[FelixPQ]

Though my CNC is not a DIY machine, the gantry (24" in Y x 32" in X) is so much out of square ~0.085" per foot that I'm not sure forcing it square using one homing switch on X and one on it's slave A and Mach3 auto squaring procedure would work, I'm afraid of loosing steps. Any advice on this process of squaring the gantry to the X axis in my case will be appreciated.

Currently, I have a 1030 extrusion connecting the 2 X axis together under the table and I want to be sure it's the right thing to do before I remove it.

Thanks in advance,

Yves

[awerby]

Try the squaring routine and check corner-to-corner on a sample square. If that didn't work, you'll need to make a mechanical adjustment to get it closer to right, and try again.

[FelixPQ]

Andrew,

could you explain or point me to the squaring routine?

[FelixPQ]

I knew I saw this before, Gerald on the MechMate forum posted this: "It is not the motors' duty to get the gantry square.". I would be incline to go with this statement from a mechanical perspective. I also saw in his post a mention about the tendency for stepper motor to settle on a full step at power on.

I can't help myself thinking that the gantry should be pretty close to being square mechanically and I have no idea of how close it should be before forcing it square using the motors. Any idea on this?

Thanks,
Yves

[ger21]

I can't help myself thinking that the gantry should be pretty close to being square mechanically and I have no idea of how close it should be before forcing it square using the motors. Any idea on this?

It should be as close as you can possibly get it.

[FelixPQ]

Gerry,

IMHO this is the best practice as it add minimal extra loads on the system and motors.

Yesterday, I check how much out of square my gantry was and it was very bad.

Attachment 311420

My machine doesn't have the fourth axis but the gantry is exactly the same.



Above, that's mine and I removed the exterior plate so you can see how it's assemble. As I said earlier, currently there is a 1030 extrusion connecting this plate to its counterpart on the other side (the 2 stainless bolts at the bottom). The 8 black bolts connect the plate to the linear bearings.

I assume that both X rails are parallel but it would be nice to know for sure, any suggestion to check this out?

The only reasons I can think of that would make this gantry out of square by so much are these plates are not flat and or the extrusion connecting both side are not cut square.

Whatever I do to fix the squareness problem in the XY plane, I already know I have the same out of squareness problem in both the XZ plane and YZ plane. I just don't see how I can fix all of these problems on this machine, there is simply no provision to adjust anything in a controlled fashion.

This is why I ask here, I don't know what to do beside giving up, maybe someone will have a great idea. I've had this machine since late 2013 and it's been collecting dust since I realize how useless it was. It reminds me of horror stories like on some cars, in order to change the spark plugs you needed to unbolt the motor and raise it by a couple inch, I don't recall exactly.

I lost my father in 2012 and I had to make his urn on a way to short notice and since then I promised myself next time I'll be ready. I decided to buy a CNC and make funeral urn as a hobby. Last summer I visited a veneer company and they gave me a ton of waste (for them) but a treasure for me. This generosity inspired me to do the same, so every urn I'll make using this veneer I'll give away to the local churches and non profit-organization. Now I have all the material I need to make at least 50 urns and I'm sure I'll still have veneer for 50 more.

I can add I'm willing to spend up to a $1000 to fix this problem. My objective would be to be able to correct the orthogonality errors (squareness error in all 3 axis planes at the same time), to a level of 0.005" per foot or less, all this in a control fashion or in other words, no trial and error method please. Lets say for example, for this gantry out of squareness problem, imagine I've setup a large precision square, say 18" and a dial test indicator on the spindle. I zero it at one end of the square, go to the other end, loosen a couple of bolts and micro adjust until the indicator indicate zero, bolt everything back. Then I run a cut air program to go to say 500 point chosen randomly, I repeat this checking and adjust process a couple of time to make sure the correction is stable. Lastly, I install mechanical stops say 0.010" off on both master and slave axis, this would allow me to verify the squareness of the gantry in less then a minute. I assume also, that adjusting the squareness in the XY plane like this would have no effect or very very little on the 2 other planes.

The main problem with my machine as it is, is you can not unbolt any part and try to make an adjustment in 1 plane without affecting 1 or the 2 other planes as well, a total waste of time.

Thanks,

Yves

[FelixPQ]

I forgot to mention there is another aspect of this machine design I found problematic.

Attachment 311426

the work table on my machine, see above, is mounted in such a way that it can and probably is not flat (that I'm sure) and not parallel with the X rails. One of the first thing I've checked on this machine is the orthogonality of the router axis with the work table. For this I use plate glass directly on the table 1030 extrusion, I immediately notice the table wasn't flat to begin and using a 5 inch diameter dual dial indicator tramming tool, I recorded an error close to 0.050" and yes it's not a typo, you read this correctly close to 50 thousand of an inch. This mean an error of 1.15 degrees out of squareness and on a 3/4" board that you can be out of square by as much as 0.015", try to make something square with this kind of error, impossible.

There is yet another possible problem, I you look on the far right of the image above you'll notice the linear bearing are installed on their side. This as the effect of loading only half the balls of the bearing which will inevitably reduce the life of the bearings by a large factor if my memory doesn't fail me to much.

All this to say, that I'm considering a partial redesign. I say partial only because I want to keep the most important and expensive part like the bearing system and the ballscrew system.

Thanks,
Yves

[ger21]

I would probably take it apart and check all the parts for squareness, and find out where the problem is. Then you can make a plan to fix it. This might involve shimming, or making new parts. It's really hard to say without knowing where the problem lies.

[FelixPQ]

Gerry,

I would be inclined to agree with you in a normal case and with a solid design. With the way this particular machine is designed, I figured the 2 inch wide extrusion making the gantry beam need to be out of square by only 0.0036" to make the entire gantry out of square by 0.1000" or in other words the design amplifies a relatively small error into a huge error. This is unacceptable whichever way you look at it and that's the real problem, the way I see it.

The same kind of problem exist where the side plate connect to the X axis bearings, here again a small angular error is amplified at least four time and what make it even worst, both the angular and say the distance movement (x rail to the bottom Y rail) are directly interlock by the side plate. It's practically impossible correct either one without affecting the other.

IMHO, what is needed here, is a complete rethinking or redesign of these side plate, as long as the design results in an amplifying effect and that one degree of freedom is interlock with another, it is futile to try to fix it.

Do I know how to correct the problem, the short answer is not at the moment. I'm open to suggestions and I'll start immediately looking at possible solutions.

Thanks Gerry,
Yves