ABL Servo probe issues

[tsteever]

I have recently upgraded to the Newest version of the Marlin firmware. Ever since I have been chasing problems! This one eludes me still. I am getting erratic errors with my servo probe. I have to go into the firmware to change the extend and retract values to get things to work correctly. Then a week later when I seem to be getting good results, the probe will start deploying at weird angles.

The issue is funny as the servo will consistently extend and retract to new degrees as if the firmware has been changed. Nothing has be changed in the firmware or physically on the printer.

I am just looking for guidance on where I would start looking to diagnose this issue. Could the servo be the cause? It isn't binding or anything. Sounds healthy when it moves, etc...

The printer was just working yesterday fine. This morning I went to start a print and the nozzle crashed into the bed cause the probe deployed at too great an angle. I hit the reset button and the restarted the print. The servo probe deployed at the incorrect angle. When I issue an extend and retract command it does so, at the incorrect angles? How can this happen?

[Roxy]

What you are describing very likely is a servo problem. Are you seeing any twitching of the probe leg when it should be sitting still?
Are you using a blue TowerPro servo from eBay? Those are of horrible quality.

The EMax-ES08A servos seem to give very good and consistent results. And they are not very expensive.

http://www.ebay.com/itm/EMax-ES08A-I...QAAOSwGtRX01j-

[tsteever]

No twitching. Yes, I am using a htx900 from hobby king which I think is the same servo. I am going to swap it out cause I cannot see what in the firmware would be causing this.

Could the wiring be a fault? It is weird that these problems all came up after making the switch to the Bugfix version of the firmware.

[Roxy]

Can you post your Configuration.h file? In the past, servo #2 was used for the Z probe. But if there was only 1 servo in the system that didn't make sense. Now the default servo is #0. You should get very repeatable action when you do a:

* M280 - Set servo position absolute. P: servo index, S: angle or microseconds

[tsteever]

It is weird cause things will be going fine for a few prints and then the next one results in the servo over extending. I then change the config file and a day or two goes by of good performance and then out of the blue the servo will under extend and I have to go back into the config file to change it to the previous setting.

Here is what currently happened. Two days ago I had to reset the config file to a new servo angle. I printed a model and things went well. Power off. The next day, I printed a file and it worked well. Power off. This morning, I again printed a file and it printed fine. Without shutting the machine off, I printed another copy and the servo under extended. This morning it printed fine! I did not shut off the machine.

I had to reset the servo angle to get the probe to extend correctly. I am going to test this out by now running a print and then run another print without shutting the power off.

Configuration.h

[tsteever]

This is getting ridiculous! So yesterday after all that I had a good print. Now, this morning I went to print and the servo is now over extending! I opened pronterface and issues some m401 and m402 commands to confirm the angles are wonky. first extend was too far. I retracted and then issues a m280 p0 s145 to check. I then did some m401 and 402 commands and it is now working. Weird? Does a m280 reset the angle in the firmware? I did try to enable the eeprom to see if that helped/hurt.

is this a servo issue?

[Roxy]

It is very possible you have a bad servo. But we can do a few things to get more confidence of that before you go to the trouble (and wait) to get a new one. M280 does not mess with your probe angles. And I had to double check but the EEPROM has nothing to do with servo angles. Those values are declared:

#if HAS_Z_SERVO_ENDSTOP
const int z_servo_angle[2] = Z_SERVO_ANGLES;
#endif

which means they can never change. (Unless you change Configuration.h and recompile the code.)

Lets do this: Do a M280 to set the leg position halfway between the two angles you have defined. And then measure how far from the bed the lowest part of the probe is. Let it just sit there for 6 or 8 hours and lets see if it moves. It might be easier to just raise or lower the Z axis so a coffee cup or something else just barely slides under it without touching it. That way you can easily check things accurately without bumping anything.

And once you get set up like that, give it an M280 P # but with no S parameter, it should tell you the servo angle. Over time, this command should keep returning the same number (to prove nothing has been done to tell the servo to go to a new angle) If you see the servo move, for sure we want to give it an M280 P # with no S parameter and see if the servo subsystem thinks anything has changed.

My guess is we might see the servo move for no good reason. And if so... the servo needs to be replaced.