PFW-1376 MMU error during mid-print does not pause/resume print #48

gudnimg · 2022-07-10T12:11:16Z

JIRA ticket: https://dev.prusa3d.com/browse/PFW-1376

If the MK3S reports an error in the middle of a print move then we want to pause the print until the error is resolved. The pausing and parking needs to be done in two main loop() iterations which is why I have split all the actions into switch-case.
If the MMU recovers from a communication timeout without the user interacting with the LCD screen, the error screen will be dismissed and the print resumed automatically.

Modify lcd_show_multiscreen_message_two_choices_and_wait_P to also handle single-screen or empty (no-clear) prompts, making other functions redundant. Saves 76 bytes. Change existing functions to simply call lcd_show_multiscreen_message_two_choices_and_wait_P with the correct arguments. This changes the prompt of existing Yes/No messages: the previous prompt would use the last two lines of the LCD, while the new prompt is using just the last line of the LCD instead. Translation do not require updates, since the Yes/No translation was already the same in both implementations.

Add an additional parameter to control the position of second choice prompt position (while defaulting to the old). This allows Yes/No prompts to be equally spaced.

PAT9125 new init sequence

gudnimg · 2022-07-10T20:29:12Z

Marking as draft again. Will try to find another solution to this. c68a02e

EDIT: fixed

Do it in multiple parts to take into account when an error appears nested inside plan_buffer_line When mmu_loop is called inside plan_buffer_line while its executing a G1 gcode it is problematic to park the extruder. We only set the state to PAUSE_PRINT and then on the next iteration of mmu_loop in the top-level loop() we've exited plan_buffer_line and we can safely park the extruder Previously this was problematic due the G1 gcode not being done executing

vintagepc · 2022-07-12T11:51:20Z

Firmware/mmu2.cpp

+    {
+    case MMUErrorPrintStates::PAUSE_PRINT:
+        start_pause_print = _millis();
+        stop_and_save_print_to_ram(0.0, -default_retraction);


I still don't follow the technical reason for needing/implementing a different and completely orthogonal pause state here. I don't think you ever want an error condition to behave like a normal "paused" print, that's going to open up a whole complicated mess of potential issues relating to the tune menu, as well as users potentially doing a remote-resume from Octoprint without attending to the actual problem. IMO all MMU errors should follow the same pausing pathway in order to keep the code simpler and reduce future bugs/corner cases.

I’m confused how to proceed. I feel like we should at minimum set isPrintPaused to true etc. So any logic in the loop() doesnt do anything unexpected.

IMO all MMU errors should follow the same pausing pathway

Do I understand correctly you mean the code shouldn’t care whether we are printing or not?

I took out the Resume and Pause states and did quick test. Didn't see any issues. 4cb9f78

Yes - in effect all errors are "while printing" or doing something to prepare to print such as a load.

Up to now we’ve only been parking while in manage_response() which is its own loop and blocks any gcodes from processing during parking. The errors which can happen at any time like the Communication timeout or Protocol error are different in that they happen in the top level loop() and we can’t block queued gcodes moves without emptying the planner/command buffer. I’m seeing usually around 6 to 12 commands queued up.

gudnimg · 2022-07-12T13:32:07Z

Parking doesn't work when communication timeout happens. Will look into it. Also stopping a print will cause the printer to crash its axis (its trying to execute some gcode)

This reverts commit 4cb9f78.

gudnimg · 2022-07-12T14:21:08Z

I reverted the change since it does not work properly. I tried emptying the planner and command queue just before parking, it didn't help (printer just ignored the parking move completely). I think it requires a full main loop() to work as Yuri mentioned. And we probably need stuff in stop_and_save_print_to_ram to save the position in the SD card file (or USB line). Either way this isn't as easy as I thought :)

gudnimg · 2022-07-12T17:37:15Z

TODOs

Add TODO in the code to remind us to block any gcode parsing while parked. We’ll look at it after thermal model branch is merged?
Reduce lcd_resume_print. We don’t need all that stuff, just restore the command queue.

And add a TODO comment

…he printer wouldn't park.

gudnimg · 2022-07-12T20:38:52Z

I implemented the TODOs and fixed one bug I found while testing.

I'm assuming we need this?

This is a draft PR showing the potential 3x retry implementation on the printer's side. It is much less code and looks more reliable than the same functionality in the MMU FW. Still, more work needs to be done: - [ ] Button is sent to the MMU even before returning from the parking position - [ ] Then the button is sent again - [ ] Then the printer runs out of retryAttempts We need to find a better spot to check for "automatic" retry and issuing of the buttons

Fix the unload procedure when the user has paused a print then stopped the print after the temperature has reached below 175°C Now the E-motor will move as expected

PFW-1375 MMU error reported by MK3S does not appear on LCD

Communication timeout and Protocol Errors are now distinguished. In case of a Protocol Error, the printer waits for heartBeatTimeout to allow filling up the input UART buffer (we expect the MMU still produces some bytes). Once the timeout elapsed, the input UART buffer is cleared and a new Start Sequence is initiated.

…ility level

Fix protocol error recovery

gudnimg · 2022-07-22T15:57:22Z

Fixed the merge conflict. The diff is a lot cleaner now.

wavexx and others added 30 commits March 29, 2022 22:37

Allow choice positioning. Balance Yes/No messages.

c63bbb6

Add an additional parameter to control the position of second choice prompt position (while defaulting to the old). This allows Yes/No prompts to be equally spaced.

Merge branch 'MK3' into MK3_MMU2

f0740f2

Use PAT9125 new init sequence

3f42db6

Unflip the hardcoded flip of the Y axis

12a5041

swi2c safe init

7795f67

Use _check instead of register read for probe()

b6ddf00

twi safe init

c3ccad8

Add disable() to swi2c as well

67e1309

Shorter TWI timeout

c4235d6

Correct sensor orientation for old pat9125 init

15956cc

Merge pull request prusa3d#3423 from leptun/MK3_PAT9125_NEW_INIT

77f5664

PAT9125 new init sequence

Filament sensor refactoring initial

8ce30e3

Fix repeated autoload menu

ebfbc55

IR volt debugging

a14f93e

Remove old fsensor init

9251bd2

Remove OQ and add fancy autoload interaction

72115d2

IR_sensor_analog voltage readings

ca5d14d

Remove old fsensor_init()

757e399

Remove dead code

2e612af

More dead code removal

249d6b3

Move more IR sensor analog stuff to Filament_sensor.h

0a05b56

Remove unused fsensor variables

5989d21

Checkpoint after a lot of removal

2a2cd9b

Revert menu hack

ffa0c44

Fsensor error state

1c44616

Separate runout and enabled settings

fdb99c8

Enable fsensor when preparing for shipping

d043a8d

Remove original marlin fsensor code

eb5b23f

Correct ifdefs for all IR sensor types

1ce3829

gudnimg marked this pull request as ready for review July 10, 2022 23:49

Separate Unparking from Resuming print state

5430986

vintagepc reviewed Jul 12, 2022

View reviewed changes

Remove the "Pause" and "Resume" print states.

4cb9f78

Revert "Remove the "Pause" and "Resume" print states."

71934da

This reverts commit 4cb9f78.

gudnimg added 2 commits July 12, 2022 20:17

Simplify lcd_resume_print call

e038ab2

And add a TODO comment

Fix an issue where if the next error is the same as the last error, t…

9a25245

…he printer wouldn't park.

gudnimg and others added 12 commits July 13, 2022 08:42

Add Octoprint Resumed

215753c

I'm assuming we need this?

An attempt at fixing the retry. Discussion needed.

781ddb5

Magic numbers are bad...

9c69827

PFW-1373

96effd5

Fix the unload procedure when the user has paused a print then stopped the print after the temperature has reached below 175°C Now the E-motor will move as expected

Fix typo

7bf003d

Merge branch 'mk3-mmu' into PFW-1375

70cb3c8

Merge pull request prusa3d#46 from gudnimg/PFW-1375

45f657e

PFW-1375 MMU error reported by MK3S does not appear on LCD

Bump MMU FW version to 2.0.19 to match/enforce the necessary compatib…

3008cde

…ility level

Merge pull request prusa3d#51 from DRracer/fix-protocol-error-recovery

375af5e

Fix protocol error recovery

Merge branch 'mk3-mmu' into PFW-1376

9d7d2fe

gudnimg added 2 commits July 22, 2022 16:07

fixup

3dc5434

Slightly shorter code, no functional difference

7a9c441

DRracer force-pushed the mk3-mmu branch from 375af5e to c53d081 Compare July 28, 2022 08:11

DRracer force-pushed the mk3-mmu branch from 55235fa to 4457c76 Compare August 4, 2022 11:31

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

PFW-1376 MMU error during mid-print does not pause/resume print #48

PFW-1376 MMU error during mid-print does not pause/resume print #48

gudnimg commented Jul 10, 2022 •

edited

Loading

gudnimg commented Jul 10, 2022 •

edited

Loading

vintagepc Jul 12, 2022

gudnimg Jul 12, 2022

gudnimg Jul 12, 2022

vintagepc Jul 12, 2022

gudnimg Jul 12, 2022

gudnimg commented Jul 12, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022

gudnimg commented Jul 22, 2022

PFW-1376 MMU error during mid-print does not pause/resume print #48

Are you sure you want to change the base?

PFW-1376 MMU error during mid-print does not pause/resume print #48

Conversation

gudnimg commented Jul 10, 2022 • edited Loading

gudnimg commented Jul 10, 2022 • edited Loading

vintagepc Jul 12, 2022

Choose a reason for hiding this comment

gudnimg Jul 12, 2022

Choose a reason for hiding this comment

gudnimg Jul 12, 2022

Choose a reason for hiding this comment

vintagepc Jul 12, 2022

Choose a reason for hiding this comment

gudnimg Jul 12, 2022

Choose a reason for hiding this comment

gudnimg commented Jul 12, 2022 • edited Loading

gudnimg commented Jul 12, 2022 • edited Loading

gudnimg commented Jul 12, 2022 • edited Loading

gudnimg commented Jul 12, 2022

gudnimg commented Jul 22, 2022

gudnimg commented Jul 10, 2022 •

edited

Loading

gudnimg commented Jul 10, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022 •

edited

Loading

gudnimg commented Jul 12, 2022 •

edited

Loading