[Fenichel]

The attachment is the complete gCode generated by PathPilot 2.4.3 and then run on an unmodified 770. It is meant to use a 0.240" thread mill (Tormach #34695) to cut M8/1.25 internal threads into two previously-drilled holes, the first at (-0.5", 0) and the second at (0.5", 0). The thread mill was held in an E32 collet.

The code has been lightly annotated with line numbers, but is otherwise unchanged. When it was run, the last pass cutting the first thread was complete, and the next steps were to

• withdraw horizontally from the cut toward the center of the hole (Line 317)
• withdraw vertically to get out of the hole (Line 318)
• move horizontally to the location of the second hole (Line 321)

Here are some selected lines from the code:

(----- Start of G-code -----)
(<cv1>)

G17 G90  (XY Plane, Absolute Distance Mode)
G64 P 0.000 (Path Blending)        ;  line 34
G21 (units in mm)
G54 (Set Work Offset)

G30 (Go to preset G30 location)
T244 M6 G43 H244

F 115.0 (Feed, mm/minute)
S 900 (RPM)
M8 (Coolant ON)
M3 (Spindle ON, Forward)

(*** Thread 1 ***)
G0 X -12.700 Y 0.000
G0 Z 0.200                        ; line 48
  . . . .
G0 X -12.596 Y 0.319           ; line 309

(Pass 30)
G0 Z 0.000                     ; line 312
G0 X -12.365 Y 0.000
G1 X -11.748
F 27.4 (Arc Feed, mm/minute)
G2 X -12.406 Y 0.905 Z -21.000 I -0.952 J 0.000 P 17
G0 X -12.596 Y 0.319     ; line 317
G0 Z 0.200                     ; line 318

(*** Thread 2 ***)
G0 X 12.700 Y 0.000         ; line 321
G0 Z 0.200

(Pass 1)
G0 Z 0.256
G0 X 13.035 Y 0.000
G1 X 13.208

When this was run, the horizontal motion of Line #321 was begun before the vertical motion of Line 318 was complete. The thread mill crashed into the side of the hole and was destroyed, and the workpiece was pulled partially out of the vise in the direction of the second hole. What had happened?

I considered the possibility that the tool had effectively lengthened during the course of threading hole #1, so that the vertical retreat was not adequate for its greater length. The collision with the inside of the hole was about 6mm below the top of the hole, so the hypothetical change in length was more than any plausible thermal expansion. The collet had been tightened with as much force as I can generate with 10" wrenches (probably 60 ft-lbs or so), and later it took no less force to open it to retrieve the stub of the thread mill. Also, the thread (except where the tool collided with it) was perfect. I don't think that the tool moved with respect to the collet. Also, the collet was well seated on the spindle before and after.

Could this fiasco have been an adverse effect of the G64 mode set at Line 34? It's all very well for it to smooth the motions of the cutter as it traces out a thread, but might that G64 have wrongly "smoothed" the vertical-to-horizontal transition between Lines 318 and 321?

All suggestions welcome.

[Fenichel]

The problem was a bonehead specification error by me. I needed to use G21 mode in order to get PP's conversational options for metric threads. I dutifully located the hole locations and the thread depth in millimeters, but I forgot to convert the Z clearance from imperial to metric.

I discovered this while talking to one of the Tormach machinists. I think I may have convinced them change the conversational thread-milling options so that regardless of the G20/G21 mode one is in, both imperial and metric threads are offered.