Yes the spindle is one of those cheap ones with the grub screw collet holder. I guess that kind of explains the run-out issue…
With regards to the belts I’ve got them quite tensioned (no way to quantify though…) Any suggestions on belt tension in general ?
If it sounds like the high note on a base it’s to tight
1 Like
It’s hard to do a “how tight” because everyone’s machine is slightly different, or a lot different. Like Tim said, if you can play musical notes, it’s too tight.
It depends on the length of the belt… my 3D printer is a higher pitch than my low rider. There is a phone app called easy tension that lets you input the belt length and then use your microphone on your phone to measure the tension.
3 Likes
Aight, so I relaxed the belts a bit (used a small torque drive to make sure they are consistent). Results start looking better now. Tried a project with Acrylic today and dimensions were +/- 0.2mm.
What I noticed was that in general external dimensions are oversize and internal (pocket) dimensions are undersized.
For example my part today had:
Nominal external width dimension 50mm, was measured 50.2mm
Nominal internal pocket width dimension 20mm, was measured 19.8mm
That suggests that there is flex in the whole gantry system? then again I am using a repeated finish pass option on Fusion 360, which in theory should take care of that right ?.. these are my assumptions… feel free to point out where I’m wrong 
In fusion 360, make sure you don’t have Stock to leave enabled. I think it defaults to 0.1mm.
1 Like
I am having a hard time thinking of what it could be if not the fusion settings. You could also try configuring probing and measure the accuracy against a 1-2-3 block using the bit as a probe with the router off. There is a backlash compensation feature that uses this m425/g425
Thanks guys!
I can confirm Stock to leave is definitely turned off during the finishing pass.
Blockquote
You could also try configuring probing and measure the accuracy against a 1-2-3 block using the bit as a probe with the router off. There is a backlash compensation feature that uses this m425/g425
I am not 100% sure what you mean here my friend… can you give me a bit more detail please ?
Being slightly over on an outside dimension is a new one, if anything it should be slightly under. Double check your bit diameter maybe. how are the diagonals, holes round? Spindle trammed?
Thanks for the suggestions Ryan! (btw your machine rocks! full stop!)
Bit diameter is questionable yes - I need to find a reliable way to measure.
Spindle is trammed indeed. Got 2.5 thau “out of vertical” over ~200mm diameter circle.
The machine is pretty square (using the dual endstops to compensate for hardware out-of-squareness). Using a 250mm pen drawn square, I cannot tell a difference in the diagonals using a standard ruler.
I’m machining another part now. Will let you know of the dimensions.
Measuring “truth” can be a real rabbit hole if you are not sure if the bit diameter and spindle runout, and a single flute is tricky to measure separately.
At a certain point you could just modify the bit diameter in CAM to reflect the effective diameter which in your case is 0.2mm undersize, and as long as it is repeatable you will get good results. Nevermind whether it’s runout or diameter.
5 Likes
To this end, I considered just running a single slot path in some material and measuring it, then calling that my tool diameter. Would that basically be the same thing for practical purposes?
interesting stuff! so I was cutting with (what I thought was) a 6mm single flute.
I set my calipers to 6mm and the cutter was easily free to rotate within the jaws. I tried to measure exactly and it must be close to 5.8-5.9mm.
I guess that explains more the problems I’m seeing!
2 Likes
It would be close enough. Use something really soft, like a piece of rigid blue foam insulation.
2 Likes
But not something like your arm… 
1 Like
Blockquote
I am not 100% sure what you mean here my friend… can you give me a bit more detail please ?
There is precious little info on the internet. I’ve got my probe setup and in the process of setting up a vise to do more metal work so this is on my radar but haven’t done it yet will post some photos if it works out well.
Here are the manual links:
Update: I actually got it going… here is some G425 V output.
May not be completely accurate as my vise isn’t screwed down and I was just using my mill bit will have to redo it with 1/8" rod. Looks like backlash could be worth correcting for.
Backlash:
Left: 0.12
Right: 0.12
Front: 0.06
Back: 0.03
Top: 0.06
Greg, I think your “left” and “right” backlash numbers are quite high. I posted this a while back:
So my average backlash was only 0.020 - 0.030mm, similar to your “back” measurement.
It looks like your X axis has some mechanical issue, like too much play somewhere or a loose belt or something. Another possibility is that your probe does not make good electrical contact on some sides. Make sure the surfaces are really clean and have no rust etc.
It shouldn’t really matter whether the vise is bolted down. If you have good contact then the force should be very tiny. You need to watch the probe as it touches the vise, it should really barely touch it and not try to move any further.
Remember I am using a four by eight foot lowrider and you are using an MPCNC. I wasn’t all that suprised by the bigger X backlash due to the design… there could be some twisting caused by pulling the heavy platform back and forth from the very edge. Anyways… I’m going to come up with a better vise installation (my carriage is almost all the way up right now because the vise is too high) and probe with a rod and give it a try and see if the number change.
@peter I did some more extensive testing and tried some of your suggestions and always seems to get the same backlash results… if you have time to have a look I started another thread here:
1 Like