Jerk Acceleration Settings and HAAS G187 Acceleration Profiles #593

Dietz0r · 2024-09-29T19:29:35Z

This adds code to implement 3rd order accleration settings over the acceleration ticks of the stepper loop.

With the default settings of 100 acceleration ticks it will update the acceleration values according to the jerk setting with each 10ms block that gets fed into the stepper loop.
If a CPU is strong enough to run 1000 acceleration ticks that would net a 1ms Jerk Stepper Loop but 100 or 200 ticks work fine.
Tested on 2 independent machines with help from @EmpyreanCNC

Known Issue:
The machine moves a bit erratically/stuttery when getting fed jog commands from either a pendant or the IOSender interface, i think this has to do with how those conitinous jog commands are fed to the controller but i wasnt able to veryify or fix the issue on my end. Normal G-Code file execution works as expected. I hope you might have more insight into why this problem arises.

Also added support for HAAS style G187 acceleration profiles currently set to 20%, 40% 60% 80% and 100% of the max settings to allow for fine control during toolpathes and improved surface finishes.
Gets called as G187 P1 for 100% speed roughing to G187 P5 for 20% slow finishing.

Hope this helps! :)

First experimental checkin for constant jerk motions

missing semicolon

bugfixing float variables

DEFAULT_X_JERK capitalization

static char added for axis_jerk

Fixed some typos and syntax errors.

Dietz0r · 2024-10-30T10:22:28Z

FYI I'll be back home in a week...

Hope you enjoyed your trip! :)

This means that there should be a setting for the default P value? And cancellation implemented?

I think the most reliable thing would be to have the default value be 100%, so the settings that are set on the axis and be modified from there with the profiles. Saves a setting slot and if you really want to just run at 80% max it's trivial to just add G187 P4 at the start of any code.

The cancellation is a good point! didnt think of that! Will look at it and add it.

This should be !gc_block.words.p to check if a value has been supplied. IMO also check gc_block.words.e, if supplied (true) error out since it is not supported?

So no one accidentally tries to use a real HAAS routine on it? yeah i see it.

settings.c

terjeio · 2024-10-30T12:51:34Z

settings.c

+ActiveAccelProfile = 1; // Initialize machine with 100% Profile
+
+//Acceleration Profiles for G187 P[x] in percent of maximum machine acceleration.
+float LookupProfile(uint8_t Profile) {


Move to planner.c as static inlined?
and the return value is either 0 or 1 since you assign the value to Profile before returning it? Use return lookup(profile); instead?
BTW I do not use camel cased or upper case names in the core.

Moving it to planner.c as static inlined would make it very hard to find if you wanted to have other profiles instead of the generic placeholder 20% ones.

I shied away from making it yet another list of $ settings because it would likely mean a whole slew of settings for however much profiles you wanted. having the lookuptable in settings.c makes it findable though.

maybe move it to the driver my_machine.h because it is a compile time setting?

…ling accordingly

terjeio

I am starting to believe that this PR should be deferred until I get an overdue settings revision done. This so that I do not run into issues later on. I'll look into this in the next few days.

terjeio · 2024-11-30T11:54:18Z

planner.c

@@ -348,9 +348,28 @@ static inline float limit_acceleration_by_axis_maximum (float *unit_vec)
        if (unit_vec[--idx] != 0.0f)  // Avoid divide by zero.
            limit_value = min(limit_value, fabsf(settings.axis[idx].acceleration / unit_vec[idx]));
    } while(idx);
+#if ENABLE_ACCELERATION_PROFILES
+    limit_value *= lookupprofile(gc_state.modal.activeaccelprofile);


The profile factor should be passed from gcode.c in the plan_block_t structure. And set it in the plan_data_init() too, perhaps to 1.0?

I rewrote it so it uses the plan_block_t structure and plan_data_init() function.

How it should now work is the machine initializes with active profile 1 - 1.0 factor. in gcode.c (including reset handling)
plan_data_init() looks up the current factor with the lookupfactor() function.
plan_block_t now has storage for the current_factor and gets it from pl_data->current_factor to calucalte adjusted acceleration limits via limit_value *= block->acceleration_factor;

hope i didnt mess that up too badly

settings.h

terjeio · 2024-11-30T12:24:27Z

settings.h

@@ -714,6 +718,9 @@ typedef struct {
    float steps_per_mm;
    float max_rate;
    float acceleration;
+#if ENABLE_JERK_ACCELERATION   
+    float jerk;
+#endif


This is another potential problem we have to be aware of, the area reserved for global settings may overflow when many axes are configured. I have already switched to claim settings storage for a second PWM spindle from above the area reseved for the core. Perhaps the jerk settings should as well? I'll have to think about that. See nvs.h.

I have just commited a version with updated settings structures and that adds a permanent field for jerk. I have also reserved ten new axis settings for the core, jerk beeing the first assigned.

If you can update your PR against this version then I will be able to merge it without risking issues with the settings.

Currently at work, will take a look and push latest revision tomorrow night or maybe wednesday!
Thanks for all the work! :)

resolved the conflicts. :)

…ction to plan_data_init, renamed lookup fuction and adjusted the acceleration limit functions to new structure

terjeio · 2024-12-26T11:04:43Z

Is this nearly ready for merge now?
FYI settings.c is out of sync with the core, one removed function is added again (set_offset_lock) and at least one modified one is duplicated by adding the earlier code (set_homing_enable).

Cleaned up out of sync functions. Most likely a remnant from resolving the merge conflict.

Dietz0r · 2024-12-26T21:15:54Z

Is this nearly ready for merge now? FYI settings.c is out of sync with the core, one removed function is added again (set_offset_lock) and at least one modified one is duplicated by adding the earlier code (set_homing_enable).

The out of sync file might have been from cleaning up the merge conflicts, my apologes.
I just did a test compile to make sure it works. Found one missing declaration.
It does compile but i didnt have time yet to verify that the new G187 code works as it should.

So apart from physical testing i guess it's ready?

edit: just wait a sec, found another issue

… data

Dietz0r · 2024-12-26T21:51:30Z

Okay, seems to be looking good now on my end.

Edit:
Managed to find a spare teensy and did some debugging on it.
Jerk works as expected on gcode files. The setting limit calculations were causing issues so i removed them and adjusted the setting description for "sane" values.
The jogging issue remains. Not sure why but the machine is adamant about going as slow as possible when it reaches the mm_remaining * speed < current_accel / jerk point on a decelleration ramp but it works fine in gcode?

G187 implementation is currently having 2 issues. It throws an error36 after the "ok" when changing p-values and the controller crashes when issuing any movements with it enabled.

It's late now and i will continue debugging later.

terjeio · 2024-12-27T06:44:13Z

planner.c

@@ -339,7 +339,7 @@ inline static float plan_compute_profile_parameters (plan_block_t *block, float
    return nominal_speed;
 }

-static inline float limit_acceleration_by_axis_maximum (float *unit_vec)
+static inline float limit_acceleration_by_axis_maximum (float *unit_vec,)
 {


Stray comma?

Yep, removed.

terjeio · 2024-12-27T07:05:32Z

settings.h

@@ -712,7 +712,9 @@ typedef struct {
    float dual_axis_offset;
    float homing_seek_rate;
    float homing_feed_rate;
+#if ENABLE_JERK_ACCELERATION 


Remove this #if/#endif, it will trigger a settings reset when enabling/disabling jerk.

…h conditional

Dietz0r · 2024-12-27T12:00:12Z

Also fixed the G187 Unused P Value error. The crash on motion command still persists. :/

Edit: Found the issue for the crashes. The plan_data_init() doesnt initialize it.

(this is rewritten code that calls the lookupfunction from inside the block)
even when written as:

#if ENABLE_ACCELERATION_PROFILES
    plan_data->acceleration_factor = 1.0f;
    debug_printf("plan_data accel factor: %f", plan_data->acceleration_factor);
#endif

it stays at 0 and the debug print doesnt get called either.
but i added it to the plan_line_data_t typescructure. and it's also in the init(). any idea why @terjeio ?

…input checking

terjeio · 2024-12-27T16:05:51Z

No crashes for me, but then I have modified your code a bit, in gcode.c:

Simplified this a bit, the FAIL macro is a return statement so no code will be executed after.
And p < 1 is not neccesarily a negative value...
gc_state.modal.acceleration_profile holds the factor, should be renamed to match plan_data.acceleration_factor.

#if ENABLE_ACCELERATION_PROFILES

        case NonModal_SetAccelerationProfile:

            if(gc_block.words.e)
                FAIL(Status_GcodeUnsupportedCommand);

            if(gc_block.words.p && (gc_block.values.p < 1.0f || gc_block.values.p > 5.0f))
                FAIL(Status_GcodeValueOutOfRange);

            gc_state.modal.acceleration_profile = lookupfactor(gc_block.words.p ? (uint8_t)gc_block.values.p - 1 : 0);
            gc_block.words.p = Off;
            break;

#endif

Copying the factor to plan_data passed to the planner is missing in your code:

    // Initialize planner data struct for motion blocks.
    plan_line_data_t plan_data;
    memset(&plan_data, 0, sizeof(plan_line_data_t)); // Zero plan_data struct
    plan_data.offset_id = gc_state.offset_id;
    plan_data.condition.target_validated = plan_data.condition.target_valid = sys.soft_limits.mask == 0;
#if ENABLE_ACCELERATION_PROFILES
    plan_data.acceleration_factor = gc_state.modal.acceleration_profile;
#endif

…e!) added the line for initilizing the acceleration_factor, renamed acceleration_profiles variable to acceleration_factor to match the rest of the code

Dietz0r · 2024-12-27T16:52:23Z

tested and working ... this took a while <3 thanks for all your help

… in G0 and G80 modal states. Ref. issue #644. Added optional support for 3rd order acceleration (jerk) and G187 gcode. Ref. pull request #593.

Dietz0r and others added 30 commits October 30, 2023 11:37

Update planner.c

c8619f5

First experimental checkin for constant jerk motions

Update planner.h

f003479

First experimental checkin for constant jerk motions

Update stepper.c

699e943

First experimental checkin for constant jerk motions

Update planner.c

2484538

Added Axis settings for jerk

4c3cf5e

Decel Ramp Calculations Updated

418e063

added missing semicolons

fb082f3

Update planner.c

f00bb4d

missing semicolon

Update settings.c

b796931

bugfixing float variables

Update config.h

9951e03

DEFAULT_X_JERK capitalization

Update settings.c

06ac1ea

static char added for axis_jerk

Update settings.c

f295c36

adjusted end of accel calculation

92e750d

Merge branch 'grblHAL:master' into master

a268962

cleaned up config and added settings description

2361610

added set_axis_setting functions for jerk settings

5bb5945

fixed typo

8667292

included extended axis setting in get_float check

04708e9

Merge branch 'grblHAL:master' into master

0789f6e

Merge branch 'grblHAL:master' into master

2930759

Merge branch 'grblHAL:master' into master

0863959

Added Min/Max calculations for Jerk

5cbb222

Implemented preliminary G187 support for up to 5 Profiles

4c28e42

Unittype and Brackets Hotfix for G187

554812d

Calculation adjustment for G187

299a7cd

Fixed some typos and syntax errors.

dec93aa

Merge pull request #1 from Expatria-Technologies/PR_Flexi_Fix

860659f

Fixed some typos and syntax errors.

Added compiletime check for added acceleration profile code

4cc7aaf

Encapsulated G-Code Function with compiletime option

9f95af1

Added compile time option for jerk controlled motion

e645ddb

terjeio reviewed Oct 30, 2024

View reviewed changes

settings.c Outdated Show resolved Hide resolved

terjeio reviewed Oct 30, 2024

View reviewed changes

Dietz0r added 2 commits October 31, 2024 21:22

added acceleration profiles to gc_state, adjusted g187 and m2/30 hand…

b7d54d7

…ling accordingly

adjusted settings.h to last commit too

88cf48a

terjeio reviewed Nov 30, 2024

View reviewed changes

distebia approved these changes Dec 7, 2024

View reviewed changes

Dietz0r added 5 commits December 10, 2024 21:10

Resolve merge conflict for new setting structure

a6ae08e

Merge branch 'master' into master

075d5ce

moved acceleration factor to plan_block_t structure, moved lookup fun…

e504964

…ction to plan_data_init, renamed lookup fuction and adjusted the acceleration limit functions to new structure

missed 2 variable names

c4969f1

file didnt want to push. found a typo

3379524

Dietz0r added 3 commits December 26, 2024 17:03

Merge branch 'grblHAL:master' into master

df6aec3

Update settings.c

254a3d1

Cleaned up out of sync functions. Most likely a remnant from resolving the merge conflict.

Update gcode.h

efcdf3f

adjusted the limit by axis maximum functions to not require the block…

ce60b8f

… data

Debugging and Compiling

c47e6a3

terjeio reviewed Dec 27, 2024

View reviewed changes

Fixed the Unused Values error, removed stray comma, removed settings.…

9d48143

…h conditional

Adjusted lookup function for array counting (start from 0) and added …

719017b

…input checking

Cleaned up the Switchcase coder after the debugging mess (thanks terj…

84657b4

…e!) added the line for initilizing the acceleration_factor, renamed acceleration_profiles variable to acceleration_factor to match the rest of the code

terjeio merged commit b0322fd into grblHAL:master Dec 30, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Jerk Acceleration Settings and HAAS G187 Acceleration Profiles #593

Jerk Acceleration Settings and HAAS G187 Acceleration Profiles #593

Dietz0r commented Sep 29, 2024 •

edited

Loading

Dietz0r commented Oct 30, 2024

terjeio Oct 30, 2024

Dietz0r Oct 31, 2024

terjeio left a comment

terjeio Nov 30, 2024

Dietz0r Dec 10, 2024

terjeio Nov 30, 2024

terjeio Dec 9, 2024

Dietz0r Dec 9, 2024

Dietz0r Dec 10, 2024

terjeio commented Dec 26, 2024

Dietz0r commented Dec 26, 2024 •

edited

Loading

Dietz0r commented Dec 26, 2024 •

edited

Loading

terjeio Dec 27, 2024

Dietz0r Dec 27, 2024

terjeio Dec 27, 2024

Dietz0r Dec 27, 2024

Dietz0r commented Dec 27, 2024 •

edited

Loading

terjeio commented Dec 27, 2024

Dietz0r commented Dec 27, 2024

Jerk Acceleration Settings and HAAS G187 Acceleration Profiles #593

Jerk Acceleration Settings and HAAS G187 Acceleration Profiles #593

Conversation

Dietz0r commented Sep 29, 2024 • edited Loading

Dietz0r commented Oct 30, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

terjeio left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

terjeio commented Dec 26, 2024

Dietz0r commented Dec 26, 2024 • edited Loading

Dietz0r commented Dec 26, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Dietz0r commented Dec 27, 2024 • edited Loading

terjeio commented Dec 27, 2024

Dietz0r commented Dec 27, 2024

Dietz0r commented Sep 29, 2024 •

edited

Loading

Dietz0r commented Dec 26, 2024 •

edited

Loading

Dietz0r commented Dec 26, 2024 •

edited

Loading

Dietz0r commented Dec 27, 2024 •

edited

Loading