Y-Axis fake(?) Crash detections - let's debug  

Re: Y-Axis fake(?) Crash detections - let's debug

I was in the same boat, getting about 10 Y crashes per benchy, but I found a solution.

When I slid my Y-axis assembly back and forth I noticed moderate resistance in the middle of the length of travel.

When I removed the Y-axis belt all this resistance went away.

I noticed that when the Y- axis belt was installed and tensioned it would cause the rear aluminum plate (the one the Y-axis motor is mounted too) to bend inward. I was able to observe this by putting the edge of my print surface (any straight edge will work) against the aluminum plate (any straight edge will work) and seeing a gap between the print surface and the rear aluminum plate. In the photo below I don't have the belt installed, so you can't see the cupping, when you test your printer make sure the belt is installed (I simply forgot to take a photo when the belt was in and the cupping was visible)

2018-09-15 14.30.26 by jculbertson562, on Flickr

I presume that this bending caused the Y-axis straight rods to bow outwards slightly (pear shaped) lending to the resistance in the middle of the length of travel.

The kind of ghetto solution

2018-09-15 16.01.22 by jculbertson562, on Flickr

2018-09-15 16.22.36 by jculbertson562, on Flickr

There is probably a more elegant way to fix this but I just built myself a screw jack to install in the printer permanently to keep the rear plate from bowing. The jack is made up of a turn buckle from Homedepot, a 5/16" bolt, a 5/16" jamnut and a command strip. You just need to install the jack when the belt is removed by dropping it in and unthreading the 5/16" bolt until the jack is tight, then tighten up the jamnut to lock it in place. Then after you tighten and tension the belt check the rear plate again with a straight edge to see if it is bowing inward still.

After that I moved the y-axis carriage back and forth along its full length of travel and found that there was no resistance.

I have since done nearly 15h of printing and have not had a single Y-axis crash.

I hope this is helpful to someone out there.

Cheers!

John

Kit or Pre-Assembled?
[X] Kit / Pre-Assembled [ ]

Firware
3.2.1-576

What Slicers did you use / did it make a difference?
[X] Slic3r
[] S3D
[] Pre-Sliced SD

No difference imo

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X = 224
Y = 237

Only Y-Axis or also other crashes? (look it up in the fail stats -> Total)

Just Y

Re: Y-Axis fake(?) Crash detections - let's debug

The above seems like a solid solution. When I get a chance I’m gonna inspect this on my machine and report back. I got spare 2020 extrusion that I’ll probably fit instead.

Thanks for the insight, john.c56

Re: Y-Axis fake(?) Crash detections - let's debug

I've been struggling with this on and off for a couple of months now. I don't get much spare time so when it started to happen I kinda got the shits and just left it for a while.

Decided to start really looking into it this week. I've got a few mods on mine, new Y motor mount and X / Y tensioners. I use igus drylin bearings but I don't think it is the tension on the U bolts - when assembling the kit I made very sure that the carriage would move smoothly.

Here's what I tried:

None of this made any difference. I created a quick 100mm x 10mm x 3mm object to place on the bed (~ 15 mins to print) to try to pinpoint where it was having issues.

Here's what I found.

It's looking like what is working is moggieuk's suggestion of changing the running/holding current, as well as the stall sensitivity (?). I used richard's settings of 28 for run/hold current (both x & y), and the sensitivity of 7.

(Note, in order to use these settings on fw 3.5.0, you need to re-enable the M911-M918 settings as per the changelog here. That means you have to download (clone) the source code, change the setting and recompile/upload the firmware).

I've got some more testing to do, I did want to ask what the ramifications were of using these settings. Is having the sensitivty so low (I think it defaults to 3) making the crash detection useless?

~~~~~

Kit or Pre-Assembled?
[X] Kit / Pre-Assembled [ ]

Firware
3.5.0 (custom compiled to enable M911-M918 gcodes)

What Slicers did you use / did it make a difference?
[X] Slic3r
[] S3D
[] Pre-Sliced SD

No difference imo

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X = 220
Y = 250

Re: Y-Axis fake(?) Crash detections - let's debug

Hi everyone,
any news since the last post ?

my new MK3S (mars 2019 ) seems to have the same problem with y-axis false positive crash detection (with sometimes additional x-layer shift )...
Depending the model i try to print, crash occurs systematically (@ same Z-height - and even @same x,y i think..

I tried everything, except firmware recompilation with sensitivity tweaks...

- Belt status (i try a lot , every time with a self-test to validate the value)
- Nothing colliding with something other
- Bearing movements are smooth on every axes
- Pulleys are ok

I tried the following point, with no success:
-disable the filament sensor
-unplug and rewire the EINSy board
-checked the U-Bolts
-remove the model from the plate during re-homing to check if the nozzle crash onto the model, but still have the problem, even in mid-air printing.
-Tried to trigger a real crash, by pinching the X-rod or blocking the bed and it works with no layer shifting (and i have to maintain a greater force to activate the crash detection btw)

and...like you...the only thing that works is to turn off the crash detection. My print are great, no layer shift at all. ( currently printing in stealth model since 3 days with no problem ...)

SO....i think it's not an assembly problem.

Did you try to ask the support ?
Did you try to change/replace the y-motor?
Did you try to change/replace the PSU ?
Did you try to change/replace the EINSY Rambo ? (TMC2130 glitch may be?)
Did you try to buy/eat more gold-bears? 🙂
or.. did you definitely give up?

Kit or Pre-Assembled?
[X] Kit / Pre-Assembled [ ]

Firmware
3.6.0 - no recompilation

What Slicers did you use / did it make a difference?
[X] Slic3r
[] S3D
[X] Pre-Sliced SD -> a lot of 1/3 benchys btw (1 hour between each crash.... 🙄 )

No difference.....

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
from loose to tight ... really 😐

Re: Y-Axis fake(?) Crash detections - let's debug

Hello Bigo.

I'm having the same issue that you described. Y axis crash coupled with X axis shifts. For me the X axis shifts only showed up after I started printing with ABS, which requires more power draw for the hot end and bed heater. I'm wondering if perhaps this is a power related issue, with the power supply momentarily browning out with multiple axis running?

Might go back to PLA and see if things improve.

Kit or Pre-Assembled?
[X] Kit / Pre-Assembled [ ]

Firmware
3.6.0 - no recompilation

What Slicers did you use / did it make a difference?
[X] Slic3r
[] S3D
[] Pre-Sliced SD

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X - 286
Y - 284

*Edit* I tried a print in stealth mode as others have suggested and the x-axis shifts have disappeared! Now I just have to fix typical goopey ABS prints.

Re: Y-Axis fake(?) Crash detections - let's debug

Hi Michael,

i was printing with PLA, never tryied ABS with this printer.

Following my last post, i tried without success these few things :

So i think it could be related to a TM2130/Einsy failure, or even a PSU problem. If it was a firmware's glitch, we should not be alone in this situation !

Next steps, i will try to turn off the bed heating, to discard a PSU failure....

Did you contact support?

Re: Y-Axis fake(?) Crash detections - let's debug

I have not contacted support yet, but plan to shortly. You may also be interested in knowing that MK3s are now being provided with a new power supply made by Delta.

https://www.reddit.com/r/prusa3d/comments/b2a699/my_mk3s_came_with_an_all_new_black_psu/

I'm really curious to see if I swapped in one of these power supplies if there is any impact to these false Y Crashes / X Skips.

Re: Y-Axis fake(?) Crash detections - let's debug

I have contacted support, but no solutions yet (they suspected a motor issue or a power panic problem ...).
I tried another test yesterday. I noticed that my benchy failed after about 1h / @ Z around 20mm (3 crashs in less 45 sec ..).
So, i cut the Gcode to begin @ z = 19mm and..wait until Z around 25mm, removing by hand the extruded filament ... but ... no crash.

I'm a bit disappointed now .... May be PSU / Motor / TMC was not at a good running temperature ...

RE: Y-Axis fake(?) Crash detections - let's debug

This has reared up again for me since upgrading to v3.7.0 - its only upon revisiting this thread did I realise I've done a custom compile of the FW to re-enable the M910/M918 service codes and have the adjustments to crash detections still in my start gcode in slic3r.

RE: Y-Axis fake(?) Crash detections - let's debug

Wow... glad I'm not the only one. Though mine seem not as bad as some. About one out of every 6 prints I'll have a random Y-crash. Just now I was actually looking right at it when it happened and saw zero reason for it. So I knew finally that I wasn't going crazy.
I wouldn't mind it so much if it weren't for the fact that it *also* was then layer shifting on X (by about 0.25mm) after doing the re-home when it detected the crash. So, for now I've disabled the crash detection.   

Still running the original MK3S (mine is an upgrade from MK3) firmware 3.5.3 and haven't upgraded to 3.7.0 yet.

RE: Y-Axis fake(?) Crash detections - let's debug

I also started getting constant Y-Axis collisions shortly after upgrading to v3.7.0, butt I also had recently put my printer into a Lack enclosure.  As everyone here has mentioned, the printer prints great with collision detection turned off.

Returning to v3.6.0 didn't solve the problem, though.  Loosening belt tension helped a little, but didn't resolve the problem.  I've had this printer for about a year, printing quite regularly with no issue up to this point. 

RE: Y-Axis fake(?) Crash detections - let's debug

Greetings,

I'd like to chime in on this as I'm having similar-like issues with Y-Axis crashing and layer-shifts. I have the updated MK3s with the updated PSU, so its still a re-occurring problem with the crash detection enabled. Oddly enough, when my Y-axis belt was out of spec (past 280), i didn't experience a crash but layer shifts. So I'm clueless.

Only way for me to get actual good prints is to disable crash-detection.

To note:

Firmware: 3.7.1.-2226

Belt Status - X 277, Y279 at the moment

RE: Y-Axis fake(?) Crash detections - let's debug

Kit or Pre-Assembled?
[] Kit / Pre-Assembled [X]

Firware
3.7.1-2266

What Slicers did you use / did it make a difference?
[X] PrusaSlic3r 2.0

I did not try other slicers yet, I don't think so. 

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X = 265
Y = 259

Which Version of the Y-axis Assembly / Heatbed Assembly do you have?

I'm not sure. The most current as of today as my machine was delivered in May, 2019

Only Y-Axis or also other crashes? (look it up in the fail stats -> Total)
Y - 10
X- 2

I recently started having layer shifts due to Y-Axis crashes. I was working near the printer when I heard a problem. I walked over and found a machine crash on the Y-Axis though nothing was wrong.

As some of my prior print results weren't ideal since I received this machine, I recently tightened my belts to the 250-270 range recommended by some. They were 279/263 or thereabouts from Prusa as assembled. The new values of 265/259 were TIGHT. The y belt was in the khz range when plucked and moved ~3 teeth which seemed too tight. I put my MK3S in an enclosure with the tighter belts in spec and was getting better quality prints with PLA.

As soon as I started high temp prints with PC the layer shifts started. There seems to be an issue here with temperature correction for the crash detection and with the belt adjustment figures. If going from 263 to 259 was tight to the point of potentially bowing the mounts (~3 teeth), then what would 200 look like (manual says 240 +/-40)?

I've adjusted the Y-Axis belt now to 277 and I guess the X-Axis belt loosened to 267. I'll try to test with the variables independent but I'm hoping loosening the belt alone will do the trick. One important note, my Y-Axis belt now at 277 is tighter than what it said from the factory (263). I wonder if they are loosening the Y-Axis after the self-test based on some other information?

RE: Y-Axis fake(?) Crash detections - let's debug

Thousands of prints, only crashes on parts with active curling.

Kit or Pre-Assembled?
[X] Kit

Firware
3.7.1 (and many prior)

What Slicers did you use / did it make a difference?
[X] Slic3r/Plicer

[X] Pre-Sliced SD

No difference imo

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X = 294
Y = 298+

But we all know these numbers don't represent belt tension.  My real belt tension hovers around 8 lbf (measured). 

Which Version of the Y-axis Assembly / Heatbed Assembly do you have?

I have used both twistwrap and textile sleeves on E, X and Y

Only Y-Axis or also other crashes? 

Just X

RE: Y-Axis fake(?) Crash detections - let's debug

I replaced the front and rear plates with extrusion- when removed, both plates were bent nearly 3mm inward.

RE: Y-Axis fake(?) Crash detections - let's debug

Hello!
I run into that problem too today. It butchered my lego head, so i look into it. Y crash.. So i read the this topic and check everything - almost.
So i printed a buddy and it went all good, its realy nice, BUT. 😀 there was 1 Y crash. So I pulled apart the bed and check on that U "nipping" and the bed holding rods (two of them) On the left rod i saw two groove.. So in the assembly i used to much force to lock that 3 U.
The two groove is !barely  visible! but they effect the smooth rolling..

So i ease them and change that two rod (with rotate) and use a worthy of mention clock/weapon oil on bearings. (They was damn dry!)
The first benchy finished with 12Y crash.
I just printing a new benchy and its around 50% without Y crash! Wish me luck and give oil to that poor soul!
Its much quiter now btw.

Maybe prusa need better bearings and just that.

RE: Y-Axis fake(?) Crash detections - let's debug

Yep, benchy is 100% without Y crash. : )

RE: Y-Axis fake(?) Crash detections - let's debug

One manufacturer says normal car engine oil will work in a pinch. But grade 2 lithium grease is best.

RE: Y-Axis fake(?) Crash detections - let's debug

This is beyond frustrating...

So, my printer was functioning perfectly when I first finished putting it together for a couple weeks, now I am having the same issue as several others above me.  I literally have nothing for the Y axis to crash into.  I have witnessed the crash several times and the hot end is usually printing right in the middle of the X an Y axis, nowhere near the end of its travel path.  Then I get Y-crash detection but X-Axis layer shifting.  My belts are not slipping (I've marked them at extremes of travel and painstakingly counted teeth just to confirm).  So without actually hitting anything, I hear a crashing sound, but it isn't hitting anything and I cannot tell which access it is actually coming from.  I suspect the software is correct is reporting only a Y-Axis crash.  So, Y crashes and then I get the layer shifting in the X-Axis and then a few layers later the same thing happens and it corrects back, then it crashes a 2nd time and after a few layers it corrects itself again.   I printed 2 parts of an Earbud case separately and the result is attached. See image...

Kit or Pre-Assembled?
[X] Kit

Firware
3.7.2 - 2363

What Slicers did you use / did it make a difference?
[X] PrusaSlicer 2.0.0

What are the Numbers on your Belt-Tension (Menu -> Support -> Belt Status)
X = 273
Y = 258

Which Version of the Y-axis Assembly / Heatbed Assembly do you have?

Not sure what this means.  My kit was delivered Early July 2019 if that helps

Only Y-Axis or also other crashes? 

Only Y-Axis crashes however I am getting X-Axis layer shift at the exact same layer every single time. See image.

RE: Y-Axis fake(?) Crash detections - let's debug

Kit built with properly greased bearings, grease occasionally reapplied to rods when they appear dry, zero crashes on any axis.

Belt tension numbers of 295 and 296.  Actual belt tension measured at 8 lbf +/-1 lbf.