Off by a Mile

[plastikdreams]

Hi AGMantz:
I assume you are making rectangles with a constant radius on the corners...is that correct?
Also, you are keeping the males and throwing away the frame with the female cutouts??

If that is the case, a couple of thoughts spring immediately to mind.

First, others have posted about the desirability of wire cutting the center holes if you want them dead nuts, and they have also posted about the uncertainty around wire pickups.
As was pointed out...it takes very little in the way of dirt or burrs or shaved wire to make a bad pickup and you can really chase your tail trying to get confidence all is as good as it can be.
When every successive pickup is different by a tenth or two it gets hard to know where you really are.
So wiring the hole as well as the outside is an obvious way to eliminate the hassle and uncertainty of wiring to a finished feature...it allows you to make the parts as accurately as the machine is capable of.

The second point is about the code:
Some of you have commented on the length and ugliness of it.
I suspect the part is being programmed without a taper command, but by breaking up the radii into small linear moves in order to get an equal radius top and bottom.
If you just tip the wire over with a taper command and run G02 or G03 in the corners, you won't get a constant radius.

All the CAM software I've tried defaults to small G01 moves and UV displacement commands to manage constant radius tapers, so the code gets huge, and if you look at what has been posted here, that's the pattern that's emerging.

There is another way called "Complex Upper and Lower" in which you stitch the code for the upper profile together with the code for the lower profile, line by line with colons in between.
It will use G02 or G03 for arc moves, and you can use it to make all kinds of weird shapes, including constant radius tapers.
I know of no CAM program that can output this kind of code, so whenever I want to do it, I have to generate the code for the upper path, do the same for the lower path and then put them together manually...a royal PITA but it gives very compact code that's easy to edit.

On a last note, I concur with everyone who's pointing to a setup or basic programming error to account for your unwanted outcome.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

Jeeze Marcus I was getting worried!

 

[SeymourDumore]

I suspect the part is being programmed without a taper command, but by breaking up the radii into small linear moves in order to get an equal radius top and bottom.
If you just tip the wire over with a taper command and run G02 or G03 in the corners, you won't get a constant radius.

Marcus

Unless I misunderstood the OP, he only has a taper on the inside diameter.
The outside profile is a straight cut.


As far as the "constant radius" and CAM ...
I guess I am lucky that Makino can do it without having to resort to linear segments and 4 axis programming.

 

[plastikdreams]

The picture of the cam must have been taken with a potato...a more detailed shot may clear things up.

 

[plastikdreams]

Single pass, 10 wire 1221 epack pm=varying adaptive control on fc=.2 fa between .07-.18. Programmed 1.000. Low carbon steel. Mits fa10s



I leave a little to dust on the grinder.

 

[Waakzaamheid]

Marcus

Unless I misunderstood the OP, he only has a taper on the inside diameter.
The outside profile is a straight cut.


As far as the "constant radius" and CAM ...
I guess I am lucky that Makino can do it without having to resort to linear segments and 4 axis programming.

Every different machine I've worked on has a code for constant radius vs. constant angle. I think Fanuc is G06 maybe? It's been a while since I've had to use it.

As for the insert, pretty sure they said they're not even touching the inside hole, only using it for pickup, and then the rectangular insert has the taper on it. Seems like they should just be using a taper command with constant radius in the corners if that's the case. Doing it 4 axis is only complicating things. I don't know if that's a result of the CAM or not, but from personal experience at least I know Esprit and BobCAD can both output constant radius code.

I will also agree with wiring that center hole. If you drill that out to maybe .125 then wire it and then the outside, you pretty much can't miss unless there's something wrong with your program.

 

[007Rob]

I just had a thought. If the outside of the part is tapered with a 5 degree angle all around, and the inside hole is out of location in relation to one side...have you measured the overall dimensions of the "bad" part at the top and compared them to what you expected?

If the taper planes/heights are not setup correctly (on the machine or in esprit depending on how you are setup) the programmed size for the largest dimensions might not end up matching up with the top surface of your workpiece.

ie if your machine thinks the top of your block is 2" higher than the actual surface of your workpiece in the tank, your part will be smaller than what you expected. and this would shift the location of the inside hole in realtion to the edge of the part.

not sure if that makes sense but just a thought

 

[implmex]

Hi wdevine:
Yep, on my machine (Chmer) which speaks Fanuc, it's G60.
But the point I was making is that the OP was programming it "fourth axis" in Esprit, and that's what the code snippet I saw was starting to show.

I just got Mastercam for Wire (I blew up a computer so it was time to upgrade) and it defaults to programming these kinds of shapes this way too.
I'm not very familiar with the software yet, but I could not find a place to input a constant radius option in the software, so it would output that magic G60 for the control to read.
I suspect it depends on how you modeled the radii in the corners when you constructed the model.

Maybe it's there and I just haven't found it yet, but when I run the defaults I get code that looks like what the OP is running too, so I suspect his Esprit does what my Mastercam does unless you deliberately tell it different and also construct the model so it has true radii on the top and bottom planes instead of splines.

I don't know Esprit at all...it's supposed to be very good for wire EDM but all I know is that it's bloody expensive.
I'm only guessing it works this way.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

 

[SeymourDumore]

It is this statement by the op:

The U and V movements are their because the ID insert has a 5 degree angle all the way around it. Normally when I'm doing just a normal 2-d cut those small U and V movements are not there.

that made me think he IS cutting inside and out.

 

[007Rob]

It is this statement by the op:




that made me think he IS cutting inside and out.

My interpretation is that the part itself is called an "ID insert" and it has a 5 degree draft around the outside edge. But I may be mistaken.

 

[implmex]

Hi All:
Yeah, I'm confused!

Of course, that state comes easily to me now

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

 

[JZ.]

Hi wdevine:
Yep, on my machine (Chmer) which speaks Fanuc, it's G60.
But the point I was making is that the OP was programming it "fourth axis" in Esprit, and that's what the code snippet I saw was starting to show.

I just got Mastercam for Wire (I blew up a computer so it was time to upgrade) and it defaults to programming these kinds of shapes this way too.
I'm not very familiar with the software yet, but I could not find a place to input a constant radius option in the software, so it would output that magic G60 for the control to read.
I suspect it depends on how you modeled the radii in the corners when you constructed the model.

Maybe it's there and I just haven't found it yet, but when I run the defaults I get code that looks like what the OP is running too, so I suspect his Esprit does what my Mastercam does unless you deliberately tell it different and also construct the model so it has true radii on the top and bottom planes instead of splines.

I don't know Esprit at all...it's supposed to be very good for wire EDM but all I know is that it's bloody expensive.
I'm only guessing it works this way.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

Hey Implex,
If you have splines on a radius, you can select the feature and then "Break into many pieces". That will let you create curves or lines
and I usually set a tolerance on what it creates. I do this a lot when programming 4axis especially when using TU and TL. Depending on how many arcs it will give you, the code can become pretty lengthy.

 

[AGMantz]

Hey everyone I once again want to apologize for being so inactive on my own post. I was out with COVID starting on Tuesday and have been really down in the dumps since then. But I'm back today so I can finally clear up some questions!

SeymourDumore
Holy Crap that code looks ....

I know you can trust Espirit with the code, but Damm that looks UGLY!
Point to point linear segments are OK, but don't see the reason for it nowadays.
Are those blocks with the U and V for your tapered holes? Look pretty darn coarse steps to me.

Yes it does look very unappealing and it's something I'm not sure how to fix to be honest. I'm going to try and clear up what exactly I am cutting here because I didn't do a great job on explaining it and there seems to be a lot of questions about it. If you look on the potato cam quality picture I posted I am just cutting the rectangles out with the curved corners(they are on a 5 degree taper) and I am not cutting anything on the inside of the rectangles. The big outside rectangle is the stock and and the small circle on the inside of the rectangle is not being wired out at all it doesn't even go all the way through the part so it'd be impossible to wire out anyway.

implmex
Hi AGMantz:
I assume you are making rectangles with a constant radius on the corners...is that correct?
Also, you are keeping the males and throwing away the frame with the female cutouts??

If that is the case, a couple of thoughts spring immediately to mind.

First, others have posted about the desirability of wire cutting the center holes if you want them dead nuts, and they have also posted about the uncertainty around wire pickups.
As was pointed out...it takes very little in the way of dirt or burrs or shaved wire to make a bad pickup and you can really chase your tail trying to get confidence all is as good as it can be.
When every successive pickup is different by a tenth or two it gets hard to know where you really are.
So wiring the hole as well as the outside is an obvious way to eliminate the hassle and uncertainty of wiring to a finished feature...it allows you to make the parts as accurately as the machine is capable of.

The second point is about the code:
Some of you have commented on the length and ugliness of it.
I suspect the part is being programmed without a taper command, but by breaking up the radii into small linear moves in order to get an equal radius top and bottom.
If you just tip the wire over with a taper command and run G02 or G03 in the corners, you won't get a constant radius.

All the CAM software I've tried defaults to small G01 moves and UV displacement commands to manage constant radius tapers, so the code gets huge, and if you look at what has been posted here, that's the pattern that's emerging.

There is another way called "Complex Upper and Lower" in which you stitch the code for the upper profile together with the code for the lower profile, line by line with colons in between.
It will use G02 or G03 for arc moves, and you can use it to make all kinds of weird shapes, including constant radius tapers.
I know of no CAM program that can output this kind of code, so whenever I want to do it, I have to generate the code for the upper path, do the same for the lower path and then put them together manually...a royal PITA but it gives very compact code that's easy to edit.

On a last note, I concur with everyone who's pointing to a setup or basic programming error to account for your unwanted outcome.

Yes, if you are referring to the small rectangles with the curved corners as the males then you are correct(i apologize I'm still trying to learn the lingo here as I'm still new). The radius on all 4 corners of the rectangle are the same. It is a punch setup so I am only keeping the rectangles with the curved corners(which are what I am referring to as the "inserts").

Question about what you are saying about my ugly code, so from what I understand my code looks like this because I am running small G01 moves? And if I change that to G02 and G03 moves it'll shorten a lot of the code up? If that is the case I'll look into that because it is a pain in the arse sifting through all that code if I want to make a small adjustment or add something.

plastikdreams
The picture of the cam must have been taken with a potato...a more detailed shot may clear things up.

I apologize for the horrendous picture quality I'll try and get a better quality one up ASAP, when I was trying to SS the picture it kept using some stupid 3rd party app the previous wire guy had on here and I didn't realize how low quality it was going to be.

wdevine

Every different machine I've worked on has a code for constant radius vs. constant angle. I think Fanuc is G06 maybe? It's been a while since I've had to use it.

As for the insert, pretty sure they said they're not even touching the inside hole, only using it for pickup, and then the rectangular insert has the taper on it. Seems like they should just be using a taper command with constant radius in the corners if that's the case. Doing it 4 axis is only complicating things. I don't know if that's a result of the CAM or not, but from personal experience at least I know Esprit and BobCAD can both output constant radius code.

I will also agree with wiring that center hole. If you drill that out to maybe .125 then wire it and then the outside, you pretty much can't miss unless there's something wrong with your program.

You'd be correct wedvine I am not touching the center hole and the rectangular insert has the 5 degree taper on it. As for this taper command you speak of sounds like a blessing I want to be part of. I'll have to contact ESPRIT and ask how I implement constant radius code because I am not familiar on how to implement that, but it sounds way better than what I am doing currently.

007Rob
I just had a thought. If the outside of the part is tapered with a 5 degree angle all around, and the inside hole is out of location in relation to one side...have you measured the overall dimensions of the "bad" part at the top and compared them to what you expected?

If the taper planes/heights are not setup correctly (on the machine or in esprit depending on how you are setup) the programmed size for the largest dimensions might not end up matching up with the top surface of your workpiece.

ie if your machine thinks the top of your block is 2" higher than the actual surface of your workpiece in the tank, your part will be smaller than what you expected. and this would shift the location of the inside hole in realtion to the edge of the part.

not sure if that makes sense but just a thought

That was one of the first things I double checked on my own Rob, I did occur to me that there was a strong possibility that I had goofed something simple up there and to be honest I thought that was the problem however, I double checked all my planes and heights and everything matches in ESPRIT with everything on the machine in that sense.

SeymourDumore
It is this statement by the op:


Quote Originally Posted by AGMantz View Post
The U and V movements are their because the ID insert has a 5 degree angle all the way around it. Normally when I'm doing just a normal 2-d cut those small U and V movements are not there.
that made me think he IS cutting inside and out.

I apologize for the confusion I am cutting from the outside in. If you look real hard at my crappy photo you'll see the small start holes on the outside of the Insert those are where I thread in and then burn to the Insert.

I want to thank you all again for all the amazing replies, tips, and troubleshooting you've given me and for putting up with my lack of knowledge and lengthy time between my replies I should be much more active the rest of this week so I'll be able to get back to your questions/posts much faster.

Many Thanks,
-AGMantz

 

[implmex]

Hi again AGMantz:
OK, so the holes in the middle of each part are blind holes so you cannot interrogate them on the wire, and you cannot skim them on the wire.
So in order to get the profiles you're going to cut centered reasonably well around those holes, you need to use the same datum that was used to put the holes in, assuming they were done on the CNC mill and not just popped in wherever on the drill press or the Bridgeport.

So the easy way for me is to shift the model in Esprit until the origin is in the same place as the datum that was used to program the holes.

This will get you as close as you're likely to get...the factors that contribute to inaccuracy will be the accuracy of the mill guy's pickup, the accuracy of his mill, the accuracy of your pickup, and the fidelity of the model he used compared to the one you're using.

Given all these uncertainties, if you can get your features within +/- 0.002" of his, you will have done quite well IMO.

With regard to the business of the code...ultimately if you can make an acceptable part...who cares how large the code was to get there.
It's not like we have to contend with only a few kilobytes of memory in the control anymore, so if the code is big...so what!

Remember, the code Esprit generates very likely depends on how the model was first constructed in CAD.
If the model was a tapered rectangle and given constant radius fillets as a separate CAD step, the edge curves top and bottom will be splines and your CAM program will break it up into small line segments by default.
If it was modeled as a top profile and a bottom profile with arcs at the corners and then lofted together, the top and bottom edges will be arcs and you can instruct the control to follow those arcs with arc commands.
If you simply create a 2D drawing with arcs in the corners in Esprit to describe one profile and then give it a taper command it will make arcs on the corners too, and the radii at the top and bottom arcs will be governed by whether you invoke the constant radius command or not.
That is done in the control, either by writing the proper GCode word from the CAM software, or by editing it into the code by hand...the control doesn't care.

So I recommend you don't get too hung up on how big the code gets quite yet in your quest for knowledge...the objective is to get out a part that's as close to the model as you can.
If you get too creative too early, you will make a part that looks pretty close to right but actually isn't quite the same as the model: a constant radius fillet modeled onto a tapered rectangular block is not the same shape as a pair of rectangles with radiused corners lofted together...the difference is small with small tapers, but it IS different.
Maybe it's not enough to matter, but maybe it is.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

 

[AGMantz]

Hi again AGMantz:
OK, so the holes in the middle of each part are blind holes so you cannot interrogate them on the wire, and you cannot skim them on the wire.
So in order to get the profiles you're going to cut centered reasonably well around those holes, you need to use the same datum that was used to put the holes in, assuming they were done on the CNC mill and not just popped in wherever on the drill press or the Bridgeport.

So the easy way for me is to shift the model in Esprit until the origin is in the same place as the datum that was used to program the holes.

This will get you as close as you're likely to get...the factors that contribute to inaccuracy will be the accuracy of the mill guy's pickup, the accuracy of his mill, the accuracy of your pickup, and the fidelity of the model he used compared to the one you're using.

Given all these uncertainties, if you can get your features within +/- 0.002" of his, you will have done quite well IMO.

With regard to the business of the code...ultimately if you can make an acceptable part...who cares how large the code was to get there.
It's not like we have to contend with only a few kilobytes of memory in the control anymore, so if the code is big...so what!

Remember, the code Esprit generates very likely depends on how the model was first constructed in CAD.
If the model was a tapered rectangle and given constant radius fillets as a separate CAD step, the edge curves top and bottom will be splines and your CAM program will break it up into small line segments by default.
If it was modeled as a top profile and a bottom profile with arcs at the corners and then lofted together, the top and bottom edges will be arcs and you can instruct the control to follow those arcs with arc commands.
If you simply create a 2D drawing with arcs in the corners in Esprit to describe one profile and then give it a taper command it will make arcs on the corners too, and the radii at the top and bottom arcs will be governed by whether you invoke the constant radius command or not.
That is done in the control, either by writing the proper GCode word from the CAM software, or by editing it into the code by hand...the control doesn't care.

So I recommend you don't get too hung up on how big the code gets quite yet in your quest for knowledge...the objective is to get out a part that's as close to the model as you can.
If you get too creative too early, you will make a part that looks pretty close to right but actually isn't quite the same as the model: a constant radius fillet modeled onto a tapered rectangular block is not the same shape as a pair of rectangles with radiused corners lofted together...the difference is small with small tapers, but it IS different.
Maybe it's not enough to matter, but maybe it is.

Cheers

Marcus

Hey Implmex,

I think I've narrowed down pretty well what I missed when I did the first Insert. You along with many others assumed it was a setup error and I'm pretty certain that is what happened. I had a extra block made up for me along with a new model since I had to redo the insert I messed up on. This time I got some extra information about what side they used to pick up and what was milled and what isn't. After programming it out in ESPRIT and re-cutting it out on the Mits I found that I was within .003", which IMO is a big difference than my previous record of .040" lol. It's still not the accuracy I was hoping to achieve but it'll work for the job and I'm sure there is many more things I have to learn before I'm getting a "professional" level of accuracy. Regarding coding and the differences between all the different radii, corners, and arcs along with throwing tapering on top of that I'm still just as lost as I was before. I'll keep reading up on it and trying out new things and I'm sure in due time I'll be able to at least have a "above average" level of understanding about it.

-AGMantz

 

[implmex]

Hi again AGMantz:
This is where we ALL started out...a whole lotta confusing arcana and not a lot of help to be found.
Your best way forward is to take it step by step and in ten short years of full time investment...hey, you'll have it pretty well figured out too.
None of it is rocket science, but every fuckup is a learning event.
Fuck up as many as us grizzled veterans and you too will be a grizzled veteran.

That's how it works in this trade...experience (often bad ones) counts for a lot!

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining