Surface Finish Calculation Time?

[Mr. Wizzard]

So, I frequently have some complex surface machining to do. Sometimes, it takes a while, but how long is too long? Right now, I have a small part, roughly 3.5" in diameter, a cavity. The surface, which I cannot show, would look similar to perfect ripples in a pond if you dropped a pebble in the water. My finish toolpath is a .020 ballnose, surface finish flowline, .001 tolerance with filter set to output all arc moves in x/y, and a .0005 stepover. So far, it's been rocking out over 6 hours. The HST are the only ones that multithread and that leaves Mastercam frozen and, pretty much my whole computer almost worthless.

Is this normal?

Does my computer just suck?

Should I just shut up and quit whining?

[peon]

That's a lot of detail to cut with a .020" ball. I'm only using a cutter that small to pick fine detail using a .0001-.0002" tolerance. Is there a reason for not using a larger tool?

[Mr. Wizzard]

I roughed out with a .125 ball, then .062, then .050, then .040, then .030. The bottom of the valley where the two steep-angled walls meet has a .015 radius on the bottom and I have to have a nice smooth finish on all surfaces. It's not very deep, less than .250.

[Mr. Wizzard]

I used an HST rest rough for the .020 ballnose to hit the stock left by the .030 ballnose first, before I run the flowline finish toolpath. It calculated last night after i left work and took 5.5 hours by checking the event logger.

[gcode]

Does my computer just suck?

I don't know.. details please

[MotorCityMinion]

Are you using a solid or surfaces? Flowline, for me anyways, seems to be one of the faster calculating tool paths. The .001 tolerance you speak of isn't too tight either. Strange.

 

OT. IMO, too many tools for the job.

 

 

[Chris Rizzo]

Does my computer just suck?

 

Are you on a pentium 3 or an i7????

[gcode]

Are you on a pentium 3

 

I installed X4 on my old P3 laptop a while back.

The install took nearly an hour to run..

and X4 took about 90 seconds to launch..

It ran OK but man was it slooooooow

[Mr. Wizzard]

I'm not computer genius, but here's what the thing is:

Dell Precision T3400

Intel Core 2 Duo

E8400 @ 3.00 GHz

2.99 GHz, 2.00 GB of RAM

NVIDIA Quadro FX 4600

 

The operation that regened over night ran over 22 hours! But, the resulting file isn't that large, only 3300K or so. I've had much bigger NC files before, but this one just seems to kill the computer.

[Mr. Wizzard]

The reason for all the tools, as I stated before, is because of the bottom of the v-shaped surface. The radius is .015 and the .020 will step across it, as opposed to a .030 just hitting it.

Regardless, I'm now at the finishing stage and can't get a "decent" calculation time on the toolpath.

[gcode]

That's a pretty good computer, and a NICE video card.

I'm guessing it's running WinXP?

It was a rocking machine in its day

This is how it rates on the High CPU chart.. you'll need to scroll down.

PassMark CPU

I don't have any idea why that particular toolpath took so long to crunch.

How does this machine do running the Benchmark file?

I'm going to guess high 2 minutes to low 3's ????

If you do work like this daily, it would be worth your time and money to buy

a new machine.

If this is not a regular thing, it would still be worth your time and money (less than $100)

to add 2 more gig of ram to that machine.

[Gary]

surface finish blend?

[Ocean Lacky™]

Solid or Surfaces?

 

If surfaces, when I run into unusually long crunch times, the first thing I do is Analyze/ Test Surfaces and check for a bad trim or surface normal suddenly reversing.

[Mr. Wizzard]

Surfaces. I tried blend, but can't get all arcs. This needs to be ALL arcs and there's not reason is shouldn't be. It's a circular pattern. I've tried solids and surfaces. I figured the easiest way to guarantee it was created 2-d geometry on the front plane (like a sectional view) on the surfaces and rotate it. All normals are good and flowline was giving me what i wanted, but not now. I set the stepover big, like .015 or so, and got the path I wanted: all arcs, smooth motion, money shot.

Then I set it to .0005 stepover for a sweet finish, it runs 22 hours overnight, and I come in this morning to find tons of linear moves and retracts.

[Müřlıń®]

Surfaces. I tried blend, but can't get all arcs. This needs to be ALL arcs and there's not reason is shouldn't be. It's a circular pattern. I've tried solids and surfaces. I figured the easiest way to guarantee it was created 2-d geometry on the front plane (like a sectional view) on the surfaces and rotate it. All normals are good and flowline was giving me what i wanted, but not now. I set the stepover big, like .015 or so, and got the path I wanted: all arcs, smooth motion, money shot.

Then I set it to .0005 stepover for a sweet finish, it runs 22 hours overnight, and I come in this morning to find tons of linear moves and retracts.

 

 

Is your surface model a revolved using splines?

[Mr. Wizzard]

Yep, gotta use splines.

[Mr. Wizzard]

Another thing: I tried a blend using two circles and the surfaces. I set it to 2d, which claims to have 2-d stepover with constant Z cuts. Instead of arcs at a constant Z height, I get linearized moves around the circle with varying Z heights (3-D moves).....

[Mr. Wizzard]

arrrggghh.....

Take original "solid"

Create curve, slice down middle from front plane

Rotate surface using splines created

all separated surfaces, but normals all good

quick flowline

mostly arcs, lots of linear moves, retracts, some surfaces completely skipped.....

[Müřlıń®]

arrrggghh.....

Take original "solid"

Create curve, slice down middle from front plane

Rotate surface using splines created

all separated surfaces, but normals all good

quick flowline

mostly arcs, lots of linear moves, retracts, some surfaces completely skipped.....

 

 

It has been my experience that in order to achieve your goal you must draw your wire frame geometry with arcs and lines instead of splines. That will decrease the number of small linear moves

 

When you create the wire frame like you are by cutting the solid and using that spline, your system tolerances are used to build the spline.

 

so I would make my system tolerances as tight as possible and my tool path tolerances as loose as possible.

 

I haven't checked system settings lately but there used to be a way to set the max number of points per spline and patches per surface.

 

Sounds like you have them set way too high and that could be attributing to the long crunch time.

 

But that will not solve the "All Arcs" you want in your NC code.

 

 

OH BTW there also used to be a way to set the cordial deviation as well and that affects how smooth your splines turn out.

[Mike Whitten CAD/CAM Contractors]

Hello Mr Wizzard,

 

It would be a good idea to try an experiment on a small area. You are now running a .001 tol and .0005 stepover. It could be getting mixed up because your tol is greater than the stepover. Just a guess as I never tried this type of thing. As peon mentioned above, the tolerance should be much smaller than you are running. .020 is a very small tool and since the stepovers are so small, I am thinking that your tolerance should be something like .00015. I always recommend starting with a scallop height calculation before anything else. I am not sure of your requirement, but it seems like you need somewhere in the range of .00004-.00012. IMO, anything less than this is just rubbing the tool and wasting time. There are many scallop height calculators on the web, but you can always just start a "Dummy" parallel cut, input your values and get an answer. You don't even have to select any surfs. A stepover of .0005 calculates to a .000003 scallop height. I would try a stepover of about .0022, which calcs to a .000061 scallop height.

 

In summary, try a test at:

.00015 tol

.0022 step

Indicate cutter to .0002 max

 

Mike

[Mr. Wizzard]

I'll give it a shot. Perhaps that is why is gives me all arcs when i "test" it with a larger stepover and starts adding linear moves when I make the stepover finer.

thanks.

[Mr. Wizzard]

Ok, just picked 4 surfaces instead of all of them. Better, but why the linear moves in the one spot? There are a few scattered randomly, but not too terrible. I'm stumped as to how or if I can get rid of them.

Right now, .001 stepover, .0005 tolerance

[Mr. Wizzard]

The radius shown above that has the linear moves is between a near-vertical surface and a near-horizontal surface. They get larger, but steeper as they step away from the center and there are 9 sets total. If i do a surface finish flowline in 3's (one steep one shallow and the radius in between) to the outer edge, the radius in between them has the same linear moves in the same spots every time. Also, the near horizontal surface gets steeper every time until the last one is like a "v-shaped" ditch with a .015 rad on the bottom.

Still stumped.....

[Mike Whitten CAD/CAM Contractors]

I think it is a combination of the tolerance and gap settings. It sounds like you need to change the gap setting to "follow surfaces" before anything else.

 

Mike

[Mr. Wizzard]

Eh, tried "follow surfaces", comes out exactly the same.