2024-10-11

Last modified: 2024-10-14 14:02:13

Sovereign Scales

I'm following the plan from 2024-10-08.

The "vice stop" is a 5mm rod pushed all the way into the bottom of a hole in the table surface, and then the brass pushed all the way up against it until it won't lean any more. I expect this would wear the hole quite quickly but should be OK for now. Pointy-end down.

Also the coolant pump is making fizzing noises, I think I need a more powerful pump.

I started running top-1 at 10:01.

surface-top takes about 4 mins.

I blundered the first one, because I set X=-5.5 when the 5mm rod would slip between the tool and the rough-cut end of the work - but I should have set X=-8.5 because it has a 3mm radius.

The pocketing toolpath leaves a nub in the centre of the half sovereign and quarter sovereign slots because the tool only cuts at the circumference.

The rectangle finishing cut could happen in 2x 5.5mm steps instead of 2x 5.0mm and 1x 1.0mm.

top-1 took 13 minutes in total.

After the first top-1 it parked at X=-7.5, I updated this to say X=-10.5 so that the others will be right.

After top-1 it should move the tool back and up so that you can more easily work with the vice.

Result of top-1:

There are some minor chatter marks on the edges, I wonder if a ballnut holder is coming loose or something. Ignore for today.

Now 2 more copies of this, but with X offset corrected.

Moving the X coordinate by 3mm did not work very well, we lost about 4.5mm of material from the other end of the part. Is my vice stop just super not repeatable at all? Or do I need to cut the material longer? Or review the gcode? I moved it back by 1mm for the third part but it seemed basically the same as the second.

OK, got 3 copies of top-1. It's 10:44 now. On to top-2!

So changes in gcode/plan for top-1 are:

1. make the rectangle finishing cut more efficient
2. fix the nub in the centre from the pocketing toolpath (maybe just adjust stepover so that it clears it out? and run a fixup job on these parts)
3. move the tool back after each job so that you can work with the vice/table
4. make a separate version of top-2 without surface-top included, to save time on the first top-2
5. take the machine apart and Loctite all the fasteners
6. remove first lap of counterweight because it doesn't cut anything
7. machine down the right-hand edge all the way to the bottom, to use it as a reference
8. run a deburring toolpath
9. work out why the coordinate offset didn't work properly.

Pugsender idea: "Turbo" button. You just hold it down during parts you know can go faster and it runs at 200% feed until you let go.

For top-2 we want shorter parallels, we only need 4mm or so of the actual part material protruding above the jaws (plus the excess removed by surface-top.

Since my rectangle cut didn't go all the way to the bottom, it's inconvenient to zero against the left-hand edge. Instead I'll touch off against the right-hand edge. Measuring another part, I get 147.29 mm for the overall length. Is that believable? Check CAD. Should be 147.0 mm, but we leave on 0.5mm for material to remove, so it looks like it's made it too small by 0.2mm?

The width measures about 15.04mm, which is more like what you'd expect. The length of the actual part measures 119.7mm, which is 0.3mm too small. Maybe something is wrong with the X axis? Check it another time.

I'll just touch off the right-hand edge and split the difference on the error.

So this X coordinate is 147mm for the length of the two parts, plus 0.5mm for the excess we left on, plus 5mm for the rod I used to touch off, plus 3mm for the tool radius, minus 0.1mm for the X error = 155.4mm.

And then does that look like it has X=0 at about the right place? Looks close!

Either there is something badly wrong with the X axis, or I made a blunder while touching off Z in surface-top and accidentally moved X instead. After completing surface-top, X=155.4 is nowhere near where it was before.

This could explain why my X offsets kept getting confused in top-1. Maybe the rapid during surface-top makes the X axis slip? Will see if I can tighten up the coupler. I tightened it slightly but it didn't feel like it should be slipping.

I've parked at X0Y0Z10, ready to run top-2, it's 11:23. So it took 43 mins to run 3 lots of top-1, and then 39 mins to faff around preparing for top-2.

14:26, I'm back.

top-2 takes about 6 mins to run excluding surface-top. So about 10 mins including.

After running surface-top on the 2nd part, the thickness is only 12.7mm instead of 13.0mm! What has happened there?? Maybe a burr sitting against the parallels? Bad. 3rd one is 13.1mm, OK.

The surface finish on the edges is looking a bit better now, is that just because I tightened up the X coupler?? It didn't even seem loose.

All 3 top-2 done at 14:58, so 32 mins for 3.

1. make a separate surface-top job that just cuts directly at Z=0, for fine-tuning the thickness
2. find the thickness by running a really short job at Z=1, waiting until that gives a 14mm thickness, then going straight into top-2 with integrated surface-top
3. remove the last lap of the counterweight, on the second side only, because the material is thinner now

I can't immediately make an engraving toolpath because Inkscape isn't working (Nvidia problem?). But I have a deburring toolpath to run, and I can probably delay engraving until after all the parts are separated if necessary.

Oops! I ran the first deburring toolpath without taking the part out of the vice, but since I had just changed the coordinate system by 0.5mm, that is wrong.

There seemed to be tiny sparks at the tip of the tool, is that because of the fizzing in the coolant? I've switched the coolant pump off this time. Nope, still there. Should I be worried about this? Maybe worth checking the spindle housing is earthed.

My vice stop system is not repeatable enough to put the deburring toolpath in the right place. This could be an issue.

Well I've possibly ruined 2 of the parts by trying to deburr them. I'll stop now and move on to slots.

Current mood:

I think positional accuracy is not super important for the slots.

Slots toolpath needs to be a lot more efficient, the majority of the motion is just in the air where the coin would be.

Tool broke instantly on contact with the work :(.

Last time I used 1mm tools I think I found that the 5mm collet worked better even though the shank is 4mm. I'll try that.

The 5mm collet seems to have done the trick. Perhaps I should just buy a new 4mm collet.

No sparks from the slotting tool. Maybe it's something to do with the surface coating on the v-bit? Or just an optical illusion because of how it cuts?

Started slotting path about 16:26, finished 16:40, so this one takes 14 mins!

We should fix this so that it doesn't spend so much time cutting air.

It might be better to run the slotting toolpath from the bottom, that way you preserve the shape of the top surface and only run half as many laps.

I've spent about 4 hours on this today, and got as far as doing top-1 and top-2 for 6 parts, and slots for 2.

If top-1 is 13 mins and top-2 is 10 mins, then let's say 11.5 mins for the top-*, and 14 mins for slots. Surprising that slots takes longer considering it is such a minor feature!

So I've done 6*11.5 + 2*14 = 97 mins of actual machining, in 240 mins of wallclock time. The rest is spent on faffing, setup, CAM, etc.

I've parked at X0Y0Z10

I'm thinking that actually I should go back to making these out of sheet material. That way I can tile the table surface with them and not have to worry about coordinate systems or even loading the vice.

Hmm, the thickest brass sheet on https://www.rapidmetals.co.uk/ is only 3mm. 3mm would work if I wanted to use a bolt-on counterweight, but I'm thinking more like 10mm if I want an integrated counterweight.