Force Sensing Auto Leveling with Heated Bed

Hi all,

I recently build a mini kossel and am having trouble getting my bed to level. It seems as though my heated bed is not mounted exactly parallel to the extrusion due to imperfections in the #D printed part. I see that the force sensors is a way to auto level your bed using the nozzel itself as the probe. However most of the examples i can find do not use a heated bed. Has anyone out there used this force sensing auto leveling WITH a heated bed??? It looks like these are rigged up through a thermistor but if someone has a headed bed i imagine the Z endstop would be used (much like the original design) I only print in PLA at the moment and am just trying to get a reliable way to level the bed everytime i run my bed around 60 C.

1. Are there specific firmwares for FSR that use the thermistor pins and others that use the Z endstop pins??
2. Is it possible to wedge the Force sensors between the glass and headed bed and still have good results?? (or does thing mess with the sensor because its wedged between)

Any thoughts on Force sensors with heated beds?????

nevermind i found a Post that helped BURIED in the back THANKS!

Quote
Ace67aod
nevermind i found a Post that helped BURIED in the back THANKS!

Can you please link to that post? I am interested in how to mount the FSRs so as to be shielded from the heat.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

This is a way more complicated issue then i thought it was going to be however i really want to get an auto leveling system working and due to my Z probe dragging due to my tilt i think this might be a useful alternative and might involve some outside of the box thinking but here are some of the links i am using to put my FSR autoleveling system to work WITH a heated bed.

Personally ive gone with a trinket system and think it is the most practical approach because it uses the Z-end stop so minimal coding changes will need to be made and it is also compatable with a heated bed (due to not using the themistor pins for operation) I currently have my parts ordered but need to wire/solder and fix the sensors. However i need to do more research on how to mount these to the glass and weather i need to insulate them from the heatbed (I plan on using cork however im not sure what kind of "give" this cork would have during the autolevling...

I plan to cover the whole process with some short youtube videos to provide some more info/help/examples of how im going about this because im having trouble finding good info on actually fixing these things and ways to insulate them....

Here are some links ive been using for my ideas

[forums.reprap.org]

[forums.reprap.org]

Information on what the trinket does and how to rig one up (though many are already for sale and are very cheap and programmed already) it probably wouldnt be hard to use some PCB and make your own.

[www.liming.org]

And Here is another post the has links that could also help

[groups.google.com][1-25]

Hope this helps and ill post links to the videos if they turn out useful

So I installed Force sensors inplace of my mechanical z probe for autoleveling on my mini kossel. Overall they seem to work loads better here is a link to a video showing the rig up and some Iterations my machine is running. seems to work we will see...

[www.youtube.com]

Is there anyway to edit Rich Cattell's marlin to reduce its tolerence from .03 to.05 my iteration level get really high and the G29 A command ends up making small changes forever do i just let it keep running or can i reduce the tolerance for a level bed see video above for more on this...

I believe you can edit the config.h file, which is what I did originally. However, you should also be able to set it in the command. I think the g-code is G30 A0.05, but I could be wrong. There is a thread on how to use the code on the delta google groups.

Thanks i got the G30A to calibrate at .05 and got the setting stored.....now to try and heat up and print... lol

would you mind to share your firmware or your conf.h file.... Ive bin searching a solution for heatbed and fsr sensor like you did. got a rumba boarded kossel xl

Quote
Moesling
would you mind to share your firmware or your conf.h file.... Ive bin searching a solution for heatbed and fsr sensor like you did. got a rumba boarded kossel xl

I'm promoting my own stuff here, but you might be interested in my differential IR Z probe. See [miscsolutions.wordpress.com] and [www.youtube.com]. Send me a PM if you are interested in ordering one stating which country you are in.

Both types of Z probe can suffer from the trigger height varying with XY position. In the case of FSRs, it is caused by the load on the three FSRs depending on where you probe. Typically, probing at the centre gives a lower trigger height than probing near the towers, because the force is shared by the FSRs equally, so you need a greater force to make the FSRs respond. In the case of the differential IR sensor, the fact that the IR probe is a short distance from the nozzle means that if the tilt of the effector varies with XY position, the height of the probe relative to the nozzle varies. In theory the effector should remain flat as it translates; but inaccuracies in the build can make it tilt slightly.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I went the FSR route and finally decided it was too unreliable approach. My goal was to get a bed level to within .05mm across the surface or that first layer can have problems when printing with a typical 0.2mm layer thickness. Compensating with a thicker first layer works but then that layer sticks out enough to see and measure.
For auto leveling I ran the G30 command but that cna take forever, assuming it even converges. If your mechanical build has too big errors in squareness, etc then G30 won't converge. After going to metal RobotDigg corners G30 would finally converge but it still took 22 runs or so. And if anything affects the system it has to be run again. Instead I added an option in the Marlin (RichRap) code G30 command to just probe three points at the towers and auto-zero the bed height measurements using the endstop values (M666 Xnn Ynn Ann). That was fast enough to run frequently. It was a compromise and the more-rigorous DC42 approach made more sense to me. But that requires the Duet board.

Here are some tidbits of my FSR 'experience':

1) The nozzle has to be completely clean when it probes or the readings are wrong. When the nozzle is hot it will ooze plastic for a long time, so THAT needs to be dealt with. I just kept runining my own 3-point auto-level until no more ooze was getting in the way. it also leaves ooze spots on the bed which have to be cleaned up each auto-level run. It's the main reason I went to the dc42 IR board.

2) I found with a heated bed the FSR would not trigger consistently so I spent endless cycles trying to get my 3-point auto-level to converge to small values. For thermal isolation I built up a double-layer of a 1/4" thick acrylic ring + cork insulation so the FSRs were well insulated from the bed heater, but that does not stop general heating of the environment.

3) Pre-loading on the FSRs can be a problem because it requires stronger probing force to overcome. The FSR approach is easier using this controller available here.
But it does not compensate for pre-load. I modified the code (I had to buy a small programmer) to compensate for the pre-load by making the trigger threshold more sensitive the harder the pre-load, which was measured once whenever the board was powered up. It's not on GIT or anything but I can give you the code if you want. It only works up to a point because the FSR pressure-resistance curve goes wonky when the pressure is too little or too much.

4) To get proper pressure on the FSR requires some kind of elastomer button between the FSR and the pressing surface. The FSRs have a boundary ring that prevents compression unless the contact interface stays smaller than that ring.
So I used a (very) stiff 1/2" rubber disk for the top and a small thin acrylic disk for the underside, all held together with double-sticky tape. But even stiff rubber pads allowed sideways forces to move the bed I did not like. Also, sticky tape does not have a strong shear-force strength which means they can allow the FSR to shift under lateral pressure.

5) After multiple tries with different designs I finally concluded that the bed assembly needs to be rock-solid, which means as few layers as possible between it and the chassis. Mounting on FSRs violates that rule, or any other design, especially an attempt to spring-mount the bed. (I tried several spring-mounted methods, including a fairly sophisticated design based on hand-made leaf-springs. It worked great and had zero lateral play but under heating conditions it would exhibit all kinds of micro-warps.)
Also, if your bed can shift at all you may encounter print shifts halfway through because the hot nozzle does hit solid sections of printed material and causes sideways forces on the bed. To counter this means either permanently mounting the FSR to the heating plate (or something like that), or use a lot of compression to fix the bed down.

6) Without the filament in a hot-end when probing, you might get accurate bed height measurements, but I was worried about how much expansion of the hot-end assembly would occur when the filament is under high forces during extrusion, so I concluded that I should probe the bed with the filament fulling loaded, which causes the oozing problems from number 1). Not proven but I thought I would throw that in.

7) Three mounted FSRs will have different triggering thresholds depending on whether the pressure is at the center or near the edge.

8) Probe points with any force on the bed boundary opposite of an FSR will actually unload that FSR because the other two act as a fulcrum line, again changing the final trigger threshold.

They can be made to work obviously, but I think the original micro-switch mechanical probe design has more benefits. Using a non-contact method with light or ultrasound or quantum tunneling or trained mice is better.

Hope this info isn't too redundant from the other thread you mentioned. Those threads are scattered all over the place so I may have missed that one.

Hello DC42,

I am very interested in the Differential IR probe for auto calibration due to the many issues with FSR and my mechanical probe. However I have had issues in the past finding correct firmware that uses auto calibration and my panel one LCD... Would i need to upgrade my electronics to duet and the panel due to take full advantage of this system?

Quote
Ace67aod
Hello DC42,

I am very interested in the Differential IR probe for auto calibration due to the many issues with FSR and my mechanical probe. However I have had issues in the past finding correct firmware that uses auto calibration and my panel one LCD... Would i need to upgrade my electronics to duet and the panel due to take full advantage of this system?

My IR sensor works as a Z probe with most electronics, including RAMPS. For Smoothieboard, you need to add a pulldown resistor. However, the fast delta auto calibration algorithm is so far only implemented in RepRapFirmware This firmware runs on Duet electronics, and there is an experimental port of it to Arduino Due + RADDS (without the web interface or software-configurable motor currents, which the hardware does not support). I have been told that a port to the Alligator board is in progress too.

One disadvantage of using an IR probe on a delta instead of FSRs is that if the effector tilt varies at all as it translates in the XY plane, then the height of the sensor relative to the nozzle will vary, so the trigger height will change too. RepRapFirmware allows you to compensate for this if you can't get rid of the tilt, but I don't think any other firmwares have this facility.

What problems are you getting with the FSRs? Do you have one of John's trinket boards?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

i am currently not using any auto calibration. i have a kit from T3DP3D it prints smaller prints fine when i do larger prints i can see my bed is not level. My problem with the FSR is getting the firmware that works with auto leveling and my panel one LCD. I need to probably try different firmware....however i'm not the greatest with software. I really like the duet + Due combo with the IR probe and have read your conversion guide one i can fund the project i plan to upgrade.... and with this upgrade i would very much like to have auto leveling worked out.

Side question with and duet +due could i upgrade to a 2020 delta or even larger with larger motors?

Quote
Ace67aod
Side question with and duet +due could i upgrade to a 2020 delta or even larger with larger motors?

I'm not sure what you mean by duet + due. The Duet is all you need, you do not need an Arduino Due as well. You only need more than the Duet if you want to drive more extruders.

Regarding using the Duet to drive a 2020 Kossel, this is exactly what I have done. See [miscsolutions.wordpress.com] and the following 2 blog entries after that one. I still use Nema 17 motors, the ones I have provide more than sufficient torque for a larger Kossel.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I was talking about the PanelDue Touch display (https://www.think3dprint3d.com/Electronics/Displays)

So would any firmware work with your IR probe with my RAMPS 1.4 and mega?

Quote
Ace67aod
I was talking about the PanelDue Touch display (https://www.think3dprint3d.com/Electronics/Displays)

Ah, I misunderstood you.

Quote
Ace67aod
So would any firmware work with your IR probe with my RAMPS 1.4 and mega?

If you mean would any firmware that supports a Z probe on Arduino Mega/RAMPS be able to use my IR board as a Z probe, then yes. When connected to an endstop input on a RAMPS board with the pullup resistor enabled, the board behaves rather like a normally-closed switch, which opens when the probe is triggered. See the commissioning instructions at [miscsolutions.wordpress.com]. The functionality you can get from a Z probe will depend on the firmware.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Do you know of any firmware that actually make calculations and corrections that still supports the Panel One display for the Mini kossel from T3DP3D?