Inverse time calculated - wrong maybe?

[GPC_CNC]

I have an IHS post for our Mazak Variaxis 630-5x2. I am doing a rectangular pocket (Swarf Cut) with an equal 4.5° draft on all 4 walls (see attached pic). My understanding from inverse time is that it is essentially the amount of time to complete a move. So to me that says that if the radii in each corner is identical, that the given inverse feeds should be relatively the same. If you look at the attached .txt file you'll see that each corner is outputting different feed given that each corner is equal and has 44 moves to complete it. So one corner averages F200.~ and the next one F553.~, etc, etc.

 

When I run this on the machine it's obvious that each corner is different. First corner takes 16~ seconds to complete and the next one 7~. I calculated the cordal length of the arc and came up with a total time of .68 seconds to complete the arc at a programmed feed rate of 108ipm.

 

My next step is that I bumped up the inverse times to match across every arc to around F500. for each move. Re-run the machine and each arc takes approx 7 seconds to complete. Go back and recalculate the feeds again all corners 500% (the max that cimco would let me bump up at one time) to give me approx F2760.~ and the machine completes each arc in around 1 second. I haven't cut an actual part yet to prove speed vs accuracy, but it seems to move very well (not choking on code or stuttering/slowing down, etc.).

 

I also tried posting with our Matsuura 5x post and I get the same differing feeds at each corner (though instead of F200.~ I would get F141.~ and a similar phenomenon at the other corners different rates...weird). I also tried moving my WCS closer to the center of the slot and that made my Feeds come out closer to each other around F300~ and F400~ respectively.

 

Am I correct in thinking that for a given arc swing (at a constant 4.5° tilt) that each arc should be the same inverse feed regardless of where it is in respect to the center of rotation? In other words if I have a pattern of these pockets, the way I'm seeing it post is that every pocket that's further from the Center of Rotation will take longer to machine. Each pocket SHOULD take exactly the same programmed time. (obviously if the machine can't move all axes that fast at a large swing, it won't be that fast).

 

What it boils down to is this: Is anyone seeing similar results? Am I on the right track in thinking that my inverse times aren't being calculated well? Am I correct in thinking that feedrates should not change based on where my WCS is located (simulated/programmed world, not what the machine can actually do)?

 

I've tried looking through the post and Machine Control Def's and nothing obvious seems to pop out to me in the way of settings that are wrong.

 

Thanks

 

Greg

52620100.txt

[crazy^millman]

So let me get this correct you have not run any code on the machine and have determined without running the code on the machine it is wrong?

 

Am I correct in thinking that for a given arc swing (at a constant 4.5° tilt) that each arc should be the same inverse feed regardless of where it is in respect to the center of rotation?

 

No if you were swinging that arc 1" from the center of rotation then you would expect to see the inverse time to be really low. But if it were 100" away from the center of the rotation then you would expect to see it go off the charts for the feed rate.

 

If the part is perfectly on center and every point is 100% the same through all 4 arcs in relation to the center of the machine with regards to X,Y,Z,B,C then ye I would expect them to be pretty close. Is that the case here no way on know since I do not know where the center of rotations is in relation to the 4 arcs.

 

Do a search about inverse time this has been covered many times on this board.

[crazy^millman]

Linky #1http://www.emastercam.com/board/index.php?showtopic=71589entry832861

Linky#2 http://www.emastercam.com/board/index.php?showtopic=63526entry716706

 

 

There are many more.

[GPC_CNC]

So let me get this correct you have not run any code on the machine and have determined without running the code on the machine it is wrong?

 

 

 

No if you were swinging that arc 1" from the center of rotation then you would expect to see the inverse time to be really low. But if it were 100" away from the center of the rotation then you would expect to see it go off the charts for the feed rate.

 

If the part is perfectly on center and every point is 100% the same through all 4 arcs in relation to the center of the machine with regards to X,Y,Z,B,C then ye I would expect them to be pretty close. Is that the case here no way on know since I do not know where the center of rotations is in relation to the 4 arcs.

 

Do a search about inverse time this has been covered many times on this board.

 

See my second paragraph in my post:

 

"When I run this on the machine it's obvious that each corner is different. First corner takes 16~ seconds to complete and the next one 7~. I calculated the cordal length of the arc and came up with a total time of .68 seconds to complete the arc at a programmed feed rate of 108ipm. "

 

Edit:

This was run with a physical part. The straight sections are nice and good finish, the corners are all dark from the tool rubbing.

End edit:

 

When I said this:

 

"I haven't cut an actual part yet to prove speed vs accuracy, but it seems to move very well (not choking on code or stuttering/slowing down, etc.)."

 

I meant that I have not run a physical part with the new parameters I edited in myself (not using the post) that achieve 1 sec per arc (as I believe it should be running). The machine appears to run well and when I cut a part I will prove whether or not the MACHINE can handle the movement accurately.

 

I've read the first link you posted before. The second one I haven't, though it doesn't seem to be what I'm looking for. I've read a lot about inverse before coming here. What I'm trying to wrap my head around here is NOT that I don't understand inverse time, it's that I think I have a fairly good idea of what it's supposed to be doing and I don't think it's doing it. I don't think it should take 16 seconds to mill a R.5716" arc and then 7 seconds on the another R.5716" arc (with a 5/8" endmill). While the rotary should be making the same 90° angle in the same amount of time (keep in mind the tilt is locked at 4.5°) per arc, the X/Y/Z axes should be moving faster to keep up the further away the feature is from the center of rotation. A given arc should be milled in a certain amount of time whether it's at the center of rotation or 20" from it at a given programmed feedrate just as a 1" line segment takes the same amount of time to move anywhere on the table in 3D space when the rotary axis are not moving (3+2).

 

Greg

[crazy^millman]

Okay missed you did run it one the machine sorry about that.

 

A given arc should be milled in a certain amount of time whether it's at the center of rotation or 20" from it at a given programmed feedrate

 

Yes that I agree with you on, but you put this way:

 

that each arc should be the same inverse feed regardless of where it is in respect to the center of rotation?

That is why I was thinking you were not that aware since you made that statement in the middle of the posting which threw me off. They are not the same thing and for inverse time they could change as for physical time no they should take the same amount of time. The reason again they could change with regards to inverse time is how is the part set up in relation to the center of rotation? Are all 4 arcs equal distance from the center of rotation to the start and ends and vectors? If not then not the inverse feed rate should change. If they are then the inverse feed rate I think should stay the same. Again looking at the supplied picture no way to tell or know.

 

I did ask these same questions which you did not answer you only pointed out that I misunderstood you, but I only misunderstood you running it on the machine.

[GPC_CNC]

 

That is why I was thinking you were not that aware since you made that statement in the middle of the posting which threw me off. They are not the same thing and for inverse time they could change as for physical time no they should take the same amount of time. The reason again they could change with regards to inverse time is how is the part set up in relation to the center of rotation? Are all 4 arcs equal distance from the center of rotation to the start and ends and vectors? If not then not the inverse feed rate should change. If they are then the inverse feed rate I think should stay the same. Again looking at the supplied picture no way to tell or know.

 

I did ask these same questions which you did not answer you only pointed out that I misunderstood you, but I only misunderstood you running it on the machine.

 

Great, now we're mostly on the same page.

 

The rectangle is NOT on the centered on the center of rotation, but what is weird (to me anyway) is that the arcs that are farther away are the ones that take 16 seconds (F200.~) and the ones that are closer take 7 seconds (F550.~). This seems opposite to me.

 

So after running the part and not being satisfied with this I did some air cuts to get my bearings with what it was doing. All of the arcs that were F200.~ I increased 275% so that they would match the other arcs. This made the time at each corner approximately 7 seconds. So that correlates (to me anyway) that given that all 4 geometries are the same and they each have 44 equally spaced line segments, that the posted inverse F# should be relatively the same in order to cut the feature in the same amount of time.

 

That being said, 7 seconds is way to long to cut a chordal length of .8"~ so I figured out that on a 2 axis table swinging that length at 108ipm should take about .68 seconds to do. I rounded up to 1 second, but when I used Cimco to adjust it would only let me bump it up to 500% larger which came out to roughly F2760. which seems to approximately cut each corner in a little more than 1 second.

 

Since posting I have cut a part and it runs well and looks smooth (won't know for sure till I pull it off) and there's no rubbing in the corners.

 

Where I'm concerned is that now that I've got something that looks right to me on the machine, why is the post so VERY far off. I believe the inverse calculations are in the PSB file, so I can't really see what it's doing. I'm very close to sending the post out to be evaluated, but I wanted to come here 1st to make sure I'm not totally crazy.

 

I've noticed this before in other swarf cuts, but they were short run and didn't matter. This will be longer running and I'd like the post to work properly.

 

Thanks

 

Greg

[crazy^millman]

You have done a very good job of figuring out the problem you are seeing and I would take this thread an send the link to the IHS post folks and ask them to address the problem you are seeing. I would think they can easily figure out what is going on and help you get it taken care of.