View Full Version : How can we “machine all over” efficiently?

01-22-2006, 06:26 PM
I’m particularly interested in learning more about jigs, techniques, suggestions, concepts etc. for making objects that need machining from every side. I’m thinking of objects that are pretty small in every direction – say perhaps up to a foot cube.

Maybe this initial concept for a jig to be used with the ShopBot indexer will get the idea across. My interest is aimed in particular at making 3D prototyping easier.

I’ve been doing some work with powerful 3D CAD (Solidworks) integrated with powerful 3D CAM (SprutCAM) and am beginning to think the following:

with suitable jigs and methodology, it should be possible to set up a combined hardware/software general approach that will let us machine every side of a block within the size of our system, and in a reliable and efficient manner.

I haven’t been using an indexer yet or any particularly fancy jigs – but if a good concept for a jig and methodology comes up it would be easier to take advantage of it by planning the machining process by putting a model of our required 3D object inside a model of the special jig I’m looking for and planning from there.

Anyway here’s a “starter for 10” concept jig for the ShopBot indexer....


1) Prepare a volume with a cylindrical peg to load in the jaws of the indexer jig. (Perhaps I should have made this model of a die without the rounded corners and added the rounds when the holes were made in each face!)


2) Machine the first face.


3) Rotate to present the second face.


4) Machine the second face.


5) Rotate to present the third face.


6) Machine the third face.


7) Rotate to present the fourth face.


8) Machine the fourth face.

So far nothing out of the ordinary. Support at the opposite end could be used with the standard ShopBot tailstock – with a custom-made form to meet the part. For example, if we had machined a flat on the face opposite the cylindrical peg when we prepared the volume to load in the jaws of the indexer jig, we could use a flat-fronted indexer tail live center.


9) Next we open the hinge in this 1st part of the jig. Our indexer is holding this hinge axis in a horizontal plane.


10) I envisage over-opening it i.e. beyond 90 degrees, placing a machined “spacer” of the required length on the indexer rail, setting the hinged part down on the machined top of this spacer and clamping.


11) Machine the fifth face.

I’m not happy with the difference in height (Z) between the machining of this face and the first four. But the important thing is...we know where it is! By that I mean that the file or files prepared to machine face 5 will be (or at least should be – I wonder what sort of tolerance we could expect from a beast like this?) at a known 3D offset from the files used to machine the first four faces.

Theoretically, with a jig of this nature, you could prepare a ShopBot “master” file that runs the entire machining of the 6 faces (at least “per cutter”) and the master file would call the individual files to run, with their 2d or 3D offset, and probably move the gantry out of the way and wait for a keyboard input before continuing – an input that you’d give when you’d made the required modification to the jig set-up.


12) Remove the (invisible) support and...


13) Close the hinge. There would be a locking device (not modelled) to keep this closed securely as required!


14) Bring in the 2nd part of the jig by sliding it along the indexer T-slot. In this model I put this 2nd part of the jig on another indexer. In the way we’re using it, it could just as well have been without the indexer! Anyway...

Because we prepared our entire machining strategy in a 3D CAD software where we worked with both the object we are planning to make and the jig set-up we’re using – including the relative positioning between jigs and part – we were able to prepare made-to-measure form-fitting surfaces to put in the clamp jaws of the 2nd part of the jig.

This might seem difficult, but in a good 3D CAD system, working with a standard procedure, the steps to design these made-to-measure surfaces can be made pretty efficient. We can set our “virtual clamp jaws” where we like – simulating what we will later do - extrude the base form up to and into the object, then cut it back by what can be called a “boolean subtract” (but different 3D CAD systems typically call it different things), and then plan the CAM to cut our set of made-to-measure form-fitting surfaces.


15) Bring it right in – our design of clamp surfaces could either by so that we simply push the 2nd part of the jig right up to meet the object (i.e. the surfaces will be supporting the object from underneath when we open the 2nd hinge) or we move the jig precisely to a known distance from the other jig (known because we modelled it all together) for clamp surfaces that will not be supporting underneath.


16) This example is the second of those types mentioned above. The clamp surfaces do not supprt from 100% underneath. In this image the 2nd part of the jig has been moved to the precise place required, but the clamp surfaces are not yet engaged. The 2nd part of the jig is then secured (i.e. so it can no longer slide along the indexer T-slot rail).


17) Now the clamp surfaces are engaged. This concept model uses 2 perpendicular sets of clamps who’s faces are always the same distance from a central plane. It may be preferable to use a type of clamping that rotates in rather than moving in linearly.


18) And with the 1st part of the jig unattached and moved away, we can open the hinge on the 2nd part of the jig.


19) Note the invisible support placed manually between the indexer aluminum rail and the underside of the jig.


20) Remove the cylindrical peg we used to load the part in the jaws of the indexer jig.


21) Machine the sixth face.


22) Close the jig...if the model requires more than one cutter we’ll have to do more work – and we’ve already cut off our cylindrical peg! Looks like we’ll have had to organise our cutting “per cutter” for faces one to five together, and then “per cutter” for the sixth face after the “hand-over” to the 2nd part of the jig.


23) Remove the finished part


24) Leave everything and go and make a cup of tea. Then start measuring to find out if its within tolerance and visually inspect it to find out if you’ll be able to hide the imperfections by hand-finishing.

01-23-2006, 09:34 AM
Chris, Looks way cool but a bit exotic. Have you considered a vacuum hold down fixture to replace the tail stock? I could envision a precision centered fixture much like your step 14 above as the primary part holder with a good vacuum on center in that jig as the tail stock. It probably would need a precision frame/ pocket in it to set the part into horizontally like your #14. You could then use the same surface as your Z Zero on all cuts and move the block anytime you need to by simply machining and rotating 90 degrees for all 4 sides, then release vacuum, remove and rotate the block 90 degrees. The 2 virgin sides could be rotated 180 degrees on the indexer to machine top and bottom of sides # 5 and 6.
With this method you could even use the center of the block as your Z zero, split the tolerance concerns, and offset from there to cut the part. Then you would be absolutely sure all sides are dimensioned from the same datum at the centerline.
A decent vacuum should hold and a precision pocket properly centered to hold the part consistently and accurately seems like it would be easier and quicker with fewer mechanical parts and dimensions to line up. Might still need the peg or square extension to insert into the vacuum tail stock but I don't see why with a properly fitted vacuum. Seems it would take a 24 step process down to 4 rotations, remove/replace, then 2 more.
Hope this makes sense.
Just a thought.

01-23-2006, 10:17 AM
Chris, I think Jerry was putting it mildly when he said it looks a bit exotic.

The hinging plate will be a headache to get exactly to 90 degrees and have no play in the hinge pin. It is for this reason that machine-shops typically use a rigid L-plate (right-angle plate) bolted to a faceplate. This L-plate can be faced by the milling cutter, and reference ridges can be left on it (or dowels drilled, etc). The workpiece is clamped/unclamped from the L-plate 2 or 3 times and the reference edges (dowels) are used to relocate it every time.

01-24-2006, 01:50 AM
There is a very simple and straight forward way to do this, as I posted elsewhere. I tried to link to the old post, the link will not work
It is a short explanation. If it is not understood, I can make diagrams. It is also a method, with a little adaption, that would allow 3D carving (or at least 2-1/2D
) on 5 of the six sides of a cube.
It requires no complicated jig. It does need a 'square' ShopBot, but any method needs that.


01-24-2006, 01:51 AM
I have copied the post mentioned above here.

To make a cube (assuming your shopbot is square)
Hold your wood on the table and even off the top surface.
Flip over and level off the under surface to the cube thickness.
Machine along one edge to the depth of the bit.
turn the wood over 180 degrees, use your machined edge against a strip of wood on the table and machine along the opposite edge, so both edge machined parts are level.
using two equally thick strips of wood on the table, turn your block over, position your machined edges on these strips.
Now level off the 3rd face.
Flip it over and machine the fourth side to the thickness of the cube
Assuming you have been machining everything in the x direction, Now machine the edge in the y direction.
Flip 180 degrees and do as before along the opposite edge
Sit again on the two strips, Machine surface 5.
Flip 180 and machine the final surface.
This works with any old lump of wood, providing you can hold it to make the first surface.

01-24-2006, 03:53 AM
This is great feedback so far for me to ponder.

Jerry - I hadn't considered a vacuum hold-down to replace the tail stock.

Certainly the concept model has the inconvenience that if you used a tail stock as I suggest in step 8 above - although it would actually be used starting from step 1 - then it would have to be removed from the indexer T-slot before step 14 and the exotic jig put into the the indexer T-slot. And thats exactly the sort of procedure I'd like to minimize.

There are some things I'm not clear about with using vacuum. Can it be used effectively on pretty rough surfaces?

I mentioned in step 1 that perhaps I should have made this model of a die without the rounded corners and added the rounds when the holes were made in each face. More useful still would be to start with a seriously unsmooth and unsquare form all over except for the cylindrical peg to load in the jaws of the indexer jig.

I also said "...general approach that will let us machine every side of a block within the size of our system" and I that together with the fact I chose a die as the model may have led Mike to think I want a jig to make cubes!

Block was not a good choice of word on my behalf:

I want a jig (or kit of separate bits plates, dowels, etc.)that can be used together with a standard procedure or methodology so that any 3D form that is "prismatic" from all sides (no undercuts)and fits within the X x Y x Z volume of our jig (or kit...) (and e.g. 300mm or 12" cube) can be made by following the procedure that will machine as necessary from each of the 6 sides to leave the desired form.

I would plan to use this jig and procedure for a part e.g. the shape of a rugby ball but e.g. 8" long, and this same jig and methodology for a part that is shaped like e.g. a scale model of an electric kettle and 6" long and also to make e.g. a 1/2 scale model of Mozart's head with his famously wavey hair. All with the same jig and methodology.

If the 3D model fits inside the virtual version of the jig (the one in the software matching the real one used) we should be able to make the part.

And I expect to position the centre of the model in the centre of the volume.

So I think Jerry has the idea but I haven't quite got my head around the details (but "you could even use the center of the block as your Z zero, split the tolerance concerns, and offset from there to cut the part" and "you would be absolutely sure all sides are dimensioned from the same datum at the centerline" sound good to me!

Gerald - I'm pretty sure your experience leaves you in position where you can't really see the point for such a jig. If someone showed you a shape you'd instantly have an idea for machining it from start to finish. A bit of paper to make notes about e.g. offsets on and you'd positively enjoy the challenge of sorting it out and doing it.

Compared with that, in a way my jig would be a "3D for Dummies" jig. No need to engage brain on a job-by-job and task-by-task basis. If the shape fits in the volume, follow the procedure.

My "Hinges" are supposed to be dowels forced to be used as reliable axes of rotation. And the "spacer" or "invisible supports" required in steps 10 and 19 above are supposed to be machine flat both ends. Maybe there's a contradiction in wanting the "hinge" to both allow rotation and serve as a precision locator.

And Mike, thanks for the procedure. It's a great procedure. Its exactly the sort of logical, methodical, trustworthy procedure or methodology I want to be using in a more "3D" way with the help of the jig I'm searching for and powerful 3D design and toolpathing software.

01-25-2006, 09:00 AM
Even as much as a 5 sided cutting jig would help out without use of the indexer.
I see something a bit different for the overhead bots.

Great idea... got me thinking.
Wish I had more time to play.

01-26-2006, 09:06 AM
Chris, While I'm no expert I would definately say the approach can be simplified.
As far as vacuuming a rough surface, not likely. you would need to start by establishing a smooth surface and perhaps even sealing it to get the good vacuum seal on the tail stock. Large bowl turners use vacuum systems on lathes so one way might be to bolt a smooth plate to the stock and later cut it off like a bowl turner would do. Or simply machine one surface to prep it then mount it.
IF you had this in place and wanted the extra measure of stability, I think you could mount a stable block or spacer below it for a place holder or maybe just a vacuum puck below the piece as well depending on the shape.
There are companies that make flexible vacuum suction cups for many automated industrial applications. Check the web and I'm sure you'll find them for pick and place machines or vacuum hold downs.
As far as your center Z Zero, What I was thinking was you could set your Z Zero at the exact center of the indexer ie; the centerline of head and tail stock as your base for all cut files.
Then create an offset in the cut file on your material set up from the top or use the bottom of the material set up as your starting Z machine surface (at that center line)rather than the top.
Then your cut file essentially recognizes the centerline as the starting z zero, moves up to the proper Home Z cut area plus its safe Z, and starts to cut. The benefit is no matter what tolerance you had for allignment and Z zero as you rotate the piece to different sides, it would always be registered from the center of it. Thereby splitting your tolerance in half compared to registering and re zeroing from the top of the block on every side. Hope that explanation makes sense.
As always my opinion is worth what you paid for it, so always pilot new and experimental techniques on a dry/air run before mounting the stock or starting the spindle.
Can't wait to see photos of the final creation.

01-28-2006, 04:58 AM
Jerry, Thanks for that input. I didn't know some large bowl turners use vacuum systems - interesting. But I'm pretty sure I won't use vacuum when I get round to making this "machine all over" jig.

There are some elements of the "starter for 10" concept jig I detailed above that I'm happy with and others I'm not.

I am happy with:

i) It will make a part by machining it from the six faces of a cube (but I'm open to other suggestions).

(to be rigourous, it's the concept of machining in both opposite directions from the three perpendicular planes That I'm happy with - so it could be rectangular on one or more faces. But at present I expect I'll end up making one that machines a cubic volume).

ii) The geometric center of the part will be positioned at the geometric center of the cube. In this way we get in practice the "split the tolerance" built in to the methodology.

I tend to see it in terms of 6 identical Z-zeroes, each one at the surface of a face of the cube: If the center of the part was NOT at the center of the cube, then the machining depth (-Z) on one face would be deeper than it needed to be (compared with if the part had in fact been centered in the cube), and we will get better results if we machine less depth. But on the scale we're discussing, it's not going to amount to much!

iii) The approach used is the same from every direction.

iv) There is one simple initial preparation of the material for initial part-holding and one "hand-over" to a set of "made-to-measure form-fitting surfaces" that we have put in the clamp jaws of the 2nd part of the jig. It might seem strange to be happy with such a complex plan, but to machine a shape on all six sides there is no escape from at some point holding it by an already machined side.

I am not very happy with:

i) Machining far off the centerline of the indexer. Maybe with a chunkier indexer than the standard Shopbot one for more holding power, maybe if the concept used 2 face-to-face standard Shopbot indexers to increase holding power, and maybe some other solution!

ii)The difference in height between the machining of the first 4 faces and the other two.
Ideally for me all six faces would be presented in exectly the same place at exactly the same height!

But having been through the exercise of making the concept model (and a quick thanks here to Gene Marshall for giving me the incentive to get the concept out of my head and to its first development destination: the WRONG SIDE of the computer monitor) there are some ideas being bandied about, mulled over, both in waking time and asleep, that I think will result in something I'll be sufficiently happy with to actually make.

For now, discussion and proposals is perfect. And if anyone has a concept but is holding back because they think they'd have to present it the way I did at the start of the thread, well no! Just put it in words. If it helps it helps. And if its amazing I'll take care of making the 3D CAD model and posting it.

01-28-2006, 03:00 PM
Hi Chris,
That's a very fancy fixture design, and I'm sure you could make it work, but I'm thinking that your biggest problem is going to be the setting of datums (X/Y/Z).
As soon as you start to open your hinge mechanism you will lose your currently set datum in at least X & Z, but if the fixture is rotated, you will lose it in Y as well.
If you are designing and making this fixture from scratch, have you thought of using a 'trunnion' style fixture instead? Using this type of setup you can use a single datum for X/Y & Z and simply rotate the model in SprutCAM to get your 5 axis positioning.
I'm going to try and upload a picture of what I mean, if it doesn't work I'll e-mail it to you :o)


01-29-2006, 11:19 AM
I feel I must be missing something here, because you appear to be making something very simple, very complicated, or my thinking is not right.
If you are simply looking to make a rectangular block, then this to me seems the simple solution.


Take your piece of timber, whatever shape, hold it by clamping or screwing, or both, and machine off the top surface. Now you have one flat surface.

Now flip it over and machine the other surface, to the thickness of the rectangle. For ease of explanation let us assume that the x axis goes left to right in these picture, y away from you, z from top to bottom.

Up to this point there is no x or y direction of the block. Now machine away one side of the wood to the depth of your bit. remove the block and , without moving x, machine a groove into the spoil board. Place a prepared straight edge (green) into the groove, flip the block, align with the straight edge, and remove the upper corner.

Take two (blue) prepared spacers exactly the same thickness, aligned acurately with y. Mount your block as shown, machine flat the surface. Invert the block, aligning the already machined surface one with x, machine the top surface to the desired thickness, and repeat as earlier by cutting away the corner.

Now repeat the earlier steps to machine flat the final two surfaces.
My drawings aren't perfect (some bits are not machined) but the principal is correct.
If the above threads suggest you are looking at ways to machine 2-1/2D, then you can follow precisely the same principle by making your rectangle longer than needed, machining sides 1,2,3,4, then create flat sides 5 and six.
2 -1/2D machining of side 5 is straight forward,
With any jig, holding an uneven shape to machine the last side is always going to be difficult.

Not only does this require no jig at all, only a straight edge to fit in a groove, and a couple of suitable spacers, it also produces a block to the accuracy of the ShopBot.

Now explain to me where I have gone wrong

(I am aware that it is possible to remove two of the flip moves, but for ease of explanation I have shown it as above)


01-29-2006, 12:24 PM
Mike, I don't think you have gone wrong as far as machining a cube goes (although I don't machine a groove and put a green stop in, I machine a very wide groove and put the workpiece in and push it against the side of the slot).

But Chis said he wants to machine "any 3D form that is "prismatic" from all sides (no undercuts)"


"I would plan to use this jig and procedure for a part e.g. the shape of a rugby ball but e.g. 8" long, and this same jig and methodology for a part that is shaped like e.g. a scale model of an electric kettle and 6" long and also to make e.g. a 1/2 scale model of Mozart's head with his famously wavey hair."

Actually your first drwg looks a bit like Mozart's head, and no doubt he would be starting with cuboid stock - perhaps if he used your method backwards....


01-29-2006, 01:26 PM
as far as making cubes is concerned you haven't gone wrong at all - and that last post should help people follow your description in words of your logical, methodical, trustworthy procedure to make a cube with the ShopBot.

I like that shape you started with! That's the sort of thing, at least in terms of needing machined all-over to make, that I'd like to be able to make with my jig.

And you've really hit the nail on the head with your comment "With any jig, holding an uneven shape to machine the last side is always going to be difficult."

In the concept jig, the exotic bit that comes in at stage 14 is meant to be the bit that handles that difficult task. And the clamping surfaces have been prepared beforehand, made to measure and based on geometry we have.

I know its a tall order. But it could be done.

Dave P.'s 'trunnion' style fixture looks like this...


...and it's a great idea. I'll probably model a trunnion style fixture with standard ShopBot Indexers and see how close I can get to a concept model that will present all 6 faces to be machined in exactly the same place and at exactly the same height.

This is looking promising. Dave P. is my SprutCAM "reseller", and when he says "simply rotate the model in SprutCAM to get your 5 axis positioning" that's (Mozart's) music to my ears...because I've got SprutCAM (and it post-processes directly to .sbp)

Now I remember reading in Bill's Corner about the 5th axis on the camera and the fact that the ShopBot software can handle the 5th axis. Is it ready to handle a trunnion style fixture? could it be that my only missing ingredient for "machining all over" is the jig, since both SprutCAM and the ShopBot software are both ready and waiting with this ability already built in?