[big-div]

Hi all,

I've got the latest PRO 4896 all setup and running now, only thing is i've lost half a head of hair trying to make this thing accurate !

I don't know whether im asking to much from it or not but I am having a hell of time trying to true it up. Going from Inches to MM as well.

Methods I've tried are a dial indicator, attached to a solid section and using the mach3 Cal program. Got close to repeatable, but was still .3mm off at times. Which seems like a lot !

Next I lashed out and purchased a set of gauge blocks to get a better distance up to calibrate it to and a mitutoyo test indicator. Works alright, able to get it within 0.01 with this method. I was calibrating it to 90mm with a couple of blocks together.

The real issue im having is over the space of the entire machine the error of calibration gets multiplied, so calibrating it to 90mm as I have, if i have an error of 0.001mm, over 2420mm i've all of a sudden got an error of 2.420mm, which is about what im getting ! So confusing, I don't know what Im doing wrong...

I've tried cutting a 90x90mm square using a small 2mm deep profile cut on laminate MDF, funny enough I was getting 89.78mmx89.78mm, which tells me my bit may not be exactly 6mm (Is what I set tool compensation to ) as that is what the packet said it was ( Measured 5.88mm on the venier, but compensating for that in CAM would make the square even smaller !).


In summary, am I asking to much ? Is my method of calibration not correct ?? The error I am most worried about is the fact that a 0.001mm error down one end equals a 2.5mm error down the other.

Does imputing a strange value into mach3's steps per cause issues ? eg. 68.3432342134

Using a Huanyang 2.2kw spindle, Moore& Wright dial ind. , Mitutoyo test (dog leg) ind. , measuremax gauge blocks.

Any suggestions, I'm really struggling here !

[big-div]

Attachment 258800 this method has got me pretty close , now only probably 2mm out over the span of the table (2420mm) so if you think about it, that's something like 0.0001 out when calibrating at 50mm

Sent from my GT-I9505 using Tapatalk

[ger21]

I'd calibrate it over the full length of the table, or at least 2000mm. Then see how much error you're seeing on small moves. I've never used r&p, so I don't know how precise or consistent they are. But make sure the pinion is meshing properly, and everything is tight.

[big-div]

Only thing I could use to measure that distance is a tape measure though. Seems pretty agricultural

Sent from my GT-I9505 using Tapatalk

[ger21]

I bet a tape measure get's you closer than 2mm though. And if you cut a 2m part, you check it's accuracy with a tape measure, right?

I've seen people mount a cheap USB microscope so they can read a tape measure very accurately. You can also buy precision measuring tapes, but they're expensive. This is how I intend to set up my big machine, if I ever get around to building it.

[alan_3301]

How are you figuring out total travel error when using the recommended steps per inch setting?
I think you are going to run in circles if you try to calibrate to 0.001mm when the single full step increment is something like 0.1mm.
The steps / inch have already been calculated for you, I would trust that measurement, a tape measure at 8 ft would be good to confirm.

[big-div]

I was sort of hoping the purchase of those gauge blocks and indicator would solve it.. Cost me bloody near on $370 ! Ah well.. Least I have them now. Might give the tape measure another go. You'd think a snap on one would be pretty good ! I'll remember to record my steps per mm value though before hand :P. The value in the XML file to begin with was in Inches (1736) , converted to mm give me 68.346456692913385826771653543307 and I think my calibrated value was 68.39 or something. Sort of looks like it should be in between if my current value is giving my 2mm over at 2420mm travel. I shall try that tomorrow though, don't think my neighbours would appreciate me out there at this hour !

[big-div]

What about a laser distance measurer ? That'd be pretty accurate surely !

[big-div]

Scratch that. +/- 2mm...

[Analias]

I discovered the same problem with calibrating my axis recently. Any measurement under an inch using the dial indicator was multiplied over the full extent of the table. I got much better results zeroing the one side of the table and telling Mach's calibration script to move 48 inches and using a simple tape measure to measure the actual travel.

I finally realized that when I measured under an inch using the dial indicator the error was multiple over the entire length of the table, while the much less accurate tape measurement error was divided by the length of the table. At 48 inches that's nearly two full magnitudes error adjustment working for or against me.

Sent from my Xoom using Tapatalk

[big-div]

Exactly the same as myself mate, so did you find doing it that way, working within smaller areas you maintained accuracy ? I'm trying to cut .5mm (0.0020" ) increments for inlays in Acrylic.

[ger21]

.5mm = .02", not .002"

You should be able to get much better accuracy than .5mm.

[Analias]

Exactly the same as myself mate, so did you find doing it that way, working within smaller areas you maintained accuracy ? I'm trying to cut .5mm (0.0020" ) increments for inlays in Acrylic.

Go back and retest your backlash. I'm not sure if you can expect tolerances within your backlash. Gerry could probably answer that one. But, I did notice that I had to go back and reset my backlash values before I could expect to hold any repeatability in the single digit thousands.

Sent from my SGH-I337 using Tapatalk

[big-div]

Nil backlash, this thing is tight as. Reset the steps per to the calculated value and it's within 1mm of the full table now, and that could be in the tape measure. I flattened out the rivets in the tongue of it to try and calibrate it to a guage block. I'll try a cut now. Another thing on tool bit sizes, do I set them to what the manufacturer says they are or to what my verniers say they are? Like I have 6mm endmill, measures 5.8 on the vernier

Sent from my GT-I9505 using Tapatalk

[big-div]

Would it be normal that i am getting different steps per for each axis, even though they have exactly the same running gear ?

Sent from my GT-I9505 using Tapatalk

[Analias]

Yes, that is normal. You will see differing values for the X & Y axis. In fact you would see different values along the ball screw since there are variances along the travel. I would expect this to be a another good reason to use the full travel of the table.

On my CRP-4848, the steps per unit were close but not the same. If I remember right, the difference was something like 60 steps.


Sent from my SGH-I337 using Tapatalk

[Analias]

Also, I think you need to zero out the back lash values before you test, otherwise the existing values are in effect during testing.

Sent from my SGH-I337 using Tapatalk

[big-div]

Roger, backlash never got changed and is still zero, calibrated both axis again after calibrating my tape measure with the gauge blocks. Move a bit of gear rack around , was out by .3mm compared to the other side. Cleaned the rails and bearings, I am now getting 60.01 roughly on the x axis, when doing a 60x60mm square . that's with a 6.1mm tool size in cam. Only getting 59.8 across the Y span so I may have to calibrate that again.

Its looking better and more repeatable now then it was before. Fingers crossed I can fix up the last little bit !

Sent from my GT-I9505 using Tapatalk

[ger21]

Try your test cuts in different areas, and see if the size changes.

Also, be aware that cutting direction can affect the size of the part. Climb cutting can result in larger parts, and conventional cutting can result in smaller parts, due to deflection of the tool or machine.