Accurate locating on an irregular object

Hi everyone,

I need to mill a constant depth (~0.040" deep) pattern on the face of an irregularly shaped object which varies in z-height. The shape of the object is essentially a portion of a sphere, but with irregular edges and not with constant curvature.

I have an STL of the object, which I can use to generate the toolpaths for the pattern. I was planning to mill or 3d print soft jaws using the negative of the STL for accurate location of the part on the mill. This works, but is pretty time consuming. I'm wondering if there's a quicker way of doing this?

One thought I had was to use a smaller vise with gripper style jaws that will hold irregular objects. That would work, but wouldn't tell me the orientation of the part (since the grippers will likely skew the part as they grip). I considered using photogrammetry or laser scanning of the part in the vise, then using that smaller vise as the WCS origin, but this seems inconsistent and also pretty time consuming.

Are soft jaws the only reasonable way to do this? I have a couple hundred of them to get through (all slightly different), so anything I can do to speed the process is much appreciated!

Thanks

Lee

Sent from my SM-N960U using Tapatalk

Much depends on what material and tolerance expectations for the result. If the material is a plastic, you could possibly consider some sort of router pattern that references the surface to be machined. This may not account for whatever XY-rotational variation may be present in overall part size/shape. If it is metal, I think machined soft jaws may be the only way to go.

Hi specfab,

Thanks for the input. The material is bone (yes, bone....). The tolerance is loose: +/- 0.005" would be fine. But, I'm cutting a 3d toolpath with tiny endmills, 1/32" or smaller. I'm mostly concerned with making sure the mill knows where the (convex) workpiece is to avoid large z-height discrepancies and broken endmills due to taking too much DOC

Lee

Sent from my SM-N960U using Tapatalk

Hi Lee:
How will you verify that your part is where you are hoping it is.
Can you mount the bone to something that you can use to locate it and pot it.

I've has success with this approach, using Cerrobend as my potting medium; mounting my irregular part on a post I could put tin a collet on the Bridgeport, and lowering the part into a vat of molten Cerrobend, then letting it cool in place.
I found the part origin and Z zero by interrogating the post before I removed it to do the machining.

The problem now moves to mounting the post accurately.

Can you swipe a stereotaxic frame from the neurosurgery guys and use that to orient your part relative to your post?
It won't be nearly strong enough to tolerate milling forces, but it should be plenty good for hot melt glueing a post in place in a known location.

I assume you can manipulate the position of the bone within the frame until you have it where you want it.
All the rest is comparatively simple.

BTW, you can use dental plaster as a potting medium if you don't want to use Cerrobend.
I've been successful with both.
You can even use stick shellac or Crystal Bond or jeweler's pitch or hard wax if your cuts are going to be light.

I've also been successful potting my part onto a tiltable table and then positioning after potting.
I used to be a dentist as you know, and I used this approach often for dental implant cases where I needed to align an implant and abutment blank to a line of draw I wanted to follow when building a custom abutment and prosthesis.

Cheers

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

Another thought occurred to me...since you have an STL model of your part, can you use it to 3D print a tripodized interrogation post for each part that you can temporarily attach to your sample so you can pot and align it using one of the methods above.
The 3D printed post would take the place of all the farting about with a stereotaxic frame.
I'd still use a manual milling machine quill to lower the mounted part into a box you can put your potting medium into.
You can prep a lot of parts quickly and reliably this way.
The rectangular sides of the box allow you to just grab the whole works in the vise in your Haas.

Hi Marcus,

Thanks for sharing your workflow - you identified the main reason I posted!

Mounting the part securely isn't so much of a problem - it should be easy enough to hold in a smaller vise with a versagrip or talon style jaw. I can then hold the smaller vise lengthwise in the 6" vise, and touch off on the fixed jaw of the small vise for the origin.

The issue comes with figuring out where the part is, and it's orientation, relative to the small vise. I'm not sure the potting medium / dowel pin addresses that? Unless I'm misunderstanding, for it to work, the part and the dowel pin would need to be imaged after it's mounted, and a 3d model of the whole assembly used for the toolpaths. If I need to do that, I may as well just image the vise with the part mounted and use that.

I'm hoping I missed a detail in your description - I know you've been here and done that!

Best regards,

Lee

Edit: I DID miss a key detail in your description - the suggestion of the stereotaxic frame.... That's a really interesting idea, and really makes the potting idea more attractive.

I can ask around and see if we have a stereotaxic frame available to run a few tests. Thanks!

I do not see the problem
If you 3d print soft jaws and make that you know the orientation of the outside of the jaws
Then also the orientation of the workpiece is known

Peter

Hi Peter,

It's not a problem as such. The workflow works well enough. I was asking if there's a quicker/less labour-intense way of achieving the goal.

I have a couple hundred of these to get through, and each has a different shape

Lee

Sent from my SM-N960U using Tapatalk

Hi again Lee:
My issue with using any holding system that puts pressure directly on the part is twofold:
First, as soon as the subject part is not rectilinear with two parallel sides, or at least one flat surface, the ability to clamp it securely is impaired.
Second is that with the typical clamping forces needed, the part tends to distort and also to be pushed out of position, often enough that both the shape of the part you wish to work on is not as you pretend, and also that you cannot get it into the position you intend.

So soft jaws of any kind are fraught with issues...they may or may not be relevant to your machining problem, depending of course on what you want to do and to what precision you want to do it.

So I'm a fan of potting these kinds of parts passively, and I come to that conclusion having confronted the problem many times, and tried many approaches.

Moving on to locating the part:
You imply they have all been scanned because you said you have STL models of their geometry.
That is a big benefit for you to establish their location relative to your work co-ordinate system.
What I propose is a 3D printed jig for each part, made from the STL geometry that you can slap onto the part so it can locate itself in one orientation only.

You use that temporarily attached jig to position the part in space, and for that purpose there is a cylinder extending from the three contoured feet of the jig, and a flat you can indicate.
You then mount that jig, together with the sample temporarily glued to it. in a device (a manual mill with a quill is a good device) that can let you pot it in a known position in an open topped rectangular box. so you can suspend the part in the box and then introduce your potting medium, or alternatively pour in your potting and then lower your part into it before the potting sets.

Once your potting has set, you grab the sides of the box in the vise in the milling center and then indicate the position of the cylindrical stem.
You pick up the Z height of the top of the cylinder, and then you break off the jig.

Now you can mill it knowing exactly where it is (within the accuracy of the 3D print, the glueing protocol; and the potting protocol.)

Cheers

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

Of course, another way is to simply slap your part (or a bunch of parts) into the potting box in any orientation, then put the whole works on a CMM and map where all it is; generating a point cloud for both the part(s) and the box.
You can then re-orient your STL model(s) to conform to your interrogation, put a model of the box into the STL model of the part, write your code, pick up the position of the box in the VMC and go to town

leeko said:

HI have a couple hundred of these to get through, and each has a different shape

Click to expand...

So if each one has a different shape AND you have an STL, that implies that each one is scanned?

The right way to do this would be to scan it in the vise/fixture out of the machine, then you just need to locate the (scanned) corner of that vise/fixture. Even the cheap 3D scanners are pretty good these days.

What if you bolt the soft jaws to the vice Then machine the jaws so just enough features protrude a couple of mm that can position the workpiece
Then lightly clamp the workpiece or use a spring load and fill the open space in the jaws with potting material
This way it is all done on one machine in one set up

Peter

Can you give us a quick picture of what you're talking about? I'm having difficulty visualizing it.

If possible I try to allow for some kind of indexing geometry on oddball shapes. Whether it is a through bore in an area that gets machined away, or dowel pins that will locate it on a fixture using vacuum or superglue, etc.

On some parts I have even added additional stock just so I have somewhere to tell the CAM where my WCS zero is and I can indicate/probe it in.