PIC based DRO

OK, a few things to clarify here.

1. I HAVE built a PIC based DRO. The electronics DO work. It's
actually a fairly simple programming task.

The unit displays absolute and relative positions, with a reset for
both. No other fancyness, because I just don't care. ( Metric
conversion, lathe "X2" mode, etc. )

The DRO is not yet in use because the mechanicals are a bit of a
problem. I can't get ABSOLUTELY reliable repeatability out of my
rotary encoder setup. But the electronics work just fine.

2. It IS inexpensive. A mid-level PIC, an LCD and not much else. And
LCDs are pretty cheap these days. I bought some backlit ones for $7.00
USD

I used LCDs instead of LEDs because

my shop is BRIGHT, 32 feet of flourescents in 70 sq feet and I was
worried about the LEDs washing out
it's actually cheaper ! when the cost of PCB real estate is considered
I could display text messages, like ERROR

3. The "problem" with a PIC based controller is that there is no easy
way to input all the possible the scale parameters. That is, I have
written code for use with a 2000 cpr codewheel. I cannot now take this
device and use it with the 360 cpi linear strip. I'd have to write new
code and program a new PIC.

In a strange twist, the above "problem" actually solves the floating
point "problem", since there is no way to enter "unusual" resolutions.
Things are simply custom coded for a particular resolution using integer
math.

( The PIC could contain code for the most "common" scale settings, I
suppose. )

4. Even with the math being integer math, I wimped out when doing the
DRO/Stepper controller. I used a second small PIC to convert quadrature
to clock/dir info. This relieved the main PIC of some processing. I
was a bit worried about race conditions, determinism, and my ability to
be SURE that no counts would be missed doing it all in a single PIC.

SOLVING the problem by adding a one dollar part seemed like a good
design choice.

I got a lot of "advice" (trans: crap) on RCM from people who had never
programmed offering advice on their better ideas for solving my
problem. NONE of them offered any actual code, of course, just abuse.

5. Scales are a problem, and as attractive as those Goldmine rotary
encoders are, I keep being drawn to linear strips, even though the
resolution is lower than I would like.

Both have problems, especially below .001"/count. With linear strips, a
person is relying pretty heavily on the linearity of the ballscrews in
the photoplotter that created them in the first place.

Rotary encoders have the problem of both slippage and scale. A person
cannot, by definition, machine a pulley to precisely the right diameter
(that pesky PI thing). And even if he can get it "close enough", how can
he be SURE that it's close.

Every technique I've encountered for testing the accuracy once on the
machine ( no sense testing it off the machine ) has some fundamental
assumption that makes me uncomfortable.

Frankly, my standard eBay search these days is for "+DRO +scale".



Alan


P.S. I believe that the Sherline DRO uses an 8051.

--

Alan Rothenbush | The Spartans do not ask the number of the
Academic Computing Services | enemy, only where they are.
Simon Fraser University |
Burnaby, B.C., Canada | Agix of Sparta