If you've followed along this far, you are hopefully enjoying this project nearly as much as me. Let me say again, I know the last few installments haven't been "edge of your seat" exciting, but I figure any new material is better than no new material. I have been doing a lot of busy work with Flashcut 2.0 parameters and just generally playing with the software and hardware. Got some limit switch harnesses complete, using 6-pin mini-din connectors. Nothing worth serializing, but stuff that must be done.
I did get to do a bit of machine work when my recently ordered lengths of 10mm T-bar from Bosch-Rexroth finally arrived. With my initial order from Bosch, I had only ordered 1 meter of the stuff, and along with that picked up a few dozen individual 1/4" x 20 T-nuts. I found out that the ready-made nuts were nowhere near as nice as cut sections of the T-bar, and I quickly used up my one meter bar creating things like the column T-supports, (2nd picture this page)which connect the base back plate to the column, and other fixtures where I wanted a solid, accurate T-nut.
First, an encoder saga which may help someone else. Servos make use typically of some type of encoder which provides feedback to the servo amplifier on the rotational position of the leadscrew. Without going into massive and dry theory, suffice it to say that with a typical quadrature encoder of 1000 lines, the ballscrew may be positionally resolved each revolution by 4X that amount, so a ballscrew with a pitch of 0.200", divided by 4000, equals a resolution of 0.00005"! It usually doesn't work out quite that fine, but tests so far using the homing switch have shown a resolution and repeatability of position of ~ 0.0002" or better.
Anyway, the motors that I am using are brushed DC, with a Renco RM15 encoder. When wired as shown in the Flashcut guide, I was having trouble with runaway movement, error positions, and other glaring problems. I opened up one of the Flashcut servos, which was behaving perfectly (and which conveniently also uses a Renco RM15) and found that their own servos had the channels' A and A~ wires reversed. When I made mine look like theirs, everything worked OK.
Today I was setting up a homing switch for the X-axis. I was having trouble getting the logic working properly. When asked to home to the X- switch, for example, the table would move in the wrong direction. Changing the motor signal polarity within the software fixed that, but now the direction of travel was reversed, i.e. table motion to the right became positive X motion, which is reversed from normal practice. Hmmm. Things were getting confusing. So I went ahead and physically swapped the wires on the motor leads. Instantly, I began to have the same problems with the servo that I had earlier, relative to runaway motion. Sice I had seen this before, I went ahead and swapped the channels A and A~ one more time, and the servo once again behaved perfectly, with the table moving in the correct direction, the home switch logic perfect, and everything pretty much going great. So the moral of this tale is, with brushed motors, you may have to swap channels A and A~ on the encoder when you physically change the polarity of the power leads.
The really good news is that the X-axis, after getting punished with dozens of rapid moves, has repeatedly homed to zero (there is a test-function within Flashcut which measures any error) with no measurable deviation. This is to be expected with a closed loop servo system, but it is gratifying nonetheless!
Back to the T-Bars.
|Anytime you use a T-nut on aluminum, it helps to have
as much contact surface area as possible, meaning long
T-nuts are better than short. Otherwise, as the hold-down
bolts are tightened, localized stresses and deformations
can build up between the rail and the structural aluminum
I decided, then, to use two long sections of the T-bar to back the Y-axis rail mounting to the mill base. Here, a cut, but not drilled and tapped, section of the T-bar is laid next to the Y axis sandwhich. The right-hand rail is in view. While designed for metric hardware, a 1/4" SHCS is a perfect fit in these rail holes.
|I began by drilling 6 holes 0.203" spaced by
60mm (hole to hole spacing on the rail). Here, I am
drilling through the bottom of the T-bar. Note
in the top picture that the top of the stock is grooved
to help center the drill. Since this is done in my
vertical mill, I didn't need the groove, and didn't feel
like hearing the drill flutes batter themselves against
the groove when the hole was started.
Normally center-drilling first is a good idea, but since this is not a precision setup (plenty of intentional slop between 1/4" SHCS and rail), and I am using a screw-machine length drill, I skipped the center-drilling step.
|For the first time in my machining career, I did
something which is universally frowned upon as bad
practice... I drove the tap into the holes under power! I
fully expected to break a tap, but for years I have
wondered if A) Is this possible? and B) Is it effective
and safe? The answer to both was YES, at least with this
setup, and this tap. The speed was probably about 120
With the spindle ticking over, I drove the tap downward, and let it self-feed when the cut began, one hand on the off switch and the other on the brake. Reverse took the tap out of the stock.
No problems whatsoever, but not recommended.
|With both T-bars drilled and tapped and deburred, the assembly was test-fitted. The idea here is to have perhaps 0.150" or so of lateral adjustment, so that the rails may ultimately be perfectly aligned with the longitudinal support beams. The view here is from the rear, with the column removed. Once the T-bars are in place and the column restored, the T-bars will be effectively trapped in the slot.|
This setup will provide for an excellent backing of the Y-axis rails.
You can clearly see the shape of the HSR-25 rail in this photo. Note the obvious ball-groove ground in the rail at the upper edge. There is another groove on the other side, and two more ground beneath the upper grooves; total of four grooves spaced at 90 degree intervals for an equal 4-way load potential.
For fun, I have included a picture of the 5 Bears CNC MarkII mill column. This beast arrived along with the T-bar stock and some other Bosch sundries. The dimensions are 180mm X 90mm, not including the facing and side aluminum plate which will help support the column. The 90X90 MarkI column is n the foreground, and a 1-2-3 block tossed in for scale.