Probe Calibration Renishaw MP12

More
17 Jul 2017 10:09 - 19 Feb 2021 20:23 #95875 by plopes9000
Motivation
Having had a Renishaw MP12 proble for a while, I decided it was time to write some gcode to properly calibrate it after having searched the web without success. The closest thing I could find was a post from eneias_eringer on the topic 18775-probe-for-center-of-hole here on the forum, calibracao.ngc, Buying Renishaw's calibration cycles was of course not an option for me. Running a probe without calibration does not make justice to a probe, much less to a Renishaw.

Concept
With the spindle manually aligned with the centre of the ring and with a precise ring with known diameter any measured radius deviation while touching off at a particular angle represents the trigger error of the probe plus probe run-out plus machine tolerances. When this deviation per angle is stored as calibration values and used when actually probing a part at an equal angle, accurate readings are possible. Probing should then always be carried out with the probe orientation used during calibration. Without this basic calibration any 3D probe is inaccurate, even a Renishaw.

Calibration Procedure - see video
1) A precision ring is placed perpendicular to the spindle axis on the mill table - goes without saying.
2) The spindle is manually aligned to the centre of the ring using a preferred method - I use a dial indicator mounted on the spindle.
3) MDI T99 M6 - in my tool table, 99 is the probe tool
4) The probe is jogged down in Z so that the probe tip is below the ring top surface
5) The NGCGUI routine p-calibrate.ngc is run with the precision ring's exact diameter as input parameter

Calibration Implementation
For the calibration, two routines are used, p-calibrate.ngc and aux-calibrate-probe-at-angle.ngc.

p-calibrate.ngc will probe the ring at 45 degrees steps (calling aux-calibrate-probe-at-angle.ngc for each angle) storing the results in the linuxcnc variables #621-#628. Note that for linuxcnc to persist #621-#628 one must manually add these variables to the linuxcnc.var file. p-calibrate.ngc reads ini parameters as below and it will validate that the probe tool is loaded..
[PROBE]
  BALL_DIAMETER  - precise value, eg, 5.99
  BACKOFF             - same as latch, e.g 1.5 (mm)
  COARSE_FEED    - eg. 700 (mm/s)
  FINE_FEED          - eg. 10 (mm/s)
  TOOL_NUMBER    - tool table probe tool number, eg. 99

aux-calibrate-probe-at-angle.ngc will do the actual probing based on input parameters, using some simple trig. Measured values are truncated to 3 digits of precision - should be changed to 4 for machines in imperial mode.

Probing Procedure
1) MDI T99 M6 - Load probe tool
2) Call aux-probe-at-angle-args [#<angle>] [#<distance>] [#<backoff>] [#<fine_feed>] [#<search_feed>] [#<rapid_feed>]
or
Call aux-probe-at-angle [#<angle>] [#<distance>]
3) The measured probed distance available as a return value, hence #5001

Probing Implementation
There are two base routines to probe using the calibration data.
aux-probe-at-angle-args [#<angle>] [#<distance>] [#<backoff>] [#<fine_feed>] [#<search_feed>] [#<rapid_feed>]
aux-probe-at-angle [#<angle>] [#<distance>]

While aux-probe-at-angle-args takes all values as args, aux-probe-at-angle will read most values from the ini parameters same as p-calibrate.ngc.

p-calibrate.ngc
O<p-calibrate> sub (calibrate probe)
#<diam_hole>=#1 (known hole diameter)

#<diam_ball>=#<_ini[PROBE]BALL_DIAMETER>
#<backoff>=#<_ini[PROBE]BACKOFF>
#<coarse_feed>=#<_ini[PROBE]COARSE_FEED>
#<fine_feed>=#<_ini[PROBE]FINE_FEED>
#<probe_tool_number>=#<_ini[PROBE]TOOL_NUMBER>

(info: MP12 3D probe calibration - known probe, known ring hole)

o<p-start> call

(debug, Ensure the diameters for ball and hole are precise and that the machine is in the exact center of the hole)

; capture current position
#<initial_x>=#5420
#<initial_y>=#5421
#<initial_z>=#5422
g21 g17 g40 g80 g94 g90 g49 g92.1
G0 x#<initial_x> y#<initial_y>
g91 g38.5 z-0.00001 f#<fine_feed> (check that probe is active)


;#####################################################################################
;## Measure and store trigger points in 5021..28 for 0,45, 90,135, ... 315 degrees  ##
;#####################################################################################
#<angle-step>=45
#<curr-angle>=0
#<curr-step>=0
o<base-angle-touches> do
	o<aux-calibrate-probe-at-angle> call [#<curr-angle>] [#<diam_hole>] [#<diam_ball>] [#<coarse_feed>] [#<fine_feed>]
	#[421+#<curr-step>]=#<_value>
  ;(debug, probe-at-angle [421+#<curr-step>]= #<_value>)
	g90 g0 x#<initial_x> y#<initial_y>
        #<curr-step>=[#<curr-step>+1]
	#<curr-angle>=[#<curr-angle>+#<angle-step>]
o<base-angle-touches> while [ #<curr-angle> LT 360 ]

g90 (absolute distance mode)
g0 x#<initial_x> y#<initial_y>

O<p-calibrate> endsub 


p-start - prep the probe
o<p-start> sub
(start probe)

#<probe_tool_number>=#<_ini[PROBE]TOOL_NUMBER>
o<pcheck1> if [ #5400 NE #<probe_tool_number> ]
  (abort, Please load probe - T#<probe_tool_number> M6)
  o<p-start> endsub [-1]
o<pcheck1> endif

M5 (Stop spindle)

M19 r0 (orient probe)

M65 P0 (activate probe input)
m64 P1 (start 3d probe)
m65 P1 (reset start 3d probe)
g4 p1 ; dwell

m66 P0 l0 (check if probe did activate)
o292 if [#5399 eq 1]
  (abort, probe not activated)
  o<p-start> endsub [-1]
o292 endif

o<p-start> endsub [0]
M2

aux-calibrate-probe-at-angle.ngc
;#############################
; probe x,y pos and an angle
o<aux-calibrate-probe-at-angle> sub
;input params
#<angle>=#1
#<diam_hole>=#2
#<diam_ball>=#3
#<feednorm>=#4
#<feedslow>=#5
; setup vars
#<radius> = [#<diam_hole>/2]
#<radius_b> = [[#<diam_hole>/2]-8+[#<diam_ball>/2]]
#<sin_a>=sin[#<angle>]
#<cos_a>=cos[#<angle>]
#<y-pos_b> =[#<sin_a>*#<radius_b>]
#<y-pos_ok> =[#<sin_a>*#<radius>]
#<x-pos_b> =[#<cos_a>*#<radius_b>]
#<x-pos_ok> =[#<cos_a>*#<radius>]

g90 g1 x#<x-pos_b> y#<y-pos_b> f#<feednorm>
g90 g38.2 F#<feednorm> X#<x-pos_ok> Y#<y-pos_ok>
g90 g1 X#<x-pos_b> Y#<y-pos_b> f#<feednorm>
g90 g38.2 F#<feedslow> X#<x-pos_ok> Y#<y-pos_ok>
#<x-touch>=#5420
#<y-touch>=#5421
#<radius_measured>=sqrt[[#<x-touch>*#<x-touch>]+[#<y-touch>*#<y-touch>]]
o<probe-off-300-if>if[#<radius_measured> lt 0]
  #<radius_measured>=[#<radius_measured>*-1]
o<probe-off-300-if>endif
#<probe-off-angle>=[#<radius_measured>-[#<diam_hole>/2]+[#<diam_ball>/2]]
#<probe-off-angle>=[fix[#<probe-off-angle>*1000]/1000]
g90 g1 X#<x-pos_b> Y#<y-pos_b> f#<feednorm>

(debug, probe-at-angle #<angle>= #<probe-off-angle>)
o<aux-calibrate-probe-at-angle> endsub [#<probe-off-angle>]

aux-probe-at-angle.ngc
;#############################
; probe x,y pos and an angle
o<aux-probe-at-angle> sub
#<angle>=#1
#<distance>=#2
#<backoff>=#<_ini[PROBE]BACKOFF>
#<rapid_feed>=#<_ini[PROBE]RAPID_FEED>
#<search_feed>=#<_ini[PROBE]COARSE_FEED>
#<fine_feed>=#<_ini[PROBE]FINE_FEED>

call o<aux-probe-at-angle-args> [#1] [#2] [#<backoff>] [#<fine_feed>] [#<search_feed>] [#<rapid_feed>]
o100 if [ #<_value> == -1 ]
  (abort, failed to probe ....)
o100 endif 

o<aux-probe-at-angle> endsub [#<_value>]


aux-probe-at-angle-args.ngc
;#############################
; probe x,y pos and an angle
o<aux-probe-at-angle-args> sub
#<angle>=#1
#<distance>=#2
#<backoff>=#3
#<fine_feed>=#4
#<search_feed>=#5
#<rapid_feed>=#6

#<diam_ball>=#<_ini[PROBE]BALL_DIAMETER>
#<probe_tool_number>=#<_ini[PROBE]TOOL_NUMBER>
 
;(debug, aux_probe-at-angle-args)

M5 (Stop spindle)
o<pcheck1> if [ #5400 NE #<probe_tool_number> ]
  (abort, Please load probe - T#<probe_tool_number> M6)
  o<aux-probe-at-angle-args> endsub [-1]
o<pcheck1> endif

m66 P0 l0 (check if probe did activate)
o292 if [#5399 eq 1]
  (abort, probe not activated)
  o<aux-probe-at-angle-args> endsub [-1]
o292 endif

#<initial_x>=#<_x>
#<initial_y>=#<_y>
#<initial_z>=#<_z>

o<probe-off-301-if> if [#<distance> LT #<backoff> ]
  (abort, distance smaller than backoff value, can not make move, distance #<distance>, backoff #<backoff>)
  o<aux-probe-at-angle-args> endsub [-1]
o<probe-off-301-if> endif

#<offset_variable>=[fix[#<angle>/45]]
o<probe-off-302-if> if [ [#<offset_variable> NE [#<angle>/45]] or [#<offset_variable> LT 0 or #<offset_variable> GT 8] ]
  (abort, angle not in calibration #<angle>)
  o<aux-probe-at-angle-args> endsub [-1]
o<probe-off-302-if> endif

#<probe_offset>=#[421+#<offset_variable>]
o<probe-off-303-if> if [[#<probe_offset> EQ 0] or [#<probe_offset> GT [1] ] or [#<probe_offset> LT [-1] ]  ]
  (abort, probe has not been calibrated for the angle #<angle>, off number #<offset_variable> [421+#<offset_variable>] )
  o<aux-probe-at-angle-args> endsub [-1]
o<probe-off-303-if> endif

;(debug, offset_variable=#<offset_variable>, offset=#<probe_offset>)

#<sin_a>=sin[#<angle>]
#<cos_a>=cos[#<angle>]
#<y_pos_b> =[#<sin_a>*#<distance>]
#<x_pos_b> =[#<cos_a>*#<distance>]
#<y_backoff> =[#<sin_a>*#<backoff>]
#<x_backoff> =[#<cos_a>*#<backoff>]
#<y_backoff2> =[#<sin_a>*[#<backoff>+0.3]]
#<x_backoff2> =[#<cos_a>*[#<backoff>+0.3]]
#<y_comp> =[#<sin_a>*#<probe_offset>]
#<x_comp> =[#<cos_a>*#<probe_offset>]
g91 g38.3 x[#<x_pos_b>] y[#<y_pos_b>] f#<search_feed>
#<x_touch>=#5420
o100 if [ #5070 eq 0 ]
  g90 g0 x#<initial_x> y#<initial_y> f#<rapid_feed>
  (msg, probe did not make contact ...)
  o<aux-probe-at-angle-args> endsub [-1]
o100 endif

g91 g1 X[-#<x_backoff>] Y[-#<y_backoff>] f#<rapid_feed>

g91 g38.3 X[#<x_backoff2>] Y[#<y_backoff2>] f#<fine_feed>
#<y_touch>=#5421
o200 if [ #5070 eq 0 ]
  g90 g0 x#<initial_x> y#<initial_y> f#<rapid_feed>
  (msg, probe did not make contact ...)
  o<aux-probe-at-angle-args> endsub [-1]
o200 endif
#5001=[#<x_touch>+#<x_comp>]
#5002=[#<y_touch>+#<y_comp>]

M67 E0 Q#5001 (set hal analog output to x pos)
M67 E1 Q#5002 (set hal analog output to x pos)

;(debug, x=#<x_touch>, x-comp=#<x_comp> y=#<y_touch> y-comp=#<y_comp> )

g90 g0 x#<initial_x> y#<initial_y> f#<rapid_feed>

;(debug, probe-at-angle #<angle> with offset #<probe_offset>)


o<aux-probe-at-angle-args> endsub [0]
M2


Integration with Probe Screen
the files xminus, xplus, yminus, yplus would look like below, one just changes the first argument, the angle or approach.
To get probe screen to use the compensation values from the calibration, I had to use hal analog outputs set by the aux-probe-at-angle-args, M67 E0 Q#5001 (set hal analog output to x pos) M67 E1 Q#5002 (set hal analog output to x pos).
The pins are created in probe screen and connected to motion.analog-out-XX on the postgui hal file.
Finally, the probed_position_with_offsets function was modified to use these pins values instead.
This seems to work well, though to be honest just got it working today 19.02.2021 ...


o<xplus> sub
o<aux-probe-at-angle-args> call [0] [#<_hal[probe.ps_probe_max]>] [#<_hal[probe.ps_probe_latch]>] [#<_hal[probe.ps_probevel]>] [#<_hal[probe.ps_searchvel]>] [#<_ini[TOOLSENSOR]RAPID_SPEED>]
#<ret_code>=#<_value>
g4 p0.5 ; dwell
o<xplus> endsub [#<ret_code>]
M2

o<xminus> sub
o<aux-probe-at-angle-args> call [180] [#<_hal[probe.ps_probe_max]>] [#<_hal[probe.ps_probe_latch]>] [#<_hal[probe.ps_probevel]>] [#<_hal[probe.ps_searchvel]>] [#<_ini[TOOLSENSOR]RAPID_SPEED>]
#<ret_code>=#<_value>
g4 p0.5 ; dwell
o<xminus> endsub [#<ret_code>]
M2


o<yplus> sub
o<aux-probe-at-angle-args> call [90] [#<_hal[probe.ps_probe_max]>] [#<_hal[probe.ps_probe_latch]>] [#<_hal[probe.ps_probevel]>] [#<_hal[probe.ps_searchvel]>] [#<_ini[TOOLSENSOR]RAPID_SPEED>]
#<ret_code>=#<_value>
g4 p0.5 ; dwell
o<yplus> endsub  [#<ret_code>]
M2

o<yminus> sub
o<aux-probe-at-angle-args> call [270] [#<_hal[probe.ps_probe_max]>] [#<_hal[probe.ps_probe_latch]>] [#<_hal[probe.ps_probevel]>] [#<_hal[probe.ps_searchvel]>] [#<_ini[TOOLSENSOR]RAPID_SPEED>]
#<ret_code>=#<_value>
g4 p0.5 ; dwell
o<yminus> endsub [#<ret_code>]
M2




postgui_common.hal
net probe-x-touchoff => probe.x-touchoff <= motion.analog-out-00
net probe-y-touchoff => probe.y-touchoff <= motion.analog-out-01
net probe-z-touchoff => probe.z-touchoff <= motion.analog-out-02

probe_screen.py changes to __init__
        self.halcomp.newpin( "x-touchoff", hal.HAL_FLOAT, hal.HAL_IN )
        self.halcomp.newpin( "y-touchoff", hal.HAL_FLOAT, hal.HAL_IN )
        self.halcomp.newpin( "z-touchoff", hal.HAL_FLOAT, hal.HAL_IN )


probe_screen.py changes to probed_position_with_offsets
    def probed_position_with_offsets(self) :
        self.stat.poll()
        probed_position=list(self.stat.probed_position)
        coord=list(self.stat.probed_position)
        g5x_offset=list(self.stat.g5x_offset)
        g92_offset=list(self.stat.g92_offset)
        tool_offset=list(self.stat.tool_offset)
#        print "g5x_offset=",g5x_offset
#        print "g92_offset=",g92_offset
#        print "tool_offset=",tool_offset
#        print "actual position=",self.stat.actual_position
#        print "position=",self.stat.position
#        print "joint_actual position=",self.stat.joint_actual_position
#        print "joint_position=",self.stat.joint_position
#        print "probed position=",self.stat.probed_position
        for i in range(0, len(probed_position)-1):
             coord[i] = probed_position[i] - g5x_offset[i] - g92_offset[i] - tool_offset[i]
        angl=self.stat.rotation_xy
        res=self.rott00_point(coord[0],coord[1],-angl)
        #coord[0]=res[0]
        #coord[1]=res[1]
        coord[0]=self.halcomp["x-touchoff"]
        coord[1]=self.halcomp["y-touchoff"]
        return coord
Last edit: 19 Feb 2021 20:23 by plopes9000. Reason: Updated post to include integration with probe screen taking into account calibration data.
The following user(s) said Thank You: andypugh, robh, gmouer, jo_key, verser, fixer, pommen

Please Log in or Create an account to join the conversation.

More
17 Jul 2017 16:39 #95923 by andypugh
Do you have a feel for how much more accurate this is than my approach? I centred the probe tip using a DTI with the spindle turning.

Please Log in or Create an account to join the conversation.

More
17 Jul 2017 16:57 #95927 by plopes9000
I see variations of up to 0.06mm depending on the approach angle. The issue is not probe runout, I have my probe tip well centered using a DTI as well and in the end I rotate the probe by hand to verify. That probe runout will be cancelled by nature of the calibration is just a bonus. What really can't be compensated for without calibration is the various trigger forces required depending on the angle of approach due to the three point design of most probes. I suppose on strain gauge probes calibration won't be as required. Does 0.06mm matter? That depends on machine tolerances and required precision.

Please Log in or Create an account to join the conversation.

More
17 Jul 2017 18:17 #95935 by plopes9000
Check out resources.renishaw.com/en/Renishaw+touch...gy(6854)?direct=true
It goes through all aspects of touch probe accuracy.

Please Log in or Create an account to join the conversation.

More
18 Jul 2017 14:00 - 18 Jul 2017 14:09 #95980 by eneias_eringer
Yes , without a calibration the error can be 0.03 to 0.06mm !!

PLopes, are you familiarized with the Renishaw Inspection Plus Cycles??? 9800 9814 9802 ??
Last edit: 18 Jul 2017 14:09 by eneias_eringer.

Please Log in or Create an account to join the conversation.

More
18 Jul 2017 14:52 #95983 by plopes9000
Eneias, I'm afraid I am not. I wish I could get a look at the g-code renishaw uses in their probe cycles.

Please Log in or Create an account to join the conversation.

More
18 Jul 2017 16:13 #95988 by eneias_eringer
Ok, i have a Inspection Plus convertion for linux CNC, but my Renishaw converted code is so hard to understand, becouse Linux cnc dont have a Simple GOTO , just a conditional GOTO..

But the use is identical to the Siemens Inspection plus routines

Please Log in or Create an account to join the conversation.

More
19 Jul 2017 21:27 #96104 by robh
hi
thanks for sharing i did see your youtube videos
this saves me abit of work to rewrite the renishaw ones.

does this do the Z also? as i have a MP700 which is true 3D probe

for Z in the renishaw all i have todo is set a datum as a known true zero reff.
then set the probe as a rougth tool length and the cycle will calibrate the length, and then the probe travel

just need the same for the tool probe now :)

and yes this does make a huge differnce on probe errors. as you dont have to clock the probe up as much just make sure you put it in the same way around all the time.. or store it in the ATC :)

thanks for the sharing your hard work

Please Log in or Create an account to join the conversation.

More
20 Jul 2017 07:53 #96125 by plopes9000
Hi Rob,

I haven't done the z calibration yet, XY plane probing was more critical to me, since it matters when flipping a part over - but yes z height calibration is on the todo list.

You know how it is, there are always 10 jobs being done in parallel, the one most needed or most liked gets immediate attention, the others wait ...

I'm more than happy to share, the linuxcnc community is great, a big thank you from me to everybody here at linuxcnc forum !

btw will be sharing as well the TS27r cycles soon - need some time/priority to iron out some details - atm busy with classicladder for the impact wrench power drawbar and in-between distracted with setting up ethercat.

Cheers

Please Log in or Create an account to join the conversation.

More
27 Sep 2017 09:20 #99509 by llrjt100
I'm currently working on getting a Renishaw OM40-2 probe and OTS toolsetter with OMI-2T interface working with LinuxCNC on my Tormach mill. I'm facing the issue of converting Renishaw Inspection Plus macros to LinuxCNC - would you be able to share your conversion? I would save me a huge amount of time! I'd also appreciate your advice as to whether the Renishaw Fanuc or Siemens macros are easiest to convert to LCNC.

Many thanks,
Richard

Please Log in or Create an account to join the conversation.

Time to create page: 0.132 seconds
Powered by Kunena Forum