[Grant Nicholson]

I recently cut a 45" circle on my LC4896. I then cut a negative circle.
When I put the positive in the negative, I found that the circles only fit one way, because they aren't true circles.

Since then I have discovered about .033 difference in distance between the x axis and the y axis when cutting a distance of 47.5".

I don't know for sure how to determine which axis is out. By eye and a ruler, it seems like the Y is longer.

I have the formula to account for this in the M scale in the interface.

I cut a negative and a positive in the two different directions and measured the difference when the positive was put into the negative with calipers to arrive at this measurement. Is there a better way?

[cabnet636]

send a pm to pflug or ed lang, they will not see your post unless they check this forum as it is a new thread

jim

[Grant Nicholson]

I looked in the members list, couldn't find pflug - is there something I'm missing?

[Montabelli]

check your axies for play, I have had it from both the thrust bearing popping loose from its collar, and the nut at the thrust bearing coming a little loose. be careful not to over tighten if it seems loose. I got to buy a new lead screw once because of that. if there is no play in the axis, then you will need to use the backlash compensation in the controller software.

If you make a square toolpath and run it on a piece of scrap big enough to make it easy to measure , then check for accuracy on each axis and then corner to corner for square you should be able top determine where any changes need to be made.

[Pplug]

I agree with montebelli, those would be the things to do! Oh, I forgot to tell you to check and make sure the numbers in the interface match the numbers on the label attached to the leg or axis.

check your axies for play, I have had it from both the thrust bearing popping loose from its collar, and the nut at the thrust bearing coming a little loose. be careful not to over tighten if it seems loose. I got to buy a new lead screw once because of that. if there is no play in the axis, then you will need to use the backlash compensation in the controller software.

If you make a square toolpath and run it on a piece of scrap big enough to make it easy to measure , then check for accuracy on each axis and then corner to corner for square you should be able top determine where any changes need to be made.

[RandK]

If I'm reading you right .033/47" that's around one half a thousandths off which is pretty good for a router over that distance. There is a reindicate document on the Techno support site.

Start by sweeping a dial indicator in the spindle collet to table. If that's off everything else is. The aluminum tables are not machined flat so you'll just have to average things. If you have the beam and squares you can check the other axiis. Backlash can be measured with the indicator against a vise or something clamped to the table. The compensation in the contoller doesn't work in jog, only when a program is run. You can measure by jogging but you'll have to write a small program to G01 then pause to check.

I think I'd square the spindle and work with pocketing 2" / 4" squares and measure with calipers. Get those right and call it a day, it's not a vmc.

[joejared]

This doesn't sound like a calibration issue, but rather bad thrust bearings, as mentioned earlier, which are producing an enormous amount of backlash. If the backlash is non-existant, it's also possible that one of the carriage bearings is damaged, causing it to bind up a little in one direction. I've experienced both problems, but the thrust bearings are the most likely cause.

I get angular contact bearings from vxb and if you have a bearing set that is based on spherical bearings, have designed a replacement bearing housing to make all bearings standard. For compatibility with the spherical bearing design, you'll need a spacer on the shaft approximately 10mm long to maintain position on the shaft, but in a nutshell, angular contact bearings are the way to go. The more modern machines are easily upgradable because all bearings are press fits into the motor side housing, eliminating the need for custom bearing housings.

Important details about replacing the thrust bearings. First, you will want a magnet near the nuts of the home switch mounting screws, or spare 2mm nuts for when you lose them. Before disconnecting the coupling, belt or otherwise, I also recommend that you mark the pulleys and belt (or coupling)with a sharpie, so that when you reinstall the servo, you're returning it to its original condition. Techno's software doesn't report the difference between the zero reference of the servo and home switch, but for repeatability, it should be close to 2.5mm, or approximately 0.1", or 1/2 revolution of the motor. When replacing the bearings, the label or narrower outer race of the bearing should be on the inside, as putting them the other way around will cause the inner race to collapse out of the bearing because of their design.