How accurate should the bed level sensor be?

General
1

I have had nothing but difficulty printing a reasonable first layer for a very long time. I just cant get it consistent. It seems good at the end of one day and in the morning it is almost time to start over. I have done this several times. I’m beginning to wonder if maybe the bed leveling sensor is just not good enough. I did three tests just now and wrote down the values for the center 9 positions that it reads and some are very off. Is this a problem? or should I just look somewhere else.

[ 7, 8, 9, 12, 13, 14, 17, 18, 19]
-.03, -.07, -.04, -.09, -.07, -.01, -.03, .01, .02
-.07, -.02, -.09, -.09, -.01, -.04, -.01, .07, .01
-.11, -.05, -.01, -.07, -.06, -.13, -.03, .01, .00

Range
.09, .06, .09, .03, .07, .13, .03, .07, .03

I picked the center most locations because most things print in that center block. As you can see 6 of the 9 spots vary by greater than .05mm

2

When I got my first machine (Creality CR10 Smart Pro) … I bounced around trying to get decent first layer. I found that there are 2 main driving forces. (1). z-offset (2) bed mesh.

(1) z-offset
For the CR10 … z-offset is set manually with paper / adjusting knobs … which sometimes feels more ‘in control’ than the auto setups of more modern machines imho.

(2) The bed mesh is established by the machine running the ‘auto level’ capability.

The one big lesson I learned was to do 1 followed by 2. And then printing. In other words … set the z-offset manually … then do the bed mesh … then let the printer do the print. Do not do any manual knob adjustment after running the bed mesh. Ask me how I know.

Since then … I have had no problems with getting a good (enough) first layer.

Hope this helps.

3

It also depends on the type of sensor being used and the type of bed being used on

Some that use magnetic field detection, might not work on a glass bed
They can also be overpowered by too strong a magnetic force
Example of this is the Beacon Probe
Although this probe is classed as one of the best available for all metal plates
It cannot handle the Mandala Roseworks Magnetic Bed
Most flexi plates have a magnetic sticker placed on the hotbed, that has a small magnetic force
The Mandala hot bed has neodium magnets built directly into the bed.
And is strong enough to throw off the Beacon results

The BLTouch and clones use touch sensor and small magnet as trigger
This can be affected in an enclosed printer, as the heat will affect the tiny magnet

Then you have other Inductive Probes which wont work on Glass or other NON-Metal beds
Capacative probes work in the same way as Inductive, but can work on non-metal beds

Then you have the good old Microswitch sensor like Euclid and Klicky
These are the most accurate of the bunch

AFTER all these options considered, you then look at your Z-offset
Most people use the paper method to tune in the Z-offset
I prefer using a set of feeler gauges. Specifically a gauge the same as my first layer height

Then you need to look at your nozzle … has it worn down since first use
Printing ABS on a Brass nozzle will wear the tip down over time, as will glow-in the dark filaments, Nylon and CF will do so even more
If so you will need to redo your Z-Offset

As for probe accuracy
As long as the variance at each point is LESS than your layer height, you are good to go

I have a Euclid Probe on my V-Core 3 and my variances (min probe to max probe) are around 0.16 across 81 probe points (9 x 9 bed mesh)

Another thing to remember
When doing your Z-Offset and Bed Mesh
ALWAYS perform test at Print Temp
This will allow for minute expansion in nozzle and bed

4

Do you have a ways or means to check the bed surface to see if it is level? Or Air gaps? Do it as a cold bed then do it as a hotbed. My first thought is the bed is warping more than likely from heat to be with in tolerances. I don’t know of a way to tweak the bed as a fact on how to do that or correct it. I.E. Apply pressure here in the up motion or there in a down motion at this point or at that point on the bed surface… You almost have to have some black smith skills to know where and how to tweak the surface so it stays with in tolerances for both cold and hot bed surfaces. You may have to simply replace the bed surface and hope the MFG tolerance creates a better leveling surface for you. OR you can try this idea - before you start a print go and do the preheating of the bed to a higher temperature and then let it cool down to the normal print temps. Then start you printing days work. Thinner bed plates can warp more when heating and cooling… Good Luck! I did the preheating on one of my past machines, so its not unheard of. Some machines will have the ability to do the Z and or speed self adjustments when printing the first layer to control the material flow rate to compensate for the actual level. That is a fair amount of programing skills to be done on the first layer leveling. Again Good Luck! Glass plates are good but not always the best choice. Glass retains heat better…

5

This is from Klipper documentation but I would think it would apply regardless of firmware:

If the results of the test show a range value that is greater than 25 microns (.025mm) then the probe does not have sufficient accuracy for typical bed leveling procedures. It may be possible to tune the probe speed and/or probe start height to improve the repeatability of the probe.

6

Thank you, everyone that has responded. I think I found the issue. I replaced the Sonic Pad with the firmware I was using previously, mriscoc professional firmware and did a series of bed levels, all at the bed temperature of 80*C. I did 6 last night and 6 this morning, to see if turning the machine off and time had any effect on the result. The variance values were {.19, .20, .21, .19, .19, .21, .22, .19, .18, .19, .21, .22} With the Sonic Pad I would occasionally get these variance values, but more often they would be .3x, .4x. .5x and even .6x.

I will turn off the machine and do the test again in a few hours to be sure. At this point I suspect the Sonic Pad is the issue. I suspected it almost from the beginning, I just didn’t want to admit I bought a short cut and mad a mistake. Since the beginning when I was attempting to change the bed temperature it would respond with 30 when I attempted to type in an 8 for example. It has many other problems, just poor engineering. I can’t understand how Creality can even sell it still. It’s been 2 years since they updated the firmware.

Anyway, thanks to all who have responded in this thread. I now have to add klipper the correct way.

7

I know I repeat myself, but a bed level sensor does not save any time.

You level the bed once and than it will be set for years. The problem is that people do not preheat their printers. As a result the bed is bend. Once the printer is started the first few prints fail instantly, until the metal gets to temperature, but by then the user turned all the wheels to compensate something, which cannot be compensated and as a result the printer is not calibrated.

If you preheat the bed for 15 Minutes and then do a proper bed leveling by hand. All it takes is preheating the bed before the first print.

I level the beds of my printers after disassembly only, because there is no need.

On the other side there are bed leveling sensors. Which compensate the cold/hot metal bending. No preheat required. But they come with a cost. More weight at the print head, more cables. Additional (flimsy) components, that fail over time. Additional cables, that may break and also have their weight. Calibrating that sensor and learning about the results. Time to ask on forums. and if something breaks you have an additional set of hardware that may caused the problem.

The ONLY advantage a sensor has in my opinion is that you can change the build surface from sheet to glass and it will automatically fix the Z height difference. However. I have a physical knob on my printers which allows to change the z offset of the entire print bed. Usually this takes a few seconds, while a brim is printing and depth is adapted again properly.

8

Well, that was premature. This is all true until the first time I try to print. I have the Z Offset set to a value that scrapes a piece of thermal receipt paper when I pull it out. It might be too tight, but for this test it is just fine. Once I try to print anything the gap between the nozzle and the bed is so great that I can see the plastic run out of the nozzle, maybe a couple of mm. After I print I can no longer get the variances I showed above, they are .3x, .4x. …

I have tried 3 slicers, SuperSlicer, Bambu Studio, Cura and get the same result. I have tried 2 different firmware versions, Klipper and mriscoc and get the same result. Unless someone has an alternative explanation, I think this is a machine issue. I’ve looked into the option of replacing the mother board, but given I’m not sure that will solve the problem, and am not sure this is useful to do. SWMBO is tired of my continuous money im spending on this thing.

Not sure what the next step is.

9

Thanks for your response. Just to be clear, I have gone to preheating the bed to 80*C so it just sits here with the bed warm way before I level it, while I’m reading my morning email.

10

Just to be clear. I do not set the Z offset in the firmware. I basically move the z Endstop with a screw. Actually by turning the screw the head goes up or down and the switch hits it earlier or later when moving down.

When I need to calibrate the bed. I print something hugh I have already in my printers octoprint library. I start the print and depending on the layer I twist the three bed wheels or the z offset. Of course the z offset requires a restart of the print as z needs to be adjusted.

Do no not make the error and change all four screws for the bed. Four bed screws are stupid. Get the bed somewhat even and then select a bed screw you never touch again and only turn the other ones.

While the printer is driving around laying the first layer I look at the result and twist the bed knobs until the newly lines are evenly.

Calibration done. Takes usually about five minutes and I have years afterwards where I don’t need to do anything. In fact I start and preheat my printer remotely all the time and mostly without even walking into the printers room. I just check if the bed is clean, if the first lines are sticking and that there is filament using the camera feed.

here with the bed warm way before I level it, while I’m reading my morning email.

Remember to preheat the printer before printing next day/week/month, too. That is the key. It brings your printer into the same state you did the calibration.

About your z distance issue. That is the most complex problem. Forget the paper people are using. Create a print with an huge brim. That gives your calibration time.

The main issue is to know if you are too close or too far away from the print bed. It usually gives the same result. The filament gets loose in both cases. In the case you are to close the nozzle will kick the previous line of the bed.

Best as a beginner is to start high and slowly go down.If the filament sticks, feel it. It should be round on top and a little squished at the print surface.

If you can roll it off the build plate with a gentle touch, you are still a little to high. It should withstand “feeling” the surface sideways.

It is hard to describe, but with a little practice you will get it right.

OctoPrintAnetA8
OctoPrintAnetA8588×441 20 KB

ABS Print in progress.