G code A,B,C interploation

[llilrex]

I am not sure if this is the right place to post this but here it goes. I am trying to figure out how the interpolation of A,B, and C axis are carried out. for example:

when x,y,z are moved the path is at a contant interpolation with respect to actual incident speed of the tool. but say you want to move X,Y,Z,A. how is this acheaved with respect to the actual incident speed of the tool path? do you treat the A,B,C axis seperatly from the X,Y,Z? by selecting a rotational interpolation? do you define a maximum radians or degrees per minute? are they combined like the X,Y,Z?

[ger21]

Probably depends on the control that you are using.

[Geof]

On Haas machines the XYZA motions are all combined to get the correct resultant feedrate. The angular A motion is converted to an equivalent peripheral linear motion using a diameter that is entered into the Parameters or specified in the program.

Or at least this is the way I interpret the explanation in the Haas manual.

[ger21]

Mach3 basically works the same way. You tell it the diameter of the part and it uses that info to calculate the required feedrate.

[HuFlungDung]

I think Geof has it right.

Whenever simultaneous axis motions are commanded, they must all begin and end in unison. So the controller calculates a proportional speed for each axis in order to make this happen.

If a linear and rotary XA simultaneous movement is commanded, the linear feedrate is 'boss' because the linear movement has an easily defined duration, and the rotary axis rate has to be proportioned so it is done when X is done.

When you have a solitary rotary movement commanded, but a linear feedrate is in effect, this is when the controller must know what the circumference of the part is supposed to be in order to convert inch/min to degrees/min. So it is theoretically possible on the Haas to have XA moves execute properly at a linear feedrate, but single axis A movements could be too fast/slow if the diameter input is way off.

Other types of controllers require degree/minute feedrates for all rotary movements, so then the rotary feedrate becomes 'boss' when A is active, but then you need to command linear feedrates whenever there are only linear axis involved. Personally, I like the way that Haas handles this, but it is understandably handling only comparatively simple motion situations. If you throw in a B and/or C axis, then you've probably got little choice but to use inverse time feedrates for all movements. That is where the duration of every movement is calculated and then inverted to give an inverse time feedrate, which is called by a particular gcode (I think?....I dunno, it has been a while since I studied inverse time).

[llilrex]

so, let me see if I can throw this out there without looking like a total moron.

if linear interpolation can be defined as:

Is = interpolation speed
t = time of move
x = vector component magnitude (G01 x#.###)
y = vector component magnitude (G01 y#.###)
z = vector component magnitude (G01 z#.###)

Is = sqrt(x^2 + y^2 + z^2) / t

add the a-axis:

a = vector component magnitude (G01 a#.###) = (angle/360) * 3.14 * r * 2

Is = sqrt(x^2 + y^2 + z^2 + a^2) / t

sorry for the computer version of the math, I don’t know how to insert actual mathematic notation.

so I suppose that it depends on the controller but will a controller automatically calculate the rotating path based on the tool location relative to the defined rotating axis or do you actually define a radius for the rotating cylinder? I read somewhere that there is a G2## code that converts an a-axis to a planer equivalent. mainly for laser marking on a cylinder, but I am talking about a 5 axis machining center.

by the way thank you very much for the quick reply. Tis great to have such a friendly resource. I hope that one day I can contribute and return the favor.

[ger21]

Originally Posted by llilrex so I suppose that it depends on the controller but will a controller automatically calculate the rotating path based on the tool location relative to the defined rotating axis or do you actually define a radius for the rotating cylinder?

Depends on the control. As Geof and I said, with some, you define the radius. But I don't know how all controls handle this, and I'm sure they don't all do it the same way.

[Heinz Reimer]

http://www.linuxcnc.org/docview/html...ng_center.html

1.2.5 describes how EMC2 handles this.

[llilrex]

Thanks for the input.

So I read how EMC2 does it. It makes sense form a controller point of view, complex kinematic relationships are hard for a controller to do on the fly and keep everything timed right. So I suppose that if one was cutting a compressor in 5-axis the CAM used would probably calculate the tool path and force an inverse feed rate mode in the code to define the complex coordinated move explicitly with the time of move rather than the rate of move.

This was so much faster than me trying to figure it out myself. Thanks guys!

[Torsten]

Our Postprocessor kicks out a inverse Feed Time command G-code for all multy Axes commands. Tell you the truth I don't know exactly how this is calculated. But anyway the burden of computation is on the Post and the controller just steps this down to ensure all Axes complete motion at the same time. The thing about this is that looking at the F-code dose not realy give you any clue at how fast a move you gone get it is all tied to the length of the move, you gone have to trust your Post to do the right thing.

[Ydna]

^ yeah, that's how many CAM programs do the 5 axis interpolation (or continuous-motion 4 axis too). The post processor will take the toolpath then segment it into a number of very small linear moves in X,Y,Z,A,B, say each 1/8" in length. The G93 code determines the amount of time it takes to reach the next block (depending on the spindle speed).

The feedrates of these moves are all different, but they'll probably be close to one-another. The tool's surface speed changes depending on where it's at in relation to the center of rotation for the A/B/C axes, so it has to slow down / speed up all throughout the movements.

I attached a sample part of a program for an impeller, it shows the movements pretty well. The program is for one small part of the program where it roughs the blade surface.

[llilrex]

thanks for the file, it clears up allot of fragmented thoughts in my head. also it is nice to know what the norm is, considering that a controller might actually do multi axis cooperative feed rate, the cam programs dont assume that they do and dont take advantage of the feature. at least this is what I gather. I dont have a +3 axis cam so I dont know for sure. does any one know a good realitively inexpencive cam software that can handle 5 axis or 6 axis? possibly compatible with solidworks?

thanks again.