Auto bed levelling - IR, inductive or nozzle probing

Hey reprappers!

I'm currently working on a new delta project wich is based on the Atom 2.0. I'm still figuring out what auto bed levelling sensor to use, and I'm getting no further.

The Atom itself uses it's nozzle with to probe, wich levers on to a microswitch. I think the benefit of this is that, even if your nozzle expands at certain temps, you always have the perfect distance from nozzle to bed.

As I'm going to use Duet electronics, I also saw the IR sensor. What are the main advantages using this over an inductive sensor? How accurate is it?

What do you recommend and why?

Thanks,
Andreas

I manufacture the IR sensor, but I'll try to give unbiased advice.

Nozzle probing: on a delta this has the advantage that the trigger height is not affected by the tilt of the effector. If your printer geometry and rod end joints are perfect then there will be no tilt, but in practice it is hard to achieve zero tilt. Rather than use a microswitch, I would use force sensitive resistors under the bed supports or a strain gauge built into the effector.

The disadvantage of nozzle probing is that you either have to probe with the nozzle hot (which may damage the bed surface) or you have to remove any filament that is stuck to the tip of the nozzle before you start probing.

IR sensor: very small so you can mount it close to the nozzle to reduce the effect of effector tilt. Works very well on most opaque bed surfaces except bright aluminium. When used with a transparent bed surface such as glass or PEI, it is important to use a black backing to minimise the reflection from the back.

Inductive: larger and heavier than an IR probe, so not very suitable for a delta printer. They detect metal and most of them don't have enough range to work with glass beds on top of aluminium. The cheaper ones tend to be temperature-sensitive.

---

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].

Another disadvantage of nozzle probing is, you have to use a strong spring ( compared to other microswitch probes ) because otherwise the nozzle will lift to easy and can't build up downforce to print the first layer.

An option worth considering is using an IR sensor for bed leveling but keeping the microswitch to detect nozzle contact. This way you only need to threaten the purity of you build surface at one position but while not having to put in lots of work every time you change the nozzle/build stage/build surface film etc.. There would probably be some software changes needed to perform nozzle distance separately from bed leveling/compensation

I do disagree with dc42 however on the removal of filament before probing as this must be done to set Z height by any method including the classic paper under the nozzle method.

As far as other methods go: I used FSRs in the distant past (not on 3D printers) and found them to suffer from fragility, insensitivity, temperature problems and aging. Capacitive and inductive probes have no advantages over IR probes

I am a great lover of piezoelectric disks but would not consider them for retro fitting to an existing Delta printer as you would likely need a sub frame to carry the build stage with its sensors as well as which there are no pre-existing amplifiers on the market for them*. Much info at [forums.reprap.org]

Mike

*Moriquendi may have some, see the last entry on the above thread

p.s., o_lampe's point is also true but if the microswitch probing is limited to a single position on the build stage it should not be too much of a problem.

Quote
leadinglights
I do disagree with dc42 however on the removal of filament before probing as this must be done to set Z height by any method including the classic paper under the nozzle method.

The best way I've found doesn't require the removal of filament (well OK, maybe right at the start when commissioning to get into the ballpark, but then you use the following method)

Firstly get a good sensor. I'm using a servo deployed contact sensor with very low trigger force (optically triggered). This has no issues with any bed surface, works hot or cold, zero nozzle offset, excellent repeatability, etc. But this post isn't about that. Any sensor that is repeatable can work with this method.

Once you have your sensor set up and in the right ballpark, then start a print using your sensor to level the bed beforehand.

While the skirt is printing, navigate to the babystep menu, and twiddle the babystepping value so that you get a perfect first layer. It helps here if you are printing a wide skirt, but after some practice you can get a feel for it with only a few loops. You're after slight dips between lines. If the nozzle is too low, then you get a flat surface between successive loops, with possible bulging out the sides and interference between lines. If the nozzle is too high, then you can see the print bed between adjacent lines (the lines aren't touching). There is a region of a few 0.01's of a mm where it looks good.

Let the print finish, then do another couple of prints. You will get a good feel for the value of babystepping that produces the perfect first layer.


Now when you have a moment, go into your firmware, and adjust the Z probe height by the value that you have come up with by experiment. Make the Z probe height larger if you need to squish the filament more, and make the value smaller if your nozzle already is too low while printing.

On the next print, you should get a perfect first layer without needing to babystep.


I do this whenever I change nozzle, and it doesn't require a clean nozzle, but it does require a little bit of watching and knob twiddling for the first 10 seconds of the first few prints after a nozzle change. It works well for me. It also doesn't require any special measuring procedure when you change nozzle, you just change nozzle and print your next piece. Then when you have a spare 5 minutes you can drag out your laptop, connect to the machine, and adjust the Z probe height using the EEPROM settings menu.

My experience with inductive probes is that they are cheap and if you get the right one reasonably easy to use. If your bed is designed so that metal is within sensing range of the sensor and the sensor isn't too far away from the nozzle, the repeatability is actually quite good. Downsides, get the wrong sensor and you have to mess about to make it work, they are large and reasonably heavy so on a delta not that practical.

Capacitive sensors are very temperature dependant which makes them problematic with heated beds. Also the same as inductive sensors they are big, reasonably heavy and fiddly to setup.

IR probe is a good solution if you mount it really close to the nozzle and your bed is IR reflective and not regularly coated in anything uneven. It's light, repeatability is good.

I've also got a delta with fsrs, they work very well I have had none of the issues a lot of people say will happen, I used the ultibots kit. However mounting the bed so that it has enough freedom to trigger the fsrs and yet not be able to shift if your nozzle hits a blob is a bit of a balancing act with the bed mounts. Piezo disk sensors (I've got one of Moriquendi's piezo boards but haven't got around to trying it) , you still have to suspend the bed in some way to allow it to be pressed and trigger the sensors. But nozzle base probing assuming you have a suitably robust build surface is quite satisfying, not having a sensor offset is quite desirable.

I like Nebbians touch probe it's a clever design all the benefits of nozzle probing with fewer of the drawbacks.

How big is your delta? I just built a micro delta and calibrated manually using DC42's least squares method and that works well and only takes about 10 minutes.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Thanks for the input!

The inductive is surely a no-go then. When I was looking into IR sensors I read they are very light sensitive, can this cause a problem?

The idea of using IR and nozzle is very good I think, I must have to look into this.

One thing that I think is a big drawback on nozzle probing with a microswitch, is that the hotend is not completely fixed, it is partially locked but allows movement, at high print speeds I think this could cause vibrations of the hotend?

My differential IR sensor is not sensitive to ambient light, except that bright sunlight (no clouds) reflecting directly off a shiny bed into the phototransistor can saturate it, which the sensor detects and responds to by blinking the LED. But there are other IR sensors on the market that are sensitive to ambient light, including sunlight, incandescent and especially halogen light.

---

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].

That is super nice. Is black glass plate a good option for an IR sensor?

If the glass itself is black, that should be an excellent bed surface for the IR sensor because it is not transparent, so the only reflection will be from the surface.

---

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 have my dc42 installed and does the probes but when starts printing the nozzle is too high and is printing in the "air", the sensor triggers when the nozzle reaches the 1.90 mm high, I can't figured out how to set the distance in my marlin, I have also the Print Bite bed.

Edited 2 time(s). Last edit at 12/08/2016 11:15PM by slanwar.

Quote
slanwar
I have my dc42 installed and does the probes but when starts printing the nozzle is too high and is printing in the "air", the sensor triggers when the nozzle reaches the 1.90 mm high, I can't figured out how to set the distance in my marlin, I have also the Print Bite bed.

Any idea how high it is? If it's printing 1.2 mm too high, then the command is:

M851 Z-1.2
M500

Of course you need EEPROM turned on for this to work.


Full instructions for Marlin are here: [www.thingiverse.com]
Note that this is for my Z sensor that is deployed/retracted with a servo, so you can skip the servo related parts.

Thanks, I will try when I get home.

...didn't work! Actually doesn't matter whats the value I use because always prints about 2 mm above surface, I even use the M502 before typing any code, comparing the 2 prusa I built with a mini kossel - no words!

I'm ready to try your Z sensor, I have everything except the metal rod I just need to find a bicycle spoke and the piano wire.

Remember though, try different sensors to improve accuracy, or reproducibility but you are still going to have to be able to enter a z-offset to get the nozzle to be at the right height for a consistent first layer. Or babystep down to the bed once printing starts. Make sure your #define MANUAL_Z_HOME_POS is set to allow you maybe 0.5-1mm of extra travel, especially when calibrating. The latest marlin version will not allow you to go below what it thinks is z=0, even if this is 2mm above the build plate when it isn't quite setup right.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

@slanwar - you dont need to run M502 before printing or typing code. M502 simply loads the values that are hard coded - ie the values in the code when it was compiled.

When you have made changes to settings using M commands like M201 M204 M205 etc, you store these settings into eeprom using M500 command. If you then use the M502 command you are effectively resetting the values you modified back to the hard coded compile time values. Be very aware of this, as you might think you are running with the values you modified during tuning, but will in fact be running with the hard coded values, and this will confuse the hell out of you.

---

Flex3Drive.com

To be honest, I don't think that my sensor will help things. As the good Dj said, you have another issue in your firmware setup. I'd muck around with the firmware settings and get the existing setup working properly before deciding on whether a better Z sensor is the way to go.

I changed Z_PROBE_OFFSET_FROM_EXTRUDER value and worked, that's what I do with my other printers so I decided to give a shot.

Quote
leadinglights
I am a great lover of piezoelectric disks but would not consider them for retro fitting to an existing Delta printer as you would likely need a sub frame to carry the build stage with its sensors as well as which there are no pre-existing amplifiers on the market for them*. Much info at [forums.reprap.org]

Mike

*Moriquendi may have some, see the last entry on the above thread

I do still have some boards available to anyone who wants them.

I disagree with them being difficult to retro fit, I found them very easy to fit under the bed of my kossel mini. I've been using them for a few months now without any need for adjustment and they've yet to let me down.

I'm probably biased but Piezo sensors are the best solution I've tried so far (Vs FSRs and mechanical probes)

Moriquendi

There is also BLTouch or 3DTouch Auto Leveling Sensor.
Advantage: It works on any kind of build plate
Disadvantage: The mount should be designed adjustable to adapt to different hotend/nozzle length.
(I bought one of these but I didn't test it yet)

I'm using nozzle probing on my CoreXY and I'm satisfied.
I can confirm the disadvantage (removal of old filament before probing), but with a CoreXY I don't need to level the bed every day.
Anyway, a wire cutter is always close to the printer

---

Best regards / Viele Grüße
Till

Quote
Moriquendi

Quote
leadinglights
I am a great lover of piezoelectric disks but would not consider them for retro fitting to an existing Delta printer .................................................

..............................
I disagree with them being difficult to retro fit, I found them very easy to fit under the bed of my kossel mini. I've been using them for a few months now without any need for adjustment and they've yet to let me down.
..............................
Moriquendi

Moriquendi and elenhinan both use piezoelectric disks attached to the surface of a stress member, somewhat as a strain gauge would be used. I had not previously explored this option but in the last few days I have done some trials with a piezoelectric disk attached to a short beam and can confirm that this is a very sensitive detector. It also seems to be less subject to crosstalk - picking up false readings from movement in the X and Y planes. The only downside is that there is inherently more compliance than there is with the mechanism that I have been using - but this increase seems to be only a few microns.

Mike

Edited 1 time(s). Last edit at 12/15/2016 04:04AM by leadinglights.

Hi Mike,

will you be adding a photo of the new design to the original piezo thread? I'm still hoping to implement this for my next printer so if the new approach seems to be working better than the original layout I'd be interested to see that. Peizo goes on the top of the beam so that it stretches when the bed is displaced downwards?

I made up a very simple switch based probe from this thread and was really impressed by the ease of use improvements from repetier's abl and distortion compensation, so I'm all fired up for having bed probing in the next printer.

Hi James,

At the moment there is not a new design as such, I simply epoxied a 27mm piezoelectric disk to a bit of 1.6mm FR4 PCB laminate. I will try to put a report with some measurements on the original piezo thread but it may be a while as I have been working on several projects (some are even unrelated to 3D printers) and am now being distracted by Xmas.

A couple of notes though from what I have observed:

• The beam should not be restrained on both ends as the disk will be bent in a sigmoid shape and output will be lower than if restrained on one end only.
• Although the disk works best on the upper surface, piezo disks give quite a lot of output from changing temperatures so should be insulated from radiant heat from the build stage heater.

A simple switch based probe such as you have pointed to has a lot to commend it. A simple but heat tolerant design that could be fitted to a hot nozzle and used with a hot build stage would be a great idea.

Mike