Z-axis suddenly going nuts

I linked a video of it happening. First time it happened was after returning to the top after a print (which otherwise was flawless) and went nuts to the point that the left brass nut came out of the x-idler bracket.
After readjusting everything, I did 2 prints fine, and then this happened again, twice now.
Strangely, it doesn't always happen, and seems to never happen when using Z-Move in Octoprint to manually move it up/down.But it also always happens now even if I load an older gcode that printed fine before, so it's not something in the slicer.
Edit: It always starts right after homing and going down about ~3cm now, with any gcode I try.
If I manually home and then use manual controls, I can run it up and down even at 150% speed without any issue.
Also, motors and stepper driver don't get hot.

It's like the damn thing heard my considering upgrading or replacing it...
Please see the video here (careful, may be loud):
[youtu.be]

The buzz is coming from the motors.
Any ideas?
Here's also the printer terminal part of the print starting and being cancelled:

Recv: T:98.59 E:0 B:61.9
Recv: ok
Send: N4 M82*29
Recv: ok
Send: N5 M82*28
Recv: ok
Send: N6 G28*21
Recv: ok
Send: N7 G1 X0 Y98 F9000*112
Recv: ok
Send: N8 G1Z0.05*97
Recv: ok
Send: N9 M92 X80.50*74
Recv: ok
Send: N10 M92 Y80.60*112
Recv: ok
Send: N11 G92 E0*119
Recv: ok
Send: N12 M104 S190 T0*26
Recv: ok
Send: N13 M140 S[first_layer_bed_temperature]*40
Recv: ok
Send: N14 M109 S190 T0*17
Recv: T:96.1 E:0 W
Recv: T:96.3 E:0 W
Recv: T:97.1 E:0 W
Changing monitoring state from "Printing" to "Cancelling"

Edited 7 time(s). Last edit at 07/19/2020 05:01PM by Arakon.

Perhaps it needs a little oil or light grease on the lead screws and the bearings if it is sticking and stalling the motors. That would make them buzz.

Have you updated the firmware at all? You could have slicer settings that are too fast for Z but the firmware previously capped it.

A loose connection on the Z motor wiring could cause them to stall and buzz.

---

[www.hydraraptor.blogspot.com]

I made no change to software or firmware whatsoever, and previous g-codes that were printing fine are showing the exact same issue.
The lead screws are well greased and run completely smooth when manually homing or moving the z-axis up and down.

I already ordered new bearings and brass nuts, though, just to be completely sure.
I actually rechecked the wiring last night, redid the wire ends/screw terminals and all, checked continuity, etc.. limit switches are also still good.

Strangely, if the printer is unplugged from power for an hour or so, it will usually perform one print just fine.. then when returning to top afterwards, it goes crazy again.

What drives me crazy is the fact that manually running the z-axis up and down using Octoprint's controls, it never causes this issue. I can run the axis up and down 5x in a row with absolutely no problem, not a single hitch.

Could this be a dying stepper driver or something? I think it's unlikely to be the motors, since both exhibit the same behavior.

Edited 1 time(s). Last edit at 07/21/2020 07:38AM by Arakon.

Additional things tested:

- Lifting the entire Y-axis off the brass nuts while this happened to check if the motors are only struggling under load: No change.
- Printing from SD-Card to make sure it's not Octoprint/the raspberry causing the issue: Print started, then went nuts again upon returning to top.

It seems that if I leave the printer powered down for several hours, I can print once.. but once the print is complete, when returning to top, it happens again. Never during the print itself.

I wonder if the power supply is failing. Have you tried measuring the 12V rail?

I have never known a stepper motor to fail. The only moving parts subject to wear are the ball bearings and the Z axis doesn't do much mileage compared to X and Y.

I had a weird fault on my first DiBond machine a few months ago where it would print some layers with far too much plastic. It took me ages to work out what was happening. There was a dry joint on the stepper motor chip pin that selected x16 microstepping, so occasionally it would switch to x8 and extrude twice as fast. I managed to resolder the pin. In general the build quality of the PCB has been excellent, I only know of one other dry joint in all the kits we sold.

You could try putting a little pressure on the chip to flex the PCB slightly while the axis is moving to see if you can provoke it to go wrong. Just grasping at straws really.

---

[www.hydraraptor.blogspot.com]

12V rail is fluctuating slightly when heating, but that's normal, I believe. It makes no difference whether the heatbed and nozzle are heating or not, the behavior remains the same, so I'm thinking that's not it either.
I also tried pressing and pushing the stepper driver and components surrounding it while moving up/down manually, made no difference, still moving smoothly when running manually and going nuts when printing.

What is actually happening on the video? At first there is a loud whine and no movement then repeated stop-start movement.

The closest I can get with my machine and Octoprint is repeated 0.1mm movements but they are not as close together as yours.

When you print the Z speed in your example is 9000mm/m but it is capped by the firmware, in my case 4mm/s. When you move with OctoPrint the Z speed is specified, as 200mm/m in my case, which is 3.3mm/s. My acceleration is 150 and the jerk 0.4.

Hot motors won't go as fast as cold motors because the winding resistance increases, so it takes a little longer for the coil current to build up on each step. My guess is the firmware rate is bordering on too high for some reason and hot motors push it over the edge.

Also you gcode has an error in it. The temperature marker has not been replaced.

Send: N13 M140 S[first_layer_bed_temperature]*40

---

[www.hydraraptor.blogspot.com]

This is happening immediately after the printer starts moving down to the bed to start the nozzle heating process. What happens is getting worse now, too.. it will repeatedly start and stop moving, somethings go "brrrrrrrt", and randomly one motor will suddenly spin at high speed, but not the other, then the other spins and the first doesn't.
Motors are actually cool to the touch, same with the stepper driver at that time (since it only just started moving the motors at all).
Neither the firmware or the printer profile/material profile have changed, and this exact thing happens with months old gcodes that printed flawlessly in before.

When the motors are wired in series, if one stalls the other tends to keep running. If they are wired in parallel then if one stops they both tend to stop, but you get half the torque that way.

What happens if you change line 8 of your start gcode to G1 Z0.05 F200?

I can't see how that file ever printed correctly because the bed temperature will not be set.

---

[www.hydraraptor.blogspot.com]

That's actually a remnant of an outdated start gcode, the bed is pre-heated and the z-move to the bed starts only after that is done. I have removed that gcode part since with no change.

Here's the beginning of a current gcode:

;Generated with Cura_SteamEngine 4.1.0
M140 S60
M105
M190 S60
M82 ;absolute extrusion mode
G28  ; home
G1 X0 Y98 F9000 ; Go to the middle of the front
G1Z0.5 ; close to the bed
M92 X80.50
M92 Y80.60
G92 E0
M104 S190 T0
M109 S190 T0
G1 X50 E5 F200 ; make a thick line to prime extruder
G1 E-1 F1200
G1Z0.3 ;lift Z
G92 E0
G92 E0
G1 F3600 E-5
;LAYER_COUNT:120
;LAYER:0
M107
M203 Z4
G0 F9000 X7.458 Y-10.216 Z0.35 ;adjusted by z offset
G92 Z0.3 ;consider this the original z before offset

I still haven't excluded a possible physical issue, my linear bearings are still in the mail. Once they arrive, I will take apart the entire X and Z axis and see if I can find i.e. a wrecked bearing. I'll also try your gcode suggestion soon, currently the printer is half disassembled after it popped out both brass nuts during the most recent reproduction of the issue.

I noticed you changed the X & Y steps per mm in gcode. Is it possible you loaded some g-code that changed the Z motion parameters? I noticed in your video that the Z acceleration sounds lower than normal.

The firmware I supplied did not have EEPROM enabled but it would remember new settings until you powered it down or pressed the reset button. Or have you loaded a version with EEPROM enabled in the past?

What does M503 show?

Edited 1 time(s). Last edit at 07/23/2020 02:01AM by nophead.

---

[www.hydraraptor.blogspot.com]

Nope, even the old gcodes that printed successfully cause it now.
No eeprom either.. it's using the original firmware supplied way back then.

M503 just responds with "OK".

That said, I started a complete rebuild. I bought a cheap Ender 3 to print parts for the Mendel (which works surprisingly well, PETG prints were instantly successful), got a BTT SKR 1.4 Turbo 32bit board, and took it all apart.
What I found is that the linear bearings were all fairly rough and the brass nuts had some "peeling" of the thread, so I think that may be the main reason why Z went crazy.
Still doesn't explain why homing and manual control was fine, but it may play a part.

Now to wait for new brass nuts, since I ordered M8 instead of M6..

Just ran a motor test with the new board. It's NOT the driver, board, or anything. When the motors are free to spin (i.e. X-axis not installed at all), the left motor is vibrating loudly and spinning very slow, while the right motor is spinning fast and smooth.

Edit: Actually, the motors switch at times.. just now, right was vibrating and moving slow, while left was spinning fine.

Edited 1 time(s). Last edit at 07/26/2020 08:11AM by Arakon.

Just seems like you are trying to run them a little too fast. That is exactly what they would do. And other than the supply voltage being too low I can't see what else would cause it.

---

[www.hydraraptor.blogspot.com]

But why would this occur now, 6 years after buying the printer, with the exact same speeds I always used? Speed is limited to 4mm/s on Z too via firmware, I triple checked.
And if I connect a different, single motor to the Z-axis stepper, that one spins perfectly smooth, too.
The current on the new board is set to 800mA on Z, should that be higher? On the old Melzi I had increased the Z pot a bit with no change.

Edit: Using F200 DOES make them spin fine.. I don't get it. Why did it change after years?
Edit 2: Increasing current from 800 to 900mA shows no change at all.

Edited 3 time(s). Last edit at 07/26/2020 09:37AM by Arakon.

The Melzis supplied were set to 1.25A but current affects torque, so with no load on the screws there should be plenty of torque to spare.

When one motor stalls it stops generating back EMF, so there is more voltage for the other. If that is spinning it implies there is not enough voltage for both but there is enough for one and voltage affects the maximum speed.

Try explicitly moving them at 4mm/s with G1 Z100 F240.

---

[www.hydraraptor.blogspot.com]

At F240, they buzz and hum like crazy.

Edit: Increased current to 1200mA, no change.

Edited 1 time(s). Last edit at 07/26/2020 09:42AM by Arakon.

In my experience motors and drivers don't degrade over time but power supplies do wear out eventually. The capacitors dry out and you get more ripple on the supply lines. That could be hard to detect without a scope but the average voltage might go down.

It should be about 12.2V when the bed is off and about 11.8V when the bed is on.

---

[www.hydraraptor.blogspot.com]

Seeing 12.12 when bed is off, 11.6 when bed is currently heating.

Seems close enough to me and it with the bed off it should definitely be enough.

If the motor current is set too high then they could warm up and increase in resistance but you said they run cool.

The other things that could affect it are the jerk and acceleration settings but if you haven't reloaded the firmware then I can't really explain it.

It is odd that it can home OK, because the homing speed should be the same as the max speed. It does home Z first, so I think the first time you home the X and Y motors might be off, so there might be a little more voltage available. Having said that I have never known it to be critical. My Z axes have always worked reliably. Power supplies tend to shut down when they are old because the ripple causes over voltage.

---

[www.hydraraptor.blogspot.com]

It still happens if I ONLY control the Z motor, with Y and X and everything else off.
I set the Max Z Speed to 3.5 in firmware now and it works with that speed. 4.0, it doesn't.
Also, it still does it on the new board, which is running Marlin 2.0.5.4 bugfix release, instead of the stock original firmware.

I am at a loss to explain it. I think logically you have eliminated everything apart from ripple on the PSU and the motors. I have never known stepper motors to slow down with age. The max speed is mainly a function of the winding inductance, which is dictated by the number of turns, so hard to see how it could change. If the magnets lost some strength I think that would affect torque rather than speed. But I can't see why they would do that and nobody else has reported a problem.

I have the oldest kit, which was the prototype and also printed many of the kit parts for production and I still use it today, so it has a lot of mileage on it. I also have much older 3D printers and haven't had a problem with their motors.

I did find this with Google "Typically one can expect 10,000 operating hours for stepper motors (approximately 4.8 years, running one eight‐hour shift per work day" but I don't believe it. I have run machines for 24/7 for more years than that and the Z motors get less use than the others.

Do you have the stock acceleration and jerk settings?

---

[www.hydraraptor.blogspot.com]

Yep, everything exactly as in your firmware github.
I also redid the wiring on the motors, just to be sure it's not a bad connection.. no change.
Wouldn't make sense either for it to work only at lower speed that way.

I give up.
Got everything set up again today.
Not only does it still do it (and now that I have silent drivers, I can hear them vibrate at times even just moving down without visible issues yet), it also started doing it at 3.5 and 3.0 mm/s speed. I'll now try replacing the motors, that's the last thing I can replace.. literally everything but the frame, the rods, the motors and the extruder/hotend/hotbed has been replaced. Rods are smooth and the threaded rods run fresh nuts smoothly up and down.

To top it all off, the first test prints ended up with massive layer shifting.. first 2-3 layers are great, after that it's printing the support all across the model already and some layers later it's spaghetti all over and it's printing far to the front left of the starting area.
It also feels like it's printing faster than it should, and overextruding, which is weird, because I measured, and when I tell to to extrude 10mm of filament, it does exactly 10mm.. Belts are tight (but not overly), profile is identical to the one using the old board and setup. I've spent every single evening for two weeks now rebuilding and trying to find the issue to absolutely no avail, nothing seems to fix it.
I attached my marlin 2.0.5.4 bugfix config, maybe someone can see an error in it. I had to reverse all motors in the firmware since for some reason, the wiring didn't match up.

Edited 1 time(s). Last edit at 07/29/2020 05:45PM by Arakon.

If the axis isn't level and is only supported by one nut they other side can make a vibrating noise due to the nut rattling on the threaded rod.

I don't understand how motors could be fading fast like that. They would have to be losing magnetism very quickly or the windings getting so hot they have shorted turns, or the ball bearings have dried out and are failing. They are standard bearings, so could be replaced, but I have never known them fail.

I would always calibrate the extruder with 100mm to get it accurate. If it is over extruding then the nozzle might catch and cause layer shifts. If that isn't the case then I would suspect the PSU if all the motors are getting weaker. That is assuming the current is set correctly and the silent drivers are happy with 12V. Most stepstick format drivers can't deliver 1.25A but the Melzi we supplied can because it has good copper heatsinking under the PCB. Melzis from other sources and later versions don't have such good heatsinking because PCB copper fill is interrupted. That is why you see some with heatsinks on.

Check the speeds in the gcode if you think it is going faster than intended. I can't see how it would be wrong unless steps per mm was wrong but then it would be bigger as well as faster. Clock speed could be wrong but I would expect that to affect baud rates as well.

---

[www.hydraraptor.blogspot.com]

Axis is level and supported properly on both sides with the nut firmly stuck in the trap.. I reprinted all the parts just to be sure everything is good. But it also happened without the X-axis installed at all. The motors are barely warm in any situation. And I'd expect all motors to develop issues if it was the PSU, but it's always exclusively the Z-axis.
The thing is, it always happens now, before motors or drivers even have a chance to heat up.

I ordered 2 new motors and will replace the Z ones.. I just can't think of anything else, and I basically replaced everything else.

My z-axis was always "singing" when moving all the way up and down.. I had assumed that was down to resonance from the nut on the rods, but I read something recently that a too high current setting could also cause steppers to whine.. they didn't get hot, but maybe getting too much current for whatever reason caused them to fail?

Edited 2 time(s). Last edit at 07/30/2020 09:20AM by Arakon.

The Z motors are more likely to be affected by low voltage because they are in series, so only get half the voltage but the full current, so they have the same torque as X and Y but a lower top speed.

The Z motors make a high pitch noise simply because they run at a high step rate and are well coupled to the axis and the base for sound. The other axes have belts to isolate the step vibration from the axis.

1.25A is definitely not too much for the motor because that is 1.25A peak and the motors are rated for 1.68A RMS, which would be 2.4A peak. At their full rating the temperature rise would be 80C and the wire is rated to take that. At half the current the temperature rise is 1/4, so they should be about 20C above ambient when it on for a long time. I.e. warm to the touch but not hot.

---

[www.hydraraptor.blogspot.com]

Well.. I've been experimenting with the motors seperately on my desk now, using the old Melzi board to drive them. They both move normally, but one of them is constantly gently vibrating when you put your fingers on it (even when swapping drivers, it's always the same one vibrating).. it's also slightly more noisy than the other when moving.

I'm seeing 1.8 ohm between coil pairs on that one, 2.5 Mohm between the other pins.
1.7 and 1.9 ohm on the other motor, and 2.5 Mohm on one pair, 3.5 Mohm on the other..

So I'll just hope it's indeed a bad motor and the replacements will fix the issue.

Edited 1 time(s). Last edit at 07/31/2020 03:14PM by Arakon.

It vibrates when stationary? What sort of frequency is the vibration?

I can't think what would cause that because in series they both see the same coil current, so should have the same forces applied to the rotor. To be vibrating you would think the current must be modulated, but that would affect both equally and would come from the driver or the PSU. I can't think of a fault in the motor that would cause it to self oscillate when a constant current is applied.

---

[www.hydraraptor.blogspot.com]

Yes, while stationary, just sitting on the desk. It starts as soon as the motor is turned on the first time (i.e. gets a command first time) and remains that way till powered off. The vibration is fairly low frequency ("rumbling", not "buzzing") and also very faint, you can only feel it by holding the motor between your fingertips.. but the other motor does NOT do it and the vibration is specific to that one motor, even when switching drivers. The motors were not wired in series for this test, but seperate. The PSU was also a completely different one, not the one from the printer.
If I can drag myself up later without dying from heat stroke, I'll wire them up in series and try again to see if that one motor is still vibrating by itself.

Edit: Just tested.. the motors, when wired in series, are both noticably "knocking" most of the time, but not every time. Once per rotation or so, there's a "bump" in both motors at the same time, although the shaft looks to be rotating smoothly. But . every now and then, they will both run smooth without the knocking (twice out of 6 tries so far). They also get mildly warm when not running.

Edited 1 time(s). Last edit at 08/01/2020 01:12PM by Arakon.