More upgrades to the table

To address the bend in the table, I added another 75 by 25 mm truss that runs lengthwise under the middle of the table, and two more legs so that the X-axis doesn't span as far. The dip is gone, and the table is stiffer.

I've also added a jig for holding the wood in place while dovetailing it.

The two auto-adjust clamps don't need to be adjusted for different thickness of wood, and do an excellent job of keeping the wood in place. The clamp on the right is attached to the new truss that runs lengthwise under the middle of the table. It's also a lot easier to line up the vertical piece with the horizontal piece with the jig, because locking in in place is just a matter of closing the clamp.

The one unexpected surprise was that clamps will push the extrusion that the wood is centered on. I fixed this by attaching it to the new truss.

Z-Probe dovetail results

Today I used the Z-probe to zero the spindle for all three axes. The results were not as good as I'd hoped. The biggest problem was that thanks to an accurate depth of cut, I was cutting more deeply than before and there was a lot of tear out from the dovetail bit.
There were also problems with how far the cut went along the X-axis --- it needed to go about another 4mm than it did. I'm not sure how that went wrong.

Z-Probe works!

One of the wonderful items that came with the X-Controller was the Z-probe.

Tonight, I tested it. It worked fine --- and there were only two surprises.

The first was that the default units for the probing seem to be English units.

The second is that the probe's mount for the carriage is not thick enough to make it all the way through my 1/4 aluminum plates. I've updated my model so that the next time I mill a plate there will be a larger, 13mm 6.25mm deep hole for it. For now I can get along without it, because the plug is snug enough for zeroing the Z-axis.

Latest GRBL settings

These are the GRBL settings The x steps are different than the Y because the X has a GT-3 belt, and a different microstep, than the Y axis. The Y axis has GT-2 belts.

X microstepping is set at 1/4 (on, off, off), Y microstepping is set at 1/8 (on, off, on) and Z microstepping is set at 1/2 (off, on, off).

$0 = 10    (step pulse, usec)
$1 = 255    (step idle delay, msec)
$2 = 0    (step port invert mask:00000000)
$3 = 6    (dir port invert mask:00000110)
$4 = 0    (step enable invert, bool)
$5 = 0    (limit pins invert, bool)
$6 = 0    (probe pin invert, bool)
$10 = 3    (status report mask:00000011)
$11 = 0.020    (junction deviation, mm)
$12 = 0.002    (arc tolerance, mm)
$13 = 0    (report inches, bool)
$20 = 0    (soft limits, bool)
$21 = 0    (hard limits, bool)
$22 = 0    (homing cycle, bool)
$23 = 3    (homing dir invert mask:00000011)
$24 = 25.000    (homing feed, mm/min)
$25 = 750.000    (homing seek, mm/min)
$26 = 250    (homing debounce, msec)
$27 = 1.000    (homing pull-off, mm)
$30 = 12000.    (rpm max)
$31 = 0.    (rpm min)
$100 = 11.111    (x, step/mm)
$101 = 40.000    (y, step/mm)
$102 = 320.000    (z, step/mm)
$110 = 8000.000    (x max rate, mm/min)
$111 = 8000.000    (y max rate, mm/min)
$112 = 500.000    (z max rate, mm/min)
$120 = 500.000    (x accel, mm/sec^2)
$121 = 500.000    (y accel, mm/sec^2)
$122 = 50.000    (z accel, mm/sec^2)
$130 = 740.000    (x max travel, mm)
$131 = 790.000    (y max travel, mm)
$132 = 100.000    (z max travel, mm)

Dovetail math

I want to get the math down for cutting half blind dovetails down in one place, so here we go.

First, the variables from the diagram

• h --- the height of our cut
• bw --- the width of our dovetail bit, at the bottom
• p --- how far apart we cut dovetails
• a --- the angle of our dovetail bit
• top width --- how much wood is left at the top 

We need a few intermediate values, the first of which is the amount of wood the dovetail cuts.

d(h) = h * tan(a)
One quick sanity checks for the math is to see what happens with a straight bit: tan(a) = tan(0) = 0, which is what you'd expect. Another check is for a mythical 45 degree dovetail bit: tan(45) = 1, also as expected.
Now we need to figure out the top width, which is a function of h and p.

top width(h, p) = p - bw + (2 * d(h))
top width(h, p) = p - bw + (2 * (h * tan(a)))

Substituting for a straight bit we get the sensible value of p - bw.
Setting the top width to be equal to the bit width we get:

bw = p - bw + (2*h*tan(a))
2bw - (2*h*tan(a)) = p
p = 2(bw - (h*tan(a)))

Now we have to figure out how we want to round out the pins.

In this case it is just a matter of making two arcs of the same radius as the bit to get nicely rounded pins.