Roger Herriot Award

At the Joint Statistical Meetings (Aug 2011), accepting the Roger Herriot Award for Innovation in Federal Statistics, I tipped my hat to pen-source software and three mentors.  I use the software (R, OpenBUGS, and MediaWiki) every day at work to model contaminant occurrence in drinking water.  I use R every month at home to write CNC machine code.

My three mentors are:
  • Franklin Smith (Research Triangle Institute, retired) who was the first to gently push me in the direction of mathematical statistics,
  • Harvey Gold (North Carolina State Univ, prof emeritus) who brought out the Bayesian in me, and
  • Andy Clayton (Research Triangle Institute, retired), who was my go-to statistician, back when I was a simple scientist.  Andy is still my model statistician. 
The Roger Herriot Award is precented annually by the Government Statistics Section, Social Statistics Section, and The Washington Statistical Ssociety of the American Statistical Association. 


Making it Better

A number of improvements this week.  First, I added the Y-axis' PVC base, moving the rails closer together.  Then I cut a piece of aluminum and put it between the rails to catch any dripping oil.

The PVC board came out just right.  Here, you can see that the cotout gives good access to the lube fitting, so it will be easy to keep the screw oiled.

Next, I drilled a hole behind the X-axis screw, so I can easily oil that screw, too.  To make oiling easier, I plugged the oiling hole with a tapered dowel.

Then, after squaring the axes and leveling the table, I cut a circular disk out of 5/8" PVC.  The big machine does a much better job than the little machine.  Below, a disk cut on the little machine is shown on the right, while the big machine's disk is on the left.  The little machine needed to cut the disk in three passes, moving about 10mm/min.  The big machine cut the disk in one pass, moving at 30mm/min.  Notice how smooth the edge is on the left.  The big machine has almost no noise (wiggle) and far less bias due to torque.  The little machine moved as much as 2mm with force (torque), but the big machine moved only 1/2 mm with much more force.

Below is a 15-second clip of the big machine cutting the disk.


Big Machine in Action

The router mounts work just fine, but the vertical plate holding the mounts & router is only 3/8" and flexes a little bit.  So...the big machine's first job is to make a thicker replacement plate.  Here's a photo of the machine in action with room lights on & off.  Neat, eh?


Making Mounts for New Router

I used the little machine to cut 88.8mm diameter circles in 5/8" PVC.  Then I used the little machine to scratch in the cutout lines.

After cutting with the scroll saw, I find the mounts fit PERFECTLY!  Next - get nuts and bolts to fasten the router to the mill.  In the photo below, the router is upside-down...but you get the idea.


Rebuilt from ground up...

First step was to build a solid steel-framed table.
Next was to level the table after marking location on the floor with red tape.
Then I moved the machine onto double layer of 3/4" plywood and "trued" the three axes.

  • Squared the X and Y axes by moving one end of the gantry (with other fastened by one screw).
  • Used shims to make Y axes move parallel to the platform (photo below).
  • Fastened both ends of gantry to tabletop.

Added stronger 24V power supply and switch and mounted with 3-axis controller on DIN rail.

Moved the computer and monitor.  It is a bit cramped - may move computer to a separate table or stand.  

Finally, added an emergency stop switch.  The switch (left-front corner of tabletop) was a spare from repairing a table saw.  All I have to do is hit it and the motors are disconnected.  The only trouble with this is that the computer thinks the screws are turning, so the zeros need to all be reset every time the E-stop is hit.  Hopefully, I won't need to do that very often.


Taking Small Steps

Added a switchboard (switch for stepper motors and switch for rotary tool).
Replaced plywood tool plate with 1/2 inch PVC.  Drilled starter holes with the small machine, then larger ones using the drill press.  
Next, ordered 24V power supply with enough amps for steppers and rotary tool.  Also ordered a steel frame to replace the desk.  That should make for a very solid tool stand (36x36in top).  Once this arrives, I will rebuild the machine from ground up, making everything just a little bit better.  


Limit Switches

Finished running wires, then soldered them to the six limit switches.
Next, tested each switch.  The all work!  Red wires are all connected to 5V supply and green wires are brought to 5V when a limit is hit.  Each pair of switches has its own port on the board, so the computer will know which axis has hit the limit - but not which switch.  When a switch is triggered, the CNC routine will stop.

Usually, with this many wires and connections, I find I did something wrong.  This is worth celebrating.


Wiring Motors and Stops

This morning, I finished wiring the motors.

Roughed out the stop wiring - two wires to each of six stops.


Motor Controller

Made a DIN-rail mount for the controller.

Hooked up the 24-V power supply and printer cable.
Tested the motor, identified the common pins, then pins 1-4.  (1=Green, 2=Yellow, 3 = Red, 4 = Blue)
Hooked up the motor (Z-axis) and tuned it.  160 rpm works well without load.


Mill in Machineroom

Cut the baseboard (39" by 45" on 3/4" plywood)
From Big Machine
Attached the large ballscrew and rails.
From Big Machine
Moved to desk in the "darkroom."
From Big Machine


Today's progress.

The old machine had a 7" by 7" work area.  The new machine will have 21" by 30", so about ten times as much area in XY plane.  The vertical travel will be greater, too (about 7" vs. 2.5" for the old machine).

Today, I'm cutting a sheet of walnut plywood for the base, then will mount the Y rails and ballscrew.  After that, will carefully keep everything square and add the gantry, which holds the X and Z rails and screws.

Some yet-to-be-resolved issues:

  • Need to get new 3-axis controller, as it will just be too much hassle to borrow from the old machine.  Besides, the old machine is still quite useful.
  • Need to figure out what kind of motor, chuck, etc., to do the carving.  High torque?  What voltage (steppers are 12V)?  Will probably need a separate power supply.  How to control motor speed?
  • Another computer?  Maybe I can have the machines in same room, then use a 15-foot printer cable to drive the big one.  Is 15-foot cable too long for good signal / control?
  • Is it time to buy Mach3 software?  So far, my programs have been less than 50 lines of code.  Can go up to 500 with free version of the software, but fully-loaded software has lots more bells and whistles.
  • Run to RadioShack to replenish supply of wires, parts for switch box, E-stop switch.  Rail mounts?


Building a Bigger Milling Machine

My small (7" by 7") milling machine just isn't big enough...so I'm using it to build a bigger machine.  If I did this right, the link below should take you to the photo album.  The machine still needs lots of work: 
  • beefing up the supports
  • adding a strong base
  • strengthening the horizontal piece
  • adding stepper motors
  • adding two stop switches per axis
  • adding emergency stop switch
  • hooking up all the electronics
Big Machine