G29 Leveling Errors

[Hurtz2P]

I'm having an issue with my Printrbot Play. Currently using RevF_v5 firmware.

It appears that when I issue a G29 code I get either a working calibration or an non-working. The non-working appears to be consistent:

Working:

Send: G29
Recv: Bed x: 20.00 y: 190.00 z: 0.81
Recv: Bed x: 20.00 y: 10.00 z: 0.84
Recv: Bed x: 90.00 y: 10.00 z: 1.15
Recv: ok
Recv: echo:endstops hit: Z:1.15

Failed:

Send: G29
Recv: Bed x: 20.00 y: 190.00 z: 0.75
Recv: Bed x: 20.00 y: 10.00 z: 0.79
Recv: Bed x: 90.00 y: 10.00 z: 4.79
Recv: ok
Recv: echo:endstops hit: Z:5.79

When working a M119 will show Z triggered and failed will show open.

I've been trying to replicate steps that will force to work or not work but best I can tell it is random. Issuing a second G29 immediately will generally correct.

Is there something I am doing wrong? Is there something I should be adding to my start codes to ensure a correct calibration?

My start code includes a G28 X0 Y0 and G28 Z0 immediately before the G29.

Any help or suggestions would be appreciated.

Thanks!

[RetireeJay]

Perhaps you have a failing wire from the sensor to the board. For some reason, it seems to be a common issue affecting the "sense" wire (but not the power wires) in the cable supplied with the probe.

When a copper wire is bent back and forth at a single point, it will eventually break (called a "fatigue failure"). The insulation around the wire will be intact, so the failure is not visible. After the wire has broken, then in certain positions and in certain conditions it can still conduct electricity - if the two ends of the wire are touching each other. But if the ends of the wire get pulled apart (even by microns) then there's no conduction.

So look for a place in the wiring harness where the wire gets bent back and forth "at a point". To verify that there is a fatigue failure there, try manually pushing, pulling, bending the wire while frequently issuing the M119 command. You may need to do this with the probe "close" to metal and again with the probe "far" from metal. The exact conditions to make the failure show up depend on whether an open circuit (broken wire) simulates "triggered" or "open", but from your description it seems that maybe "open" corresponds to the broken-wire condition.

In the long run, you should look at finding ways to spread the bending out so that your wires gently bend over a large distance rather than bending "at a point". Cable Chain is one way to do this (several posts about it are in this forum; don't get hung up on the early discussion about the philosophy of how to do cable chain best. For you, it doesn't matter if you get 100 million cycles or 10 million cycles; your failure is coming after 10000 cycles!).

[Hurtz2P]

Thanks for the quick reply - this was one of my thoughts also (seems to be a bit of chatter about this via Google searching).

On a failed G29 the final point does not reach the bed similarly to a good one and will stay a number of mm above the bed. During a failed G29 if I issue a G28 Z0 it will lower to the bed position and then I can reissue a G29. This second G29 does appear to produce good results.

I'm definitely not counting out that there may be in fact a broken wire. I am just hopeful it isn't something worse...

So then part 2 of a question: Is the probe a standard LJ12A3-4-Z/BY PNP found on eBay or AliExpress? Can it be directly installed or are dividing resistors required? I ask because I can get 10 for the cost of a single Printrbot one - if this is a recurring issue then it will be good to have spares (and for other DIY projects ).

[Sagbot]

Yes if the P/N matches it is (being sold as) a plug in replacement. What quality you get from these stores can vary of course.

How about taking your probe out, and set it up with 12 Volts and see if you can bend the wires and verify that the problem is with the probe wiring.

(EDIT: Jay already described what to do. Sorry Jay.)

Also, if you wish, you can still print without a probe. You have to be SURE not to run G29 or G92. You'll have to level the bed. On the rev D board I can use no Z switch at all and as long as I don't run G29 or G92, it will remember Z position until turned off. You'll have to level your bed and it's convenient to leave the bot on all the time so it remembers its position and keeps a consistent first layer.

The rev F boards might not like having no connection on Z end stop, so if that's true you can leave the probe connected and drape it over your neck or something.

[RetireeJay]

Is the probe a standard LJ12A3-4-Z/BY PNP found on eBay or AliExpress? Can it be directly installed or are dividing resistors required?

Why are you talking about replacing the probe? Just fix the wire. If you don't have the tools or skill to splice the wire, you can probably find someone who does (see if your town has a "makers group".

[Hurtz2P]

Why are you talking about replacing the probe? Just fix the wire. If you don't have the tools or skill to splice the wire, you can probably find someone who does (see if your town has a "makers group".

I'll look and I do have the skills - my fear is the break is going to be quite close to where the wires enter the probe. I'd just like to make sure I have a spare on hand when I try to fix.

It's looking more like a wire problem. If I move the head around and check M119 I can get a Z triggered without the probe lighting up.

Thanks again all for the prompt replies. I will update this thread with my solution once I have one.

[Hurtz2P]

Well this sucks...

Set up a new probe and plugged it in. Hit the power and light would come on when close to the bed. When I tried a M119 the light went out and will not come on again regardless of where the probe is.

Plugged in the old one and light will not come on either...

Did the board fry?

Any ideas?

[RetireeJay]

The M119 command simply reads the state of the input pin for X, Y, and Z limit switches. In fact, on the Rev F board there's even a buffer transistor in between the probe's output and the Atmel's input pin - so even if your firmware were corrupted and the M119 command somehow tried to send an output instead of reading an input, there's no way in the world that executing an M119 command could send a damaging charge back up through the sense line and fry your probes. The probes' going dark has to be a coincidence, or something caused very indirectly by your command.

So we've got to look for something else. I would say to check the connections - are you sure the probes' connectors were completely seated on the board? Is there any possibility that the jumper on the board that sends power to the probe is not fully seated? I wonder about the power supply: after the probes go dark, is the green light on the board still lit? Does the board itself continue to respond to commands that you send it? Can you exercise the motors?

You might try wiring 12V directly to the probe's supply and looking at the output voltage.

[Hurtz2P]

The board still operates - I use Octoprint and it's terminal to issue commands. I manually slid the bed to end and rechecked M119 - the y end toggled. Thinking possibly that the cable was broken somewhere else I unplugged the extension wire and tried the sensor directly into the control board with no luck.

The sensor has a plug shrink wrapped about 5 inches from the sensor - I think the original problem was right at the interface where the shrink wrap stopped. I cut off the shrink wrap and made my new sensor the same length as the original using the original extension wire.

All except the z endstop appear to be working - traffic appears via the Octoprint terminal.

I was quite diligent in ensuring the wires were plugged in correctly but what would happen if they were reversed? Fry sensor? Fry board?

[RetireeJay]

The probe specs say it's protected against reversed power supply and against short circuits.

[Hurtz2P]

Well something happened - going to break out a meter and do some checking (and a power supply to check sensors)...

[Hurtz2P]

OK...

I just recently ordered a new Printrboad and probe, something did fry on the original. Both installed and initially appeared to be working - although the Z leveling height appeared high.

Before the board fried my Z was set at Z-0.75.

With the new probe I've got it installed at about 1 mm above the aluminum plate and it finally started printing when I set at Z-3.0. Recently changed filament and had to reset the Z at Z-2.4.

I've confirmed that I can consistently trigger all end stops. I'm finding that doing a G29 I get the first probe point erratic (numbers appear to toggle between about 2.0 and 3.0)

Also the Z will stay triggered until about Z 4.5.

Is this a probe error? Have I installed it too high? Too low?

Thanks for your help.