[Emc-users] latency calculation
I need know how to calculate latency in EMC2's latency-test. could anyone tell me the latency test's algorithm? -- ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] lantency calculation
sorry,,I mean the base and servo thread.. how do we calculate them? -- ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] How to create custom kinematics module?
Hello, folks!
I have several questions about following part of my kinematics module
(I have added the numbering at the beginning of lines, so that it is
easier to understand, which phrase or line is being discussed):
1 #include rtapi_math.h
2 #include kinematics.h /* these decls */
3
4 const double head_offset = 200;
5
6 PmCartesian old;
7
8 in t kinematicsForward(const double *joints,
9 EmcPose * pos,
10const KINEMATICS_FORWARD_FLAGS * fflags,
11KINEMATICS_INVERSE_FLAGS * iflags)
12 {
13 double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
14 pos-tran.x = joints[0] - head_offset * sin(joints[5]);
15 pos-tran.y = joints[2] + head_offset * sin(joints[4]);
16 pos-tran.z = joints[3] + cos(joints[4]) * cos(joints[5]) * head_offset;
17 pos-a = asin(sin(joints[4]) / cos(xy_tan));
18 if xy_tan = 90 + (180 * i), where i is not 0, but is 1, 2, 3, etc
or -1, -2, -3, -4 etc
19 then pos-a = asin(sin(joints[5]) / sin(xy_tan));
20pos-c = joints[6];
21pos-u = joints[7];
22pos-v = joints[8];
23
24return 0;
25 }
26
27 int kinematicsInverse(const EmcPose * pos,
28double *joints,
29const KINEMATICS_INVERSE_FLAGS * iflags,
30KINEMATICS_FORWARD_FLAGS * fflags)
31 {
32double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
33joints[0] = pos-tran.x + (head_offset * sin(joints[5]));
34joints[1] = pos-tran.x + (head_offset * sin(joints[5]));
35joints[2] = pos-tran.y - (head_offset * sin(joints[4]));
36joints[3] = pos-tran.z - (cos(joints[4]) * cos(joints[5]) *
head_offset);
37 OR joints[3] = pos_.tran.z - (cos(pos-a) * head_offset);
38joints[4] = asin(sin(pos-a) * cos(xy_tan));
39joints[5] = asin(sin(pos-a) * sin(xy_tan));
40joints[6] = pos-c;
41joints[7] = pos-u;
42joints[8] = pos-v;
43
44return 0;
45 }
1) What would correct syntax for line 18 and 19 look like?
2) Is the current assignement between XYZ axis and joints[1], [2], [3]
and [4] going to work correctly? Is there anything specific to be
changed in HAL for these kinematics to work?
3) How do I include these lines in xy_tan (it is the tangent for the
nozzle movement in XY plane) calculation:
For extremes:
If X - old.X = 0, and Y - old.Y 0, then C = 0
If X - old.X = 0, and Y - old.Y 0, then C = 180
If Y - old.Y = 0, and X - old.X 0, then C = 90
If Y - old.Y = 0, and X - old.X 0, then C = 270
All the rest 4 quadrants are appropriate value of tangent functions as follows:
If X - old.X 0, and Y - old.Y 0, then 270 C 360
If X - old.X 0, and Y - old.Y 0, then 0 C 90
If X - old.X 0, and Y - old.Y 0, then 90 C 180
If X - old.X 0, and Y - old.Y 0, then 180 C 270
4) Are the formulas correct at all? Yesterday I was convinced that
they are perfect, but now I am not that sure, because of the fact that
sin and cos values repeat every 180 degrees, values of tan and cotan
functions repeat every 180 degrees and the more I think about them,
the more I get confused.
Please, share Your thoughts, opinions and comments! They will be
really appreciated, as they can help get this thing working correctly.
Thank You in advance!
with best regards,
Viesturs
P.S I have attached whole module file, if anyone is interested.
/
* Description: ABkins.c
* Kinematics for a 5 axis gantry-style waterjet for dynamic taper compensation.
*
* Derived from a work by Fred Proctor Will Shackleford
*
* Author: Viesturs Lācis
* License: GPL Version 2
* System: Linux
*
* Copyright (c) 2010 All rights reserved.
*
/
#include rtapi_math.h
#include kinematics.h /* these decls */
const double head_offset = 200;
PmCartesian old;
int kinematicsForward(const double *joints,
EmcPose * pos,
const KINEMATICS_FORWARD_FLAGS * fflags,
KINEMATICS_INVERSE_FLAGS * iflags)
{
double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
pos-tran.x = joints[0] - head_offset * sin(joints[5]);
pos-tran.y = joints[2] + head_offset * sin(joints[4]);
pos-tran.z = joints[3] + cos(joints[4]) * cos(joints[5]) * head_offset;
pos-a = asin(joints[4] / cos(xy_tan));
if xy_tan = 90 + (180 * i), where i is not 0, but is 1, 2, 3, etc or -1, -2, -3, -4 etc
then pos-a = asin(joints[5] / sin(xy_tan));
pos-c = joints[6];
pos-u = joints[7];
pos-v = joints[8];
return 0;
}
int kinematicsInverse(const EmcPose * pos,
double *joints,
const KINEMATICS_INVERSE_FLAGS * iflags,
KINEMATICS_FORWARD_FLAGS * fflags)
{
double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
joints[0] = pos-tran.x + (head_offset * sin(joints[5]));
joints[1] = pos-tran.x + (head_offset * sin(joints[5]));
joints[2] = pos-tran.y -
Re: [Emc-users] How to create custom kinematics module?
Viesturs Lācis wrote:
1) What would correct syntax for line 18 and 19 look like?
could be something like :
18 for (int i=1; i limit; i++){
19 if (xy_tan == (90+(180*i)) || xy_tan == (90-(180*i)))
19 pos-a = asin(sin(joints[5]) / sin(xy_tan));
20 }
where limit is the maximal value I can take.
this code is not optimised, because for large limit value, il will
take a long time to iterate. But should work right for a limit value of
5 or 10.
--
ThinkGeek and WIRED's GeekDad team up for the Ultimate
GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the
lucky parental unit. See the prize list and enter to win:
http://p.sf.net/sfu/thinkgeek-promo
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] latency calculation
The latency-test runs the following realtime component: http://git.linuxcnc.org/gitweb?p=emc2.git;a=blob;f=src/hal/components/timedelta.comp Each time a realtime thread runs, the time interval del=(now-last) is computed. The minimum and maximum del ever seen are computed in min_ and max_. The jitter is the bigger of max_-period or period-min_. period is the promised realtime interval in nanoseconds, and rtapi_get_time() is the measured wall time also in nanoseconds. The timedelta component also measures long term error in 'err' and 'avg_err'. These values are not shown in the latency-test GUI but can be used to determine if there are long-term differences between expected and actual intervals. Jeff -- ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] How to create custom kinematics module?
2010/6/11 Przemek Klosowski przemek.klosow...@gmail.com:
1 #include rtapi_math.h
2 #include kinematics.h /* these decls */
3
4 const double head_offset = 200;
5
6 PmCartesian old;
7
8 in t kinematicsForward(const double *joints,
9 EmcPose * pos,
10 const KINEMATICS_FORWARD_FLAGS * fflags,
11 KINEMATICS_INVERSE_FLAGS * iflags)
12 {
13 double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
14 pos-tran.x = joints[0] - head_offset * sin(joints[5]);
15 pos-tran.y = joints[2] + head_offset * sin(joints[4]);
16 pos-tran.z = joints[3] + cos(joints[4]) * cos(joints[5]) *
head_offset;
17 pos-a = asin(sin(joints[4]) / cos(xy_tan));
18 if xy_tan = 90 + (180 * i), where i is not 0, but is 1, 2, 3, etc
or -1, -2, -3, -4 etc
19 then pos-a = asin(sin(joints[5]) / sin(xy_tan));
OK, but what is 'i'? or are you saying that the test is for
(xy_tan-90) being an integer multiple of 180? without understanding
your geometry, this doesn't make immediate sense to me...
If You recall trignometry from elementary school, tangent values (just
like for sin and cos functions) repeat every 180 degrees. For example
tangent of 45 degrees is 1. Just like it is for 225, 405 etc. General
way to write it is tan(45 + 180*i) = 1, where i is integer - 1, 2, 3,
4, 5 etc.
Actually I seem to have made this bit more complicated as necessary -
for me xy_tan should be in range from 0 to 359 (including both of
these numbers), so instead of writing 90 + (180 * i), I could have
said 90 or 270.
By the way, how should I limit the xy_tan in the mentioned range from 0 to 359?
For example, for deltaX = 1, deltaY = -1 in Excel for atan2 function I
get value -45 degrees, while I would like it to be 315. This way each
position for xy_would be described with only one possible value
instead of several. Like in the given example, it can be -45 or +315
etc.
because you whip the value of pos-a betweenthe values based on joints[5] and
joints[4]--normally I would expect the pos values to be continuous.
Just to make sure that I understand You correctly - do You see the
probpos-a = asin(sin(joints[5]) / sin(xy_tan));lem in the fact that I
want to switch from this formula pos-a = asin(sin(joints[4]) /
cos(xy_tan)); to this formula pos-a = asin(sin(joints[5]) /
sin(xy_tan));?
If this is the case, then how should I treat the situation that for
xy_tan values 90 and 270 the cos(xy_tan) = 0 and dividing by zero is
error. How to solve that? In those situations the nozzle will be
moving along Y axis and all the tilt should be in joint[5], joint[4]
should be in its zero point. So in this situation joint[4] = 0, but
pos-a must not be zero. The same situation is for xy_tan values 0 and
180 - joint [5] should be in its zero point and all the tilt should be
in joint[4].
Are there any suggestions?
Just to remind - I am trying to implement kinematics, where cutting
head with A and B rotary axis (rotating around X and Y axis
respectively) are behaving as a cutting head with A and C rotary axis
(rotating around X and Z axis respectively) where A angle (amount of
the tilt of the head) is given in g-code, but C is calculated as a
tangent of the head movement, so that direction of tilt is ALWAYS
perpendicular to direction of movement.
It seems to me that I have managed implementation of calculating the
tangent, and that the struggle is with transforming these A-C style
kinematics into A-B style cutting head.
2010/6/11 yann jautard brico...@free.fr:
Viesturs Lācis wrote:
1) What would correct syntax for line 18 and 19 look like?
could be something like :
18 for (int i=1; i limit; i++){
19 if (xy_tan == (90+(180*i)) || xy_tan == (90-(180*i)))
19 pos-a = asin(sin(joints[5]) / sin(xy_tan));
20 }
where limit is the maximal value I can take.
this code is not optimised, because for large limit value, il will
take a long time to iterate. But should work right for a limit value of
5 or 10.
Thank You for the code!, I will try it.
with best regards,
Viesturs
--
ThinkGeek and WIRED's GeekDad team up for the Ultimate
GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the
lucky parental unit. See the prize list and enter to win:
http://p.sf.net/sfu/thinkgeek-promo
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] How to create custom kinematics module?
If atan2() returns a negative result, add 360 degrees (or the equivalent in
radians) to the result.
Neil
2010/6/11 Viesturs Lācis viesturs.la...@gmail.com
2010/6/11 Przemek Klosowski przemek.klosow...@gmail.com:
1 #include rtapi_math.h
2 #include kinematics.h /* these decls */
3
4 const double head_offset = 200;
5
6 PmCartesian old;
7
8 in t kinematicsForward(const double *joints,
9 EmcPose * pos,
10const KINEMATICS_FORWARD_FLAGS * fflags,
11KINEMATICS_INVERSE_FLAGS * iflags)
12 {
13 double xy_tan = atan2(pos-tran.y - old.y,pos-tran.x - old.x);
14 pos-tran.x = joints[0] - head_offset * sin(joints[5]);
15 pos-tran.y = joints[2] + head_offset * sin(joints[4]);
16 pos-tran.z = joints[3] + cos(joints[4]) * cos(joints[5]) *
head_offset;
17 pos-a = asin(sin(joints[4]) / cos(xy_tan));
18 if xy_tan = 90 + (180 * i), where i is not 0, but is 1, 2, 3, etc
or -1, -2, -3, -4 etc
19 then pos-a = asin(sin(joints[5]) / sin(xy_tan));
OK, but what is 'i'? or are you saying that the test is for
(xy_tan-90) being an integer multiple of 180? without understanding
your geometry, this doesn't make immediate sense to me...
If You recall trignometry from elementary school, tangent values (just
like for sin and cos functions) repeat every 180 degrees. For example
tangent of 45 degrees is 1. Just like it is for 225, 405 etc. General
way to write it is tan(45 + 180*i) = 1, where i is integer - 1, 2, 3,
4, 5 etc.
Actually I seem to have made this bit more complicated as necessary -
for me xy_tan should be in range from 0 to 359 (including both of
these numbers), so instead of writing 90 + (180 * i), I could have
said 90 or 270.
By the way, how should I limit the xy_tan in the mentioned range from 0 to
359?
For example, for deltaX = 1, deltaY = -1 in Excel for atan2 function I
get value -45 degrees, while I would like it to be 315. This way each
position for xy_would be described with only one possible value
instead of several. Like in the given example, it can be -45 or +315
etc.
because you whip the value of pos-a betweenthe values based on joints[5]
and
joints[4]--normally I would expect the pos values to be continuous.
Just to make sure that I understand You correctly - do You see the
probpos-a = asin(sin(joints[5]) / sin(xy_tan));lem in the fact that I
want to switch from this formula pos-a = asin(sin(joints[4]) /
cos(xy_tan)); to this formula pos-a = asin(sin(joints[5]) /
sin(xy_tan));?
If this is the case, then how should I treat the situation that for
xy_tan values 90 and 270 the cos(xy_tan) = 0 and dividing by zero is
error. How to solve that? In those situations the nozzle will be
moving along Y axis and all the tilt should be in joint[5], joint[4]
should be in its zero point. So in this situation joint[4] = 0, but
pos-a must not be zero. The same situation is for xy_tan values 0 and
180 - joint [5] should be in its zero point and all the tilt should be
in joint[4].
Are there any suggestions?
Just to remind - I am trying to implement kinematics, where cutting
head with A and B rotary axis (rotating around X and Y axis
respectively) are behaving as a cutting head with A and C rotary axis
(rotating around X and Z axis respectively) where A angle (amount of
the tilt of the head) is given in g-code, but C is calculated as a
tangent of the head movement, so that direction of tilt is ALWAYS
perpendicular to direction of movement.
It seems to me that I have managed implementation of calculating the
tangent, and that the struggle is with transforming these A-C style
kinematics into A-B style cutting head.
2010/6/11 yann jautard brico...@free.fr:
Viesturs Lācis wrote:
1) What would correct syntax for line 18 and 19 look like?
could be something like :
18 for (int i=1; i limit; i++){
19 if (xy_tan == (90+(180*i)) || xy_tan == (90-(180*i)))
19 pos-a = asin(sin(joints[5]) / sin(xy_tan));
20 }
where limit is the maximal value I can take.
this code is not optimised, because for large limit value, il will
take a long time to iterate. But should work right for a limit value of
5 or 10.
Thank You for the code!, I will try it.
with best regards,
Viesturs
--
ThinkGeek and WIRED's GeekDad team up for the Ultimate
GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the
lucky parental unit. See the prize list and enter to win:
http://p.sf.net/sfu/thinkgeek-promo
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
--
http://www.pixpopuli.com
--
ThinkGeek and WIRED's GeekDad team up for the
Re: [Emc-users] lantency calculation
thanks for the answer Jeff,, now, how do we calculate base_period and servo period? Jeff Epler jep...@... writes: The latency-test runs the following realtime component: http://git.linuxcnc.org/gitweb?p=emc2.git;a=blob;f=src/hal/components/timedelta.comp Each time a realtime thread runs, the time interval del=(now-last) is computed. The minimum and maximum del ever seen are computed in min_ and max_. The jitter is the bigger of max_-period or period-min_. period is the promised realtime interval in nanoseconds, and rtapi_get_time() is the measured wall time also in nanoseconds. The timedelta component also measures long term error in 'err' and 'avg_err'. These values are not shown in the latency-test GUI but can be used to determine if there are long-term differences between expected and actual intervals. Jeff -- ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
