Greetings all;
Using the g38.2 on a lathe for a potential auto-zero turned out to be a
lot more painful that it was on my mill, so here is a blow by blow that
at some point I'd like to see in the wiki in case someone else has to
re-invent this contraption.
Background:
I made a gage out of some 25mm by alu blocks. The probe contact faces
are double sided pcb material super glued to the gages faces. Its
base was milled to fit the front V-way and just lay on the rear of the
bed for x location, with a L shaped piece where the L is rotated 90
degrees CW so it can project to the right in a position behind the
mounted cutting tools. Another 25mmx25mmx75mm block of alu was super-
glued to the front face of the 'L' and its right end face serves as the
mounting face for another piece of PCB material which will be one of the
Z axis contacts.
Bolted to the front face of the 25mmx25mmx75mm block is a smaller alu
piece, about 30mmx10mmx10mm with half the thickness, nominally 5mm, cut
away for about 10mm, and suitable sized bits of pcb material lining the
cutaway, so this, when swung down so it overhangs the right end of the
25mmx25mmx75mm block, thereby supplying a probe contact surface for a
smallish boring bar facing the operator.
A 12 foot headphone extension was purchased from the shack, and cut in
two such that one end was terminated at the probe in terminals on the
drive power box, the other end was stripped and the active wire was
soldered to the 4 pieces of PCB material. I did not ground the
shielding on this gage, thinking that the lathe frame was grounded
through its line cord, and the computer & power boxes were also grounded
at their power cords, but it appears there is enough noise from the 40
kilohertz 2M542 amplifiers impressed on the lathes frame by motor winding
capacitance that I have quite a bit of noise, up to 3 or 4 volts, being
induced back into the probe input, so when contact is solidly made, the
LED on the C1G interface card is still about 2% of normal brightness.
I need to get one of those monster cables & figure out a way to star
ground everything without making any ground loops, which should be fun.
At that point I raised the servo-thread to 2 khz, and added a debounce
module with a delay of 4 into the circuit, which seemed to be sufficient
to give solid logic levels, as opposed to pocket comb teeth on the hal-
scope display without it.
[snippet of .hal to do this]
--------------------------
# Now, we need some help for probing
net probe parport.0.pin-10-in-not
debounce.0.0.in # feeds x,z
net probe-debounce debounce.0.0.out
and2.0.in1 and2.1.in1
setp debounce.0.delay 4
# for both axis's
--------------------------
I might try delay at 3 or even 2 there eventually because this delay
enhances another problem, that of tripping off when attempting to move
back away from contact. The debounce holds it true too long.
I have ordered another screw for the z, but it will be mid-February
arriving.
The Z probe is only interested in establishing a stop point while moving
left, and because of the nominal 21 thousandths of backlash in that
drive, I thought it best to enable probe detection only when moving left.
To that end, the .hal code to develop a direction signal from backlash-vel
[snippet of .hal file]
---------------------
# Z only here, uses comp2, and2.0, and2.1, or2
setp comp.1.hyst 0.000001 # for z only, makes dir
sig from backlash move
net probe-disable1 axis.2.backlash-vel
comp.1.in0
setp comp.1.in1 0.000000
net probe-disable2 comp.1.out
and2.0.in0 # z dir, valid left only
net probe-disable3 and2.0.out
or2.0.in0 # z for combined probe
---------------------
Because of the slight hyst value, it remembers the polarity of the backlash
move, opening the path for the probe signal when moving to the right. Works.
But since the X needs to be sensitive in both directions, I chose to enable
detection only when it was actually moving.
[snippet of .hal file]
------------------------------------
# Now, we need a similar function for x but bidirectional
net x-probe-control0 axis.0.joint-vel-cmd
abs.3.in # x is moving signal
net x-probe-control1 abs.3.out
comp.2.in1 # + for either direction
setp comp.2.in0 0.000001
# threshold of movement
net x-probe-control2 comp.2.out
and2.1.in0 # now we have true/false
------------------------------------
But then I couldn't back away from contact once it was made, so a disable
during backlash was needed. Backlash there is nominally .001", its a ball
screw.
[snippet of hal file]
------------------------------------
net probe-x0 and2.1.out
and2.3.in0 # needs 1 more sig yet
# now, kill x sense during backlash
net probe-x1 axis.0.backlash-vel
abs.4.in # vel pulse while bklsh mv
net probe-x5 abs.4.out
comp.3.in0 # now uni-directional
setp comp.3.in1 0.00000001
net probe-x2 comp.3.out
and2.3.in1 # false while bklsh mv
net probe-x4 and2.3.out
or2.0.in1 # x for combined probe
# and hook this rube goldberg up to the target, motion.probe-input
net probe-final or2.0.out
motion.probe-input #final probe control
------------------------------------
Complex, and probably "simplicatable", but seems to work, the disable
signal is about 38 milliseconds long on the halscope. It also does
not work at all well if I move the 1 in the setp one space left. It is
indeed that critical, needing every millisecond I can coax out of the
vel signal which peaks at about 40 millivolts on the halscope.
Thanks for your reading patience.
Comments welcome.
Cheers, Gene
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
My web page: <http://coyoteden.dyndns-free.com:85/gene> is up!
My views <http://www.armchairpatriot.com/What%20Has%20America%20Become.shtml>
WHERE CAN THE MATTER BE
Oh, dear, where can the matter be
When it's converted to energy?
There is a slight loss of parity.
Johnny's so long at the fair.
I was taught to respect my elders, but its getting
harder and harder to find any...
------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122412
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
