Mirror Toolpath - Stock to Leave Optirest

[Metallic]

Here's a new one. My part may be kinda screwed but oh well, I have 2 parts that are mirrored copies of one another and I have 100 toolpaths written for this part, thus I do not want to re-write or re-chain the toolpaths and I have been using transform toolpath. The toolpath has 0.02 for both Wall and Floor stock to leave. Parameters on the copied/mirrored path show that. I have dropped appx 10 hours programming this to get a first article across the finish line and well I don't love the idea of doing it all over again.

 

 

HOWEVER backplot clearly shows this violation of stock, and thus my part on machine is looking unfinished after finish passes. Some of the paths are driven by a stock (original part) model. Would that be the cause for it not leaving me stock?

 

[crazy^millman]

Transform Mirror for this? Sorry, but I got bitten too many times to trust that without doing some testing or checking. Running parts without doing so it use at your won risk in my humble opinion. If you can get submitted to QC send it to them so they can review and hopefully use it to make this work better.

[Metallic]

Yea Ron you would think something like mirroring NCI would work easily in August 2021 but I guess I have too high a hope. This attempt was not even using NCI data, I literally had it transform and create brand new ops. And checked each op (adjust as needed), appartenly not well enough tho. Knowing this I probably will never again mirror something that isn't 2D contour.

In my case I have some luck in that this part is not hot...but it sure as heck ain't simple.

[Leon82]

When I have to do this I usually save as bring the new model in and reselect or rechain.

[Metallic]

24 minutes ago, crazy^millman said:

Transform Mirror for this? Sorry, but I got bitten too many times to trust that without doing some testing or checking. Running parts without doing so it use at your won risk in my humble opinion. If you can get submitted to QC send it to them so they can review and hopefully use it to make this work better.

I know that you are quite good with 5 axis so I will pose a more genera related question ( this may apply to 3D machining as well)

 

how do you deal with multiaxis part families where you have identical (mirrored) or similar features (could be different sizes/locations/scales) etc? Do you create a toolpath template where your parameters are fixed and you’re only chaining or selecting geometry and avoidance? Similarly how do you deal with axis control lines or tool vectors?
 

it seems unproductive to write an A+ program only to have to go do it all over again 

[Leon82]

Sometimes I feel multi-axis paths do not like to be copied.

[crazy^millman]

2 hours ago, Metallic said:

I know that you are quite good with 5 axis so I will pose a more genera related question ( this may apply to 3D machining as well)

 

how do you deal with multiaxis part families where you have identical (mirrored) or similar features (could be different sizes/locations/scales) etc? Do you create a toolpath template where your parameters are fixed and you’re only chaining or selecting geometry and avoidance? Similarly how do you deal with axis control lines or tool vectors?
 

it seems unproductive to write an A+ program only to have to go do it all over again 

Thanks for the kind words.

Brute force hard work if needed. I will use the file as a seed file and import operations into the file where needed. I have used mirror toolpaths, but would never take 100 operations and use one transform to do it. Maybe a combination of both things, but many times since you have already figured out one part it becomes much easier to do the mirror. Trust me I wish we had the easy button one click method to take 40-200 hours of work and we are done, but in my experience just never that easy.

 

[Metallic]

19 hours ago, crazy^millman said:

Thanks for the kind words.

Brute force hard work if needed. I will use the file as a seed file and import operations into the file where needed. I have used mirror toolpaths, but would never take 100 operations and use one transform to do it. Maybe a combination of both things, but many times since you have already figured out one part it becomes much easier to do the mirror. Trust me I wish we had the easy button one click method to take 40-200 hours of work and we are done, but in my experience just never that easy.

 

Great advice. I will break down my workflow thought process:

 

I developed my 100 toolpaths for the first part. Great. I realized I can’t mirror all 100 just that easily so I did it via chunking - initial roughing, Optirough and then the 5 axis toolpaths were all separate toolpath groups and separate transformations. So it the end I had like 7 new toolpath groups consisting of the transformed toolpaths. Everything “looked” good but under closer inspection obviously was not. So I wasn’t necessarily doing a single transform across the board. 
 

im sure my method might work to a certain extent but in this case needed some extra scrutiny. Live and learn!

[Colin Gilchrist]

14 minutes ago, Metallic said:

Great advice. I will break down my workflow thought process:

 

I developed my 100 toolpaths for the first part. Great. I realized I can’t mirror all 100 just that easily so I did it via chunking - initial roughing, Optirough and then the 5 axis toolpaths were all separate toolpath groups and separate transformations. So it the end I had like 7 new toolpath groups consisting of the transformed toolpaths. Everything “looked” good but under closer inspection obviously was not. So I wasn’t necessarily doing a single transform across the board. 
 

im sure my method might work to a certain extent but in this case needed some extra scrutiny. Live and learn!

As part of your process, I'd suggest taking each "group" and running it through Verify, then using the Comparison function to color-shade the remaining stock amounts. This give you a visual that you've got good roughing, semi-finishing, and finishing paths. I typically do this, but skip any tapping operations.

I find it works best to just bite the bullet, and mirror the geometry and operations, then go through the groups and check each one. Sometimes you need to modify boundaries or surfaces to get good paths on the opposite side.

Keep in mind this fact: if you are mirroring any geometry which happens to be a Spline entity, that Spline is based on mathematical equation of nodes points and knot vectors. These man not give you the results you expect when mirrored. Especially if the Spline is a "trimmed Spline".

[nickbe10]

On 8/13/2021 at 3:44 PM, Metallic said:

it seems unproductive to write an A+ program only to have to go do it all over again 

Depends, the basis for a good continuous improvement program is a robust "post and go" source file.

If you are never going to see the part again, all bets are off.

 

[Metallic]

4 hours ago, nickbe10 said:

Depends, the basis for a good continuous improvement program is a robust "post and go" source file.

If you are never going to see the part again, all bets are off.

 

Yeap this is a prototype part not a production item. By the time I see it again the design will be fully revved and my old paths won't be useful anymore

[Chally72]

If you're willing to shoot me a file link via PM, I can take a look at the mirroring and attempt to figure out what's going on. 

One thing that is a challenge (depending on your mirror choices) when mirroring paths is determining how to solve all the circular dependency issues created if you're referencing stock within your operations. So what Metallic has probably done, and what you need to do in these cases, is to create a separate mirror for every string of operations that exists between stock models in the tree, and then sometimes create separate stock or rough operations for critical stock-aware parts of the program before continuing on. This issue arises because we're programmatically mirroring an operation which references the original, non-mirrored stock, and now we have a dilemma about if the original reference was the intent or not for the mirror, etc etc. 

Mirroring tools don't always allow the power we need to explore all scenarios that may exist, especially on a complex part like Metallic has posted, so it often takes a bit of time and workflow adjustment to get it right in a case like that.