Fourth Axis Machining feed rate

I am trying to machine a part that looks like a cup with varying heights of the side wall around the cup. the part is held on my fourth axis and the program basically varies x as the part rotates on the A axis. The problem I am having is getting a consistent feedrate in the cut. For G1 commands where X is constant and only A is changing, I need a feedrate of F130 because the feedrate is needed in degrees per min, but for G1 commands where X is changing, linuscnc wants the feedrate in inches per minute. I can get most of these changes done, but I run into problems when both X and A are changing. It appears that when X changes, the ipm rate takes over, but if the change is a small X increment over a large A degree change, the A axis feedrate goes way up and the overall linear rate of the tool movement on the part is high and I am snapping tools. I need to be able to get a constant surface feed rate whether the G1 command is changing X or A. Remember this is on a mill's fourth axis. Any suggestions?

I have computed the delta-X (the axis distance moved) and the delta-S (the circumferential distance moved) for each G1 command and I think I can use this information to compute a different feedrate for each G1 command, but I am not sure exactly where I am going with this. Has anyone ever run into this problem?

Sorry if this seems trivial given your pretty good understanding of machining, but have you looked at your ini file parameters to verify your linear and angular settings? Did you use Stepconf or Pncconf to create your mill profile without making any additional edits?

I believe that my linear and rotary setting are good. I can easily machine any 3D part and specify the feed rate in inches per minute. I have also machined circumferential slots in a tube and for those cases I specified the feedrate in degrees per minute.

My current problem arises when I want to move both in the X and the A directions, such as for when you would machine a spiral around the circumference of a rod. Actually I think I could figure out how to machine a spiral since it would essentially have a constant lead angle and I could simply determine the feed rate by running the program without a tool and then just eyeball the feed rate.

Unfortunately my problem is more complex since my profile is not a simple spiral. I have attached a jgp of the part. The blank has been turned on the lathe and currently looks like a simple cup. Now I want to rotate and mill it on the fourth axis in order to create the shape of the two flanges on the circumference. My problem arises because the "lead" ( the change in x relative to the change in A) varies and thus even a slow constant feedrate results in rapid accelerations in the tool movement relative to the surface of the part. This appears to happen when the cut is moving at f2 in the X direction, but the profile requires the part to do a rapid 30 degree rotation for just a little movement in the X direction. The resultant movement is a high surface velocity and that snaps the tool.

I am now looking at recomputing my profile using a constant 2 degree rotation for each G1 command and computing the associated X value for each 2 degree rotational step. Maybe this will stop the accelerations of the tool.

I am open to any other suggestions. Is it too bad I can not program in surface feet per minute like on a lathe. That would take into account the movement in both the X adn A axis.

Are you using a Cad program with post-processing capabilities? Sounds like you are doing the computation by hand, but the drawing looks like CAD.

Check into G93 (inverse time mode) in :
linuxcnc.org/docs/html/gcode/g-code.html#gcode:g93-g94-g95

Jon

The part was originally modeled using ProE and that CAM file was created using BOB-Cad. I did take the CAM file and hand edited it to included machine specific commands, but the actual G1 commands comprising the profile were generated by BOB-Cad. There is nothing wrong with the profile, to only problem is that I do not seem able to control feed rate when the mill is moving in both the X and A directions.

I did write a small computer program which took the original profile and created a new profile with constant steps of 2.0 degrees for each G1 command. I have not tried out the new CAM file yet, but if it does not provide smooth operation I may try it with the suggested G93 inverse time mode.

Thanks for the help.