Hi!

I just set up auto bed leveling and it could work much better.
The problem is that my servo is wobbling a couple of degrees back and forth, especially during stepper movements (z-axis).
It seems to me that some current is leaking somewhere on my RAMPS board and so the servo is inaccurate during movement.
If you wish, i could measure a bit with a scope and post pictures. Do you need a video or is it clear what I mean when I say wobbling?

The main question is how can I fix this?

Things I tried:

-swap servo .... Didn't help
-use feriite core .... Didn't help
-separate stepper wiring from servo wiring ... Didn't help


I have considered using a capacitor to prevent voltage peaks in the servo signal cable, but I have no clue if that is a good idea, nor do I know which cap should be used.
Any other ideas?

Thank you so much!

Just learned that servos are pwm driven, so maybe the capacitor is not a good idea, I thought they are voltage driven.

Any othe thoughts?

Make sure the signal line to the servo is shielded and that shielding is grounded at (only) one of the ends - the one from board. Do not try use the shielding for any other purpose. No caps on the signal line, but can try a good size electrolitic cap between +5v & Gnd placed right near the servo. Also you could try using a different 5v regulator to servo supply. Or make its own +5v supply from a local regulator.

Hi, what do you mean by shielding?
And by local regulator you mean a 5v source that has nothing to do with the ramps at all? I could use a USB power supply for that...
Another option would be using a 7805 or similar...
What would a good sized decoupling cap be? Still in the uF range?
Thanks!

Edited 2 time(s). Last edit at 05/18/2014 06:10AM by Yellobello.

Uh ok, shielding is a cable that has an outer layer of wire mesh or metal foil and that outer mesh is connected to ground, and this way it protects the inner wires from any outer external field which could cause signal interference. Look up "Faraday cage" and shielded cable. The cable shielding is only grounded to one side, preferably to the side closes to the psu. If its grounded at both sides it creates a ground loop which is not beneficial. Shielding is a must have on all signal lines, servo signal but also thermistors and endstops which run along other stepper and heater wiring which creates intereferences. This will probably solve your wobbling, and if it does then you wont need anything else further.

For a motor cap you can test in a reasonable range like 100-220-470uF cap between the +5v to Gnd which are connected to the servo, that is as close to the servo motor as possible. Check if you see any difference with or without this cap. This is for the power supplied to the servo and its sort of a capacitor for motor, which purpose is to help when the servo draws peaks of current. Again no cap on signal line.

Local 5v source means a 5v voltage regulator, (7805 or similar), which needs an input higher than 5v, so it can be supplied with 12v (not usb), and it outputs 5v. If you seach "7805 datasheet" you will find schematics in there. But hopefully shielding or at least a motor capacitor will solve your issues so you wont need to try patching in a local regulator.

Ok, thanks for your hints, this is what I tried now:

-Used a shielded USB Cable (cut the plugs off and soldered the mesh to GND) ----> No change
-.Used 100uF and 470uF caps (10 and 35v) -----> No change in behaviour

Grrrrrrr!1!111!!!!

I doubt that clean 5V could fix the issue, but it´s worth a try.. i only have a 7905 around, which makes -5V, but that´s ok, right?? If i just swap the poles??

I serously start to think that either my arduino or my ramps board produce this issue...

There´s digital and analog servos around.... judging from the low cost my servos were, maybe it´s worth trying a digital servo?

If the shielding and a motor cap did not solved the issue, then its probably not interference and probably not a power issue either, although the latter would be better confirmed with a local 5v regulator, but still. Anyway not 7905, but 7805, a lower positive value one, or some adjustable one, with a reference made by two resistors.

Try move the servo by hand just a tiny bit, maybe it has something with its feedback reference. Or try changing the signal pin and instead of that one, use some other free pin from somewhere else.

Edited 1 time(s). Last edit at 05/18/2014 03:12PM by NoobMan.

By any chance, you do not use positions like 0 or 180 degree (in case of 180 servo), that would be too close to its limits. For the best, use the two positions like 40 degrees and 140, so its spaced away from 0 and 180 limits - or something more or less like that. Then you can fix the axis positions to fit these angles instead.

Ok, first off, i did not manage to solve the problem in a traditional way. All the good hints did not change a thing. Strange, right? What I did not do yet is using another pin, maybe that will help. Also I got a digital servo for trying, I'll let you know if that helped.
However, I found a way around this. Marlin has an option to shut off servos after some miliseconds. Fiddling arround with that value I found a moment where no other axis is moved for a couple of ms between probing. At a value of 600ms there is a moment without interference, so when I cut off the servo at that value there is no wobbling and it always goes to my programmed 135 degrees. That works for now, but is just a workaround without solving the actual problem.
Maybe thats why I am an engineer ;-)

Good thing you solved it. Probably even better if you hacked it around.

Thanks. The servo deactivation delay fixed my wobble problems!

you had the same kind of trouble?

I am thinking now that this is some RAMPS related issue since it is very well possible that the servo lanes are too close to the stepper lanes on the board, causing interference...
RAMPS has so many issues, a 1.5 would be a really good thing.
I had trouble with the PTC fuses, the MOSFETs get too hot and some other minor issues. Since RAMPS is a very commonly used board, the issues with it seem to huge ....

Take a look, this guy has the same problem:

Youtube link

Even though he did not complain about it, It seems that this is a common issue. I just found it on a ABL video in YT