v3.6.4-pre1

This is a test release for a problem that has affected several people - potentially everybody who uses RMT stepping. A change in the underlying ESP-IDF toolkit from Espressif led to an uninitialized variable that, in turn, cause the RMT hardware to randomly use the wrong clock. Sometimes it would work right, then after you reboot, it would be wrong and the step pulses would be too long by a huge margin.

v3.6.3

New Features

• Added $Motor/Enable ( or $ME) command to compliment the $Motor/Disable (or $MD) command
• $30 will now respond with the max speed on the speed_map (emulates Grbl). This is for compatibility with some gcode senders.

Enhancements and fixes

• GPIO Input Reading Overhaul. This fixes the occasional problem with reading limit switches.

• Fixed motion stuttering when using I2S pins on spindles.

• Fixed CoreXY Homing.

• Created new SPI driver. This fixes issues with talking to Trinamic drivers.

• More accurate test for Trinamic motor communications. It tries to read the driver "version" to see if it matches the configuration.

TestLimits

This test release fixes some 3.6.2 problems with limits (which can make motors not run) and corexy homing. Give it a try and report back.

v3.6.2

New Features

• LittleFS is now supported for the local filesystem. The default is still SPIFFS but you can switch to LittleFS per http://wiki.fluidnc.com/en/support/filesystem_migration. In some future release LittleFS will become the default.
• Reserved GPIO pins, like those used for FLASH communication, will now report an error in the startup messages.
• You can now use multiple WebSockets and open multiple WebUI pages.
• Tool numbers can now go up to 99999999

Fixes

• Fixed resume after feedhold
• Dynamixel servo fixes
• Fixed laser staying on during hold
• Fixed decoding of WebUI real time characters (overrides, etc)
• Fixed some parking issues

There is a change to the WebUI. You should load the new one to pick up some bug fixes. It is in the wifi folder of the release.
http://wiki.fluidnc.com/en/installation#upgrading-firmware

v3.6.1

Features

• Safety door and parking improvements. It is now configurable.
• Spindles now have a config option to stop during an alarm.
• Dual motor single switch homing is now supported. The axis is not squared.

Enhancements and Fixes

• Numerous homing and limit switch improvements.

v3.5.1-pre-FS

Test release for file improvements. It supports both SPIFFS and LittleFS for the local filesystem. If you have an SD card, you can switch to LittleFS as follows:

$localfs/migrate=littlefs

That will backup all of your local files to the SD card, reformat the local filesystem to LittleFS, then restore the backup onto the newly-formatted local FS. You can go back the other way with:

$localfs/migrate=spiffs

TestPullups

This is a test release for a bug that only affects active-high limit pins with large-value/weak pulldowns. The problem is that the internal pullup resistor on limit pins was activated even if you did not say :pu in the config file. If there is a weak pulldown, the combination of the pullup and the weak pulldown can make the pin float at an intermediate voltage which could be interpreted as high even if the external switch is not active, so there would never be a low-to-high transition.

TestParking2

This is a test release for the newly-revised parking code, which is now configurable via the YAML file. Parking is a feature that works in conjunction with the safety door switch. If the safety door is opened and parking is enabled, the spindle will pull out in the +Z direction by "pullout_distance_mm" at rate "pullout_rate_mm_per_min" before stopping the spindle, then will move to the Z position "target_mpos_mm" in machine coordinates at rate "rate_mm_per_min". Normally that movement is in the Z axis, but the axis can be changed by "axis". Here is an example parking config section:

parking:
  enable: true
  axis: Z
  target_mpos_mm: -5.000
  pullout_distance_mm: 5.000
  rate_mm_per_min: 800.000
  pullout_rate_mm_per_min: 250.000

The machine needs to be homed, otherwise the "target_mpos_mm" position is ambiguous.

v3.6.0

This release gets us back on track after the debacle that started with v3.4.5, where we switched over to the new versions of the Arduino Framework, ESP-IDF SDK, and C++ compiler. It has a few targeted fixes and one major shakeup of the code. The targeted fixes are

• #575 - from @grittershub - Better connection test for UART-connected TMC2208 and TMC2209 chips
• #528 - from @breiler - XModem fixes - CR and packet length
• #559 - from @brianredbeard - logic error in cycle/hold state test (also found by @Priyanka328)
• #551 - from @brianredbeard - $I output was incorrect after multiple uses of $I
• #538 - from @LukasGossmann - OPT message reported Mist status incorrectly
• #576 - from @forbin1ab - WallPlotter rate calculations were wrong

The major shakeup is #562 . It revamps the main protocol state machine to use an event queue instead of polling status variables, which allows limit switches and control pins to be handled with events. This resolves a longstanding problem with limits. Direct polling of limit switches is slow if you have a lot of switches (possibly several per axis), but handling switch changes in an interrupt routine can give incorrect results due to problems in the ESP32 edge detection circuit. In the new scheme, an event is issued as soon as a pin becomes active, then further pin state changes are locked out until later, after the main code has responded to the event and the pin has had time to stabilize.

This scheme works well with the testing that has been done so far, having resolved some problems reported by users.

As an adjunct to the new event/limit scheme, the homing code was rewritten to use an event driven model. In the process, some pulloff problems were resolved and single-axis/multi-axis homing for CoreXY machines is now supported.

It also fixes some control pin problems like delayed cycle start after feedhold and undesired auto-repeat of macro pins.

In summary, the revamp features:

• Stable operation of limit switches
• Homing pulls off initially-active limit switches on either end
• Asymmetric final pulloffs work
• Single/multi axis homing works in CoreXY
• Cycle start after feedhold works right
• Control pins like macros do not auto-repeat when held

EventQueue2-Test

We have been working to stabilize FluidNC built on the new Espressif and Arduino core code. This is a test release for our latest work.

The main focus of this version is improvements to limits and homing. We think that we have fixed problems with noisy / bouncing limit switches. The homing code has been revamped to use an event-driven model and to combine Cartesian and CoreXY homing into a single framework. CoreXY now supports multi-axis and single-axis homing.