Z-height sensor - inductive?

First let me state I'm definitely NOT trying to undo any of the good work done on none contacting z height sensing.

I was just going through my 'old projects parts box' and there it was an inductive sensor from RS Components
RS 249-2489 [uk.rs-online.com] which is actually a Pepperl + Fuchs Part No. NBB0,8-4M25-E0
It 4mm diameter and has a sensing distance of 0.8mm.

These are switches, so just an on/off output (with LED indicator in the barrel) so you don't get any distance indication, just a switched action at a fixed distance from the target.
I tried it hand held with the aluminium tape under the Kapton and it appears to work very well. So apart from the high cost £55+VAT of this solution is this the 'ultimate' sensor technology. Not affected by temperature, humidity, stray light or stray capacitance.
OK I'm jumping the gun, I could not find temperature data and I also don't know at the moment the stability of the switching point.

So across the Forum, we have reflective light, ultrasonic and capacitance being investigated. Magnetic dropped below the radar. But is inductive an option if the price is right?

I'm still in favour of Daves Modulated IR design with a top hat light guard, but I'm interested in thoughts on the inductive option.

Is there an easy way to interface a position switch such as this to the existing interface circuit and software?

---

Ormerod #007 (shaken but not stirred!)

Hi Treth,

Thanks for bringing a new idea to the problem of height sensing.

The modulated IR sensor works well, except in the case of direct sunlight falling on the bed, and even that might perhaps be fixed by suitable shielding. But we really want a sensor that works anywhere on the bed. That is why Radian & I have both been working on ultrasonic and more recently capacitive sensors. So an inductive sensor that relies on aluminium tape wouldn't offer any great advantage that I can see.

However, an inductive sensor that senses the aluminium under the glass could be interesting. The main disadvantages I can see of inductive sensing is that it would probably require a hand-wound coil, and would need a reasonable area in order to have good detection range at 4mm or so - although probably not as much area as a capacitive sensor. It might be necessary to move the bed compensation points a little further in from the corners, bit that is also true to some extent for ultrasonic and capacitive sensing. By using an on-board microcontroller as I am currently doing in kit 4 and 5 of my hot end board, I think the cost could be quite low, not counting the cost of winding the sense coil.

---

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].

From here the next logical step for the Z sensor has to be Sensor Fusion

regards
Andy

---

Ormerod #318
www.zoomworks.org - Free and Open Source Stuff

Hi Treth. The capacitive scheme has been working very well for me in the last couple of weeks, but I suspect changes in humidity are going to require more frequent recalibration than an inductive approach - which is probably the least affected by environmental variables. As things were moving surely, if a little slowly, with the other techniques I didn't want to get sidetracked by exploring yet another proximity detection solution - but now you bring it up I guess now might be the time

I would say that it's definitely viable. In the past I've used a small ferrite cored inductor in place of a crystal to clock a microprocessor at a rate that varies when brought close to a ferrous object. This wasn't required for a critical measurment so I used the CR watchdog timer as the reference timebase. However, the PIC chip I'm currently using for the capacitive sensor has secondary xtal amplifier that can clock an on-board 16 bit timer so this could be used with a LC tank leaving the microprocessor clock xtal as the timing reference. Rats, I guess I'm going to have to test it out now.

---

RS Components Reprap Ormerod No. 481

Quote
kwikius
From here the next logical step for the Z sensor has to be Sensor Fusion

regards
Andy

Interesting. The first method mentioned in the link was "stereoscopic vision" which is another technique I was playing around (in my thoughts) with! This or a close range triangulation method 'fused' with the modulated IR beam and the.........., could be the solution!

I still like the inductive sensor concept and my thoughts have moved on to a pancake (flat) type coil and a simple BFO type detector or similar as used in low cost metal detectors. Here we could detect the aluminium heat spreader in any position on the bed. Sadly I think temperature will then become an issue. So no better than capacitance or ultrasonic.

Looking back at the capacitive design, I do like the fact (mentioned in the posts) that at the close range sensing most of the dielectric is glass so it could be a stable option.
I do however wonder if the capacitor plate should have a 'ring guard' or similar as used on capacitive switches to reduce stray capacitance effects, but perhaps this will reduce the sensitivity to the desired capacitance too much.

I digress, but the inductive option could still be good in comparison to the IR sensor which both require a target, or if the range went through the glass could be another option as pointed out by DC42.

Thanks for your thoughts.

---

Ormerod #007 (shaken but not stirred!)

Quote
Radian
Hi Treth. The capacitive scheme has been working very well for me in the last couple of weeks, but I suspect changes in humidity are going to require more frequent recalibration than an inductive approach - which is probably the least affected by environmental variables. As things were moving surely, if a little slowly, with the other techniques I didn't want to get sidetracked by exploring yet another proximity detection solution - but now you bring it up I guess now might be the time

I would say that it's definitely viable. In the past I've used a small ferrite cored inductor in place of a crystal to clock a microprocessor at a rate that varies when brought close to a ferrous object. This wasn't required for a critical measurement so I used the CR watchdog timer as the reference timebase. However, the PIC chip I'm currently using for the capacitive sensor has secondary xtal amplifier that can clock an on-board 16 bit timer so this could be used with a LC tank leaving the microprocessor clock xtal as the timing reference. Rats, I guess I'm going to have to test it out now.

Hi Radian, this was just another technique that came to me and I do not want current efforts diluted!
I like the capacitive sensor, even though the capacitance is small, in my view the fact that the glass is there could be its success.
I also like the ultrasonic method, I have never seen the sub cycle detection method and think this quite novel, but not so sure about temperature and humidity effects so much.
The modulated IR beam will probably work for me as I can reduce the sunlight impact, so this is my favourite at the moment.
The inductive switch from RS looked amazingly good with the tape and I felt I should just mention it for discussion......

---

Ormerod #007 (shaken but not stirred!)

My experience so far with capacitive sensing is that it is somewhat sensitive to bed temperature. However, that may be because I am not using a temperature-stable reference capacitor; or there may be a gap between the glass and the aluminium near the centre of the board where I am taking measurements, and that gap may be expanding; or the fact that the glass bed is coated with solvent cement may be affecting the reading. I hope to try out a more stable reference capacitor later today.

The capacitance being measured is small. Assuming that the sense plate is 1mm higher than the nozzle, when the nozzle is touching the glass, the capacitance of my sense plate (15mm x 20mm) is about 1.7pF. Stray capacitance in my setup is about 30pF, and although some of that is due to the chip I/O capacitance, some of it will be due to the PCB traces running most of the length of my sensor board that are connected to the sense plate. With a PCB optimized for capacitive sensing, this could be reduced.

One of my concerns with capacitive sensing is that it may be sensitive to coating on the glass, for example Kapton tape, PVA glue, the solvent cement that I sometimes use, and possibly even the traces of vinegar used to prime the bed when printing directly on the glass. Inductive sensing should be less sensitive.

When I come to try out inductive sensing, instead of using the inductor in an LC oscillator circuit, I'll probably try measuring the phase shift of an LR or LCR circuit. The losses in the inductor will increase substantially as it gets nearer the plate, so there should be a measurable amount of phase shift.

I have two Arduino-based metal detector designs, and it is tempting to try adapting them for this application. However, the induction balance one needs a complicated search coil, and I think a design based on the pulse induction one would be most unlikely to meet CE EMI standards.

---

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].

dc42, I'd be interested to see how you get on with improvements to the capacitive sensor in the other thread. Temperature stability doesn't seem to be an issue using the PIC CSM - it applies and removes a constant current to the sensor plate alone so there's no additional external parts involved.

A couple of years ago I spent (wasted?) hundreds of hours on a project to design a powerful pulse induction metal detector. For our application even a drastically cut down version would still require a minimum amount of precision analogue circuitry so I'm pretty sure we can count it out on the grounds of cost/complexity if nothing else. My detector project showed me just how rapidly diminishing the returns are (figuratively as well as literally) when trying to increase detection by increasing field strength. It did however fork into a useful contactless battery charger for my car!

---

RS Components Reprap Ormerod No. 481

Quote
Radian
A couple of years ago I spent (wasted?) hundreds of hours on a project to design a powerful pulse induction metal detector. For our application even a drastically cut down version would still require a minimum amount of precision analogue circuitry so I'm pretty sure we can count it out on the grounds of cost/complexity if nothing else. My detector project showed me just how rapidly diminishing the returns are (figuratively as well as literally) when trying to increase detection by increasing field strength. It did however fork into a useful contactless battery charger for my car!

Reminds me of the wasted days in military research attempting the long-range detection of things that resonate at the frequency of an AK47 rifle. Still, failures are also instructive.

Dave
(#106)

I've been looking into the various height sensing options. Here are my thoughts so far:

1. Manual calibration: lower the nozzle slowly to the bed, usually with a sheet of paper in between. This is accurate but time-consuming. It cannot be used for automatic bed compensation. The nozzle needs to be clear of solidified extruded filament, which can be achieved by heating it to 150C if the filament is PLA.

2. Unmodulated IR: poor due to high sensitivity to ambient light (sunlight and incandescent light). OK in e.g. a basement lit by fluorescent or LED light. Generally needs reflective targets; however Kim B. has reported successful use using red glass and no targets. It is not clear whether the IR is being reflected from the surface of the glass, scattered in the glass, or reflected from the heat spreader.

3. Modulated IR: generally satisfactory. Slight sensitivity to ambient light (sunlight and incandescent light), but 30 times less than unmodulated IR. Still need to avoid direct bright sunlight shining on the bed. Main disadvantage is that it requires reflective targets (but see note above). Use of Kapton, solvent cement etc. on top of targets is likely to alter the reflectivity and require recalibration.

4. Ultrasonic, based on sensing interference from ultrasound reflected from the surface of the bed. Works anywhere on the bed. Main disadvantage is that it is temperature sensitive, due to variation in speed of sound with temperature. Works well with the bed cold and at a stable room temperature. If used with the bed at an elevated temperature, then convection currents and draughts affect the reading, making it less accurate.



5. Capacitive. Senses the capacitance between a plate mounted on the hot end and the aluminium heat spreader. The capacitance is only about 1 to 2pF, assuming a reasonable-sized (300 sq.mm) sense plate 1mm above the nozzle and a 3mm glass bed. May be affected by Kapton tape, solvent cement etc. on the bed surface and also by humidity in the air. Any slight surface conductivity on the bed is also likely to cause an increase in effective capacitance. Need to have good contact between the heat spreader and the glass across the whole surface. Stray capacitance in the sensor circuit may be many times higher than the sensed capacitance, and any temperature sensitivity in this stray capacitance will cause inaccuracies. I have found the temperature sensitivity to be too great on my prototype, however I am using a hacked sensor board and the stray capacitance in PCB traces is large (about 50pF). The stray capacitance could be reduced with better board layout. Radian has reported that his prototype is temperature-stable. In order to achieve height accuracy of 0.05mm, the capacitance needs to be measured (and the stray capacitance stable or compensated for) to within about 0.04pF, which is less than 1 part in 1000 if the stray capacitance is 50pF. If the stray capacitance can be reduced to 10pF, then only 1 part in 250 accuracy would be needed.



6. Inductive. Using a practical size sense coil (18x13mm), with the coil used in an L-C oscillator circuit, I find that the inductance change is about 1.6%/mm with the lower cheek of the sense coil between 4mm and 6mm from the heat spreader plate. To achieve 0.05mm stability, the inductance will need to be measured to 1 part in 1250. If the inductance is measured by using the inductor in an oscillator and measuring the frequency, then the tuning capacitance and sense coil inductance need to be stable to better than 0.08%, and the crystal oscillator used to measure the frequency needs to be stable to better than 0.04%. This should be achievable using a crystal, but probably not using a ceramic resonator. A larger and/or ferrite-cored sense coil should be more sensitive, but there is limited space available. It may be possible to achieve greater stability using a double sense coil in conjunction with techniques such as induction balance; however even a single sense coil winding is difficult to mass-produce. Need to have good contact between the heat spreader and the glass across the whole surface.



7. Force sensing. Lower the head on to the bed and detect the reduction in weight of the head when it touches, e.g. by monitoring the sideways force on the x-runner bearing support. Not tried yet. Would work anywhere on the bed, but like manual height setting, it would be sensitive to any solidified filament on the end of the nozzle.

Thoughts or suggestions, anyone?

---

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].

Anther option is to use a z-axis position detector like for the x and y axis, either a microswitch or an opto break beam system, ideally in a dark box

regards
Andy

---

Ormerod #318
www.zoomworks.org - Free and Open Source Stuff

Quote
kwikius
Anther option is to use a z-axis position detector like for the x and y axis, either a microswitch or an opto break beam system, ideally in a dark box

regards
Andy

True, but that's only OK if the bed is level and very stable. If like me you use several different glass plates, then you really need to sense the height above the glass, not the height of a fixed switch. I have seen a picture of a microswitch on the hot end that rotated into place when needed.

---

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].

It seems to me that all of the suggestions have downsides and trying to get any of them stable enough for consistent Z height and bed levelling is proving very difficult.

I wish I had more expertise and could contribute more because to me, the rotating micro switch would be the answer to providing consistent results.

---

appjaws - Core XYUV Duet Ethernet Duex5
firmware 3.1.1 Web Interface 3.1.1
Ormerod 1-converted to laser engraver, Duet wifi
OpenSCAD version 2020.07
slic3r-1.3.0, Simplify3D 4.1.2, Cura-4.4.1

I only use option 1, manual with a feeler gauge. Your comment of 'accurate but time consuming' is rather odd. The time taken to set Z0 in centre of the bed is about 2 or 3 minutes. I usually set the bed to warm up whilst doing it, so no time is lost before printing. I appreciate most people don't want to faff about with a 3D printer, but how often are most people printing? I would imagine we are only talking about a few times a week. I really don't mind a few minutes every few days.

Other option I would like to use is 7. In the other sections of this forum people are using force sensing resistors. I would love to have this option and just watch the nozzle touch the bed in a few places and its done.
Seriously, THAT is the option people need to be developing for the Ormerod. I can't as I have very little knowledge in that sort of thing. But as seen by the other sensors, other people do have the ability.

But for now I'll spend 3 minutes by hand.

How about a super cheaper and simple touch probe? This is the method I use on my small CNC machine to mill PCBs. The probe side would be pretty simple, maybe make it spring loaded like a biro tip. For the earth plate foil tape might work, or maybe cut the corners off the glass and use the aluminium spreader?

Regards,

Les

---

Pointy's Things
Pointy's Blog

Firstly very nice summary DC42.

Quote
dc42

Quote
kwikius
Anther option is to use a z-axis position detector like for the x and y axis, either a microswitch or an opto break beam system, ideally in a dark box

regards
Andy

True, but that's only OK if the bed is level and very stable. If like me you use several different glass plates, then you really need to sense the height above the glass, not the height of a fixed switch. I have seen a picture of a microswitch on the hot end that rotated into place when needed.

A mechanism that can be moved below the extrude nozzle for direct contact sensing and out the way, i.e. just above the extrude nozzle height while printing does appear to have many advantages.

Having looked at various ideas and products since I built my printer (which sadly I haven't had time to use for a few months now!) my 'best' thoughts are around this simple concept.

Essentially it is a probe consisting of a none rotating shaft free to move in z direction.
The shaft has a hole which is to allow position detection by a slotted-opto.
The slotted-opto is used as an analogue sensor such that it detects the varying IR as the sensing slots moves across its own sensing axis.

The probe is held loosely on a spring attached to a small solenoid or stepper motor used to lift it clear while printing.

Obviously this is just a sketch and the probe should be in a sleeve and the slotted-opto screened from stray IR, but it should work perfectly with Daves modulated IR sensing.
The precision of the sensing is set by the mechanical distance between the tip and its sensing hole, so this should be stable and the reference point for the 3D printer is the slotted-opto mounting which will be precise. So this should be very stable.

---

Ormerod #007 (shaken but not stirred!)

Quote
dc42

Quote
kwikius
Anther option is to use a z-axis position detector like for the x and y axis, either a microswitch or an opto break beam system, ideally in a dark box

regards
Andy

True, but that's only OK if the bed is level and very stable. If like me you use several different glass plates, then you really need to sense the height above the glass, not the height of a fixed switch. I have seen a picture of a microswitch on the hot end that rotated into place when needed.

I'm not too sure that different plates is a problem. If each plate is labelled and you use a different setbed macro for each one. (Ideally the setbed macro could take a label as a parameter but initially just a rename of the setbed.g file would do).

Have to say I like the idea of working from one fixed point of reference. In fact you can then use the fixed point as a known reference for other z-height sensors. You can then make more precise measurements to see how each part of the bed moves as the bed heats up for example

regards
Andy

---

Ormerod #318
www.zoomworks.org - Free and Open Source Stuff

Quote
Treth
Firstly very nice summary DC42.

Quote
dc42

Quote
kwikius
Anther option is to use a z-axis position detector like for the x and y axis, either a microswitch or an opto break beam system, ideally in a dark box

regards
Andy

True, but that's only OK if the bed is level and very stable. If like me you use several different glass plates, then you really need to sense the height above the glass, not the height of a fixed switch. I have seen a picture of a microswitch on the hot end that rotated into place when needed.

A mechanism that can be moved below the extrude nozzle for direct contact sensing and out the way, i.e. just above the extrude nozzle height while printing does appear to have many advantages.

Having looked at various ideas and products since I built my printer (which sadly I haven't had time to use for a few months now!) my 'best' thoughts are around this simple concept.
[attachment 31710 Zheightpositionsensor.PNG]
Essentially it is a probe consisting of a none rotating shaft free to move in z direction.
The shaft has a hole which is to allow position detection by a slotted-opto.
The slotted-opto is used as an analogue sensor such that it detects the varying IR as the sensing slots moves across its own sensing axis.

The probe is held loosely on a spring attached to a small solenoid or stepper motor used to lift it clear while printing.

Obviously this is just a sketch and the probe should be in a sleeve and the slotted-opto screened from stray IR, but it should work perfectly with Daves modulated IR sensing.
The precision of the sensing is set by the mechanical distance between the tip and its sensing hole, so this should be stable and the reference point for the 3D printer is the slotted-opto mounting which will be precise. So this should be very stable.

Perhaps you could use a variation of this instead of the force sensor. BTW Sincere apologies I havent done any work on this. I should do something rather than just talking about it!

regards
Andy

---

Ormerod #318
www.zoomworks.org - Free and Open Source Stuff

Quote
kwikius
... BTW Sincere apologies I havent done any work on this. I should do something rather than just talking about it!

regards
Andy

It is these discussions that help us progress....
I talk too much...., but as long as we have some 'do-ers' we are a good team.

Just to add, I have used these slotted-optos for position sensing many years ago (using a 6502 uP !!! that dates me.) and the positioning accuracy using an analogue measurement was very good, but I can't provide 'figures'.

It would also be better if the solenoid or positioning device was only powered while doing the sensing task.

---

Ormerod #007 (shaken but not stirred!)

thinking out of the box a bit....

Electrical circuit, positive print head, negative heat bed, nozzle touches, completes the circuit, set Z0.
Glass is an insulator, but I believe Kapton tape conducts, so you would need to use Kapton.
The circuit would only need power whilst sensing Z.
But could this be another option to consider?

Quote
balidey
thinking out of the box a bit....

Electrical circuit, positive print head, negative heat bed, nozzle touches, completes the circuit, set Z0.
Glass is an insulator, but I believe Kapton tape conducts, so you would need to use Kapton.
The circuit would only need power whilst sensing Z.
But could this be another option to consider?

That's what I suggested half a dozen post up. Simples!

Regards,

Les

---

Pointy's Things
Pointy's Blog

So you did, sorry, totally missed it when I was reading.

Quote
balidey
So you did, sorry, totally missed it when I was reading.

Hehe, no worries.

An obvious problem is the fact that the probe would have to be lower than the nozzle., so you would have to moved it out of the way somehow after probing.

Regards,

Les

---

Pointy's Things
Pointy's Blog

There used to a 'touch probe' for any material.
RS sold a robust diamond tip switch in a small tubular housing that had repeatable micron switching accuracy.
Basically a very robust accurate switch. Probably too expensive for us if they are still produced, but as with other techniques you need the 'move out the way' while printing and this must not impact on the accuracy.

The simplest method I can come up with is a solenoid pushing down the sensor against a spring that locates the sensor against the precise end-stop. Then do the contact measurement. Power off solenoid to retract.

---

Ormerod #007 (shaken but not stirred!)

I was thinking you could use the nozzle as the probe, not a separate one. Then no need to move anything or have additional hardware, just a feed to the print head, ground to the Kapton.

Quote
balidey
I was thinking you could use the nozzle as the probe, not a separate one. Then no need to move anything or have additional hardware, just a feed to the print head, ground to the Kapton.

I like this idea, but I think the Kapton Polyimide tape is a good insulator.
Also any hot melt extrude at the tip will be an issue.

---

Ormerod #007 (shaken but not stirred!)

I too was thinking about using a solenoid to extend or retract a contact probe. Mechanically, the simplest arrangement I came up with was to use a solenoid with spring return action, so that the spring extends the probe towards the bed. In normal use, the solenoid would be energised to retract the probe. When the probe is not retracted, the inductance of the solenoid could be measured to determine the amount of extension. The inductance change should be very large and so easy to measure. This would avoid the need for an opto sensor or switch, and I am much better at designing electronics and software than mechanical parts. The disadvantage is that power needs to be applied to the solenoid most of the time.

---

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
dc42
...The disadvantage is that power needs to be applied to the solenoid most of the time.

..but not as much holding power as the homing-pull requires - BTW, would be rather simple to make a flip-flop version with a horizontal (compared to the vertical movement) mounted spring

Erik

Quote
balidey
I was thinking you could use the nozzle as the probe, not a separate one. Then no need to move anything or have additional hardware, just a feed to the print head, ground to the Kapton.

I doubt it would be at all reliable due to the presence of plastic on the nozzle preventing it from making electrical contact much of the time.

While automatic Z homing is definitely desirable, unless it can be achieved fairly easily and is very reliable, I'll stick to the manual paper method, which takes me less than 30 seconds to carry out now that I have a level bed with no compensation and only need to take one measurement, so I'm not particularly bothered about the lack of automatic Z height adjustment.

With a stable level bed, my thoughts are definitely gravitating toward doing any automatic Z homing via a simple optical or mechanical switch on the Z post - I think it would be accurate enough for the purpose. I think it is more fruitful to focus efforts on achieving the mechanical stability that makes it unnecessary to measure bed height directly. As another poster said, if you have different thicknesses of glass, you can simply have a different Z home macro for each glass, but I cannot see I will need more than two glass plates (one to be currently printing and the other to strip off the last prints and prepare for swapping after the current print without delay), and it should be possible to get two glass plates cut that are exactly the same (within the tolerance needed).

In any case, many of the methods discussed would also not give the same height for different thicknesses of glass.

Dave
(#106)

Quote
dmould
In any case, many of the methods discussed would also not give the same height for different thicknesses of glass.

That's a good point.

I was going to develop a prototype inductive sensor next, however I'm leaning towards a solenoid-operated contact probe now. This will be independent of glass thickness and work anywhere on the bed. I've located a couple of solenoids that look suitable, and I am making inquiries.

There remains the issue of how to do x-homing. One possibility is to use a microswitch on the x-motor mount, although that would involve yet more wires running back to the Duet. Another is to keep an IR sensor on the hot end board just for IR homing. A third possibility is to mount a microswitch on the hot end board, such that its actuator lines up with the z-nut trap.

---

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].