the Table or Gantry Y axis for constant depth of cut (Assuming you already have a constant X depth).

Since the distance from the center line between the Y axis rails to the bed is known. A bar of near the same length could be hung straight out from the Y axis rails(running on the Y axis carriage) to indicate on top another bar laid across the X axis rails. Assuming the bar across the X rails were straight, the indicator should show any twist in the Y axis rails. For now, this does not prove the Y axis rails are going to hold the Z axis perpendicular to the table, but that can be proven after a skim cut is done on the table. As long as the spindle itself is trammed to the XY plane, it will cut that consistent to the level of the X, Y rails. After a true surface is obtained on the table, the Y axis can now be set to bring the Z axis perpendicular to the bed and the spindle reset to tram accordingly. Only after the X, Y and Z are trimmed in, could one confidently set the XY square.

4) Follow the "yet to be determined method" to square the Y axis.

I've already addressed one method for this in a previous post in this thread. The other might be the circle, diamond, square to see what it cuts like. It just makes it hard to address all of the conditions contributing to mis-alignment mixed together.

5) Once squared, bore 2 holes for dowel pins into the table at each end of the Y travel so you can re-verify the squareness with an indicator in the future.

Sure, if you can do this in an area that does not get abused. I don't think particle board is stable enough. The nearest metal horizontal plate at the table ends might be best?

DC

[Motivator-1]

Thanks again DC for some very useful advice.
I am going to look further into my machine design to see if I can incorporate better fine tuning functions for this 3 dimensional alignment issue. I'd like to hear from others who have already been there, or have some other clever ideas.
John

[simso]

Ive finished building my machine last month and its really not that hard of a thing to do. Make sure that whatever your building your item on is nice and flat, then drill both items at the same time, eg stack them on top of each other and drill all the way through. When It comes to squaring your base you cannot beat ba tape measure across the angles, Drill screw barce as required. Same goes for assembly of the gantry ect. When the item is completely finished put a dial indicator on the the bottom of your z axis and drive it to the four corners, this will tell you exactly how aligned everything is. Drill and tap holes in your base plate and insert grub screws so that way you can wind your base bed up or down to compensate for the variations in the machines travel. Run your unit again with dial indicator and wolla perfectly squared item ready for some cutting
Steve