Question on the bed switches

[BackwardEcho]

I have a stupid question that might have been thought of already and quickly thrown out: First of all, what are all the bed switch pins on the mainboard for? And more importantly to this stupid question, could they be repurposed / combined into some other form of input?

[aegean-odyssey]

It's a good thought, but the three bed switches all go to the same digital input on the CPU. So, if for some reason the bed switches are unused, then the one I/O pin could be used for something else.

All of the I/O pins of the 48-pin CPU part are committed, but it's possible that a few of the pins could be repurposed. I list them below. Except for the most simple of applications, though, I think that there are better options -- for one, adding a print server like OctoPi (OctoPrint) would allow for additional I/O, and I think the server could be configured to respond to commands embedded in the print job's g-code.

CPU	I/O	alternate function	note
PIN12	PA2	USART2_TX (TIM15_CH1, ADC_IN2)	PA2, PA3 seem to terminate at an unpopulated 2-pin header on the controller PCB. The firmware intializes the pins as USART2 ( though the port is unused by the firmware).
PIN13	PA3	USART2_RX (TIM15_CH2, ADC_IN3)
PIN34	PA13	SWDIO	PA13, PA14 terminate at an unpopulated 4-pin header on the controller PCB which includes 3.3v and ground. It serves as the CPU's debugging and programming port. Though the two pins could be used as general purpose I/O, some extra care may be needed to initialize and use these pins properly.
PIN37	PA14	SWCLK
PIN43	PB7	I/O (TIM17_CH1N)	All of the bed switches are connected in parallel to PB7 and the firmware configures the I/O as a digital input. One could configure PB7 for other uses such as I2C or USART pins, but these are essentially useless as other pins needed to complete an implementation are not available.

[BackwardEcho]

Well, the reason why I asked is because I was wondering how plausible it would be to somehow wire a BLtouch to the mainboard, maybe even just using the z-stop wires without the servo wires if it was possible (with my 5 minute understanding of the wiring for that sensor). But if I were to do that, I'm now thinking it could just be easier to create something with a regular microswitch. Granted, I am sure someone came up with an idea like that already... time to google around.

[PurpleHullPeas]

I have seen people mount BLTouches to the MPMD, but their results were always pretty crappy. Deltas don't do well with probe offsets, and you're not doing yourself any favors with added effector weight.

I have seen a few people mention in passing about using a detachable pressure probe or even simple microswitch setup similar to the MP Delta Pro. I.E., remove the probe after calibration. I haven't seen any in-depth guides or anything like that.

[BackwardEcho]

Personally, I wasn't even thinking of any offset but a Z-offset. My idea would have been to: dismount the plastic housing holding the hotend from the effector plate, attach another housing with the abl / microswitch in the exact spot as the nozzle, get the data, and then reattach the hotend. Ideally, it would be a temporary hardware swap that wouldn't have an adverse effects on the data and be less of a pain than inputting a permanent probe offset as it would just be using the already existing "probe point" of the nozzle. It just wouldn't be using the bed hitting the switch but the switch hitting the bed.

[PurpleHullPeas]

Yep. That's what I was referring to with part 2 of my comment.

I have seen a few people mention in passing about using a detachable pressure probe or even simple microswitch setup similar to the MP Delta Pro. I.E., remove the probe after calibration. I haven't seen any in-depth guides or anything like that.

[BackwardEcho]

Oh, sorry. I was imagining like a clip on switch that would still need an offset where I was thinking mine would be a total nozzle replacement.

…

On Tue, Mar 9, 2021, 10:16 AM PurpleHullPeas ***@***.***> wrote: Yep. That's what I was referring to with part 2 of my comment. I have seen a few people mention in passing about using a detachable pressure probe or even simple microswitch setup similar to the MP Delta Pro. I.E., remove the probe after calibration. I haven't seen any in-depth guides or anything like that. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#67 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQYJ74SZIEZ4D46WI5T2UADTCZCWXANCNFSM4YXOM5MA> .

[BackwardEcho]

So I found this a while ago and I'm wondering if the maker is wrong in that it can't be used for the stock mainboard. Granted, I would still want to try and design my mount mount / poker.

[aegean-odyssey]

Hmm, an alternate bed leveling mechanism is probably best to implement on the main board (as opposed to my "use a printing host" suggestion). The BLTouch seems to require two digital I/Os, one input and one output, plus ground and +5v power. Using the BLTouch probably could be done -- technically, the I/O is available; Marlin "supports" the BLTouch (it's a little bit of work to configure a "SERVO0_PIN" and likely patch Marlin's BLTouch code, though); and the project's latest tool chain creates a smaller binary (so there's probably enough code space for the changes). Overall, adding the feature would be "quite involved", I think.

As @PurpleHullPeas pointed out, any X or Y offset complicates the calibration on a delta printer, so I think a bed probe that relies on the nozzle somehow and interfaces more directly as a switch is a simpler solution -- something like a force sensitive resistor (roughly along the lines of https://hackaday.com/2019/09/01/force-sensitive-resistor-takes-the-pain-out-of-bed-leveling/).

[BackwardEcho]

I'm just thinking of something like this but turned upside down, with the metal rod centered where the nozzle would be, and totally (and temporarily) replacing the actual hotend assembly on the effector plate. Using this as a base model and just remixing the needed parts.

[aegean-odyssey]

Yes, looks like it could work and your idea has the advantage of not requiring firmware changes.

[BackwardEcho]

While wondering if I'd need to worry about my z height differences with the probe vs nozzle, I realized I have to worry about which z-endstop would trigger first: the bed or the probe. By that I mean, which button resists being pressed more than the other. It's gonna be a good while before I can test this even when I complete my basic model as I ordered a few switches from ali(not so)express. When I get them, do you know if the current bed switches are Normally Open or Normally Closed?

[BackwardEcho]

Should I try to make the height of the switch adjustable? or just hard mount it?

[BackwardEcho]

Well, I made a thing... too bad I can't test it yet as I don't have any micro-switches to use (some are in the mail)... Aside from that...

I present... the Poor Mans MP-Touch

It uses 2 printed parts, six M2 screws(about 12mm in length), a Binding Post fastener, and a micro switch. It's supposed to attach directly to the effector plate, totally removing the hotend assembly while the printer calibrates. I modified ksihota's MPMD Heatsink Casing v8 part and centered the binding post with the socket meant for the bowden coupler. I wanted it to be usable with other micro-switches / "pins", but I honestly have no idea if I am that capable of a modeler to account for everything like that. So I at least made the switch's height adjustable.

[aegean-odyssey]

I printed and use 3 bed chocks to keep the bed from tilting. During a print, I slide a chock under the bed at each switch just after the initial "bed probe". For your testing, the chocks may provide a simple way to keep the bed from moving and the bed switches from making contact.

[BackwardEcho]

Would the bed actually tilt though? Assuming the nozzle doesn't jam into the bed during the initial layer and all.

[aegean-odyssey]

For the machine here, the bed can tilt (move) during printing. I crudely modded the printer by affixing a 3mm thick glass plate to the original bed with double-sided tape. I think the extra weight of the glass plate makes the bed switches a bit more sensitive -- enough so that when the nozzle is extruding and moving away from the center, the tiny bit of drag from the molten material can bed can tilt the bed. The effect is more pronounced as the print's height increases, for short, relatively small models, the bed usually does not tilt at all.

I first ran into this problem trying to print a model that pretty much used the entire build volume. It would print more or less fine until the upper part of the build volume, but at some point the bed would rock back and forth as the nozzle moved across the build area. Consequently, the result was a horrendous looking print.

I created the bed chocks as a quick and simple way to keep the bed from tilting. Far from a perfect solution as the bed must tilt once during the Z probe in the start-up g-code. So I slip the chocks in place just after the probe (the g-code pauses for a few seconds before priming the nozzle). It's simple enough to do and I've made it part of my standard procedure for printing with the Monoprice MP Mini Delta.

Of course, there's a catch: the bed chocks can easily fall into the base of the printer through the openings at the printer's towers. After a few times of pulling a chock away from the bed only to have it fall into the base of the printer, I decided to use these tower shields that I reworked from a thingiverse model. The tower shields are great add-on in any event (I highly recommend covering the tower openings in some way), and the combination of tower shields and bed chocks work fairly well here.

Now if I understand your design, there will be no need for the bed switches (the bed never needs to move) and the bed can be more permanently immobilized. All in all a better solution I think, as opposed to trying to improve on the bed switch design. It also looks like your modifications may offer some avenues to improve part cooling on the printer as well. So maybe you'll improve on two significant short comings of the printer.

[BackwardEcho]

I was thinking there would still need to be active bed switches, but they wouldn't be relied on for anything except G29 P0. The 'probe' is meant to be removed after the mesh and M665/6 values have been found and made. Once I get my printer to get past the thermal runway problem it is having (which I hope isn't a thermister as I am now trying with r13 instead of r14 and r15) I will try to print the parts needed for the MPMD and my other printer which lost a stepper mount.

I am curious how it could improve part cooling: just another design idea that people didn't think of before to incorporate into their own? This part isn't anything more than a micro-switch with a screw/nail/binding-post (in this case I chose a binding-post screw as it is smooth all around and has a smooth top for the switch to be triggered with) in place of a hotend assembly.

[aegean-odyssey]

I was thinking that your part might eventually fit into the twin fan mod.

[BackwardEcho]

Well, the switches came in the mail.... now just to get the heating problem talked about in #64 fixed and I should be able to print the pieces to test. Not to mention to print a stepper mount for my larger printer, it's original acrylic one broke under tension from the belt...

[BackwardEcho]

Ahh yes, it works. I bought some replacement bed switch wires from Gigdigit (because I still don't know what the actual connector is or where to find it) and soldered some solid 20 gauge wire to it, red to red and blue to white. I believe the bed switches are Normally Open for those curious.

So now I just need to resolve my heating issue then I should be good to go. @aegean-odyssey any luck with resolving the other thermal runway issues originally reported in #64 ? Mine is most likely a hardware failure, but I'd like to rule out software entirely first before altering my heater cartridge wire for the 40w I bought.

[PurpleHullPeas]

but I'd like to rule out software entirely first before altering my heater cartridge wire for the 40w I bought.

Flash Marlin4MPMD 1.3.3 and see if it goes away.

[aegean-odyssey]

@BackwardEcho, I not sure if there's any particular problem with issue #64, but I do want to understand why there seems to be so much variation, printer to printer, in the temperature control -- and maybe work out what needs to be tweaked to get the printers to behave more uniformly.

[BackwardEcho]

Since I have my delta printing again I printed the 2 pieces for the MP-Touch, although I've had a couple of hiccups which should have been addressed already in the model. If not, feel free to yell at me.

But here is a list of materials I bought for this:

-replacement Bed Switch wires from Gigdigit, for the connections
-electrical tape and heat shrink. to protect the soldering job
-small gauge wire, it's just a microswtich
-a microswitch
-a Binding Post screw, the one / side without any exposed thread and a smooth top (no divots for a screwdriver). Got mine from Ace.
-four M2 x12 screws, to hold the bracket
-two M2 x20, to hold the switch
-two M2 nuts, to hold the switch

I actually don't have the last two things yet as I spaced the required size, but I shall get them soon.

MP-Touch.zip
In case either of you want the model. I'll also post it on thingiverse as a remix if the results are better than stock.

[BackwardEcho]

Hmmm.... Might not be the "be all end all" I was hoping it to be....

Anyone want me to try something specific?

[BackwardEcho]

Take this too. This is with the MP-Touch

Without Chocks

With Chocks

[aegean-odyssey]

These look quite to me. How does the first layer print? (the ultimate test)

[BackwardEcho]

so....
a couple of funny things

The first: disregard the info as my belts might not have been tuned. I retuned them after I put in my new heater cartridge as I needed to take a tower out.

The second: I made some adjustments to the model to "raise" the "nozzle" a bit as I noticed the adjusted max Z height, if I was understanding the data correctly, was coming back smaller than normal. So I made it so the pin sits at the same height as a normal stock nozzle.

The third: my bowden coupler broke on my cloned BMG I had reversed in place to allow clearance for my enclosure. So I had my old titan clone installed and just now tuned. Back to reprinting the touch....

[BackwardEcho]

Wired two of my bed switches in parallel to free up a port on the board. Since they all go to the same digital input as @aegean-odyssey stated, all of them occupying a different header is meaningless. The wire I added for the probe works just fine in the newly opened slot.

[BackwardEcho]

I hate words, have a drawing that took me a million years in MS Paint. How does the probing work?

[aegean-odyssey]

It's more like your option 2. Since the printer's "home" position is at the top of the build volume (as opposed to "home" = Z = 0), a probing sequence moves down until detects the build plate (probe switch asserted). The firmware uses this distance from the home position and the M851 value to figure the build volume height, M665 H. In short, the firmware removes the probe offset from the measurement.

A difference between the M851 value and the true probe offset moves where the firmware puts zero on the Z-axis -- changing the distance from Z=0 to the "home" position (i.e., top of the build volume) -- so M665 H is affected as well. Btw, the endstop offsets setting (M666) also affects M665 H. Consequently, the computed M665 H value can move around a little bit in the normal course of calibrating the printer.

[BackwardEcho]

I reeeeeeaaally don't want to lose that build volume. It is a peeve of mine when I see a cool printer mod and it takes up space. I will only accept a mod that doesn't loose travel or even gains a little.

To regain that space, since I expect to lose it during a calibration with the probe, would it be fine to do this process?

1. calibrate with probe and lose height (105)
2. manually adjust H to about 125
3. remove the probe
4. perform a G33 P1 with the normal bed switches to acquire new and calculated build height

5a) use the printer like before without the probe and rely on the bed switches for Z

5b) after acquiring new height without probe, get a value for M851 with probe and rely on the probe for Z

Though I need to figure where a G29 can fit

[aegean-odyssey]

The height (H) should be consistent regardless of which probe method (bed switches or probe attachment) is used. I think the process is:

• Set M851 Z to match the probe -- approx. 0.5 for travel in the bed switches or -15(?) for your probe attachment;
• Calibrate with G33 (or with G33 P1 to compute an H value);
• Create a bed level mesh with G29.

If you use the bed switches to probe for the height at the start of a print, M851 Z should be adjusted to around 0.5 to match. The M851 Z value should reflect the actual offset of the probe in use. Again, I think this is the way Marlin does things.

[BackwardEcho]

If you use the bed switches to probe for the height at the start of a print, M851 Z should be adjusted to around 0.5 to match. The M851 Z value should reflect the actual offset of the probe in use. Again, I think this is the way Marlin does things.

Now that I'm home this makes a bit more sense, though now my concern is this: do I need to input a M851 Z value before doing a calibration? Will it add that offset to the final H or will it still come out to about ~105? I might just need to do this whole process and explain what I did, because I sure as hell don't understand how marlin works enough to explain what I am trying to accomplish with the talk about the H value.

How does M666 effect H? would any value set in M666 no longer be true if H is changed manually?

[BackwardEcho]

There's continuity, boys! I am using a now heavily modified v6 effector by Zek as the "base" so you can get right to printing after removing the detachable probe.

[BackwardEcho]

rebuilding now

[BackwardEcho]

another short video

[BackwardEcho]

@PurpleHullPeas @aegean-odyssey Is G29 P0 supposed to overwrite M665 H#? I have the probe running on the printer for the first time and M665 H seems to change every time I run a G29 P0.

[BackwardEcho]

Actually, does it even matter? Could the printer print out of its defined M665 Height? I was able to move the effector to Z125 and then all the way back down to Z1 after I found a good M851 offset (which is easier to find than I thought it would be) when the H was set to 109.26.

[BackwardEcho]

Altered Denni's 'Carbon Paper Test' gcode as it used G30 which needs a Z probe touch. That isn't happening with this mod, so manual movements it is! Now I think I can calibrate like I did before, or close to.

[BackwardEcho]

@PurpleHullPeas , using your python script to calibrate, is a deviation of "0.01098" per run good? I think it's good. It looks good.

[PurpleHullPeas]

Yes, that looks very good 👍 .

I would say you have reached a point where you cannot reliably measure additional improvements. Filament tolerances are advertised as +/- 0.03mm. The cheap and/or worn nozzle probably is not exactly 0.4mm. Shaky hands while measuring imperfect carbon paper circles on a not perfectly flat piece of paper? That is just a long way of saying that you have reached a point where many things are out of your control, so you should be happy with what you have achieved 😁

[BackwardEcho]

I think I got the process down, now I just need to write it out for other people. Though I wonder how many people would actually be interested in installing it (where if I were to make a guide and upload the files, they would install it and not just think "neat").

[PurpleHullPeas]

I'm just saying from my own experience working as a professional software developer and as a hobbyist, be careful about making any promises to anyone until after it has actually been tested. We have seen it happen way too many times,. specifically for this printer - when some seemingly standard/basic 3D printing tutorial does not work the same with the MPMD because of stock firmware weirdness/inconsistencies. Sometimes, even the support provided directly from the manufacturer does not work. Remember, they couldn't even use the same basic Cura def for both the V1 and V2. I had to make a new one.

[BackwardEcho]

I don't plan to say anything but, "given the hardware similarities between v1 and v2, this mod should in theory work. But I do not have v2 to test with. Try at your own risk"

Hell, I plan to say that with anyone that has stock firmware on the v1. I did not test it, but given how I think it works, it should work just fine.

While typing this out, the one thing I can think of that would ruin this for the v2 would be if the bed switches WERE NOT tied to the same input pin like on the v1. If they had their own individual pins and were individually used to calibrate the printer with their specific location. Then I can see this mod not being compatible.

[PurpleHullPeas]

I had to write completely different python calibration code depending on whether I was calibrating stock firmware or mpmd_marlin_1.1.x. Certain Gcode commands just do not work like you would think on stock firmware. I had to write slightly different python code for Marlin4MPMD vs mpmd_marlin_1.1.x. If you are only going to test the mod on mpmd_marlin_1.1.x on a V1, then my suggestion would be to be very upfront about that and mention that is up to the owner if they want to try making it work with a different setup.

[BackwardEcho]

That is the plan. I don't want to be the cause of someone unknowingly damaging their printer beyond repair.

Once I write a bit more of the documentation I'm thinking I'll share it here before I click publish on thingiverse. See if I missed anything, warning or explanation wise.

[BackwardEcho]

Here is what I have so far before getting into the installation process.

[BackwardEcho]

How is this for a flat bed?

[BackwardEcho]

I decided to publish it under the "work in progress" label because otherwise asking for feedback would be a pain in the butt

@PurpleHullPeas @aegean-odyssey care to read what I put so far?

[PurpleHullPeas]

Avoid referencing specific page numbers in my document. Reference sections, instead. While it is true that I barely update it anymore, it is a dynamic document that could be changed from time-to-time. For example, when I add a reference to this tutorial, it may shift page numbers past that point. If they're unable to do a simple ctrl+F in my document or read the table of contents, then they are definitely not going to be able to complete your mod.

I'd also suggest posting a bed leveling test print. Of course, at this point, you have done a lot to your machine, but a simple, "see, I DO know what I am doing" goes a long way for credibility. Side rant: I will never forget a couple of people who came to the Facebook Group posting super flashy mods (all printed on a different printer) and convinced a bunch of people to try their mods...even though they had not (and still have not) printed successfully with their own highly-promoted mod.

Otherwise, looks good, so far.

[BackwardEcho]

Oh man, I do not want be one of those guys. I have been able to print just fine in general, but I have a 0.25 nozzle so the pressure is on to find my perfect Z-offset for that bed leveling test. This printer is currently my only working printer. I forgot I have access to my sister's printer. Which I will say I did print parts on it before for my mpmd, but I have since reprinted on the mpmd. I was worried about any unaccounted for inaccuracy "contaminating" the effector mod at the time.

As of now I am printing a bed level test and I have changed the "page numbers" to section titles. I do also have a few more photos to upload, but I'm slower to do those as I need to compress / resize the them until they fit inside the 500kB limit.

[aegean-odyssey]

You sure have put a great deal of effort into the mod, and I think your modifications really do address some of the shortcomings of the MPMD printer with good results. Alas, though, our V1 printer is at the end of its life and Monoprice is even less keen on allowing tinkering with the firmware and modifications with its V2 model. And while your modifications may make for better prints (even better than the V2), I think it will be tough to get "buy in" from users if the V1 is scarce and the V2 in more and more a "closed" machine.

Your mod strikes me as the partial design of a what could be a truly opensource version of the MPMD. The MPMD has been a fun printer, but it's not as "open" as Monoprice claims. Wouldn't it be nice if there was a printer like the MPMD whose parts and design were readily available and accessible?

[BackwardEcho]

Your mod strikes me as the partial design of a what could be a truly opensource version of the MPMD. The MPMD has been a fun printer, but it's not as "open" as Monoprice claims. Wouldn't it be nice if there was a printer like the MPMD whose parts and design were readily available and accessible?

My man

I have thought about remaking the mpmd but with aluminum extrusions, linear rails, and the like, so don't tempt me. However I am stopping myself as I want to do two things before I attempt: 1) Finish my corexy overhaul. I have a tronxy x5sa-400 that I got linear rails for and left it on a hiatus because I am going just as balls to the wall with it as I am doing with my mpmd. 2) I want to learn CAD proper. I am currently using Blender, which is fine for personal use, but if I were to seriously design something like that I should use the proper tools.

[BackwardEcho]

@PurpleHullPeas and or @aegean-odyssey , I think I am done with the hardware side. Can one or both of you give it a read and tell me if I missed anything? I am sorry for the asking this sizeable task of both of you.

[BackwardEcho]

Made the firmware / gcode side. Also tested with stock firmware as I had my original loud controller board installed since I was testing the glue I used to reattach the heatsink since I gave the thing an alcohol bath. Printed a fair bit to make sure the glue would be fine before wanting to put my quiet board back in.

[PurpleHullPeas]

@BackwardEcho I added it to my Roadmap document maybe ~1 month ago in the alignments section next to Dennis's switch replacement. I think it looks like a great upgrade and the proof is in the pudding that it works.

[BackwardEcho]

I saw that last night, made me giddy. :D

Also, what exactly didn't save on the stock firmware? Was it M665 R? Because it appears to have saved for me during my testing of the v44 firmware. Couldn't tell you why it was able to be saved though, maybe it was a glitch during original installation that was fixed when I reflashed it.

[PurpleHullPeas]

M665 R does not save on stock firmware, even if it says that it does. Reboot the machine, try printing, and your results will be crap. There are old posts in the Facebook Group about it if you dig far enough.

[BackwardEcho]

Oh, then I have been fooled. I didn't let it stop me from putting the parameters in the start gcode though, I put everything in there for the stock firmware.

[BackwardEcho]

Just printed a cube on stock firmware with a 0.25 nozzle. Sure it is a small print that was centered in the bed... but holy shit that first layer came out perfect.

[PurpleHullPeas]

The whole point of putting the stuff in your Start Gcode is to get around the bug, so you won't see any problems if you have M665 R, there.

I realize I sound very stubborn old man here, but I'm generally not going to change a stance on something Dennis Brown tested.

[PurpleHullPeas]

Here is an old post demonstrating one instance of the M665 R bug. https://m.facebook.com/groups/mpminideltaowners/permalink/2059674647381344/

[BackwardEcho]

I wasn't trying to convince you otherwise, this is just my first time dealing with it. I thought the new value just wouldn't save in eeprom and would show the original value. So I was a bit confused hence why I asked.