I am looking for information on how any non-contact ABL Z probes automatically determine the Z offset. I have seen many advertisements and reviews stating or implying that a probe or a printer sets the Z offset between the nozzle and the sensor without the user having to measure it, but little about how this is done.

To be clear, I only need information on the automatic determination of the Z offset for non-contact probes. Nozzle contact probes for piezo, force gauge, accelerometer, FSR, nozzle microswitch, nozzle electrical contact, nozzle capacitive, and sensorless probes are pretty well covered.

Sufficient information would be helpful, such as URLs or a brief description of how this Z offset is obtained for what type of probe. Answers such as "Your life will be improved by using a BLTouch probe" or "Nobody needs ABL if they make their printer right" are not wanted.

Background: I am writing an article about methods and sensors for automatic bed leveling of FFF printers. This article will be for publication in a journal and may be used to update the RepRap Wiki on Z probes. I have bought or made probes using just about every technology used on FFF printers, but I know nothing about how any non-contact probes get the magical Z offset value.

Mike

Hi Mike,

it can't be done without further preparation or modification. What I can think of is using a thin piece of the material the contactless probe can detect, let's say steel.

The contactless probe detects the steel build plate but does not know about any additional layer on top, e. G. PEI sheet. You need a small reference piece of that material somehwere on the PEI layer. Presuming you (and this the printer) know it's thickness the probe can use it to determine the z offset by probing it, probing the bed and subtracting that fixed distance.

Keep in mind that e. g. a piece of aluminum tape might work but has different material properties than a steel sheet so you may also need to get some factor on top of the tapes thickness to account for that.

So basically you need to set that value manually once and then you can use any additional layer on top of the bed which does not really affects the probes behavior in some way (e. g. a metalized foil)you want to.

But... in the end it always needs some form of calibration with a reference if you don't directly use the nozzle for probing, e. g. here with Klipper and a mechanical switch for probing and auto z calibration via a reference pin:

Klipper Auto Z Calibration

Edited 1 time(s). Last edit at 03/19/2024 09:01AM by Demolux_D1.

Quote
Demolux_D1
....

it can't be done without further preparation or modification. ...................

Agreed, I haven't had any evidence that "Automatic Z Offset measurement" is anything but a marketing phrase. Only nozzle contact methods can obtain the Z position and also map the build surface, though few nozzle contact solutions are without problems - Duet3D's Smart effector is the closest to a good solution but is hard to extend to non-Delta printers.

So far, the best solution I have found is to use two sensors: A single underbed sensor to obtain a reference Z datum from a nozzle contact, and a second sensor for mapping the bed surface. Two versions of this can be seen at [www.youtube.com] and [www.youtube.com]

Mike

Edited 1 time(s). Last edit at 03/22/2024 10:21AM by leadinglights.

Well, there is a possibility left where you can claim it though: If a z offset calibration is not needed in the first place, e. g. because of a very precisely and sturdy built machine with a factory-one-time-calibration.

But I don't know any FDM printer that has this, especially in the consumer world. It may exist in professional machines and it does exist in the resin printing world. I suspect because there you don't really have that heatbed-warping influence named heat.

Demo

I don't know of any precise machine of any reasonable complexity that has a reliable factory-one-time calibration - and I am not just talking about 3d printers.

A 3D printer with a consistently flat print surface and a reasonable degree of thermal design can achieve and maintain fair accuracy when manually adjusted with a piece of paper or a feeler gauge, assuming fair accuracy of about 200µm, that the prints are all of similar size and materials, and that room temperature does not change by more than ±5°C.

In the real world, however, users will expect to print thin layers on large beds in environments of wildly varying temperatures; and to do this while changing nozzles and print beds between prints.

The most practical way of doing this is to have some automation that will both warn of inconsistencies in the build surface and as far as possible, compensate for the inconsistencies.

Mike

Edited 1 time(s). Last edit at 03/23/2024 03:51PM by leadinglights.

Well, I would not be that generic as e. g. there are laserprinters or other 2d printers which operate without any calibration. Sure, you can calibrate printheads and stuff but the machine does work reasonably well without it.

In contrast, an unleveled or uncompensated warped build plate usually leads to a failed print. Yet I am on your side considering a lot of other machinery, especially when you go precise and large at the same time, where heat and expansion plays a major role (things I design for currently in my large format 3D printer, where e. g. the gantry could bind or deform if I would just mount it like in a regular, small 3D printer, simply because it extends in the range of millimeters for a given temperature difference instead of some microns).

I have a large resin printer which came precalibrated and no platform levelling was needed initially (changed the film so I did need to do that however later on).

I agree that compensating for the inconsistencies is the most practical way in any case - and in most use cases compensating for it via software/sensoring is the most versatile and cheapest method.

Yet compensating for it mechanically sometimes can be beneficial too, e. g. in precision measuring devices or large CNC machines, which usually compensate on multiple levels (including restrictively by demanding a specific ambient temperature). You could e. g. use Invar or other exotic materials, or watercooling and such, to compensate for a temperature differential and a very sturdy construction to minimize any flex. The latter probably would be enough in a small format 3D printer. Take a thick slab of milled aluminium or even granite as a heatbed and even that would warp within thight enough tolerances. But yeah. Why if you have ABL in the first place.

Agreed with everything except the first line:- Almost all inkjet printers perform a little calibration dance when they are first switched on. Along with the function checks, there are usually several calibration checks such as total carriage run. A calibration check and adjustment will be done with every sheet of paper fed and replacement of ink cartridges or print heads will usually involve a calibration check, which may or may not need user assistance.

Back to the original subject. When I was doing some background work for a paper, I came across a non-contact Z-sensing method that could give an accurate nozzle position and map the bed. This used the nozzle as part of a capacitative sensor and the non-linear relationship of distance between the capacitor elements to predict the position of the bed without actually touching it.

I seem to have lost the notes I made at the time and am searching for any info on this. If it rings a bell I would like to get more info on this. It may have been a large company, possibly Stratasys.

Mike

True, I completly forgot about that dance when they turn on ^^

Would be interesting to know more about this. But - in the end it uses the nozzle in some way and I suspect the print surface also plays a role so changing it (or the nozzle) would maybe throw off the measured values using a capacitive method without a touch. So you get the magical Z offset for that specific configuration (but then you could also use other non-contact probes and a pre-configuration from factory) - just a guess, though, never heard anything about such a method.

Let me know if you find some info!
Thx

Demo

Edited 1 time(s). Last edit at 03/24/2024 06:52PM by Demolux_D1.

From my possibly faulty memory of the article, I believe the method should work for any conductive nozzle and any conductive surface. However, I suspect that there are all kinds of problems in getting this to work with the sort of accuracy and predictability that is needed.

Very roughly, the capacitance between the nozzle and the build surface conforms to the capacitance equation where the capacitance varies with the inverse of the distance between the nozzle and the build surface. As this is a predictable curve, the values at three distances can be used to extrapolate the position where the nozzle will contact the bed.

The problems I foresee are that the external conductive path (wires) is likely to be long and may vary unpredictably. Gunk on the nozzle, and plastic film and gunk on the bed will also make it messy.

I am annoyed that I didn't keep better notes on everything I read, particularly things I didn't quote in the references.

Mike