Jay's Technical Talk » router

Z-axis cart and adjustment assemblies

Jay — Fri, 02 Jan 2009 01:39:25 +0000

Drilling the holes for the adjustment screw

I cut out the parts for my Z-axis assembly using a jigsaw, and while butting the machine edges worked fine the other edges were a bit sloppy. I took advantage of an in-law’s shop to finish them up with a table saw and now the edges are much nicer. The base dimensions of my Z-axis are 11 inches by 10 and 1/2 inch, and the sides are 2 and 1/2 inches high. My 12 inch black pipes are mounted with their centers 7 and 1/2 inches apart using adjustment assemblies made out of an old plastic cutting board. My Z-axis cart is just slightly over 5 inches square.

The adjustment assemblies are 2 and 1/2 inch squares with four machine screws threaded through the edges to use to adjust the position of the pipes. I took advantage of the in-laws shop again to use a drill press to put all the holes in the pipe adjusters. Most of the holes could have been done with a hand drill, but the edge holes would have taken extra care.

Because I mounted the adjustment assemblies outside the Z-axis assembly I lost about an inch of my planned 5.5 inch travel and my current z-axis travel is only 4.5 inches.

CNC – First 2.5D cutting

Jay — Sun, 30 Nov 2008 02:12:34 +0000

I have designed a bracket that I used as my first attempt to manufacture something with my CNC mill. Because I’m still using python code I wrote myself and have not yet calibrated my stepper coordinates into real-world measurements, I had to design the bracket in Inkscape, print it out, and manually move my router bit to each point to determine the stepper coordinates along the path I wanted to cut out. This really sucked, and I plan on switching over to the parallel port controller board and EMC2 very soon so that I do not have to keep writing code. (I had to add code to cut out the bracket that interpolates between two endpoints so that I could route semi-accurate 60 degree lines.)

The piece of wood was secured to my sacrificial base with two wood screws, which worked out quite well. I am using a Black & Decker RTX rotary tool (low setting is 12,000 rpm) which is a relatively high powered tool, not the type of thing you buy at harbor freight for $10. Even so, my X/Y table moved too fast for it (causing my Dremel Drill press mount to be pulled out of alignment , as well as some damage to my bracket in a non-critical area).

So much for thinking that my rate was too slow! So I slowed the motion of my X/Y table by a factor of 2 and tried again. By only lowering the bit 1/32nd of an inch at a time I was able to cut out the wood without too much smoke and with a good deal of accuracy.

Needless to say, this was a LONG procedure. Each lap around the bracket took over six minutes, and it took about 25 of them to cut out the piece of wood. Once I got greedy and tried to do more than 1/32nd of an inch and knocked the dremel out of the holder with the extra pressure, so after re-calibrating to the starting point I sat down for the long haul. The entire procedure took around two hours. This makes me want two things: 1) A working Z-axis, so the computer can lower the dremel 1/32nd of an inch each time around without me, and 2) A much stronger spindle motor (trim router?) so I can cut out wood parts faster!

Wife Acceptance Factor (WAF):
The noise (especially when I was mistakenly using the 30,000 rpm setting instead of the 12,000 setting) was very loud, and the wife and cat ran upstairs to escape from it. After I was done, she noticed one good thing about the wood smoke I had produced. “[the smell] Makes me think I’m in a log cabin, [with a] burning fireplace.” If I want to transplant this device to my office, I am definitely going to need a sound dampening enclosure. (Also needed if I want to keep it at home.)

Below are a few pictures of the part and mechanism, as well as a movie of the cutting procedure in action:

First performance evaluation

Jay — Sat, 22 Nov 2008 02:46:49 +0000

Checking Accuracy

Scan of the circle test

Using my simple router code I had the X/Y table plot a square with a circle inside of it (and then added some extra code to draw a smaller circle inside that). The first time I tried this I found that one side of my X/Y table was significantly (1mm or more) lower than the other. I added a washer between the table and the piece of wood that attached it to the Y-axis drawer pull and that improved things quite a bit ( the washer is on the bottom left corner of the test pattern). Using a felt tip marker gives a good indication of the level of the work platform, as the tip will be pushed up (and spread out, leaving a wider mark) if the table gains height, or get smaller if the table drops. One of my corners (top left in the image) has a distinct dip in it, but because I started the plotter at that corner and adjusted the height as it was moving I need to repeat the test. The giant ink spot in the top middle is because I left the pen sitting on the paper while writing the code to draw the inner circle.

My test pattern (and working area) is approximately five and one quarter inches on a side. By measuring the distance from opposite corners you can also check to see how perpendicular your X and Y axis are. Using my “wide” felt tip marker, the corner distances are both 8 and one-eighth inches. I may find a small difference if I use a more precise marker, but for now I’m happy with the “squareness” of my X/Y axis system. The pattern you see here took over ten minutes to draw, so the IPS of my system is relatively low. The video below shows my plotter drawing the test pattern (The video is sped up to 3 times actual speed.)

I changed the test pattern code to start at the top middle so I could adjust the pen somewhere other than the top left corner and found that the top left corner was also low.

Test two (starting in the top right corner)

So, I added a washer to that corner as well. I then re-did the test pattern. This time I did not draw the circles. I plan on using OpenCV to do some line detection on this four line box to see if I can get a highly accurate measure of the distance of the diagonals. It appears that my washer (shims) are slightly wider than I need and the “washer” side is now slightly higher than the non-washer side. However, it is better than it was before. Once I get my Z-Axis working, I may re-visit this issue to prop up the other side with some very narrow shims (heavy paper?), but I expect this is level enough for cutting out parts to make a 2nd gen router.

Box after washers were added

New Serial Controller

Jay — Wed, 19 Nov 2008 02:44:30 +0000

I received the new and improved serial stepper controller board (version 4) from Ocean Controls. (Cost: 65 AUD) Although it supports up to a 6kHz stepping rate, I found that my steppers can only be driven at the RATE 12 setting, or approximately 745 steps a second. I’m not sure if this is because I’m not providing enough voltage or power (currently at 6v and 1.66A) or if that’s just the maximum rate of the steppers. When I try to use RATE 11, the steppers make nasty noises and are obviously loosing steps here and there. I don’t think it’s a current issue, as the usable RATE doesn’t change when running one stepper or two steppers at the same time. At some point in the future I’ll find a higher voltage power supply and see if that allows the motors to step faster.

This speed is still a small improvement over the original stepper controller, which worked at 605 steps a second maximum speed (despite the fact that it was supposed to run at 1000 s.p.s. or 1kHz). The documentation for the SSCB says that rate 12 is equivalent to 755.9 steps/second, so my measured 745 is relatively close, and probably is due to the built in acceleration and deceleration at the start and end of motion.

The real benefits of the version 4 board is the ability to start multiple motors at the same time, and get a prompt back over the serial line when the motors stop moving so I don’t have to poll the position (which still results in the controller board dropping steps). The new board also includes two command controlled relays, which will be perfect for turning on a spindle motor via program control, and perhaps enabling the stepper motor power supply.

The new multi-motor start command and feedback when finished greatly simplified the python code to interface with the stepper controller, and the serial control board no longer drops characters (perhaps because I am sending less characters and waiting patiently for motors to finish moving.)