Artillery Genius temperature-calibrated inductive Z probe mod

prusaprinters

EDIT: I don't recommend this! Even with temperature compensation, I found the Z probe to be just too unreliable. Get a PINDA! :D I've been enjoying using the stock inductive Z endstop as a bed probe on my Artillery Genius, but was suffering a bit from its temperature sensitivity: I was noticing that at different times of day, and after printing, the necessary Z probe offset was wildly different. Marlin has the capability to add a temperature calibration to Z probe values using the G76 command, but to do this we'd need some way to monitor the temperature of the inductive probe. (This is what the Prusa MK3 does using the PINDA V2 probe.) I happened to have a couple cheap thermistors lying around, and after my Neopixel mod, I knew that I still had a couple wires free down in the case I could use to connect them to the motherboard. Why not try to hack together my own PINDA V2-style probe? Print new probe holder I designed a probe holder with a channel down the side that would fit the thermistor, allowing me to get a pretty good guess as to the temperature of the probe. It's designed to fit into my modified BLTouch mount for the BMG V6 fan shroud, but it should also work with the Waggster mount in a stock Volcano hot end. The STL file includes some break-away supports, so it can be printed without any additional support. Install thermistor and inductive probe The thermistor comes with very long wires. We only need four five inches. I cut it short, then crimped on new Dupont connectors. Plug the thermistor into the W/B pins on the BLTouch header (order doesn't matter on this side). The inductive probe should be connected to the B/R/Y yellow pins, with the Blue/Brown/Black wires, respectively. Fit the thermistor all the way into the channel, then apply some thermal paste for good thermal contact with the probe. Install the probe with a short M3 screw to capture the thermistor into the channel, then the whole assembly onto the fan mount bracket. Motherboard wiring Inductive probe Because the GND (B) and 5V (R) pins on the Genius hotend header have their own connections to power, you only need to worry about relocating the wire that was connecting the BLTouch header to the servo control pin. On my board, that was a single white wire ending in a Dupont connector, that was connected to pin D11 in the SERVOS1 header. I moved this white wire over to pin D18 on the Z- endstop connector. Thermistor The MKS Gen L board has one extra temperature input, nominally for monitoring a second extruder. I used this to monitor the probe thermistor. After my Neopixel mod, I still had two individual wires free: a red one, terminating in a 3-position JST connector; and a black one, terminating in a single Dupont connector. These wires have continuity with the W and B pins on the BLTouch header, respectively. IMPORTANT: the black wire ALSO has continuity with the common ground! This means there is only one orientation that will work for plugging these into the T1 header. On the motherboard, the pin closest to the corner of the board is internally connected to GND. The black wire can thus go to that pin, while the red wire can go to the other pin (adjacent to the serial connector to the TFT). To make sure I got this right, I re-crimped these wires into a new JST connector, which enforces the right polarity of the connection. To double check before repowering the board, I tested the hotend header with my multimeter, and made sure the B header was connected to ground, while the W wire was not. Firmware Getting this working in firmware was a bit of an adventure. You can see my working firmware branch on GitHub here and modify it to fit your setup if you desire. First of all, in addition to the changes required to enable the inductive Z probe I had to enable the new sensor in Configuration.h, and add a line to tell the board which pin to use for it: define TEMP_SENSOR_PROBE 1 #define TEMP_PROBE_PIN TEMP_1_PIN Then, in Configuration_adv.h, I needed to enable and modify the thermal probe compensation section, uncommenting the lines: if HAS_BED_PROBE && TEMP_SENSOR_PROBE && TEMP_SENSOR_BED // Enable thermal first layer compensation using bed and probe temperatures #define PROBE_TEMP_COMPENSATION Because the stock inductive endstop probe appears to have much higher thermal sensitivity than expected, the calibration values at higher probe temperatures exceed the sanity check in the firmware. Thus, I also reduced the number of samples, so that the max probe temperature compensation values was 60 degrees: // Probe temperature calibration generates a table of values starting at PTC_SAMPLE_START // (e.g. 30), in steps of PTC_SAMPLE_RES (e.g. 5) with PTC_SAMPLE_COUNT (e.g. 10) samples. #define PTC_SAMPLE_START 30.0f #define PTC_SAMPLE_RES 5.0f #define PTC_SAMPLE_COUNT 6U Finally, the most recent 'stable' release of Marlin (2.0.5.3) has a bug in the G76 calibration code. This is fixed in the bugfix-2.0.x branch, so you need to make sure to base your firmware off of this branch. Results It works a little better! But still not perfect. Considering PINDA V2 sensors aren't very expensive, I'll probably just order one of those! Print instructionsCategory: 3D Printer Parts Summary I've been enjoying using the stock inductive Z endstop as a bed probe on my Artillery Genius, but was suffering a bit from its temperature sensitivity: I was noticing that at different times of day, and after printing, the necessary Z probe offset was wildly different. Marlin has the capability to add a temperature calibration to Z probe values using the G76 command, but to do this we'd need some way to monitor the temperature of the inductive probe. (This is what the Prusa MK3 does using the PINDA V2 probe.) I happened to have a couple cheap thermistors lying around, and after my Neopixel mod, I knew that I still had a couple wires free down in the case I could use to connect them to the motherboard. Why not try to hack together my own PINDA V2-style probe? Print new probe holder I designed a probe holder with a channel down the side that would fit the thermistor, allowing me to get a pretty good guess as to the temperature of the probe. It's designed to fit into my modified BLTouch mount for the BMG V6 fan shroud, but it should also work with the Waggster mount in a stock Volcano hot end. The STL file includes some break-away supports, so it can be printed without any additional support. Install thermistor and inductive probe The thermistor comes with very long wires. We only need four five inches. I cut it short, then crimped on new Dupont connectors. Plug the thermistor into the W/B pins on the BLTouch header (order doesn't matter on this side). The inductive probe should be connected to the B/R/Y yellow pins, with the Blue/Brown/Black wires, respectively. Fit the thermistor all the way into the channel, then apply some thermal paste for good thermal contact with the probe. Install the probe with a short M3 screw to capture the thermistor into the channel, then the whole assembly onto the fan mount bracket. Motherboard wiring Inductive probe Because the GND (B) and 5V (R) pins on the Genius hotend header have their own connections to power, you only need to worry about relocating the wire that was connecting the BLTouch header to the servo control pin. On my board, that was a single white wire ending in a Dupont connector, that was connected to pin D11 in the SERVOS1 header. I moved this white wire over to pin D18 on the Z- endstop connector. Thermistor The MKS Gen L board has one extra temperature input, nominally for monitoring a second extruder. I used this to monitor the probe thermistor. After my Neopixel mod, I still had two individual wires free: a red one, terminating in a 3-position JST connector; and a black one, terminating in a single Dupont connector. These wires have continuity with the W and B pins on the BLTouch header, respectively. IMPORTANT: the black wire ALSO has continuity with the common ground! This means there is only one orientation that will work for plugging these into the T1 header. On the motherboard, the pin closest to the corner of the board is internally connected to GND. The black wire can thus go to that pin, while the red wire can go to the other pin (adjacent to the serial connector to the TFT). To make sure I got this right, I re-crimped these wires into a new JST connector, which enforces the right polarity of the connection. To double check before repowering the board, I tested the hotend header with my multimeter, and made sure the B header was connected to ground, while the W wire was not. Firmware Getting this working in firmware was a bit of an adventure. You can see my working firmware branch on GitHub here and modify it to fit your setup if you desire. First of all, in addition to the changes required to enable the inductive Z probe I had to enable the new sensor in Configuration.h, and add a line to tell the board which pin to use for it: define TEMP_SENSOR_PROBE 1 #define TEMP_PROBE_PIN TEMP_1_PIN Then, in Configuration_adv.h, I needed to enable and modify the thermal probe compensation section, uncommenting the lines: if HAS_BED_PROBE && TEMP_SENSOR_PROBE && TEMP_SENSOR_BED // Enable thermal first layer compensation using bed and probe temperatures #define PROBE_TEMP_COMPENSATION Because the stock inductive endstop probe appears to have much higher thermal sensitivity than expected, the calibration values at higher probe temperatures exceed the sanity check in the firmware. Thus, I also reduced the number of samples, so that the max probe temperature compensation values was 60 degrees: // Probe temperature calibration generates a table of values starting at PTC_SAMPLE_START // (e.g. 30), in steps of PTC_SAMPLE_RES (e.g. 5) with PTC_SAMPLE_COUNT (e.g. 10) samples. #define PTC_SAMPLE_START 30.0f #define PTC_SAMPLE_RES 5.0f #define PTC_SAMPLE_COUNT 6U Finally, the most recent 'stable' release of Marlin (2.0.5.3) has a bug in the G76 calibration code. This is fixed in the bugfix-2.0.x branch, so you need to make sure to base your firmware off of this branch. Results It works a little better! But still not perfect. Considering PINDA V2 sensors aren't very expensive, I'll probably just order one of those!