All_surface scanning z-probe

Sindarius

@gloomyandy I'd probably consider something a little more like the Euclid probe where you can pick up and drop off. Would be neat to see especially for folks who use glass/mirror beds.

T3P3Tony

Also a ball is not necessarily the optimal contact shape. even a relatively large radius ball is going to have minimal contact area so you are either going to have to scan many close together lines, miss areas, or have a huge ball. probably a roller is better as scanning only needs to be on one axis. Manufacturing and mounting a smooth and concentric roller is its own challenge!

R4ffers

@Sindarius Euclid is how I do it

o_lampe

Thanks for the interesting replies. They made me think...
IMHO a small contact area would be best to find the lowest valley in the surface. A roller or cylinder would average out what's underneath.
OTOH a ball might miss the highest point, but that's less of a problem when it comes to 1st layer adhesion.

Retracting the ball is a thing I haven't thought about. It would be like a BLtouch, but here the sensor board would need an actor to lift the ball.
Maybe a parking macro for the probe could retract/unretract the probe. Like with the oldfashioned allen key switch probes.

If we'd use a capacitive sensor the ball could be made of POM, which is lighter. And has less impact, if we don't retract it and it bumps into the print.

A magnetic sensor (*) would need a mechanic to translate the vertical motion into rotational. (maybe backlash problems)
*) the new MT6835 has a super high resolution of 21 bit = 0,00017 degree/step and has an SPI clock of 16MHz

o_lampe

@T3P3Tony
I just read the discussion in the official thread about the need of a separate Z-probe for calibration...
What if outside of the coil are two protruding contacts and the steel washer is used to close the contact? Voilá, z=0 datum set

nikscha

@o_lampe You could retract/deploy the ball with an electromagnet. Altough I wouldn't use a ball but a rod with a spherical end. I don't think this kind of zprobe needs a "rolling" element.

Surgikill

@o_lampe I was considering mechanical devices that could be used for non-metallic build plates, like glass.

You could use a FSR or a load cell on a ballpoint pen style roller, this would probably cause issues with heat drift.

You could use some style of linear potentiometer/hall effect. This would probably drift with temp also.

You could use an optical gauge with a slider. This would probably drift with temp too.

I think the biggest issue is eliminating the temp drift.

o_lampe

@Surgikill Temp drift can be an issue, but when we scan the bed really fast and contact material has a low thermal conductivity (ceramic ball or POM) it won't matter.
The inductive probe is especially sensitive, but some other sensors just expand a tiny bit. The range of motion we have to deal with is so small, their thermal expansion is below the radar.
Other issues like AD-conversion noise and interface delay can also mess up the result.

oliof

What about adapting a digital dial gauge? Something like https://github.com/stawel/dialIndicatorToSerial ... I assume the challenges are a) getting readings fast enough, and b) finding a digital dial gauge that's precise enough without costing an arm and a leg.

o_lampe

@oliof There are linear hall sensors, but I don't know how sensitive they are. Passing a small magnet along the sensor would act like the rotary mag-sensors, but without the need of a motion translation.