Request "M425 - Backlash compensation" enabled
Closed this issue · 24 comments
I would like to request that "M425 - Backlash compensation" to be enabled on X, Y, and Z axes.
I do mainly "engineering" type of prints and would like to be as accurate as possible on my prints. Backlash compensation would be a HUGE benefit for me. I have tuned my axes as best as possible with a dial indicator, but the lack of backlash compensation is frustrating, for me.
I have this currently enabled in an internal build and will roll it into the next release assuming I don't encounter any showstoppers.
OMG! Thank YOU!!
I was just starting to setup a fork and try to figure out how to do my own build. (It's been a LONG time since I done any coding)
I also require some further input here as there are multiple settings involved, depending on far you take it:
M425 is but one component of Backlash compensation. Technically you can have the capability enabled but not include M425 (BACKLASH_GCODE). This seems stupid to me because you'd end up with firmware hardcoded values so it was a natural leap for me to also enable M425.
Things got a little more tricky as a went further into the options available to the capability. Namely, CALIBRATION_GCODE which further extends M425 but has some specific hardware requirements (see below). I do not believe this is a wise setting for us.
//
// Backlash Compensation
// Adds extra movement to axes on direction-changes to account for backlash.
//
#define BACKLASH_COMPENSATION
#if ENABLED(BACKLASH_COMPENSATION)
// Define values for backlash distance and correction.
// If BACKLASH_GCODE is enabled these values are the defaults.
#define BACKLASH_DISTANCE_MM { 0, 0, 0 } // (linear=mm, rotational=°) One value for each linear axis
#define BACKLASH_CORRECTION 0.0 // 0.0 = no correction; 1.0 = full correction
// Add steps for motor direction changes on CORE kinematics
//#define CORE_BACKLASH
// Set BACKLASH_SMOOTHING_MM to spread backlash correction over multiple segments
// to reduce print artifacts. (Enabling this is costly in memory and computation!)
//#define BACKLASH_SMOOTHING_MM 3 // (mm)
// Add runtime configuration and tuning of backlash values (M425)
#define BACKLASH_GCODE
#if ENABLED(BACKLASH_GCODE)
// Measure the Z backlash when probing (G29) and set with "M425 Z"
#define MEASURE_BACKLASH_WHEN_PROBING
#if ENABLED(MEASURE_BACKLASH_WHEN_PROBING)
// When measuring, the probe will move up to BACKLASH_MEASUREMENT_LIMIT
// mm away from point of contact in BACKLASH_MEASUREMENT_RESOLUTION
// increments while checking for the contact to be broken.
#define BACKLASH_MEASUREMENT_LIMIT 0.5 // (mm)
#define BACKLASH_MEASUREMENT_RESOLUTION 0.005 // (mm)
#define BACKLASH_MEASUREMENT_FEEDRATE Z_PROBE_FEEDRATE_SLOW // (mm/min)
#endif
#endif
#endif
/**
* Automatic backlash, position, and hotend offset calibration
*
* Enable G425 to run automatic calibration using an electrically-
* conductive cube, bolt, or washer mounted on the bed.
*
* G425 uses the probe to touch the top and sides of the calibration object
* on the bed and measures and/or correct positional offsets, axis backlash
* and hotend offsets.
*
* Note: HOTEND_OFFSET and CALIBRATION_OBJECT_CENTER must be set to within
* ±5mm of true values for G425 to succeed.
*/
#define CALIBRATION_GCODE
#if ENABLED(CALIBRATION_GCODE)
//#define CALIBRATION_SCRIPT_PRE "M117 Starting Auto-Calibration\nT0\nG28\nG12\nM117 Calibrating..."
//#define CALIBRATION_SCRIPT_POST "M500\nM117 Calibration data saved"
#define CALIBRATION_MEASUREMENT_RESOLUTION 0.01 // mm
#define CALIBRATION_FEEDRATE_SLOW 60 // mm/min
#define CALIBRATION_FEEDRATE_FAST 1200 // mm/min
#define CALIBRATION_FEEDRATE_TRAVEL 3000 // mm/min
// The following parameters refer to the conical section of the nozzle tip.
#define CALIBRATION_NOZZLE_TIP_HEIGHT 1.0 // mm
#define CALIBRATION_NOZZLE_OUTER_DIAMETER 2.0 // mm
// Uncomment to enable reporting (required for "G425 V", but consumes PROGMEM).
//#define CALIBRATION_REPORTING
// The true location and dimension the cube/bolt/washer on the bed.
#define CALIBRATION_OBJECT_CENTER { 264.0, -22.0, -2.0 } // mm
#define CALIBRATION_OBJECT_DIMENSIONS { 10.0, 10.0, 10.0 } // mm
// Comment out any sides which are unreachable by the probe. For best
// auto-calibration results, all sides must be reachable.
#define CALIBRATION_MEASURE_RIGHT
#define CALIBRATION_MEASURE_FRONT
#define CALIBRATION_MEASURE_LEFT
#define CALIBRATION_MEASURE_BACK
//#define CALIBRATION_MEASURE_IMIN
//#define CALIBRATION_MEASURE_IMAX
//#define CALIBRATION_MEASURE_JMIN
//#define CALIBRATION_MEASURE_JMAX
//#define CALIBRATION_MEASURE_KMIN
//#define CALIBRATION_MEASURE_KMAX
//#define CALIBRATION_MEASURE_UMIN
//#define CALIBRATION_MEASURE_UMAX
//#define CALIBRATION_MEASURE_VMIN
//#define CALIBRATION_MEASURE_VMAX
//#define CALIBRATION_MEASURE_WMIN
//#define CALIBRATION_MEASURE_WMAX
// Probing at the exact top center only works if the center is flat. If
// probing on a screwhead or hollow washer, probe near the edges.
//#define CALIBRATION_MEASURE_AT_TOP_EDGES
// Define the pin to read during calibration
#ifndef CALIBRATION_PIN
//#define CALIBRATION_PIN -1 // Define here to override the default pin
#define CALIBRATION_PIN_INVERTING false // Set to true to invert the custom pin
//#define CALIBRATION_PIN_PULLDOWN
#define CALIBRATION_PIN_PULLUP
#endif
#endif
FWIW, I'm looking forward to playing with this a little myself with only BACKLASH_COMPENSATION, BACKLASH_GCODE, and MEASURE_BACKLASH_WHEN_PROBING enabled.
Yeah.... It's a rabbit hole with the autocalibration. Which I don't even now if it's possible with the S1/Pro. You would have to have some conductivity from the nozzle to ground. Which I don't think is there.
But you can use the probe to auto calibrate Z, which is great and I think is turned on from Creality (https://github.com/CrealityOfficial/Ender-3S1/blob/s1_pro/Marlin/Configuration_adv.h). But having the ability to just enter values for X, Y would be a godsend
Yup, I'm with ya. Good to know you are not looking to enable that final option as I would have to spin up a new variant to support it.
I'll be enabling 3 options listed in the update right above yours.
But you can use the probe to auto calibrate Z, which is great and I think is turned on from Creality (https://github.com/CrealityOfficial/Ender-3S1/blob/s1_pro/Marlin/Configuration_adv.h). But having the ability to just enter values for X, Y would be a godsend
nah, they have it all turned off.... the entire section is disabled.
https://github.com/CrealityOfficial/Ender-3S1/blob/s1_pro/Marlin/Configuration_adv.h#L975
Ugh. I was wondering about that. It's like that they started, but then quit.
anyway, I think I have what I need to proceed at this point. Expect to see something firmware wise in the next couple days. there's a bug related to zoffset while printing from SD I want to fix too and will bundle these together.
here you go: https://github.com/synman/Ender-3-S1-Pro-Firmware/tree/main/bugfix-2.1.x
Will leave this open for a couple days to collect feedback / etc.
Just want to confirm that you did not enable "MEASURE_BACKLASH_WHEN_PROBING" for the Z-axis.
I still have to do that manually.
BACKLASH_COMPENSATION, BACKLASH_GCODE, and MEASURE_BACKLASH_WHEN_PROBING are enabled. The only thing I did not enable is CALIBRATION_GCODE
and I went with the defaults for MEASURE_BACKLASH_WHEN_PROBING:
#define BACKLASH_MEASUREMENT_LIMIT 0.5 // (mm)
#define BACKLASH_MEASUREMENT_RESOLUTION 0.005 // (mm)
#define BACKLASH_MEASUREMENT_FEEDRATE Z_PROBE_FEEDRATE_SLOW // (mm/min)
Z_PROBE_FEEDRATE_SLOW computes out to 240mm/min
That's what I noticed when setting the backlash, 5 micron increments.
On then next go-round, could you set it to 1 micron?
#define BACKLASH_MEASUREMENT_RESOLUTION 0.001 // (mm)
When you use a good analog dial indicator, it's easy to "guestimate" 1 micron measurements. I just want to see if it actually makes a difference. (This doesn't work with an electronic dial indicator, because most of them only show hundreds of a millimeter)
And on a that note, for grins-and-giggles, could you also set:
SERIAL_FLOAT_PRECISION to 3 or 4 in the advanced configuration section?
BTW, I've reduced of my "elephants foot" by 99% by setting the Z backlash :) I'm still working on the X, Y, settings and seeing if there really is any measurable differences in prints.
Sure thing. Will possibly get to it tonight.
Take your time. I'm fighting with the new version of Octopi and how it broke my camera setup.
Big changes are on the horizon for Octoprint and cameras.
I dropped mjpeg-streamer on my more recent x64 builds of it.
It will soon be 100% plug-in based.
I didn't think it would affect my camera (Aruducam 2.0 electronic/manual focus). So now I can't use the Arducam focusing plug-in.
I've been meaning to kick the mjpeg-streamer. Just hadn't because I eventually got it working perfectly.
Oh, to top it off, I accidentally ripped out the ribbon cable going to my Bigtree touch display trying to get the microSD card out. Trying to reconnect it, I fsck'd the connector on the display so I had to order a new one. (It was NOT a good day for me).
There are a bunch of "breaking" changes in the latest bullseye builds.
Fact is, we've reached a point where the legacy camera apis really need to die and are no longer functional.
The new ones are still evolving so it's all around fun fun.
Playing around with this tonight with its built-in calculations with probing Z, I decided to run a probe repeatability test to 1. generate the backlash statistics for Z and 2. measure its impact.
Probe Repeatability Test - 10 runs (backlash disabled)
M48 P10 X100 Y100 V2
...
Recv: Mean: 0.056085 Min: 0.045 Max: 0.063 Range: 0.018
Recv: Standard Deviation: 0.006873
...
Send: M425
Recv: Backlash Correction inactive:
Recv: Correction Amount/Fade-out: F0.00 (F1.0 = full, F0.0 = none)
Recv: Backlash Distance (mm): X0.00
Recv: Y0.00
Recv: Z0.04
Recv: Average measured backlash (mm): Z0.04
So I then load that measured backlash value in and turn on compensation (I prefer quality over accuracy):
Send: M425 Z F1 S3
Recv: ok
[...]
Send: M425
Recv: Backlash Correction active:
Recv: Correction Amount/Fade-out: F1.00 (F1.0 = full, F0.0 = none)
Recv: Backlash Distance (mm): X0.00
Recv: Y0.00
Recv: Z0.04
Recv: Average measured backlash (mm): Z0.04
And re-run my probe test:
Send: M48 P10 X100 Y100 V2
Recv: M48 Z-Probe Repeatability Test
...
Recv: Mean: 0.051417 Min: 0.049 Max: 0.053 Range: 0.003
Recv: Standard Deviation: 0.001346
This feels significant to me. The range of the tests dropped from 0.018mm to 0.003 and check out the std deviation:
0.006873
0.001346
^^^ That is pretty freaking significant.
That's awesome.
But in reality, for the Z axis, you want accuracy over quality (if you are doing any type of Z hops). It's only on the X, Y axes that you would set quality over accuracy.
You know what would be really cool, is find a way to reorientate the probe temporarily for the X and Y with a temporary fixed stop and have a way to probe the backlash for X and Y.
So... when can I play with it!?!?!?!? :)
I just need to push the commit... built the binaries a couple minutes ago. I was doing some reading and apparently z backlash can cause layering issues. I interpreted as a Z axis problem but I also read about it causing gaps and blobs on x/y.
its out there now
I'm going to close this as an issue and move it to Discussions with the Feature label.