Looks Like I Joined the Club

[verhoevenc]

Hmmmm... This is news to me. Guess I'll have to give this a go!

Chris

[verhoevenc]

Sorry for the slow responses... life has been busy. I ran these tests awhile ago but am just now getting around to posting them up.

In the below when I use the term "offset" I mean the line, in all directions, was moved inward or outward. Therefore, the width of a neck, or pocket, would be double the offset amount.

I cut 3 necks:

• One full sized neck out of pine 2x4 to exact width, no "final pass"
• One .010" undersized offset pine neck, no "final pass"
• One full sized maple neck milled .010" offset oversized and then finalized to exact width in one final full-depth pass

I then took some scrap ash and cut the following routes:

• .010" offset undersized : Nothing fit this pocket
• Enlarged to a full-width pocket in a single finishing pass: Only the under-sized pine neck fit
• Enlarged to a .010" offset oversized pocket in a single finishing pass:
  • The maple neck fits in and I can slip two thicknesses of standard weight paper in next to it
  • The undersized pine neck fits, but has TONS of slop and wobble as expected. 4 thicknesses of standard weight paper seems to be the slop amount here
  • The standard sized pine neck also fits decently well, only 1 thickness of paper will slip in there

I mic'd the paper at .0035" thick. This leads me to, I believe, the following answers:

• The "final pass" concept leads to a better fit, but it's not perfect still.
• Since the maple neck and final pocket both had finishing passes and can still fit 2 paper thicknesses... This means, to me, on each side of the next I left .010" extra and can fit a .0035" paper in. So .0065 of over-tightness/undercutting. This undercutting is shared by two parts, so a total of .00325" per cut.

To me 3.25" thousandths seems like a fairly decent amount of undercutting?! If anything I'd expect OVER-cutting since this is a LONG (read 2" cutting length) 1/4 endmill. I'd have expected some deflection there. This is especially annoying as the Shopbot desktop claims a resolution of .25" thousandths, a whole 3 thou better than I'm apparently getting.

Thoughts?

Chris

Note: This was an extremely interesting read. Apparently climb cuts (what I've been using almost exclusively) are prone to undercutting. Meanwhile, conventional passes do the opposite, the pull the bit in an over-cut: http://www.talkshopbot.com/forum/showthread.php?22362-Parts-0-050-quot-to-big-in-X-and-Y&highlight=undersized

[Prostheta]

That 0.25/1000ths will simply be stepper/leads crew resolution. I doubt that translates to the cutter. Still, that's just me being pedantic.

Slop in the mechanism and climb cutting will undersize from the cutter trying to drag itself into the work.

[curtisa]

If the cut were being executed perfectly I'd expect nothing to fit together if the dimensions of the neck and pocket were exactly the same. There needs to be some give somewhere - either the neck needs to be a shade smaller or the pocket a shade bigger. How much difference will depend on what is achievable for your machine, how the material behaves when cut, the amount of flex in the cutter, the amount of backlash and slop in the axes, how true each axis' motion is etc...In all but the most expensive machines I would have thought that this would be something that could only be determined by trial and error.

There's a video I saw on Youtube recently by the Fender Custom Shop. They were explaining the reasons why they use CNC for their bodies and necks. At one point the guy demonstrates the supposed exacting nature of fitting a CNC'ed neck into a CNC'ed pocket. The neck was such a poor fit that it would fall out in a stiff breeze. It's entirely possible that you'll get far better results if you CNC the neck and use more "traditional" methods to route the pocket by hand.

[Prostheta]

I'd like to see that!

If the CNC were perfect, and parts came out to absolute perfection you'd still need a degree of intentional slop even for an interference fit. When I do furniture work, an enclosed mortice/tenon joint needs slop just to allow the air to escape. My Roubo workbench had a problem with a leg fitting because I made the joinery too well. Once glue was added into the joint, the trapped air couldn't escape even with four pipe clamps trying to close it up. I almost considered drilling a hole through the other side to relieve the pressure. Absolute insanity, but I learnt from it.

That said, being able to dial in exactly what you want and have parts come out as you expect is what we're looking at here. Whether knowing your machine's specific weirdnesses is what you have to deal with, or whether they are problems that can be rattled out is probably common at this price point for CNCs. This is why that 0,25/1000th" precision sort of bothers me. It's a number that likely cannot be met in reality. This is entirely the difference between precision and tolerance. Being able to aim precisely at a 0,25/1000ths point is undermined if you have a tolerance of +/-10/1000ths because of slop or whatever the root cause is.

[Prostheta]

Blatant excuse for a shop-shot, however this is what reminds me of that lesson. It was literally like trying to push a hydraulic ram, and just bounced around....

[demonx]

I think you'll find that even though offsets are way too small once you start allowing for harder timbers and then spraying paint etc

I did all my original mockups in pine and then found out I had to change a lot of things once going to hardwoods. Not only that, but your speeds, cut depths, overlaps and everything will be different in pine vs hardwood.

 

[verhoevenc]

I feel ya @demonx. I was playing with inlays yesterday... Cut a piece out of some Pickguard material and it dropped right into a cavity of the exact same size in ash. Then I went to go do some real inlay...

I just the pieces out full size and then cut out the pockets with incrementally smaller and smaller 'versions' of a 1/16" bit. By that I mean everything was cut with a true 1/16" bit, but I would set up the CAM to think it was 1 thou smaller, then 2, etc. so the bit would cut a larger pocket. Ended up needing to tell the machine the bit was 10 thou smaller than it is before the pieces would go into their pockets in wedge!!!! 10 thou! It was 0 thou in ash..

Never would have seen a lot of this stuff coming before I got in there and started testing.

Chris

[Prostheta]

What is the reasoning for harder woods requiring different offsets? That they do (and consistently so) would indicate there being one. Inquiring minds need to know! Is this down to chatter in the cut edges creating interference in the mating? Kind of like the ripples left in wood if you throw it through a thicknesser at light speed? I don't see how this would happen with shearing cut endmills since they're in constant contact. Just thinking out loud....

[ScottR]

How about bit deflection? 1/16" is a pretty narrow bit and I imagine each different hardwood machines differently. As you say, certainly they plane differently.

SR

 

[demonx]

On 6/6/2016 at 5:32 AM, Prostheta said:

What is the reasoning for harder woods requiring different offsets?

It is probably a combination of what you said and probably more to the point something soft will "help" itself in where something hard wont have the same amount of give. For example it'd be a LOT easier to shove a 10mm crayon into a 10mm hole than a 10mm drill bit. (To use an exaggerated example)

[verhoevenc]

So I finally made my 3D CAM decision. Although MeshCAM didn't have every feature I would have liked it to... It had enough and was extremely friendly to use! In contrast, my trial with CAMBAM just had me ripping my hair out in clumps.

MeshCAM was so easy I didn't even bother buying CutViewer with it and have run code on active builds with just looking at the paths in MeshCAM.

Chris

[10pizza]

That's a really nice toy you have there Chris!

I'm sure there's a lot of time going into preparing the models, but once you learn how the machine reacts to your models and setup. But each next one will be easier I guess.