[Garban]

We recently purchased a used Toyoda FA550 II with a 15M control. The machine has been working great except for profile milling accuracy. It's rounding off corners and making sloppy blends into arcs at all feed rates. It does improve slightly at lower feed rates under 40.0 IPM. The only thing that helps is using a G9 at the end of every finish pass line of code which causes a noticeable dwell mark and high cycle time. G62 doesn't help and G61 does the same as G9 but modal. Hole position accuracy is excellent (positional 0.003 over 11 inches). The company we purchased it from never did precision profile milling, only face milling and precision holes. Is there a parameter that controls X/Y axis accuracy during milling only? Any suggestions would be greatly appreciated! Thanks in advance.

[txcncman]

What does your actual code look like? Outputting line segments for arcs? 20 points to cover 90 degrees of a 1" arc?

[Garban]

I use standard G2/G3 with an arc lead in:

X0Y16.6861Z15.0G43H12S10000M3M8
Z0.1
G1Z-0.375F70.0
G3X0.125Y16.6861R0.0625
I-0.125J0
X0Y16.6861R0.0625
G0Z3.0

You can see a "cusp" between the lead in / lead out arcs on the part about 0.020 tall. If I add another "I-0.125J0" after the first it cleans the cusp out but increases the cycle time.
Thanks for your any help.

[txcncman]

I suggest an overlap of your start and end points of your contour. I have never had a machine that did not leave a cusp such as this when only programming 360 degrees of sweep. Some were less, but they all did it.

[dcoupar]

Just curious... you say "if I add another I-0.125J0 after the first it cleans the cusp but increases the cycle time". How long does it take to interpolate a .125 radius circle at 70 IPM? I'm kind of surprised it doesn't mill a square.

[psychomill]

Have you tried using G8 or G5 instead?

Also, having the servos tuned will go a long way to fixing your issues. A .020 tall cusp for that program is a mile off IMO...

[angelw]

I use standard G2/G3 with an arc lead in:

X0Y16.6861Z15.0G43H12S10000M3M8
Z0.1
G1Z-0.375F70.0
G3X0.125Y16.6861R0.0625
I-0.125J0
X0Y16.6861R0.0625
G0Z3.0

You can see a "cusp" between the lead in / lead out arcs on the part about 0.020 tall. If I add another "I-0.125J0" after the first it cleans the cusp out but increases the cycle time.
Thanks for your any help.

Hi Garban,

I'm not in favor of using "R" format in circular interpolation under any circumstance, I just don't see the point. Not so in your example, but if either the start or end point is incorrect relative to the the correct shape tool path, the control merely shifts the arc centre to allow the tool path trajectory to pass through the two given points with no alarm being raised. Accordingly, an erroneous tool path can result without it being obvious until the part is finally inspected. When using I, J and K format, an alarm would result if the end point is not on the trajectory specified by the start point and the I, J and K arguments.

Following is an extract from a Fanuc Manual, where they themselves qualify "R" format as being possibly inaccurate in some ircumstances.

"When an arc having a center angle approaching 180° is specified, the
calculated center coordinates may contain an error. In such a case, specify the center of the arc with I, J, and K."

In your example "R" format is being used with a 180° centre angle.

Given that you have the possibility of an inaccurate tool path even if you do everything correctly, the possibility of an erroneous tool path without it being obvious if either the start or end points are incorrectly stated, and that you can't program a complete circle with just one block, I fail to see the relevance of "R" format in most Milling applications.

I agree with psychomill that 0.020 is extreme, and I don't believe that the "R" scenario stated above would cause such an error, but I would eliminate every possible cause of errors. Although the tangent point of two circles is the same whether the the relationship of their diameters is close or vastly different, better results are achieved if the approach and exit path is not abrupt as shown in the attached picture on the right.

Regards,

Bill