Auto Bed Leveling - inductive vs microswitch

So I've come to the point where I'd like to install an auto bed leveling feature. Not that I really need it but I find it a really cool thing to have Plus, I won't have to level the bed every now and then.

So, I've done some reading and have already found out that there are countless ways to do it. However, using an inductive sensor or using a microswitch seem to be the most popular options. So what are the up and downsides of each alternative? What would you recommend?

I seem to have some doubts about the microswitch option because it needs a separate motor and some kinematics to let it go up and down (like the landing gear of a plane). So I wonder if it is going to come down every time to the exact same Z position relative to the hot end nozzle, so that the hot end barely touches the heated bed every time at Z=0. So maybe the inductive sensor would be more reliable and it seems to require less space.

My setup is in my signature.

Thanks in advance for any advice

---

Self-sourced Mendelmax 2.0-based Reprap Machine -- Ramps 1.4 & Mega 2560 -- DRV8825 (Z@1A, [email protected], [email protected], E@1A) -- genuine E3D v6 direct setup -- 350W custom silicone heated bed -- ABS 1,75mm -- Marlin 1.1.0-RC7 -- Cura 15.04.6

In most auto leveling schemes, you have to add extra mass and extra wiring to the extruder carriage. The extra mass may affect/limit maximum print speed (which doesn't mean much to me), and the wiring adds some complexity and additional points of failure (which bothers me a lot). If your printer doesn't require leveling and zeroing with every print, I wouldn't bother with it. If you printer does require leveling and zeroing with every print I would look at fixing the printer rather than going to autoleveling. If the printer can't maintain level and zero settings from print to print it probably suffers from print quality problems that will be remedied when you fix it so it doesn't need constant releveling and zeroing.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I don't have issues with bed leveling and I don't need to level it often at all. I just find auto leveling interesting.. I have a bowden setup so there is not much weight in the X carriage. The wiring and additional mechanics depend on the auto bed leveling type...

---

Self-sourced Mendelmax 2.0-based Reprap Machine -- Ramps 1.4 & Mega 2560 -- DRV8825 (Z@1A, [email protected], [email protected], E@1A) -- genuine E3D v6 direct setup -- 350W custom silicone heated bed -- ABS 1,75mm -- Marlin 1.1.0-RC7 -- Cura 15.04.6

The disadvantage with the servo, is that it has a tendency to have unwanted movement. The movement during printing can cause the servo to hit what's being printed and the steppers lose steps, ruining the print. Also, it can also move a bit during the auto-leveling, throwing off the measurements. Currently I have the servo setup and the movement problems. I'm going to try a proximity sensor, once the ones I ordered arrive. Until they arrive I won't know if I have to replace the current glass & aluminum bed with a MIC-6 bed.

I've seen some repeatability results from microswitches showing 2-3 orders of magnitude more spread in their data than inductive or capacitative sensors, whether that really means they do the job less well I don't know. I've tried inductive with an aluminium bed which worked, and capacitative with a glass bed and it works too.

That's being said I use it more now to see if my bed needs manually levelling and for optimising first layer adhesion than I rely on it to print on a unlevel bed.

I've also used force sensing resistors on a delta which work brilliantly well, being nozzle based there is no offset and the nozzle is the probe.

I'd like to explore nozzle based probing on moving bed designs, I have a Chinese hot end with an fsr built in but haven't tried it yet, and I'm designing a prusa i3 x-carriage with a hinge which will move just enough to click a microswitch when the nozzle touches the bed. If you have a metal bed you could ground a signal through the nozzle to the bed as long as you were fairly scrupulous in cleaning your nozzle or always probe with a heated bed and nozzle heated enough to allow any residual plastic to be pushed away to make contact but not enough to cause ooze.

Excuse me for mentioning my own product, but the other popular option is infra red. My differential IR sensor has been adopted by both E3D and Think3dPrint3d for their 3D printer kits. The sensor is much smaller and lighter than an inductive or capacitive sensor or a servo-deployed switch. Follow the link in my signature for more.

---

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 looking at DC42's IR sensor, and that looks like a very good solution to me. [miscsolutions.wordpress.com]

I once saw a design for a servo based abl probe where the servo was used only to lift the probe out of the way when not in use. When probing the servo did not affect the position of the probe at all, so there was no need to worry about variation in servo position - a very clever design. Sadly I don't have a link to it. It used an opto endstop which the probe could pass straight through when stowed.

From investigations some years ago I believe that capacitative, inductive and microswitch sensors all offer repeatabilities in the range of a few microns to a few tens of microns. In 3D printers microswitches are easily made unreliable by exceeding their over-travel limits.
I am presently trying out DC42s differential IR sensor and am very impressed - repeatability is better than the feeler gauges and mechanical DTIs that I have checked it with. The only thing that it lacks, as do most of the others, is that it needs to be adjusted if anything happens to change its position relative to the nozzle - such as nozzle replacement.
I am looking at force sensors in the build stage mounting or the nozzle mounting to detect the nozzle touching the bed. This may be only for providing a datum or for bed leveling as well.

Mike

I have just opened a thread on PrinterHead-based single FSR probe and I am exploring some possibilities. I'd like to somehow replicate the "hinged microswitch in the hotend mount " solution but with No hinge, No microswitch (FSR will be used and very LITTLE extra movement associated with the nozzle hitting the bed. I am also thinking of using the FSR in a NC state by preloading it at the beginning and breaking circuit when the nozzle hitting the bed will tend to push the whole mount away from the pre-loaded FSR.

If anyone interested please chime in. If my understanding is correct and this will work, we're looking at a sub $10 solution (assuming no extra electronics are necessary to wire the FSR).

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I haven't played with fsrs, but a quick look suggests you'd want to build it into a voltage divider circuit and then use a comparator with a cermet reference to get a tunable digital output. No big deal, and it would be interesting to see what the repeatability is. I suspect the interesting part is the mechanical design to get the range of force to match the characteristics of the sensor. It would be nice to probe directly with the nozzle for sure.

Which is better thicker and shorter or narrower and longer? I see 6mm in diameter but sensing distance is 1mm :-/


Then again there are these square ones:


Which are the ones we should be looking for for 3D printing purposes?

Did anyone split open such an inductive probe? I wonder if the 3 circuits it's made of are delimited by wires or they are single circuit board and only the schematics shows 3 distinct circuits.

Edited 3 time(s). Last edit at 02/01/2016 04:00PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

The fsrs I use come with a "trinket" (their name not mine) board which interprets the fsr signal and provides three wire output for direct connection to z min endstop. I presume it helps when you have 3 to smooth out the signal, but I'd probably use one even for a single fsr, it's probably a few $ and purpose designed for the job.

Quote
DjDemonD
The fsrs I use come with a "trinket" (their name not mine) board which interprets the fsr signal and provides three wire output for direct connection to z min endstop. I presume it helps when you have 3 to smooth out the signal, but I'd probably use one even for a single fsr, it's probably a few $ and purpose designed for the job.

My interest in FSR is the following: I will place it under load by default, just like under bed, where the weight of the bed keeps it activated. This should be the zero. Then, when the nozzle pushes into the bed, that very slight push-back onto the hotends' clamp will release just a tiny bit of pressure from the FSR. How small of a difference can the FSR sense? I am aiming at as little movement as possible, if possible no movement at all but for that the sensitivity of the FSR should approach the strain gauge performance (we started discussing that on the other thread so I am not hijacking this one).

SO what is the typical break force for FSR (I read 20g)?
HOW much are those FSRs you are talking about cost? I would be interested if they come with a board and are not very expensive

Edited 1 time(s). Last edit at 02/01/2016 05:08PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I bought mine from ultibots they work well ($41). Board available for $10 here johnsl fsr board. So 1 fsr and board $16.95.

I don't have a direct way of measuring the force require to trigger one but I could put my "jewellery" scales on top of my kossel bed and press down see what it takes to trigger it.

Not a very scientific test but it seems 50g approximately but the fsrs are under the glass bed, aluminium heat spreader and sandwiched between two cork discs,and wired to a johnsl board.

Edited 2 time(s). Last edit at 02/01/2016 05:41PM by DjDemonD.

I bought one of these also fsr hotend just wish I could find the time to try it. Might need the controller board off my kossel to output a useful signal and doesn't appear to be rebuildable. But I bought it as it seemed like a neat idea might be worth copying in some way. Fsrs are meant to be quite heat sensitive (78 deg C max, hence cork discs come with ultibots kit to insulate from bed temperatures) so using one in a hotend might be a problem but in your hinge type idea this isn't an issue.

Edited 1 time(s). Last edit at 02/01/2016 05:29PM by DjDemonD.

I'll take the risk of cross-posting because my other thread doesn't seem to have too much success ...I'll link my idea here:



Now, ideally, without an additional board the pink FSR in the image should have the two wires coming out one in the Z-min and one in the Ground (or whatever the wiring of an FSR is) on the board. In the Kossel FSR Wiki they say that "We use three FSR wired in parallel, connected to the Z_MIN endstop pin. This allows calibrating the print surface with 0.02 mm repeatability, which is good enough for printing PLA[...]The endstop (digital) input pin works just fine, without any additional parts."

So in theory this should work although the repeatability should be better.

Considering that the Strain Gauges work in a similar fashion (am I wrong?), I wonder why the output from those can't be sent directly to the Z_MIN . Is the output too small that it can't be picked up as a difference in resistance?

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

This looks interesting. My understanding of fsr's using a controller board - it samples the resistance under the normal load i.e. the weight of the bed on the fsr's, and outputs 0v on the signal pin as long as the resistance remains at this level, then when the resistance drops it switches from 0v to 5v on the signal pin to the z min endstop. So it operates normally open, using the controller board. Without the controller board the fsr in your example will show resistance in the 1-5 kohm range according to the graph I have attached from this guide fsr guide assuming your hot end/carriage is applying 100-500g of weight on the fsr. When you remove the load as the nozzle touches the bed the resistance will increase perhaps to 15 kohm or higher. So perhaps it is possible to run 5v through the fsr to the z min endstop signal pin, set it as normally closed and then use it as you have described. But I am not sure if you will be able to discriminate between open and triggered without a bigger difference in resistance, or a controller board configured to operate essentially in reverse.

In your kossel example above which does not use a trinket board or Johnsl board the FSRs are preloaded by mass of the bed then force is increased when probing, not decreased. So will it work? I can try it with my FSR's later on this evening if I get the time.

How will you manage mounting your hot end to allow a little movement on bed contact but no movement during printing? This is where I feel having a hinged print head will work better, the mass of the whole head will maintain the nozzle position under its own weight unless you push upwards during probing when it will move a tiny amount. I can see an advantage to this if you get a blob and do not have nozzle lift enabled in your slicer there is less chance of knocking the part off the bed as the head can rise over the blob if needed. My idea using a microswitch could easily be converted to an fsr, this would drastically cut down on the amount of movement afforded to the head during printing. The CEL robox has a system like this using metal contacts which tilt back on probing and make a circuit with the smooth rod. Given its generally great reviews this movement in the print head must not affect print quality much.

Edited 2 time(s). Last edit at 02/02/2016 05:59AM by DjDemonD.

Quote
realthor
I'll take the risk of cross-posting because my other thread doesn't seem to have too much success ...I'll link my idea here:

Now, ideally, without an additional board the pink FSR in the image should have the two wires coming out one in the Z-min and one in the Ground (or whatever the wiring of an FSR is) on the board. In the Kossel FSR Wiki they say that "We use three FSR wired in parallel, connected to the Z_MIN endstop pin. This allows calibrating the print surface with 0.02 mm repeatability, which is good enough for printing PLA[...]The endstop (digital) input pin works just fine, without any additional parts."

So in theory this should work although the repeatability should be better.

Considering that the Strain Gauges work in a similar fashion (am I wrong?), I wonder why the output from those can't be sent directly to the Z_MIN . Is the output too small that it can't be picked up as a difference in resistance?

The strain gauge works by its resistance being related to its "length". It needs to be stretched or compressed along its longest side. Forces acting on its sides (large area) as you show will not work.
If measuring between 2 parts each end needs to be secured. For your proposal I suggest you fix it (epoxy) to a cylinder slightly longer than the sensor and then clamp that.
I am not familiar with the auto levelling feature although I know its in the firmware, but I don't see that you have to use the nozzle at all. You can run a simple code to raise the hot end a specified distance, fit a spacer of known size to your sensor (extends below the nozzle), test the various locations. Then remove the spacer and calculate the offset to put in your slicer. You may need to develop a fudge factor but once you have worked it out it shouldn't change.

My drawing above refers to a FSR not a strain gauge. But I am thinking about a strain gauge solution as I learn that I need an additional op-amp of small board of sorts which would then raise a digital pin on the 3d printer's controller board.

Regarding leveling, all this is about the auto-leveling feature, where the printer does this by itself, without me placing a z-zero probe on the bed like CNCs and other large/specialized machines do. Those don't need calibration very frequently because of their size, their weight, their rigidity and good materials/parts. 3D printers nowadays are more like flimsy skeletons compared to those. If I would be able to build my printer of steel bars and could have a granite bed and it would be square and all... i wouldn't probably need calibration but twice a year. But we kind of need it every other day or every new print for some printers.

Hence my exploration of a cheap (<$15) way to achieve this: inductive probe needs metal bed, capacitive probe is prone to humidity and temperature, IR probe is too expensive, FSR and Strain Gauge probes are the last attempt to achieve something reliable and cheap. With these two, FSRs are cheap but not very reliable (maybe those piezo are but no idea of the cost of those) and strain gauges need very careful mounting on a steel beam of sorts (to build a load cell).

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

For a strain gauge, you need to use an instrumentation amplifier, not an ordinary op amp. You will also need up to 3 precision resistors, depending on the configuration of the strain gauge. So it's not a cheap and easy option. If you want something really cheap, I suggest a microswitch.

---

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].

Quote
MCcarman
I am not familiar with the auto levelling feature although I know its in the firmware, but I don't see that you have to use the nozzle at all. You can run a simple code to raise the hot end a specified distance, fit a spacer of known size to your sensor (extends below the nozzle), test the various locations. Then remove the spacer and calculate the offset to put in your slicer. You may need to develop a fudge factor but once you have worked it out it shouldn't change.

The issue is that offset probes of any type have limitations, you are unable to probe any area on the bed (unless you can mount your probe a few mm from the nozzle - almost impossible), with deltas you can get effector tilt meaning the probed points might not translate to above-the-bed-heights for the nozzle, they occupy space and add mass to your print head/effector.

Using the nozzle as the probe satisfies many requirements, the exact x,y position you probe gives you an exact z height for that point for the nozzle itself, you can probe the whole bed, generally there is little or no mass added to the print head/effector (thinking fsr's under the bed for a delta).

I have an i3 and a kossel mini, the i3 uses a capacitative proximity sensor at a significant offset 37mm in x direction, the kossel uses fsrs under the bed. I know which printer performs better when it comes to bed adhesion and generally has fewer problems with the bed being level - the kossel. As such we are trying to work out the simplest and cheapest nozzle based solution for a cartesian printer. I think it probably would be to pass a current through the nozzle to a conductive bed surface like aluminium, so long as the nozzle was clean (or hot) when probing. But you're then limited to a metal bed, I prefer printing onto printbite or glass.

Quote
DjDemonD
The issue is that offset probes of any type have limitations, you are unable to probe any area on the bed (unless you can mount your probe a few mm from the nozzle - almost impossible), with deltas you can get effector tilt meaning the probed points might not translate to above-the-bed-heights for the nozzle, they occupy space and add mass to your print head/effector.

Using the nozzle as the probe satisfies many requirements, the exact x,y position you probe gives you an exact z height for that point for the nozzle itself, you can probe the whole bed, generally there is little or no mass added to the print head/effector (thinking fsr's under the bed for a delta).

I have an i3 and a kossel mini, the i3 uses a capacitative proximity sensor at a significant offset 37mm in x direction, the kossel uses fsrs under the bed. I know which printer performs better when it comes to bed adhesion and generally has fewer problems with the bed being level - the kossel. As such we are trying to work out the simplest and cheapest nozzle based solution for a cartesian printer. I think it probably would be to pass a current through the nozzle to a conductive bed surface like aluminium, so long as the nozzle was clean (or hot) when probing. But you're then limited to a metal bed, I prefer printing onto printbite or glass.

You've summarised well the issues that led me to develop the mini differential IR sensor:

- Needs to be light enough to use in a delta without appreciably adding to the weight on the effector

- Needs to be small enough so that it can be mounted close to the nozzle, to minimize the effect of tilt on a delta and maximize the area of the bed that can be probed. On my delta, the sensing area is just 14mm from the nozzle of my genuine E3Dv6.

The IR sensor works very well with PrintBite, because it's opaque to IR.

---

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 don't doubt the effectiveness of your sensor David and it does seem to satisfy all of the requirements for a z probe (albeit with a small offset - perhaps that could be reduced by miniaturising the IR sensor and mounting it on the heatblock with some insulation? - just blue sky thinking there). And I am sure that for all practical purposes this type of offset sensor is more than sufficient.

I just can't help but feel there is a purity in using the nozzle as the probe, and I share Realthor's desire to find a cheap, repeatable and practical nozzle based solution for moving bed printers (since under-bed FSR's satisfy this entirely IMO for fixed bed printers - I am printing ABS on the Kossel with FSR's and still getting excellent results).

Quote
dc42
For a strain gauge, you need to use an instrumentation amplifier, not an ordinary op amp. You will also need up to 3 precision resistors, depending on the configuration of the strain gauge. So it's not a cheap and easy option. If you want something really cheap, I suggest a microswitch.

Hi David, I was getting quite excited about the strain gauges, which is a new idea to me. They seem cheap (about $1 each in lots of 5), conceptually simple and robust (once epoxied to a suitable structure), and self compensating for temperature if used in pairs. If we could find a carriage design that we could fit two pairs for a full bridge then no extra resistors are needed and we're just left with the circuit design. Does it really need an instrumentation amp to get a simple tunable on/off signal? I know that the signal conditioning requirements are fairly steep for getting accurate measurements, but our needs are much more basic than that. My first thought was that a simple op-amp or comparator might work if we're dealing with a signal of a few mV? My electronics is coming up short, I can imagine the (trivial) circuit for using a comparator with a half bridge, but not the full bridge. I guess the full bridge is also trivial if it has 0 offset at rest, but that's an unlikely special case, and I can't quite see how to make the trigger point tunable. Some sort of feedback?

Edited 1 time(s). Last edit at 02/02/2016 08:41AM by JamesK.

I have just mentioned in another thread [forums.reprap.org] that the Sparkfun HX-711 breakout board will interface to a full bridge strain gauge load cell and is really cheap (£2.99 in the U.K. and £0.99 post free from China.)
Difficulties are:-

• The output from this is a two wire serial output not a digital on/off signal. There may be some supporting software in the article [www.instructables.com] but I haven't yet had time to read it.
• The biggest single problem is to make or buy a load cell element that will support the extruder & nozzle firly while providing a good sensitivity

So I am not accused of stealing anothers thunder, this can be found in realthors first submission.

Mike

Quote
leadinglights
I have just mentioned in another thread [forums.reprap.org] that the Sparkfun HX-711 breakout board will interface to a full bridge strain gauge load cell and is really cheap (£2.99 in the U.K. and £0.99 post free from China.)
Difficulties are:-

• The output from this is a two wire serial output not a digital on/off signal. There may be some supporting software in the article [www.instructables.com] but I haven't yet had time to read it.
• The biggest single problem is to make or buy a load cell element that will support the extruder & nozzle firly while providing a good sensitivity

So I am not accused of stealing anothers thunder, this can be found in realthors first submission.

Mike

No worries @leadinglights if 200 people on this forums would write about this we would have a workable solution in one month. I don't want to take credit for anything but maybe insisting too much on this .

I wonder if we need a load cell at all. Basically we can use an aluminum strip and epoxy the strain gauge to it. I have mentioned in my original thread some already ready-to-buy strain gauge + board maybe those can be used as an interface with the printer's controller. Please read it over on the other thread I am sorry for transforming this " inductive vs microswitch" thread into FSR vs Strain Gauge

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

The HX-711 seems a bit overkill for what we need. A lot depends on the sensitivity, but after some more thought, it looks to me like two strain gauges on opposite sides of the flexing structure to form a half bridge, with a cermet pot to form the other side of the bridge and provide offset adjustment and a comparator to generate the digital signal would be a good place to start some experiments. I was hoping we might be able to build the flex structure into the carriage design, but I don't know how practical it would be to glue the gauges to fdm plastics. Non-contact IR still seems pretty appealing in a lot of ways, not least being that we know it works.

Edited 1 time(s). Last edit at 02/02/2016 08:32PM by JamesK.

Amazingly, no one that I saw mentioned the BLTOUCH. I have this working now on my FT 2020 for $33 shipping included from S. Korea. It took about 2 weeks to get here. [www.ebay.com]

My biggest problem was that, while it worked right out of the 'box' with my laptop hooked up, it required a separate 5V supply on the ramps board. Some people can run it off of the Ramps power and others like me can't.

---

Folger Tech 2020 i3 and FT-5 as well as modified JGAurora A5 with direct drive E3D/Titan. All running the BLTOUCH.
Great kits. Having fun and running the heck out of them.
Running Marlin 1.1.0 RC8 on the i3 and FT5. Custom firmware on A5.
Folger Tech Wiki board >[folgertech.wikia.com]

Quote
tjnamtiw
Amazingly, no one that I saw mentioned the BLTOUCH. I have this working now on my FT 2020 for $33 shipping included from S. Korea. It took about 2 weeks to get here. [www.ebay.com]

My biggest problem was that, while it worked right out of the 'box' with my laptop hooked up, it required a separate 5V supply on the ramps board. Some people can run it off of the Ramps power and others like me can't.

We're kind of talking sub $15 solutions here. I'd go for DC42's optical sensor for anything else (besides lasers, electron beams, etc). Moreover, I haven't really saw it adopted on a large scale after the indiegogo campaign. It's so hard to find these days too.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Just throwing this out here: what is in IR sensors that we are looking for?

There is this Polulu IR proximity sensor that sells for $5.95.
What about this chinese one?

I want to understand the difference between them and what makes them worth considering.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff