So, what I’ve got now is essentially a hybrid: MK2 frame, motors, board, and power supply, with an MK42 bed (but with magnets and a sheet) and an MK2.5s hotend all on 12v. I think I spent under $50 in total to do all these updates.

After all this, I had to sort out the firmware. I downloaded the latest version for the MK2.5S (3.14.1) and made a few modifications. I’m running a RAMBo13a board, so I compiled with that profile (download link: MK25S-RAMBo13a), but here are my changes if you need to replicate this for a different board.

In MK25S-RAMBo13a.h (travel limits after homing, adjusted by trial and error):

#define X_MAX_POS 246
#define X_MIN_POS 0
#define Y_MAX_POS 210
#define Y_MIN_POS -2.2
#define Z_MAX_POS 200 // MK2 had 200 mm Z height
#define Z_MIN_POS 0.15

In mesh_bed_calibration.h (adjusted bed reference for MK42):

#define BED_ZERO_REF_X (-22.f + X_PROBE_OFFSET_FROM_EXTRUDER)  // -22 + 23 = 1
#define BED_ZERO_REF_Y (-0.6f + Y_PROBE_OFFSET_FROM_EXTRUDER)  // -0.6 + 5 = 4.4

#ifdef HEATBED_V2
  #define BED_X0 (14.0f  - BED_ZERO_REF_X)   // 14.0 - 1.0 = 13.0
  #define BED_Y0 (10.8f - BED_ZERO_REF_Y)   // 10.8 - 4.4 = 6.4
  #define BED_Xn (217.0f - BED_ZERO_REF_X)   // 217.0 - 1.0 = 216.0
  #define BED_Yn (206.4f - BED_ZERO_REF_Y)   // 206.4 - 4.4 = 202.0
#else
  #define BED_X0 (13.f - BED_ZERO_REF_X)
  #define BED_Y0 (8.4f - BED_ZERO_REF_Y)
  #define BED_Xn (216.f - BED_ZERO_REF_X)
  #define BED_Yn (202.4f - BED_ZERO_REF_Y)
#endif

In mesh_bed_calibration.cpp (modified mesh points + probe radii so it doesn’t hit endstops):

#ifdef HEATBED_V2
  const float bed_ref_points_4[] PROGMEM = {
    115.f - BED_ZERO_REF_X,  10.8f  - BED_ZERO_REF_Y,  // 10.8 - 4.4 = 6.4
    216.f - BED_ZERO_REF_X,  108.8f - BED_ZERO_REF_Y,  // 108.8 - 4.4 = 104.4
    115.f - BED_ZERO_REF_X,  206.4f - BED_ZERO_REF_Y,  // 206.4 - 4.4 = 202.0
     13.f - BED_ZERO_REF_X,  108.8f - BED_ZERO_REF_Y   // 108.8 - 4.4 = 104.4
  };
#else
  const float bed_ref_points_4[] PROGMEM = {
    115.f - BED_ZERO_REF_X,   8.4f - BED_ZERO_REF_Y,
    216.f - BED_ZERO_REF_X, 104.4f - BED_ZERO_REF_Y,
    115.f - BED_ZERO_REF_X, 202.4f - BED_ZERO_REF_Y,
     13.f - BED_ZERO_REF_X, 104.4f - BED_ZERO_REF_Y
  };
#endif

...............

#define FIND_BED_INDUCTION_SENSOR_POINT_X_RADIUS (4.f)
static const float search_radius = 4.f;
#define IMPROVE_BED_INDUCTION_SENSOR_POINT3_SEARCH_RADIUS (4.f)

Additionally I had to tweak a few things in xyzcal.cpp, because the 32x32 calibration matrix would run out of the printers travel boundaries and lock the X and Y axies (loose steps, no calibration possible). To much to publish here, but I changed the xyzcal_scan_pixels_32x32_Zhop() function, so that it would initialize the pixels[] variable with zeros and only measure pixel/position 4-28 in X and Y direction. This finally gave me the new 32x32 scan calibration feature, without loosing steps

Hi wolf, is it possible for you to explain more in depth what you did in this step? I am trying to do the same thing and I am stuck.

1. I still use the old MK42 12V Heatbed. It works, money is tight at the moment.

2. I still use the PINDA version 1, I never had version 2 and SuperPINDA ist to expensive for me at this point.

I already printed with a slightly modified 3.2.3 Firmware. The sole change was

#define Y_PROBE_OFFSET_FROM_EXTRUDER 5  // MK2/S 9

Then I changed the firmware to 3.10.1 with minor changes to test most functionality:

- Added Configuration.prusa.h from MK2.5S variant
- Added TACH1 in temperature.cpp, pins_Rambo_1_3.h for a 5V tacho print fan

The printer works and fans cycle in self-test with autofantest. Calibration and other stuff failed obviously.

So I changed the necessary values in mesh_bed_calibration.cpp (I never noticed until now the misspelling of levelling ;-) )

const float bed_ref_points_4[] PROGMEM = { /* changed */ 33.f - BED_PRINT_ZERO_REF_X - X_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_X, // 33 - 2 - 23 - 0 = 8 + 4 = 12 17.4f - BED_PRINT_ZERO_REF_Y - Y_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_Y, // 17.4 - 9.4 - 5 + 4 - 0 = 7

236.f - BED_PRINT_ZERO_REF_X - X_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_X, // 236 - 2 - 23 - 0 = 211 + 4 = 215 17.4f - BED_PRINT_ZERO_REF_Y - Y_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_Y, // 17.4 - 9.4 - 5 + 4 - 0 = 7

236.f - BED_PRINT_ZERO_REF_X - X_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_X, // 236 - 2 - 23 - 0 = 211 + 4 = 215 213.4f - BED_PRINT_ZERO_REF_Y - Y_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_Y, // 213.4 - 9.4 - 5 + 4 - 0 = 203

33.f - BED_PRINT_ZERO_REF_X - X_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_X, // 33 - 2 - 23 - 0 = 8 + 4 = 12 213.4f - BED_PRINT_ZERO_REF_Y - Y_PROBE_OFFSET_FROM_EXTRUDER - SHEET_PRINT_ZERO_REF_Y // 213.4 - 9.4 - 5 + 4 - 0 = 203};

and mesh_bed_calibration.h

#define BED_X0 (9.f - BED_ZERO_REF_X) // 12 - 4 + 1
#define BED_Y0 (11.4f - BED_ZERO_REF_Y) // 7 - 4 + 8.4
#define BED_Xn (212.f - BED_ZERO_REF_X) // 215 - 4 + 1
#define BED_Yn (207.4f - BED_ZERO_REF_Y) // 203 - 4 + 8.4

Additionally I had to tweak a few things in xyzcal.cpp, because the 32x32 calibration matrix would run out of the printers travel boundaries and lock the X and Y axies (loose steps, no calibration possible). To much to publish here, but I changed the xyzcal_scan_pixels_32x32_Zhop() function, so that it would initialize the pixels[] variable with zeros and only measure pixel/position 4-28 in X and Y direction. This finally gave me the new 32x32 scan calibration feature, without loosing steps.

Also, since the changes on the hotend change the printer head geometry the furthest point left (= X_MIN) is not outside of the print bed coordinate system, but already at 2-3 mm right of the MK2/S zero pint. This means, that the available printer size shrinks in X direction to 0-246 mm, so you will loose 4 mm and the printer will always start at +4mm X axis with every print. So to stop the printer to go out of bound set the travel limits in Configuration_prusa.h

#define X_MAX_POS 246    // was before 250

You could manage to print at position , but then the X end-stop is already depressed, so you should avoid that.

With these changes the self-test runs without problems, nice calibration point maps are measured, calculated and send to the terminal. Also, mesh bed levelling with 3x3 measurement points works flawless (cold and hot), obviously without steel sheet the 7x7 option doesn't. I have a variance of 0.69. Since I still use the PINDA version 1 probe I think there are some problems with the mesh bed levelling, because my first layer calibration looks horrible, but I still haven't made enough tests yet. Obviously the PINDA version 1 uses different correction factors, PINDA version 2 ist actively corrected with a thermistor readout and SuperPINDA has a less variance correction table or even none. So this (and a bed straightening with washers) is on my ToDo list.

So, happy modding.

Posted by: @hillinator

I don’t know if anyone still cares about this topic, but last night I successfully modified the mk2.5S firmware to work with the mk42 bed! I can post a compiled .hex file if anyone cares! I might just make a new forum post and link all of the mk2.5/S .hex files depending on your setup.  

Any chance you ever made that post about the changes you made. I am doing a similar setup and I'd love to get a jump on the necessary firmware changes.

Can you advise what changes were necessary to make the mk2.5 firmware work with an mk42 heatbed?

Cool! I did this awhile ago before upgrading to MK 2.5.  As I recall, the MK 2.5 PINDA mount is in a slightly different X-Y location than the 2S , so it failed the XYZ setup. What I did was to alter the frame geometry a little so that the PINDA could find the sense points.  This method reduces the available build volume a little bit, so the firmware fix is probably better.

I attempted to commit out the two lines you suggested, but the compiler will not compile because it breaks the code. I am afraid there is a lot more that has to be done for this firmware to work with the mk42 heat bed.