Up until my attempt at the kossel I have lived with just normal old school manual bed leveling. For the most part its still the best way to go but the idea of a probe near ish to the hotend to get things fine-tuned has been growing on me. I still don't view it as a good idea to run before every print - things just should not change that much to require constant probing ( or your printer is just shit - sorry ).
So far the only methods I have tried involved microswitches. My biggest complaint with these is the trigger pressure and movement speed. Sure it works fine for your X, Y and maybe even the Z axis though I admit I don't like the run of the mill microswitches. The variance I have noticed in the z min switch has been enough to actually make a difference - the kind that fails the print out-right or worse. I have also tried the IR sensor but those have issues as well - what happens if the light refracts funny or the angle of the ir led and/or sensor are off? I've trashed build plates thanks to these damn things!
On the z min side of things I have recently switched to a different microswitch from omron - the SS-5-FT to be exact. They have a 50 gram trigger force vs others that are 150 gram or higher. So far this switch is great - I've actually been getting reliable results to date and I've had ~200 print hours with them so far - probably 30 or 40 prints. The other option is to use either a hall effect or optical sensor both of which I have and have tested off box. On a funny note I found a really neat use for those old banned buckyball magnets though I have not yet set that test up.
My latest foray into the probe based tuning centers around the BLTouch. This is actually a really sweet sensor - they did a great job designing it. Super light, small and seems very well thought out. I've had a few oopsies and nearly broke one - bent the pin a little bit and it needed to be 'fixed' ( eg I had to take pliers and bend it back ). Their firmware on the sensor has given a visual indicator ( blink ) when things are not right and has been super helpful.
My printer has a 300x300 ish build plate of which I get approx 285x285 to print on. I put together a set of 5 test plans to execute though now that I am writing this I could add at least 2 more and get some really interesting data out of this. The plan is to use the sensor to probe various points around the bed both on the perimeter and inward towards the center. The idea here is to get a plane and make all probed points within .05mm of each other in terms of travel from a set height to the bed. Once I have a plane figured out ( running all test patterns ) I'll then go tweak the bed and iterate until its perfect. Once this is done the only issue I should ever have to deal with again is the zmin switch - the bed *should* stay level after that assuming nothing is changed or moves on its own.
Note, this is all with a Smoothie board - in my case its the X5 mini v2 from Panucat.
Set 1 probe points
M557 P0 X250 Y240 ; back right corner
M557 P1 X20 Y240 ; back left corner
M557 P2 X135 Y40 ; front ~center
results
Recv: X:250.0000 Y:240.0000 Z:13.3400
Recv: X:20.0000 Y:240.0000 Z:13.3237
Recv: X:135.0000 Y:40.0000 Z:13.4625
Recv: max delta: 0.122499
Set 2 probe points
M557 P0 X250 Y50 ; front right corner
M557 P1 X40 Y50 ; front left corner
M557 P2 X135 Y250 ; back ~center
results
Recv: X:250.0000 Y:50.0000 Z:13.5175
Recv: X:40.0000 Y:50.0000 Z:13.4175
Recv: X:135.0000 Y:250.0000 Z:13.3438
Recv: max delta: 0.173750
Set 3 probe points
M557 P0 X40 Y250 ; front right corner
M557 P1 X40 Y50 ; back right corner
M557 P2 X250 Y135 ; right ~center
results
Recv: X:40.0000 Y:250.0000 Z:13.3450
Recv: X:40.0000 Y:50.0000 Z:13.4362
Recv: X:250.0000 Y:135.0000 Z:13.3500
Recv: max delta: 0.091249
Set 4 probe points
M557 P0 X250 Y50 ; front right corner
M557 P1 X250 Y240 ; back right corner
M557 P2 X40 Y135 ; left ~center
results
Recv: X:250.0000 Y:50.0000 Z:13.4913
Recv: X:250.0000 Y:240.0000 Z:13.3438
Recv: X:40.0000 Y:135.0000 Z:13.4037
Recv: max delta: 0.147500
Set 5 probe points
M557 P0 X40 Y135 ; left ~center
M557 P1 X250 Y135 ; right ~center
M557 P2 X150 Y150 ; ~center
results
Recv: X:40.0000 Y:135.0000 Z:13.3663
Recv: X:250.0000 Y:135.0000 Z:13.3525
Recv: X:150.0000 Y:150.0000 Z:13.3625
Recv: max delta: 0.013750
Now, the probe points themselves are a little strange - the nozzle will bump into the build plate clips so I have to move things out a bit. I could fine tune the probe points and I probably will. Also note the probe is offset from the nozzle tip by -32.11 on the X and 38.96 on the Y.
The Z value aver each Recv line is the distance traveled to trigger the probe from an original height of 15mm.
My basic script for each of the probe points is as follows:
M280 S7.0 ; raise the probe
G28 ; home all
M114 ; print current positions
G1 Z15 F3000 ; move Z to 15mm
M280 S8.4 ; run BLTouch self test
M280 S5.5 ; enter zmin mode
M280 S3.0 ; drop probe pin
G29 ; run the plan
M280 S7.0 ; raise the probe
Before running that script I issue the M557 probe point lines. You can check these with M503 - the probe point defs are towards the end of the output.
So far I've been able to prove what I kinda already knew - the right front of the bed is a little low. I've found that I can print pretty well on the far left of the bed and back 2/3 of the bed though I do get some differences in layer squish. As I said above the plan is to use the sensor to very finely tune the plane of the bed. Its very likely I will unhook the wiring or even remove the probe when I'm not actually using it.
I have found the smoothieware.org zprobe doc very helpful. Minor note, on the X5 mini you need to switch the jumper to 5v instead of 3.3v power on the endstops. The normal smoothie is only 5v last I checked - the bltouch lights up fine but you need 5v to get the probe to go up/down.
