- The newest Arduino IDE 1.6.12 has recently updated to avr-gcc v4.9.2.
Unfortunately, it produces a compiled size almost 0.7KB to 1KB larger
than prior versions! This can easily cause the base build to exceed the
Arduino Duemilanove/Nano flash limit of 30.5KB. The Arduino Uno seems
to be ok still with its 31.5KB flash limit.
- Makefile `-flto` compile flag added to cut down on the horrible flash
size when using the new avr-gcc. (Edit Makefile and remove comment on
COMPILE definition). This brings it in-line with what the IDE produces.
- Functionalized repetitive tasks in report.c to try to reduce overall
flash size. Successfully cut down about 160bytes.
- Removed printFloat_SettingValue() and printFloat_RPMValue()
functions. These aren’t required and can be replaced with a direct call
to printFloat() because they don’t require a unit conversion check.
- Reverted back the serial RX count function to how it was. The
variable type was unsigned and cause an integer underflow whenever the
calculation produced a negative number. The old way was the correct way.
- Lots of minor edits to the code CSVs and markdown documents.
- Expanded on explaining feedback messages and startup line execution
feedback.
- Created a new settings codes CSV to help GUIs import the values and
meanings.
- Incremented to v1.1a, rather than keep 1.0e. This is because there
are existing v1.0 installations. Don’t want to confuse people further.
- Certain version of the Arduino IDE did not like the `inline` in the
function header. Removed from spindle_control files to fix the problem.
- Tweaked the communication protocol slightly. Added message type
indicators for all `[]`bracketed feedback messages. It’s been
problematic for GUI dev to try to determine the context of a message
and how it should be handled. These indictors should help tremendously
to remove context all together.
- Also altered how `$N` startup lines are presented when executed. They
now start with an open chevron ‘>’ followed by the line and an ‘:ok’ to
indicate it executed. The ‘ok’ is on the same line intentionally so it
doesn’t mess up a streaming protocol counter.
- Managed to save a 100+KB from refactoring parts of report.c. (Thanks
Vasilis!) Freed up room to alter the protocol a little.
- Wrote a markdown document on interface messaging to make it clear how
it’s intended to work. See interface.md in /doc/markdown
- Started to pull in some Wiki pages from the old grbl site and
beginning to update them for v1.1.
- Created new commit log for v1.1.
- Feature: Realtime feed, rapid, and spindle speed overrides. These
alter the running machine state within tens of milliseconds!
- Feed override: 100%, +/-10%, +/-1% commands with values 1-200% of
programmed feed
- Rapid override: 100%, 50%, 25% rapid rate commands
- Spindle speed override: 100%, +/-10%, +/-1% commands with values
50-200% of programmed speed
- Override values have configurable limits and increments in
config.h.
- Feature: Realtime toggle overrides for spindle stop, flood coolant,
and optionally mist coolant
- Spindle stop: Enables and disables spindle during a feed hold.
Automatically restores last spindles state.
- Flood and mist coolant: Immediately toggles coolant state until
next toggle or g-code coolant command.
- Feature: Jogging mode! Incremental and absolute modes supported.
- Grbl accepts jogging-specific commands like $J=X100F50. An axis
word and feed rate are required. G20/21 and G90/G91 commands are
accepted.
- Jog motions can be canceled at any time by a feed hold `!`
command. The buffer is automatically flushed. (No resetting required).
- Jog motions do not alter the g-code parser state so GUIs don’t
have to track what they changed and correct it.
- Feature: Laser mode setting. Allows Grbl to execute continuous
motions with spindle speed and state changes.
- Feature: Significantly improved status reports. Overhauled to cram in
more meaningful data and still make it smaller on average.
- All available data is now sent by default, but does not appear if
it doesn’t change or is not active.
- Machine position(MPos) or work position(WPos) is reported but not
both at the same time. Instead, the work coordinate offsets (WCO)are
sent intermittently whenever it changes or refreshes after 10-30 status
reports. Position vectors are easily computed by WPos = MPos - WCO.
- All data has changed in some way. Details of changes are in the
markdown documents and wiki.
- Feature: 16 new realtime commands to control overrides. All in
extended-ASCII character space.
- While they are not easily typeable and requires a GUI, they can’t
be accidentally triggered by some latent character in the g-code
program and have tons of room for expansion.
- Feature: New substates for HOLD and SAFETY DOOR. A `:x` is appended
to the state, where `x` is an integer and indicates a substate.
- For example, each integer of a door state describes in what phase
the machine is in during parking. Substates are detailed in the
documentation.
- Feature: With the alarm codes, homing and probe alarms have been
expanded with more codes to provide more exact feedback on what caused
the alarm.
- Feature: New hard limit check upon power-up or reset. If detected, a
feedback message to check the limit switches sent immediately after the
welcome message.
- May be disabled in config.h.
- OEM feature: Enable/disable `$RST=` individual commands based on
desired behavior in config.h.
- OEM feature: Configurable EEPROM wipe to prevent certain data from
being deleted during firmware upgrade to a new settings version or
`RST=*` command.
- OEM feature: Enable/disable the `$I=` build info write string with
external EEPROM write example sketch.
- This prevents a user from altering the build info string in
EEPROM. This requires the vendor to write the string to EEPROM via
external means. An Arduino example sketch is provided to accomplish
this. This would be useful for contain product data that is
retrievable.
- Tweak: All feedback has been drastically trimmed to free up flash
space for the v1.0 release.
- The `$` help message is just one string, listing available
commands.
- The `$$` settings printout no longer includes descriptions. Only
the setting values. (Sorry it’s this or remove overrides!)
- Grbl `error:` and `ALARM:` responses now only contain codes. No
descriptions. All codes are explained in documentation.
- Grbl’s old feedback style may be restored via a config.h, but
keep in mind that it will likely not fit into the Arduino’s flash space.
- Tweak: Grbl now forces a buffer sync or stop motion whenever a g-code
command needs to update and write a value to EEPROM or changes the work
coordinate offset.
- This addresses two old issues in all prior Grbl versions. First,
an EEPROM write requires interrupts to be disabled, including stepper
and serial comm. Steps can be lost and data can be corrupted. Second,
the work position may not be correlated to the actual machine position,
since machine position is derived from the actual current execution
state, while work position is based on the g-code parser offset state.
They are usually not in sync and the parser state is several motions
behind. This forced sync ensures work and machine positions are always
correct.
- This behavior can be disabled through a config.h option, but it’s
not recommended to do so.
- Tweak: To make status reports standardized, users can no longer
change what is reported via status report mask, except for only
toggling machine or work positions.
- All other data fields are included in the report and can only be
disabled through the config.h file. It’s not recommended to alter this,
because GUIs will be expecting this data to be present and may not be
compatible.
- Tweak: Homing cycle and parking motion no longer report a negative
line number in a status report. These will now not report a line number
at all.
- Tweak: New `[Restoring spindle]` message when restoring from a
spindle stop override. Provides feedback what Grbl is doing while the
spindle is powering up and a 4.0 second delay is enforced.
- Tweak: Override values are reset to 100% upon M2/30. This behavior
can be disabled in config.h
- Tweak: The planner buffer size has been reduced from 18 to 16 to free
up RAM for tracking and controlling overrides.
- Tweak: TX buffer size has been increased from 64 to 90 bytes to
improve status reporting and overall performance.
- Tweak: Removed the MOTION CANCEL state. It was redundant and didn’t
affect Grbl’s overall operation by doing so.
- Tweak: Grbl’s serial buffer increased by +1 internally, such that 128
bytes means 128, not 127 due to the ring buffer implementation. Long
overdue.
- Tweak: Altered sys.alarm variable to be set by alarm codes, rather
than bit flags. Simplified how it worked overall.
- Tweak: Planner buffer and serial RX buffer usage has been combined in
the status reports.
- Tweak: Pin state reporting has been refactored to report only the
pins “triggered” and nothing when not “triggered”.
- Tweak: Current machine rate or speed is now included in every report.
- Tweak: The work coordinate offset (WCO) and override states only need
to be refreshed intermittently or reported when they change. The
refresh rates may be altered for each in the config.h file with
different idle and busy rates to lessen Grbl’s load during a job.
- Tweak: For temporary compatibility to existing GUIs until they are
updated, an option to revert back to the old style status reports is
available in config.h, but not recommended for long term use.
- Tweak: Removed old limit pin state reporting option from config.h in
lieu of new status report that includes them.
- Tweak: Updated the defaults.h file to include laser mode, altered
status report mask, and fix an issue with a missing invert probe pin
default.
- Refactor: Changed how planner line data is generated and passed to
the planner and onto the step generator. By making it a struct
variable, this saved significant flash space.
- Refactor: Major re-factoring of the planner to incorporate override
values and allow for re-calculations fast enough to immediately take
effect during operation. No small feat.
- Refactor: Re-factored the step segment generator for re-computing new
override states.
- Refactor: Re-factored spindle_control.c to accommodate the spindle
speed overrides and laser mode.
- Refactor: Re-factored parts of the codebase for a new jogging mode.
Still under development though and slated to be part of the official
v1.0 release. Hang tight.
- Refactor: Created functions for computing a unit vector and value
limiting based on axis maximums to free up more flash.
- Refactor: The spindle PWM is now set directly inside of the stepper
ISR as it loads new step segments.
- Refactor: Moved machine travel checks out of soft limits function
into its own since jogging uses this too.
- Refactor: Removed coolant_stop() and combined with
coolant_set_state().
- Refactor: The serial RX ISR forks off extended ASCII values to
quickly assess the new override realtime commands.
- Refactor: Altered some names of the step control flags.
- Refactor: Improved efficiency of the serial RX get buffer count
function.
- Refactor: Saved significant flash by removing and combining print
functions. Namely the uint8 base10 and base2 functions.
- Refactor: Moved the probe state check in the main stepper ISR to
improve its efficiency.
- Refactor: Single character printPgmStrings() went converted to direct
serial_write() commands to save significant flash space.
- Documentation: Detailed Markdown documents on error codes, alarm
codes, messages, new real-time commands, new status reports, and how
jogging works. More to come later and will be posted on the Wiki as
well.
- Documentation: CSV files for quick importing of Grbl error and alarm
codes.
- Bug Fix: Applied v0.9 master fixes to CoreXY homing.
- Bug Fix: The print float function would cause Grbl to crash if a
value was 1e6 or greater. Increased the buffer by 3 bytes to help
prevent this in the future.
- Bug Fix: Build info and startup string EEPROM restoring was not
writing the checksum value.
- Bug Fix: Corrected an issue with safety door restoring the proper
spindle and coolant state. It worked before, but breaks with laser mode
that can continually change spindle state per planner block.
- Bug Fix: Move system position and probe position arrays out of the
system_t struct. Ran into some compiling errors that were hard to track
down as to why. Moving them out fixed it.
- Mega2560 support has been moved to the Grbl-Mega
[project](http://github.com/gnea/grbl-Mega/) to clean up the code and
make future developments easier with increased flash and RAM. All new
developments between the 328p and Mega2560 will be synced when it makes
sense to.
- OEM single file compile configuration option. Before OEMs needed to
alter three files. Provided a way to just alter the config.h file to
contain everything for a particular Grbl build. See config.h for more
details.
- Removed defaults and cpu_map directories and reverted back to
defaults.h and cpu_map.h to contain all definitions. This should help
reduce some headaches and the previous implementation inadvertently
created. Also, it makes the single file config.h possible.
- Moved (and tweaked) the invert control pin mask define and placed
into config.h, rather than in the cpu_map.h file. Makes more sense
there.
- Soft limit errors were stuck in a feed hold without notifying the
user why it was in a hold. When resumed, the soft limit error would
kick in. Issue should be fixed to behave as intended to automatically
hold and issue a soft limit alarm once the machine has come to a stop.
- Soft limit errors were stuck in a feed hold without notifying the
user why it was in a hold. When resumed, the soft limit error would
kick in. Issue should be fixed to behave as intended. To automatically
hold and issue a soft limit alarm once the machine has come to a stop.
- Planner was under-estimating maximum speeds through straight
junctions in certain cases. The calculations have been updated to be
more accurate.
- Strange sizeof() bug in the most recent releases. Manifested as an
alarm upon a power up even when homing was disabled. Fixed by declaring
sizeof() with struct types, rather than variable names, even though
they were validated to give the same value.
- Spindle speed zero should disable the spindle. Now fixed.
- New configuration option for inverting certain limit pins. Handy for
mixed NO and NC switch machines. See config.h for details.
- New pin state realtime reporting feature. Instead of `Lim:000` for
limit state reports, the new feature shows `Pin:000|0|0000`, or
something similar. The `|` delimited fields indicate xyz limits, probe,
and control pin states, where 0 is always not triggered, and 1 is
triggered. Invert masks ARE accounted for.
Each field may be enabled or disabled via the `$10` status report
setting. The probe and control pin flags are bits 5 and 6, respectively.
- Remove the now deprecated `REPORT_CONTROL_PIN_STATE` option in
config.h
- The old limit pin reports `Lim:000` may be re-enabled by commenting
out `REPORT_ALL_PIN_STATES` in config.h.
- Incremented the version letter (v1.0c) to indicate the change in
reporting style.
- Replaced all bit_true_atomic and bit_false_atomic macros with
function calls. This saved a couple hundred bytes of flash.
- G38.x was not printing correctly in the $G g-code state reports. Now
fixed.
- When investigating the above issue, it was noticed that G38.x
wouldn’t show at all, but instead a G0 would be printed. This was
unlike the v0.9j master build. It turned out volatile variables do not
like to be defined inside a C struct. These are undefined on how to be
handled. Once pulled out, all weird issues went away.
- Also changed two ‘sizeof()’ statements in the mc_probe() and
probe_state_monitor() functions to be more robust later on.
- Updated the commit logs to individual files for each minor release.
Forgot to update the generating script to account for this.
- G38.x was not printing correctly in the $G g-code state reports. Now
fixed.
- Potential bug regarding volatile variables inside a struct. It has
never been a problem in v0.9, but ran into this during v1.0
development. Just to be safe, the fixes are applied here.
- Updated pre-built firmwares with these two bug fixes.
- G28 and G30’s behavior has been updated from the old NIST g-code
standard to LinuxCNC’s. Previously when an intermediate motion was
programmed, the NIST standard would move all axes to the final G28/30
stored coordinates. LinuxCNC states it only moves the axes specified in
the command.
For example, suppose G28’s stored position is (x,y,z) = (1,2,3) for
simplicity, and we want to do an automated z-axis tool retraction and
then park at the x,y location. `G28 G91 Z5` will first move the Z axis
5mm(or inches) up, then move Z to position 3 in machine coordinates.
Next, the command `G28 G91 X0 Y0` would skip the intermediate move
since distance is zero, but then move only the x and y axes to machine
coordinates 1 and 2, respectively. The z-axis wouldn’t move in this
case, since it wasn’t specified.
This change is intended to make Grbl more LinuxCNC compatible while
making commands, like the shown tool retraction, much easier to
implement.
- Version bump requested by OEMs to easily determine whether the
firmware supports the new EEPROM reset feature. Other than that, no
significant changes.
- Tweaked the previous EEPROM restore implementation and added new
functionality.
- `$RST=$` restores the `$$` grbl settings back to firmware defaults,
which are set when compiled.
- `$RST=#` restores the `$#` parameters in EEPROM. At times it’s useful
to clear these and start over, rather than manually writing each entry.
-`$RST=*` wipe all of the data in EEPROM that Grbl uses and restores
them to defaults. This includes `$$` settings, `$#` parameters, `$N`
startup lines, and `$i` build info string.
NOTE: This doesn’t write zeros throughout the EEPROM. It only writes
where Grbl looks for data. For a complete wipe, please use the Arduino
IDE’s EEPROM clear example.
- Refactored the restore and wipe functions in settings.c to
accommodate the new commands.
- New restore setting defaults command. Only wipes ‘$$’ setting in
EEPROM and reloads them based on the defaults used when Grbl was
compiled. Used with a `$RST` command
NOTE: `$RST` is intentionally not listed in the Grbl ‘$’ help message.
- Updated new homing cycle to error out when a pull-off motion detects
the limit is still active.
- Created a limits_get_state() function to centralize it. It reports
state as a bit-wise booleans according to axis numbering.
- Updated the print uint8 functions. Generalized it to allow both base2
and base10 printouts, while allowing base2 prints with N_AXIS digits
for limit state status reports. Doing this saved about 100bytes of
flash as well.
- Applied CoreXY status reporting bug fix by @phd0. Thanks!
- Critical fix for M0 program pause. Due to its recent change, it would
cause Grbl to suspend but wouldn’t notify the user of why Grbl was not
doing anything. The state would show IDLE and a cycle start would
resume it. Grbl now enters a HOLD state to better indicate the state
change.
- Critical fix for M2 and M30 program end. As with M0, the state
previously would show IDLE while suspended. Grbl now does not suspend
upon program end and leaves job control to the GUI. Grbl simply reports
a `[Pgm End]` as a feedback message and resets certain g-code modes.
- M2/30 g-code reseting fix. Previously Grbl would soft-reset after an
M2/30, but this was not complaint to the (linuxcnc) g-code standard. It
simply resets [G1,G17,G90,G94,G40,G54,M5,M9,M48] and keeps all other
modes the same.
- M0/M2/M30 check-mode fix. It now does not suspend the machine during
check-mode.
- Minor bug fix related to commands similar to G90.1, but not G90.1,
not reporting an unsupported command.
- Homing cycle refactoring. To help reduce the chance of users
misunderstanding their limit switch wiring, Grbl only moves a short
distance for the locate cycles only. In addition, the homing cycle
pulls-off the limit switch by the pull-off distance to re-engage and
locate home. This should improve its accuracy.
- HOMING_FORCE_ORIGIN now sets the origin to the pull-off location,
rather than where the limit switch was triggered.
- Updated default junction deviation to 0.01mm. Recent tests showed
that this improves Grbl’s cornering behavior a bit.
- Added the ShapeOko3 defaults.
- Added new feedback message `[Pgm End]` for M2/30 notification.
- Limit pin reporting is now a $10 status report option. Requested by
OEMs to help simplify support troubleshooting.
- Commit history added to repo, as an easier way for people to see view
the changes over time.
- Grbl logo copyright license added. All rights reserved with regards
to the Grbl logo.
- G2/3 full circles would sometime not execute. The problem was due to
numerical round-off of a trig calculation. Added a machine epsilon
define to help detect and correct for this problem. Tested and should
not effect general operation of arcs.
- Added a grbl planner simulation tool that was written in Matlab and
Python. It was used to visualize the inner workings of the planner as a
program is streamed to it. The simulation assumes that the planner
buffer is empty, then filled, and kept filled. This is mainly for users
to see how the planner works.
- Updated some of the compile-time ifdefs when enabling line numbers.
The leaving the un-used line numbers in the function calls eats a
non-neglible amount of flash memory. So the new if-defs remove them.
- ALPHA status. - Multitasking ability with run-time command executions
for real-time control and feedback. - Decelerating feed hold and resume
during operation. - System abort/reset, which immediately kills all
movement and re-initializes grbl. - Re-structured grbl to easily allow
for new features: Status reporting, jogging, backlash compensation. (To
be completed in the following releases.) - Resized TX/RX serial buffers
(32/128 bytes) - Increased planner buffer size to 20 blocks. - Updated
documentation.
