Hall effect sensors

[lohiaprateek 0]

http://reprap.org/wiki/Hall-%CE%98

Does any one have any experience with the hall-o sensor on the rep-rap page, fine tuning the z axis on the pot seems to make it fairly precise in the variations.

 

[gr5 1,292]

Other people claim the z limit switch is repeatable to better than .001mm so no need to do hall effect or optical sensor as existing design works fine.

 

[msurunner 2]

I would agree. Most of the inconsistencies I find in the first layer are a result of prying the previous part off the build platform, causing the bed to come out of level.

 

[mastory 39]

A flatter surface to print on will go alot further than revising the switch gear. This was the best benefit I have found from adding a heated bed. Now I am printing on the near perfect flatness of window glass (generally at room temp with blue tape by the way)

A 3 point leveling system is also a big help

A Z switch positioner like this one or similar is also a big help. There are several options on Thingiverse. I have one that I will soon upload to Youmagine.

Since making these changes along with installing the HBP, I have been impressed with the stock switch's accuracy. I haven't measured, but I think it is in the range of +/-.002"(.05mm) or better repeatability.

I can print a plate of separate parts and expect each one to start without having to babysitting the machine at the start of each model.

 

[lohiaprateek 0]

makes sense, not that I'm complaining, but I guess I do need that HBP upgrade

 

[jumpmobile 0]

What about using this sensor for automated bed leveling?

http://www.thingiverse.com/thing:89146'>http://www.thingiverse.com/thing:89146'>http://www.thingiverse.com/thing:89146'>http://www.thingiverse.com/thing:89146 approach (http://www.thingiverse.com/thing:89146) is using the Ultimaker endstop switch mounted to the printhead. It seems to work but the hall sensor would do an even better job since there is no need for it to touch the build platform for measuring its distance to the nozzle. Therefore it doesn't need to be retracted during the print process.

This chip allows for distance measuring up to 4 mm which i think would be enough distance to be out of the way during printing or if the nozzle even crashes into something.

For the magnetic field one could use neodymium-magnets and an iron-plate instead of an acrylic or heated aluminium bed.

 

[gr5 1,292]

But you need to measure at 3 points with an accuracy of better than .02mm. If the magnets are mounted a little high or low it won't work right.

 

[jumpmobile 0]

Oops. Why does the accuracy have to be so high?

 

[Daid 297]

The normal end-switches have a repeatability of around 2-5 micron. So you might want to get close to that.

(The wood in the UM-Original, that's another problem, that can shift 0.5mm by just tugging it)

 

[jumpmobile 0]

But isn't this resolution eliminated by the manual-leveling procedure? A M3-thread has a pitch of 0.5, so one tenth of a turn will give you 50 microns height difference.

Edit:

On magnets-positioning:

You just place a magnet behind the Hall-sensor (so, both on the print head) and the just measure the flux as a ferromagnetic heatbed is approaching.

Described here:

http://www.hall-sensor.biz/

quote:

"Mounting a Hall sensor and a magnet in one device gives new possibilities. Non-magnetic ferromagnetic objects can be detected, and distances measured. Several metals have ferromagnetic properties and can be used, including some Stainless Steel versions. To be able to use strong magnets for a large range, and to be able to see very small changes, the Hall sensors must also show a good sensitivity for small changes in a strong magnetic field."

 

[Daid 297]

But isn't this resolution eliminated by the manual-leveling procedure? A M3-thread has a pitch of 0.5, so one tenth of a turn will give you 50 microns height difference.

 

Which is why you should give a rats-ass about resolution, you should care about repeatability. Offsets from your home point can be set in software.

 

[gr5 1,292]

You just place a magnet behind the Hall-sensor (so, both on the print head) and the just measure the flux as a ferromagnetic heatbed is approaching.

 

Oh. I got confused when you said "magnets" (plural).

Ferromagnetic? So basically iron? Can iron be made flat? And will it stay flat (not warp, not chip or scar too much)?

This could work I suppose. Why not measure a grid of 80 points and have cura mess with the extrusion flow (by controlling E, not Flow) to fill in the lower spots while laying down the first layer. There is another printer out there that does this - the UP maybe?

 

[jumpmobile 0]

I wouldn't mind the bending: iron (especially steel) is harder than aluminium and therefore should withstand the forces when ripping of a printed object more easily.

Maybe this rigidity will cause trouble when the nozzle crashes into the surface of an iron/steel-sheet since it is much harder than brass also. But a crash is exactly what one wants to avoid with the sensor-system.

Iron also has a heat conductance that is two thirds lower than that of aluminium. Maybe this will inflict perfomance of a heatbed, but then again: people are using heatbeds in combination with glass wich has two hundreth heat-conduction and that also works.

Here is a video of Z-leveling with hall-probe:

 

So the idea is not that far from makeable...

 

[baasb 6]

I found another very nice and easy way to level the bed.

 

I think this is not very difficult to implement on UM.

 

[gr5 1,292]

I'm not sure what ferromagnetic means but I think Steel doesnt count. I know magnets stick to iron, but not steel.

 

[lohiaprateek 0]

I saw the semi automatic method a few weeks ago, I would have love to have something like this, but alas I cant design for nuts. Maybe some of the more experienced people on the forum will be able to do it.

 

[gr5 1,292]

I'm not sure what ferromagnetic means but I think Steel doesnt count. I know magnets stick to iron, but not steel.

 

Wait. I think that's wrong. I must be thinking of stainless steel. Never mind.

 

[jumpmobile 0]

Bought one...

Now searching for a piece of non-stainless, ferromagnetic, super even steel ;-)

 

 

[drayson 74]

I'm also very interested in semi-auto leveling. Parts needed are more or less available, 3 rods with a thread for mouting to the bed, clamps with screws are printable but...

Honestly I have no clue how to make sure the bed can move down if something is wrong with the g code. Fixed clamping points eleminate this safety buffer, moveable clamps might bring in inaccuracies to the leveling... Any clue on this???

[jumpmobile 0]

I am still waiting for my hall-o to arrive, but in the meantime i was thinking about other ways to sense the nozzle-bed-distance for a specific point.

Actually you don't need a hall sensor specifically. Ultrasonic-sound, capacitive- and induction-sensing are also ways to achieve that. I am currently working on an inductive-sensor, that works just like a metal detector. The advantage of inductive over hall is that you can -theoretically- sense the distance to any material you like and therefore do not need to replace your aluminium/acrylic bed with a ferromagnetic one.

Also i hope that inductive sensing has a better resolution than hall-effect-sensing.

@drayson: I don't understand. That bed-leveling system from Kuehling&Kuehling should work with any bed that uses screws for leveling. Is there a need to use clamps? Or are you printing on glass?

 

[gr5 1,292]

Fixed clamping points eleminate this safety buffer,

 

I was thinking the same thing but then someone over on the google groups mentioned you could modify it slightly and use 2 springs (one much stronger than the other) in such a way that you still get the safety buffer of a springed platform (so if the nozzle hits the platform hard, it will give in a bit and nothing will break).

The strong spring is the safety spring. The weaker spring is used only during the leveling process.

 

[drayson 74]

My intension is to use it with glass bed plates, so crashing into the surface shall be avoided...

[drayson 74]

Hmm... Might work but needs some space below the platform.

I started thinking about printed parts with some kind of predefined breaking area. If the nozzle hits the platform hard enough the part breaks instead of the glass plate.

[jumpmobile 0]

I recieved my Hall-O last week and began toying around with it today.

The board works well an delivers a constant (linear) change in output voltage from 2.5V to 5.1V as the magnet approaches.

I also printed a test assembly to control magnet - sensor distance.

http://www.thingiverse.com/thing:186063

It is also possible to measure the distance to an approaching ferromagnetic bed, but the change in output voltage was more like 0.3V for the 4-1mm bed distance. So to be able to measure 0.1mm you should be able to measure voltage changes of 0.01V (the signal was stable to 0.003V outside of the printer chamber).

 

[gr5 1,292]

The Arduino can measure only 10 bits from 0 to 5V so voltages that will stay between 0v and 5v can be measured directly to within 5v/1024 or about 5mv accuracy.