Eliminating/auto correcting racking on dual screw axes

[digits]

Hi guys, I am part way through building a new machine of my own design, and now that it's pretty much assembled, it seems that I have some racking issues to deal with. I could have asked this in my build thread, but I thought it might apply to other similar designs, and so any replies might help others as well.

Here's the machine:


Both the Z and Y axes are dual screw, and I am bit stumped as to how to get them both exactly square and then keep them that way. From my initial tests, the servos are doing their jobs perfectly, and what every rack it has when I power it up stays constant throught the moves, but I need a way to quickly and accurately re-square post power-down/E-stop.

From what I understand, it doesn't matter how unsquare the machcine is to the rest of the planet, as long as the spindle is square to the table and X, Y and Z are all perpendicular, it should cut true.

I can level the Z-axis by jogging to X-min and taking a spindle height measurement with my height-gauge, and then jogging to X-max and taking another height measurement. I then just need to move the Z-screws one by one to achieve perfect leveling.

Presumably, I could put an indicator in the spindle and then indicate along the T-slots on the table to get the Y-axis true, also adjusting a screw at a time until it's straight.

I could do all this adjustment by hand, but I have seen a slight slip in the Z-axes during E-stops, so I'd have to redo it everytime something went wrong. I also don't feel too happy about doing setups on the machine with the table powered up as a runaway servo with >1HP behind it could cause me a lot of damage, but I can see that torquing down vices and clamps could cause Y-axis racking.

So, I am after an auto-calibration routine that's not going to cost a fortune, and preferably can be done once a setup is on the table. I was thinking of using some sort of electronic dial indicator/DTI to measure the racking errors, and feed them back to the PC, which would then work out the required correction, and adjust a single screw and remeasure to confirm the adjustment. Does this sound sensible or better still affordable? I know I will need some electronics to dynamically gang and ungang my step and direction signals on my dual-screw axes, and obviously I will need some sort of electronic indicator - does anyone know if/how they output their readings to a DRO?

Any tips or suggestions for alternative solutions would be most welcome.

Cheers.

[Greolt]

Why don't you just use a quality home switch on each side of the dual drive axis.

If using Mach software it will resquare the axis with a homing sequence.

Greg

[digits]

Originally Posted by Greolt Why don't you just use a quality home switch on each side of the dual drive axis. If using Mach software it will resquare the axis with a homing sequence. Greg

I did think of that, but how would I ensure that the home switches are exactly parallel? I could indicate the machine square, and then ease the home switches into position until they activate, but I wouldn't be able to tell if one switch was more depressed than the other.

I suppose I could then try iteratively homing, re-indicating and adjusting the home switch position until its correct, but that sounds really tedious - hence my desire to automate it!

It is a good idea though

[Greolt]

Originally Posted by digits ..................... but that sounds really tedious - hence my desire to automate it!......................

But you only have to do it once then whenever you start your machine or after having an estop event just do a homing sequence and WALA!!!!

[digits]

Originally Posted by Greolt But you only have to do it once then whenever you start your machine or after having an estop event just do a homing sequence and WALA!!!!

True - do you have any links to sensibly priced precision home switches? I guess I need about 0.05mm/0.002" accuracy and repeatability, and I probably need switches that the axis can roll right over as my homes would need to be infront of my limits.

Cheers.

[Greolt]

There are many types of limit switches. Photo Interupters are acurate but not good in dusty environments.

Some use capacitive proximity switches. I don't have experience with them.

These type I have used with good results. Ebay Number 180141531018

The US ebay has many types on offer.
<table cellpadding="0" cellspacing="0" width="100%"><tbody><tr bgcolor="#ccccff"></tr></tbody></table>

[mhasting2004]

Here's how I would square up our flat bed imagesetters (previous job) which I am modifing as your not imaging on a clear medium.

First off I assume your machine is stiff and does not rack "dynamically".

Get some thin material (formica, printing plates..etc) and lightly stick two together face to face (ie some spray glue or thin double sided tape.

Hold this securely down to you bed and cut a big square (bigger the better).
Mark the long axis (y?)
Seperate the two pieces and flip one over so as to have the long axis still alligned to the other (face to back)

get a good lupe (magnifier)and compare the error between the two x axis edges. If you are out of square you will have cut a parallelagram by flipping the pieces you will be double your error. calculate the ratio of you square size to your gantry width and then skew the gantry the required direction and amount. Remeber this will be half of the error you measured times the ratio.

Repeat.

If your good one shot shoud make you square if not you may need to try again.

Clear as mud?

For your Z axis use the table as your zero on both sides

Cheers

Mark

[digits]

Thanks for the reply mark - I don't really want to have to cut anything if I get the Y-axis off square, but I do intend to do some trial cuts and resquare if necessary once I think it's square in the first place.

[digits]

Originally Posted by Greolt There are many types of limit switches. Photo Interupters are acurate but not good in dusty environments. Some use capacitive proximity switches. I don't have experience with them. These type I have used with good results. Ebay Number 180141531018 The US ebay has many types on offer. <table cellpadding="0" cellspacing="0" width="100%"><tbody><tr bgcolor="#ccccff"></tr></tbody></table>

Cheers - I can get switches very similar to that from RS in the UK - quicker but more expensive.

Could you please explain what the switching characteristics table is all about then -

OFmax.
350g

RFmin.
114g

PTmax.
0.4mm

OTmin.
3.58mm

MDmax.
0.05mm

FPmax.
--

OP
33.4±1.2mm

I get OFmax and RFmax as operating and release force, but what are the other distances - specifically which is the repeatabilty/hysteresis?

Cheers.

[mhasting2004]

Fair enough. I was assuming it was pretty close and you were just getting the last bits of skew out.

Depending on how your gantry is attatched to your bearing blocks you could, once you new the error, just lossend the bolts and advance one side or the other using the steppers/ servo independently then snug up the bolts again and recheck. Probably easier with a dial indicator though.

Cheers

Now I must hurry up and get mine finished!

[digits]

Well, I was inside the machine for most of yesterday - looks like one of my Z-columns wasn't even bolted down Still, it does show just how nicely balanced it all is I will re-evaluate everything once I think it is actually all square - there's way, way less movement with all the bolts in place

I am still interested in automated measurement/calibration though, but I think I will hold off buying an expensive digital-dial indicator until I can get my digital calipers to talk to my PC.

[digits]

Well, even with it all bolted together properly, my machine will run happily with 0.8mm/0.03" difference in Z across the X-axis travel I don't think there's any dynamic skew though, as it seems to return to Z=0 with whatever skew it started with after power up.

I do like the simplicity of independent home switches on each screw, but I have been looking and I can't find any precision switches which are rugged enough or have enough overtravel. I think I am looking for repeatibility of about 0.05mm/0.002" which is probably asking for too much.

I am considering fitting digital scales to each screw assembly, so that I could either read off the home errors, or feed them directly into Mach 3, but that would add about £50-100/$100-200 per screw! Anyone else got any good ideas?

Cheers.