Bruce Thompson
04-05-2022, 1:47 PM
Bosch 1618 EVS router with Rockler 1/4" two-flute straight bit; 1/2" shank, in Jessem Rout-R-Lift
Incra LS system router fence with right angle stock holder for making box joints. Accuracy of spacing of cuts 1/1000"
Making small boxes for my travel chess set project with sides 1 1/8" deep, using 1/4" box joints with 4 full cuts and one 1/2 cut per corner, so pins and slots must all be .250" to fit snugly together.
I made a trial cut on two pieces of scrap stock and found woefully sloppy joints. Measured with digital calipers, the slots measured .261" wide and pins .239" wide. A clear sign that there is significant runout at the bit flutes. I set up a dial gauge against the flutes and found .010" difference from one flute to the other as the bit is rotated by hand past the gauge, so there is apparently .005" runout; .005" under at one flute and .005" over at the other.
So, where is the error coming from; the bit, the collet, or the motor shaft? To try to find out, I marked one flute on the bit and one side of the shaft, then loosened the bit and rotated it 60 degrees while holding the shaft in place, and tightened it down again. I did so five times, advancing 60 degrees each time and measuring again with the dial gauge after each re-positioning. The results were as follows, counting the initial position as 0 degrees:
0/360º .010" difference between flutes
060º .008"
120º .0025"
180º .001"
240º .0005"
300º .0025"
So, I succeeded in finding the sweet spot to complete this project, but that makes installing bits for future box joints or dovetail joints a tedious process. I would rather fix the problem, but I don't really know how to define it based on the evidence so far. I think that if the shaft/collet were running true, then the fault would be with the bit only, but that runout should remain constant no matter how it is installed. If it is a combination of bit runout and collet runout happening to combine for the big error, I would have expected the 300º value to be close to its neighbor the 360º value, but instead, it was the same as its opposite 120º position.
What have I missed in my testing regime?
What could explain the odd distribution of test results?
What would the cure be?
Incra LS system router fence with right angle stock holder for making box joints. Accuracy of spacing of cuts 1/1000"
Making small boxes for my travel chess set project with sides 1 1/8" deep, using 1/4" box joints with 4 full cuts and one 1/2 cut per corner, so pins and slots must all be .250" to fit snugly together.
I made a trial cut on two pieces of scrap stock and found woefully sloppy joints. Measured with digital calipers, the slots measured .261" wide and pins .239" wide. A clear sign that there is significant runout at the bit flutes. I set up a dial gauge against the flutes and found .010" difference from one flute to the other as the bit is rotated by hand past the gauge, so there is apparently .005" runout; .005" under at one flute and .005" over at the other.
So, where is the error coming from; the bit, the collet, or the motor shaft? To try to find out, I marked one flute on the bit and one side of the shaft, then loosened the bit and rotated it 60 degrees while holding the shaft in place, and tightened it down again. I did so five times, advancing 60 degrees each time and measuring again with the dial gauge after each re-positioning. The results were as follows, counting the initial position as 0 degrees:
0/360º .010" difference between flutes
060º .008"
120º .0025"
180º .001"
240º .0005"
300º .0025"
So, I succeeded in finding the sweet spot to complete this project, but that makes installing bits for future box joints or dovetail joints a tedious process. I would rather fix the problem, but I don't really know how to define it based on the evidence so far. I think that if the shaft/collet were running true, then the fault would be with the bit only, but that runout should remain constant no matter how it is installed. If it is a combination of bit runout and collet runout happening to combine for the big error, I would have expected the 300º value to be close to its neighbor the 360º value, but instead, it was the same as its opposite 120º position.
What have I missed in my testing regime?
What could explain the odd distribution of test results?
What would the cure be?