Commit b7b13588 made it possible that the kinematic code could be
restarted after a flush_step_generation() call without a sufficient
delay.
Rename last_sg_flush_time to min_restart_time and use that to ensure
_calc_print_time() always pauses kin_flush_delay time since the last
flush_step_generation() call.
Also, update force_move to invoke flush_step_generation() after any
movements. This is needed to ensure there is a sufficient delay
should force_move be called on a stepper motor that is part of the
toolhead kinematics and is using a step generation "scan time".
This fixes possible "internal error in stepcompress" reports when
using FORCE_MOVE.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
The current code has the mcu report a trsync_state message every 10ms
and expects a time extension within 25ms. However, this means that if
a single mcu->host report is lost then 20ms would elapse until the
next report, which would allow for only a 5ms round-trip time before a
timeout error is reported.
Increase the trsync_state timing so that a message is sent every
7.5ms. This increases the total number of messages per second sent
from mcu to host to 133 (from 100). With this change, a single lost
message would still allow for up to a 10ms round-trip time.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
When multiple MCUs are involved in homing, stagger the scheduling of
the trsync_state report messages from each mcu. Staggering helps
spread the bandwidth, helps reduce locking contention in the host, and
reduces the chance that intermittent latency could result in a
communication timeout.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
It is not valid to call time.sleep() in the host python code (it could
causes glitches in other processing, and it does not ensure there is a
pause between operations on the mcu).
Use minclock instead of time.sleep() to ensure there is a sufficient
pause during chip startup.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Invoke button callbacks directly from the background thread. This
ensures that button notifications are delivered and delivered in the
correct order. Previously, if a callback blocked, it was possible a
new update could start before the previous update was completed, which
could lead to lost events or out of order events.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
When in debugging "batch mode", use the existing method of keeping the
last 30 seconds of history from the furthest planned move time. This
avoids keeping all moves in memory during a batch test.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Move calculation of clear_history_time to the callers of
_advance_flush_time() as a minor processing optimization.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Previously, the BatchBulkHelper class was designed primarily to
register webhook clients, and internal clients used a wrapper class
that emulated a webhooks client.
Change BatchBulkHelper to support regular internal callbacks, and
introduce a new BatchWebhooksClient class that can translate these
internal callback to webhooks client messages.
This makes it easier to register internal clients that can process the
bulk messages every batch interval.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
The APIDumpHelper class is mainly intended to help process messages in
batches. Rework the class methods to make that more clear.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
The APIDumpHelper class already ensures that the start/stop callbacks
will only be called when needed.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
All the accelerometers use a standard response for their query_status
messages. Create a common helper class to process those responses.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Commit b7b13588 changed the internal flush time tracking, but
introduced the possibility of motion restart occurring too close to
the last motion end in some rare cases. This could result in
internal stepcompress errors.
Track the last step generation flush time (last_sg_flush_time) and use
when recalculating the next print_time.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Add support for maximum filament diameter to hall filament width sensor.
If the diameter of the filament diameter is larger than the limit, the
virtual runout sensor will trigger. The default value is set to maximum
flow adjustment threshold to prevent oversized filament from clogging.
Signed-off-by: Sami Haahtinen <ressu@ressukka.net>
Introduce a new step_gen_time variable for flush_step_generation().
This allows need_flush_time to be set to future times without
interfering with flush_step_generation().
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Call toolhead.note_kinematic_activity() on each pin update to ensure
that those updates will be flushed properly.
This fixes "Timer too close" errors on SET_PIN commands that are
issued when the toolhead is idle.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Track a "NeedPrime" queue state instead of the "Flushed" state, and
continue running the background flushing timer as long as there may be
data in any of the move queues.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Rename last_kin_move_time to need_flush_time and rename
force_flush_time to last_flush_time to improve variable name clarity.
Move low-level flushing to new _advance_flush_time() so that it is
possible to flush the queues without needing to advance print_time.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Simplify the code by introducing a separate lookahead priming flush
timer. After this change, the flush_timer is not active in any of the
special queuing states.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Clarify the internal naming to make a more clear distinction between
"stalling" (input not coming fast enough) and "pausing" (the need to
hold up reading of input to avoid buffering too far into the future).
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
These internal low-level config parameters were never documented.
Going forward, developers may modify them by altering the internal
settings in toolhead.py.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
The set_register() code may block, and it therefore may be possible
that the loop in _init_registers() could occur in parallel with other
updates. That could result in a "OrderedDict mutated during
iteration" error.
Avoid the error by querying the latest value during each iteration of
the loop.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
It's possible to build and configure tmc5160 drivers with external
mosfets that support more than 3 amps. The actual maximum for tmc5160
drivers is dependent on how the board is wired and the mosfets used.
Increase the error check to 10 amps. This error checking is primarily
intended to catch "obvious misconfigurations" (eg, specifying
milli-amps instead of amps), and the new value of 10 amps should
suffice for this task.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Extends the BMxx80 category with support for the older BMP180 sensor, providing temperature and humidity output.
Signed-off-by: VAXXi Popescu <github@vaxxi.net>
The output_pin module is only capable of updating an output pin at
most once every 100ms. Add a new pwm_tool module that is capable of
queuing updates in the micro-controller and thus allowing for much
higher update rates.
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
This reverts commit 6749985302.
A defect was found in the above commit (the input shaper code calls
note_step_generateion_scan_time() for many steppers, so the
input_shaper class can't be used as the index).
Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
Kevin O'Connor