Since yesterday I'm testing Klipper at one of our printers. In particular the input shapers and their auto tuning with a ADXL345 accelerometer woke my interest.

It turns out my bed slinger printer (Bear Frame with Bondtech Extruder) has different resonance frequencies on each axis.

After the auto calibration /measurement it recommended:
Recommended shaper_type_x = 3hump_ei, shaper_freq_x = 61.4 Hz
Recommended shaper_type_y = 2hump_ei, shaper_freq_y = 57.4 Hz

I see there are input shapers on the way for RRF but I couldn't find exact infos what shaper will be used and if it will be possible to enter a shaper frequency for each axis separately.

To sum it up, here are my wishes for RRF 3.4+:

• support for a accelerometer (preferable ADXL345 as it's easy to find and buy) to measure the needed frequency. The input shapers 2HUMP_EI & 3HUMP_EI may need this as it can't be easily measured
• Input shapers: 2HUMP_EI & 3HUMP_EI
• separate frequencies and dampening for each axis (at the moment RRFs DAA uses the same frequency for both axis)

I've compiled a stable 3.2.2. firmware (Duet 3 Mini 5+) with the TMC22xx.cpp fix from 3.3b1, so sensorless homing does work without loud crashing
Now, after 2-3 full days of experimenting, time to enjoy this nice new board a bit.

If anyone needs it, here you go: https://www.file-upload.net/download-14505164/Duet3Firmware_Mini5plus.uf2.zip.html

The PINDA 2 has 4 cables.
White is the thermistor. You can connect it to Thermistor 2 on your Duet Wifi board.
Black is the signal for the probe and goes into Z Probe in.
Blue is ground an brown is +5V which you can "steal" from anywhere. I did chose to used the 5V/GRND from the expansion PINs on the board because they are right to each other.

If you use the stock PINDA position you can use my config code:

; Z-Probe PINDA
M574 Z1 S2 ; Set endstops controlled by probe
M558 P5 C"^zprobe.in" I1 H0.7 F1000 T6000 A20 S0.005 ; PINDA
M308 S2 P"e1_temp" A"PINDA" Y"thermistor" T100000 B3950
G31 P1000 X23 Y5 Z0.985 	                 ; PEI Sheet Offset C0.0010 S20 H2	
;G31 P1000 X23 Y5 Z1.315 					; Textured Sheet Offset
M557 X24:221 Y10:195 P9                     ; Define mesh grid

As for the temperature compensation.
For G31 there are S and C parameter available for that.

From the documentation:

Cnnn Temperature coefficient^2
Snnn Calibration temperature^2

2Optional parameters 'S' (temperature in oC at which the specified Z parameter is correct, default is current temperature) and 'C' (temperature coefficient of Z parameter in mm/oC, default zero) can be set. This is useful for probes that are affected by temperature. In RRF2 the bed temperature reading is used. In RRF3 you must specify which temperature sensor to use in the H parameter.

Unfortunately there is no example or way described how to use those parameters exactly.
What I did was, I chose a location where I expect no warping of the heat bed, removed the sheet and probed directly over the bottom left screw. That is the location the Prusa stock FW does also do it's temperature calibration.
Then I did measure my Z height at 35°C, 40°C, 45°C and 50°C PINDA temperature.
The problem then I had was that the inaccuracy wasn't linear but I think the "Temperature coefficient" parameter assumes that it is.
That is why PRUSA does save a specific Z offest value for the specific temperatures in a table and uses it when the temperature has been reached.

I already posted a possible solution in another thread:
https://forum.duet3d.com/topic/13516/conditional-gcode-and-object-model-variables/64?_=1592403073563

TL:DR

if PINDA_Thermistor <25°C {
G31 P1000 X23 Y5 Z0.995
}
else if PINDA_Thermistor >=25° AND PINDA_Thermistor <30° {
G31 P1000 X23 Y5 Z1.005
}
else if PINDA_Thermistor >=30°C AND PINDA_Thermistor <35° {
G31 P1000 X23 Y5 Z1.020
}
.
. and so on...
else {
G31 P1000 X23 Y5 Z2.0 ; safety height
}

What we need are the variables to make proper use of the thermistor of the PINDA.

At the moment M220 S20 (sets speed to 20%) does also slow down travel moves.
But what if we could choose whether to slow down travel moves with M220 or not.
Slowing down travel moves has a big drawback as this can increase the chance for stringing and blobs.

For example: M220 T1
The new parameter "T(1)" would then mean, that it does exclude travel moves. This setting can be set in config.g and when there is M220 Sxx in a gcode file it uses the set parameter (T1) from the config.g unless it gets overwritten with T0.

Use cases would be modifier in the slicer software. For example a print has very fine details between specific layers so we could set M220 S50 for those layers but without slowing down travel moves.

Another use case is the Palette 2 multi color unit and their Octoprint plugin. The multi material unit needs to do splices between colour changes and printing too fast may cause a failed print as the buffer cannot keep up. That's why the Plugin has a feature to slow down prints during splicing actions with M220 which does also slow down travel moves and can cause stringing / blobs.

I've upped my Config. You can copy the things you need.
https://www.file-upload.net/download-14164086/Config.zip.html

If you intent to use the whole config: You will need to add this in your start gcode in your slicer software for the filament runout sensor:

; Prime Filament Sensor for Runout
M581 P1 T2 S-1 R0 ; Filament Sensor P1 triggers Trigger2.g always (R0)  TRIGGER OFF
M950 J1 C"nil" ; Input 1 e0 Filament Sensor 
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor

The reason for this is that I wrote a filament autoload script and you can't have runout detection and autoload enabled at the same time without conditional code.

For the PINDA:
Black is the signal for the probe and goes into Z Probe in.
I connected 5V and GRND to the expansion port pins.
Thermistor is connected to E1

You can read out the PINDA thermistor temperature but we'll have to wait for conditional code being implemented as the temperature compensation feature as it is right now assumes that the inaccuracy is linear.

@cncmodeller

Yes that's the sensor.

Input shaper ZVD(D) paper:
https://www.researchgate.net/publication/316556412_INPUT_SHAPING_CONTROL_TO_REDUCE_RESIDUAL_VIBRATION_OF_A_FLEXIBLE_BEAM

EI2 input shaper:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.465.1337&rep=rep1&type=pdf

Most simple explanation: It basically cancels out ringing frequencies.

@dc42
Defining a minimum and maximum acceleration range might be good approach for input shaping.
With Klipper's (automatic) calibration measurement you can set a "max smoothing target" as too much acceleration may lead to losing fine details. It then recommends to stay below a specific acceleration, depending on the used shaper the value differs.
What Klipper does not offer though is to set a specific acceleration range. Meaning, if printing small details, the acceleration slows down as needed, so the shaper works properly.

Speaking of pressure advance. Is some sort of pressure advance smoothing planned?

The Accelerometer saves so much time and filament finding the best frequency even for the basic DAA shaper.

M593 off

And with DAA:

Can't wait for the "real" shapers

I think this thread might be the right location for this I guess because it's something I would need

Since I upgraded my Prusa MK3s to a Duet 2 WiFi board I try to enable all the features the Prusa original FW had.
This means:

• automatic loading of filament when there is no filament in the extruder
• PINDA temperature compensation table because the probe Z offset differs depending on the temperature. In my experience this is not linear and the higher the temperature the lower Z = 0 becomes meaning if you calibrate your printer with PLA, ABS will need recalibration as the nozzle will be too low.

My target is to create a config package for others so they can 1:1 swap their board and don't need to tweak much as my config uses all the convenient features, like the filament sensor, with explanation of what needs to change in the config depending of the used extruder or stepper motors (e.g. 0.9/1.8 degree).

1.
Feature to load the filament automatically.
Status: no filament in the extruder, extruder has PLA temperature (210°C)
Behaviour: Putting filament in the extruder triggers the filament sensor which then automatically loads filament.
Problem I had: The firmware only allows me to operate the filament sensor in one direction because the pin is blocked/reserved. Meaning I could only enable auto load or run out detection at once. So I had to split it in two over macros and start gcode.

So here is my (over complicated?) solution right now:

config.g

; Filament Sensor Port and Loading Feature ON
M950 J1 C"e0stop" ; Input 1 e0 Filament Sensor 
M581 P1 T2 S0 R0 ; Filament Sensor P1 triggers Trigger2.g always (R0)

Runout is not working right now. So when the printer is starting up and there is no filament and filament is put into the extruder trigger2.g is called:

trigger2.g:

; Loading Filament Script

M300 S800 P200
M581 P1 T2 S-1 R0 ; Filament Sensor does not listen anymore (S-1). Autoload is disabled. 
M291 P"Please wait while the nozzle is being heated up" R"Loading PLA" T5 ; Display message
T0 ; Activate Hotend
M291 P"Feeding filament Trigger2" R"Loading PLA" T5 ; Display new message
G1 E70 F800 ; Feed 70mm of filament at 800mm/min
G1 E55 F300 ; Feed 50mm of filament at 300mm/min
G4 P1000 ; Wait one second
G1 E-5 F800 ; Retract 3mm of filament at 800mm/min
M400 ; Wait for moves to complete
M292 ; Hide the message
G10 S150 ; Set current tool temperature to 150C
;G10 S0 ; Turn off the heater again

; Filament Sensor Port and Runout Feature ON
M581 P1 T2 S-1 R0 ; Filaent Sensor does not listen anymore (S-1). Autoload is disabled. 
M950 J1 C"nil" ; Input 1 e0 Filament Sensor 
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor

After the loading script the sensor is put into mode "runout detection ON"

But what if you turn your printer on and filament is already loaded and you directly start your print? This would mean that there is no runout detection enabled.
My solution right now is putting this into the Start GCODE in the slicer software:

; Prime Filament Sensor for Runout
M581 P1 T2 S-1 R0 ; Filament Sensor P1 triggers Trigger2.g always (R0)  TRIGGER OFF
M950 J1 C"nil" ; Input 1 e0 Filament Sensor 
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor

And there is a third scenario. What if you want to manually change your your color (filament) mid print?
After unloading the filament the sensor needs to be in mode "autoload" and after the autoloading is complete it has to go back to the mode "runout detection".

So i added this to my filament-change.g macro:

M25
G91 ; Relative Positioning
G1 Z20 F360; Raise Z
G90 ; Absolute Values
G1 X200 Y0 F6000 ; Parking Position
M300 S800 P8000 ; play beep sound
M98 Pejectfilament.g
M400

Called Pejectfilament.g:

M591 D0 P0 C"e0stop" S1 ; Filament Runout Sensor
M950 J1 C"nil"

M291 P"Please wait while the nozzle is being heated up" R"Unloading PLA" T5 ; Display message
T0 ; Activate Hotend
M109 S210
M291 P"Retracting filament..." R"Unloading PLA" T5 ; Display another message
G1 E+8 F300 ; Extrude 8mm of filament at 300mm/min
G1 E-45 F800 ; Retract 480mm of filament at 800mm/min
G1 E-15 F900 ; Retract 480mm of filament at 900mm/min
G1 E-25 F900 ; Retract 480mm of filament at 900mm/min
M400 ; Wait for the moves to finish
M292 ; Hide the message again
M300 S800 P200

; Filament Sensor gets ready for next load
M950 J1 C"e0stop" ; Input 1 e0 Filament Sensor 
M581 P1 T3 S0 R0 ; Filament Sensor P1 triggers Trigger3.g 

At the end trigger3.g:

M300 S800 P200
M581 P1 T2 S-1 R0 ; Filaent Sensor does not listen anymore (S-1). Autoload is disabled. 
M291 P"Please wait while the nozzle is being heated up" R"Loading PLA" T5 ; Display message
T0 ; Activate Hotend
M291 P"Feeding filament Trigger3" R"Loading PLA" T5 ; Display new message
G1 E70 F800 ; Feed 70mm of filament at 800mm/min
G1 E55 F300 ; Feed 50mm of filament at 300mm/min
G4 P1000 ; Wait one second
G1 E-1 F800 ; Retract 1mm of filament at 800mm/min
M400 ; Wait for moves to complete
M292 ; Hide the message

; Filament Sensor Port and Runout Feature ON
M581 P1 T2 S-1 R0 ; Filaent Sensor does not listen anymore (S-1). Autoload is disabled. 
M950 J1 C"nil" ; Input 1 e0 Filament Sensor 
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor

I know it first seems quite complex but when runout is enabled the port is blocked and can't be used for autoload.
Now I've read about the implementation of if / else statement which could make things quite simpler and also lowers the change of error as there would be no need to put code into the start gcode.

Like so (be aware, I'm no coder! )
Filament Sensor Status = 1 means no filament

func autoload {
M581 P1 T2 S-1 R0 ; Filament Sensor does not listen anymore (S-1). Autoload is disabled.
M950 J1 C"nil" ; Input 1 e0 Filament Sensor
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor
}

func runout_detection {
M581 P1 T2 S-1 R0 ; Filament Sensor P1 triggers Trigger2.g always (R0) TRIGGER OFF
M950 J1 C"nil" ; Input 1 e0 Filament Sensor
M591 D0 P2 C"e0stop" S1 ; Filament Runout Sensor
}

if Filament_Sensor state changed to "1" {
autoload
} else {
runout_detection
}

Something like that

2. PINDA Temperature calibration

The PINDA (v2) has an built in thermistor to use the Prusa FWs temperature compensation table.
RRP FW has has similar feature built in into the G31 command:

G31 P1000 X23 Y5 Z0.995 C0.0010 S20 H2

I've measured the Z trigger height with several temperatures from 20°C, 25°C up to 50°C and saw that the variance was not linear. At first the offset got bigger meaning the nozzle would be too close to the bed and after 40°C the difference were much smaller up to 50°C.
I'm not quite sure how the Prusa FW stores and uses the recorded values when the FW does it's auto temperature PINDA calibration but what would work is if you could use different G31 code lines depending of the temperature of the PINDA thermistor.

My "I'm no coder" mind thinks of the following:

if PINDA_Thermistor <25°C {
G31 P1000 X23 Y5 Z0.995
}
else if PINDA_Thermistor >=25° AND PINDA_Thermistor <30° {
G31 P1000 X23 Y5 Z1.005
}
else if PINDA_Thermistor >=30°C AND PINDA_Thermistor <35° {
G31 P1000 X23 Y5 Z1.020
}
.
. and so on...
else {
G31 P1000 X23 Y5 Z2.0 ; safety height
}

Hope it is clear what I mean

I think you could use daemon.g for this:

if sensors.analog[0].lastReading > 300
your gcode

See in the Object Browser (Plugin) which address your sensor has.

I can confirm pause does not work as intended with 3.3 beta 1.

My print does halt in place but the contents of pause.g are not run.

@dc42

Using the same frequency with ZVDD or other shaper is giving me worse results. I don‘t think I can just use the measured ringing frequency with shapers other than ZVD?

Measurement without input shaping (Y axis):

With ZVD input shaping:

The print results to also show that it is working.
I also have no issues with PA nor was it necessary to recalibrate PA.

Is it planned to to be able to use different shapers and frequencies per axis or to use an "automatic frequency and shaper finder" in conjunction with the accelerometer (like Klipper does)?

@pats-prints

I see your probe‘s offset is 15mm on Y.
If your axis is slightly tilted it’s more pronounced for the probe and the mesh is not usable.

@tinchus

Glad it works. Curve smoothing is a life safer for downloaded STLs. Those files are often low res because of size.

@tinchus

Precision: 0.02mm
Angle detection 120° each
Cutoff: 4-6mm (depending on how bad the exported model is)

@dc42

If the model is poorly exported and If I forgot to enable the "curve smoothing" feature in SuperSlicer, it's more or less a "corner" to "corner" transition for the printer (RRF) with acceleration and deacceleration for every "corner". So the speed is not constant and it has to respect the set jerk speed.

The interesting bit though is that @Tinchus mentions that the object is perfectly round printed with Marlin. Maybe Marlin's junction deviation has some kind of planner built in that assumes it's a round object and thereby prints it round? Then it would be interesting to know with what thresholds Marlin is working with so it does not smoothes out edges that are wanted.
Right now every printer in the house is running RRF (+1 Klipper) so I can't test that.

Here is an example of curve smoothing within the slicer software. Advantage is that STLs downloaded from the internet or exported from CAD can be rather small because the additional information is added within the slicer software.

Sliced without curve smoothing, F360 export setting "medium":

And with curve smoothing enabled:

@tinchus

The quality setting of the exported model isn't set to "Highest".

You may want to try the "curve smoothing" feature of SuperSlicer.

@dc42

Thanks, I can confirm the bug is not present anymore.
Used conditional gcode and also M291 in my paus.g file. The printer stays in pause as it's expected.

@noisettetbou @dc42

This does also happen with pause.g when using conditional gcode (if).
It seems messages do also cause issues.

I'm on RRF 3.3RC2 with Mini5+ (standalone).

I've already reported the bug here for 3.3b2:

https://forum.duet3d.com/post/228870

@cooknadam

Maybe something like this?

if sensors.gpIn[4].value == 0
	var triggerheight = move.extruders[0].position + 200 - move.axes[2].machinePosition
	set var.triggerheight = move.extruders[0].position + 200 - move.axes[2].machinePosition
	while move.extruders[0].position < var.triggerheight 
		if sensors.gpIn[4].value == 1
			break
	if sensors.gpIn[4].value == 0
		M226     ; calls pause.g

Needs some testing though.