Cones

[Matthew D. Hoedeman]

Are there any known issues with auto-feature cones and V3.7 MR3? Getting some funny values for the cone 'A'ngle.

[craiger_ny]

Don't know. What kind of "issues" are you having Matt? In regards to the cones I don't have time for your other issues.

[Matthew D. Hoedeman]

Well, I seem to be getting a lot of variation in the results. The cone is formed, then pierced. I have run the same cone, over and over, and I am getting up to 0.5 degrees difference from check to check. I am simply doing a CTRL-E and executing the same feature, over and over. Now, it IS a rather small cone and it is pretty shallow. It calculates out to about 0.0015" of error over the length of the cone I am checking, which isn't much, but one heck of a lot more than I am used to seeing on this machine (about 4X as much). I just wanted to know, really, if the manually learned would be 'better' than the auto-feature or if there were ANY thinig about either that I need to watch out for. I am using 4 hits/level with 3 levels

[#2#]

Other than that they are kind of a pain in the **** to get in the right spot, not that I know of.

Did you try checking the angle another way (couple of lines)?

[Matthew D. Hoedeman]

Other than that they are kind of a pain in the **** to get in the right spot, not that I know of.

Did you try checking the angle another way (couple of lines)?

Haven't tried that yet, but I did just learn one manually and it give a result of 82.248, check the same one with auto feature, it gives 83.076! All good hits, I watched them all.

[Matthew D. Hoedeman]

One thing I have noticed, the learned cone does not use the cone's side vectors for teh touches, it measures it more like 2 circles, but all hits ARE on the 'good' portion of the cone surface. 2 cones in the part, that is all I just checked:

learned cone #1 82.295
Auto feature cone #1 83.752

Learned cone #2 82.307
Auto featuer cone #2 82.375

So, one cone shows a LOT of difference depending on which method is used, the other repeats quite well, over 4mm of cone length, it is a total error amount of 0.0048mm total but the first one is .1017mm. WTF? I will try the line method next, however, that will not fix my over-all problem, there are 2 circles that get constructed from these cone at the intersecion of the cones with a plane and those intersections have a positional tolerance of +/-0.010" and a size tolerance of +/-0.002" Yeah, right! For the freaking bracket that holds the spring to the oven door.

[#2#]

One thing I have noticed, the learned cone does not use the cone's side vectors for teh touches, it measures it more like 2 circles, but all hits ARE on the 'good' portion of the cone surface. 2 cones in the part, that is all I just checked:

learned cone #1 82.295
Auto feature cone #1 83.752

Learned cone #2 82.307
Auto featuer cone #2 82.375

So, one cone shows a LOT of difference depending on which method is used, the other repeats quite well, over 4mm of cone length, it is a total error amount of 0.0048mm total but the first one is .1017mm. WTF? I will try the line method next, however, that will not fix my over-all problem, there are 2 circles that get constructed from these cone at the intersecion of the cones with a plane and those intersections have a positional tolerance of +/-0.010" and a size tolerance of +/-0.002" Yeah, right! For the freaking bracket that holds the spring to the oven door.

I would suspect the cone which the hit vectors look funny on was not learned properly.

You could always do 4 lines around the cones. Then create a point where each line pierces the plane. Use those 4 points to create a circle.

[craiger_ny]

I always use learned features. With the cone you should get good hit vectors after an edit, if you are not something is surely up. Do you edit the learned feature including the feature vector?

[Matthew D. Hoedeman]

I would suspect the cone which the hit vectors look funny on was not learned properly.

You could always do 4 lines around the cones. Then create a point where each line pierces the plane. Use those 4 points to create a circle.

A re-run of the learned cone had the touches 'normal' to the sides of the cone, just like the dcc cone. Go figure.

[Matthew D. Hoedeman]

I always use learned features. With the cone you should get good hit vectors after an edit, if you are not something is surely up. Do you edit the learned feature including the feature vector?

Nope, I just let it go with what it learned. Also, a re-run of that learned cone DID measure the hits 'normal' to the surface of the cone. Go figure (Again!).

[#2#]

Nope, I just let it go with what it learned. Also, a re-run of that learned cone DID measure the hits 'normal' to the surface of the cone. Go figure (Again!).

So did that put the angle closer to the autocone?

[Matthew D. Hoedeman]

So did that put the angle closer to the autocone?

Nope, but it does make it closer to what I know (and have checked) as the angle of the punch putting the cone into the part. The only thing I can figure is that possibly one (or more) of the DCC cone hits was making contact in a 'bad place' 3 out of 4 give good results, but that one dcc cone was just bad all the time. And, re-checking the same cone, over and over, with the DCC cone also gave up to 0.6 degrees of change, so there must be something down in the cone that I am just not able to see with my crappy eyesight. Also, these cones are for flat-head screws, so that they will not stick above the surface of this bracket and all it does is stop the oven door from opening past 'flat' and holds the spring to counter-weight the door so any feeble housewife can close it. And this thing has, for it's LARGEST tolerance, +/-0.015" and this will REALLY get you, the customer standards for the die are 0.015" of clearance PER SIDE for the pilot to the pilot hole inthe part (this is a prog-die). So, with a maximum tolerance of 0.015", there is 0.015" of slop designed into the tool.

[John Kingston]

I use vector points and constructed cones. Mainly because I need to use multiple tips to get all the areas of the cone. This method appears to work well for me.

[#2#]

I guess I would try changing the number of hits on the auto cone so that maybe it won't hit that questionable point.

Is it possible your probe has a chip stuck to it or a chip in it? Or is there maybe some grease in that hole? That could explain the wandering results.

[Matthew D. Hoedeman]

I guess I would try changing the number of hits on the auto cone so that maybe it won't hit that questionable point.

Is it possible your probe has a chip stuck to it or a chip in it? Or is there maybe some grease in that hole? That could explain the wandering results.

Nope, tip is fine, as a matter of fact, I have used 2 different tips, got simular results, and the reason for the second tip is:

I SMASHED IT! Yippie! It's been some time since I've done that. I program almost ALWAYS in metric, but this thing is in english, and I used a metric value where I should have used an english one and SMACK-O, time for a new tip!

No grease, I wipe them out. But, the surface of the cone is almost mirror-like due to the forming process and I am checking the concave portion, not the convex side and that is the side that is mirror-like from the form punch. It is too small to get my fingers in there to feel it and with the mirror-finish, you can't see for crap. It's like a Grand Illusion, man.