Question on the auto bed leveling of the UMS5

[paxpring 4]

Really short question about the UMS5: The nozzle is pressing into the glass bed during each measurement of the auto bed leveling procedure to the point where you see the bed moving downwards, is it supposed to do this?

[SandervG 1,521]

I don't think so, did you adjust the bed screws? Perhaps they are too high. But it is no mechanical error, just a misconfiguration and should be relatively easy to correct. 

[Smithy 1,146]

The behavior is more or less normal. The leveling sensor is a capacity sensor and measures the distance between the head and the build plate. The build plate moves slowly upwards and the measured value of the sensor changes during the move. Then the nozzle touches the build plate, and then the build plate moves a little bit further, but the measured value doesn't change anymore. That's the point where the printer knows the exact distance, the height of the bed when the measured value gets stable.

 

You can also see that the bed slightly bends in one direction, just a little bit, this is also normal.

 

But the bed screws are also important, they should not be too tight and also not too loose. The springs under der bed should be able to work, so choose something in the middle. On the other hand when leveling is working, I would not touch it. 

[neute 16]

Hi, if I may board this thread shamelessly -

 

if I understand the autoleveling behavior correctly, it is flawed because of some assumptions that result in an inconsistent result, that is also consistently too close (this can be measured by aborting a print after the first layer or measuring the brim of the part and comparing it to "initial layer height").

 

Please correct me if I'm wrong here! (and please tell my why!)

 

My understanding of the autoleveling is mostly like @Smithy describes; the bed creeps up to the nozzle, and as soon as the nozzle hits the bed, the distance between nozzle and bed doesn't change any more and this is measured by the sensor. In an ideal world, this would mean that the nozzle, which is held down by a spring in the printcore to ensure perfect position, gets pushed upward against the spring force. However, since nothing is ideal, not only the nozzle gets pushed in, but the bed flexes (due to the cantilever design), the x-y rods flex upward (visibly!), and maybe even the bed springs compress. If my assessment of the system is correct, this introduces a significant error. I've noticed this with one S5 and one S3 that I have access to.

 

The solution to gain a first layer that is consistent in height with what I set in Cura, is to use the Z-offset plugin at 0,1 mm extra height. Problem solved (albeit in a quick & dirty way, I shouldn't have to do this - this shouldn't be an issue in the first place).

 

However, and this is why I mentioned "inconsistent", I also often get height differences between the two nozzles that sometimes exceed one layer height (note that the printer is happy about this, no "height difference exceeds realistic values" error). Resulting, of course, in filament being squished into the former layer of support material etc.

Keeping the nozzles meticulously clean only helps so far. Also it seems as if the problem gets worse with 3rd party cores, a.k.a 3DSolex, which supports my theory that the spring in the printcore plays a role.

 

On the UM3, all of these issues were resolved by disabling autoleveling - unfortunately, UM decided that we shouldn't have this option for the newer machines.

 

If there is something wrong with my machines or if I had a misassumption, please let me know!

 

Best regards, Niklas

[paxpring 4]

2 hours ago, SandervG said:

I don't think so, did you adjust the bed screws? Perhaps they are too high. But it is no mechanical error, just a misconfiguration and should be relatively easy to correct. 

 

I have only touched the screws during the manual bed leveling procedure, meaning that I moved each screw only a maximum of 1 mm. 

 

 

1 hour ago, Smithy said:

The behavior is more or less normal. The leveling sensor is a capacity sensor and measures the distance between the head and the build plate. The build plate moves slowly upwards and the measured value of the sensor changes during the move. Then the nozzle touches the build plate, and then the build plate moves a little bit further, but the measured value doesn't change anymore. That's the point where the printer knows the exact distance, the height of the bed when the measured value gets stable.

 

You can also see that the bed slightly bends in one direction, just a little bit, this is also normal.

 

But the bed screws are also important, they should not be too tight and also not too loose. The springs under der bed should be able to work, so choose something in the middle. On the other hand when leveling is working, I would not touch it. 

 

So, you say that it is okay for the nozzle to touch the bed and pressing it down? SandervG said otherwise.. 

 

 

8 minutes ago, neute said:

Hi, if I may board this thread shamelessly -

 

if I understand the autoleveling behavior correctly, it is flawed because of some assumptions that result in an inconsistent result, that is also consistently too close (this can be measured by aborting a print after the first layer or measuring the brim of the part and comparing it to "initial layer height").

 

Please correct me if I'm wrong here! (and please tell my why!)

 

My understanding of the autoleveling is mostly like @Smithy describes; the bed creeps up to the nozzle, and as soon as the nozzle hits the bed, the distance between nozzle and bed doesn't change any more and this is measured by the sensor. In an ideal world, this would mean that the nozzle, which is held down by a spring in the printcore to ensure perfect position, gets pushed upward against the spring force. However, since nothing is ideal, not only the nozzle gets pushed in, but the bed flexes (due to the cantilever design), the x-y rods flex upward (visibly!), and maybe even the bed springs compress. If my assessment of the system is correct, this introduces a significant error. I've noticed this with one S5 and one S3 that I have access to.

 

The solution to gain a first layer that is consistent in height with what I set in Cura, is to use the Z-offset plugin at 0,1 mm extra height. Problem solved (albeit in a quick & dirty way, I shouldn't have to do this - this shouldn't be an issue in the first place).

 

However, and this is why I mentioned "inconsistent", I also often get height differences between the two nozzles that sometimes exceed one layer height (note that the printer is happy about this, no "height difference exceeds realistic values" error). Resulting, of course, in filament being squished into the former layer of support material etc.

Keeping the nozzles meticulously clean only helps so far. Also it seems as if the problem gets worse with 3rd party cores, a.k.a 3DSolex, which supports my theory that the spring in the printcore plays a role.

 

On the UM3, all of these issues were resolved by disabling autoleveling - unfortunately, UM decided that we shouldn't have this option for the newer machines.

 

If there is something wrong with my machines or if I had a misassumption, please let me know!

 

Best regards, Niklas

 

By all means!

I must admit, the first layer (and first few layers) on the UMS5 is (are) a pain indeed. I have been playing with the initial layer line width in cura, which helps only slightly. This is not an issue with my Prusa MK3S, which has auto bed leveling enabled too...

[SandervG 1,521]

29 minutes ago, paxpring said:

So, you say that it is okay for the nozzle to touch the bed and pressing it down? SandervG said otherwise.. 

 

 

I suppose the key lies in nuance. They should touch, and the bed should slightly push the nozzle up to perform the measurement. It is possible that it pushed the bed down all so slightly at the end of its measurement. But it is not supposed to push it down, which is what I understood from your initial message. 

 

I believe the space between the Z stage and the heated bed is suppose to be 14mm ideally. Perhaps you could check this too @neute?

 

 

We do not allow active leveling to be disabled on the Ultimaker S5 because the bed is significantly larger than an Ultimaker S3 or Ultimaker 3. This means that a minor offset can go by unnoticed on an Ultimaker (S)3, but can compromise bed adhesion on an Ultimaker S5. And we know that a successful first layer is key to a good print. 

 

It shouldn't be a pain and you shouldn't resort to Cura to work around it. Make sure the bed springs are at the correct tension and your print cores are clean. You can find more tips via this link, via our partner in the us FBRC8. 

[neute 16]

Thanks for your input.

I've never seen this tool to set the gap before, shouldn't it be included with new machines if this distance is that important?

 

As I've mentioned, I've never been able to get a successful first layer without tweaking the G-Code via the Z-offset plugin. That's my whole point - to me, a successful first layer means that it looks good and doesn't have elephant's foot and over extrusion. Adhesion shouldn't be achieved by forcefully squishing the filament to the print bed, but by using suitable measures such as brim, raft, adhesion medium and bed temperature (all of which usually work! If they don't, forcing the material onto the glass plate won't either IMO). And for what it's worth, a "slight touch" shouldn't make the x-y gantry flex.

 

In short; the UM auto bed leveling is flawed. Change my mind. 

 

 

PS: I don't have bad intentions or a grudge against UM, but I've never read an answer to this problem that addressed the concerns directly, only hints that I've followed in the first place (clean nozzle etc). 

[SandervG 1,521]

37 minutes ago, neute said:

In short; the UM auto bed leveling is flawed. Change my mind. 

 

 

It is not flawed, despite your personal experience. Countless of users are happy with it and it works as it should. Otherwise it would be the only thing people talk about online, outside of the couple of reports and issues some people having with it. Because it does have some dependencies, like a clean print core and the correct tension on your build plate springs (14mm). 

 

For good adhesion the first layer needs to be squished onto the build plate. An unfortunate side affect is indeed elephant feet. This can be battled by adding a negative value in initial layer horizontal expansion. Hope this helps! 

[neute 16]

Hi Sander,

 

what distance should I set between the cast aluminium bed and the plate on the S5? Or how can I measure 14 mm here? I can't insert a tool or calipers since there is an edge all around. See photo.

 

 

17 minutes ago, SandervG said:

For good adhesion the first layer needs to be squished onto the build plate.

 

Then I've had tremendous luck using the z-Offset plugin, even with "difficult" materials that used to lift off the build plate (here the proper adhesion medium did the trick, while squishing the material onto the plate did not).

 

19 minutes ago, SandervG said:

An unfortunate side affect is indeed elephant feet. This can be battled by adding a negative value in initial layer horizontal expansion. Hope this helps! 

 

What setting do I need to change to battle the overextrusion of the initial layer, which leads to material build-up and bulging next to the nozzle path?

 

Thanks for your help,

Niklas

[SandervG 1,521]

As with a lot of people, I have also been working from home for last few months and unfortunately I don't have an Ultimaker S5 at my place so I can't check. Is there not some free space near the rear? My apologies if I mixed up the Z stage design with the one of the UM3 in my mind. 

 

Otherwise I'll ask one of my colleagues who visits the office to give an approximate measurement of top and bottom of the construction. 

 

37 minutes ago, neute said:

Then I've had tremendous luck using the z-Offset plugin, even with "difficult" materials that used to lift off the build plate (here the proper adhesion medium did the trick, while squishing the material onto the plate did not).

 

It sounds like you did. And using the correct adhesive is also crucial. 

 

39 minutes ago, neute said:

What setting do I need to change to battle the overextrusion of the initial layer, which leads to material build-up and bulging next to the nozzle path?

 

There shouldn't be any over extrusion on the first layer like that. I believe that usually means the nozzle is too close or the first layer is too thick. 

[Smithy 1,146]

I measured the distance on my S5. It is a distance that works fine for me since 2 years, so don't take this as an official distance.

 

[UlrichC-DE 115]

Just happen to be the same distance apart. But no supportive digital caliper. (I just can't get used to working with kid gloves on😆 )

But also no problems with it. I once did a manual calibration to avoid taxing the algorithms too much. But it was.

[paxpring 4]

@SandervG

 

Okay, I feel like this topic has drifted slightly so I'll get back to my question. I understand better now, I guess. To sum up: it is procedure for the nozzle to touch the bed for the purpose of measuring bed-nozzle-distance. However, downwards moving beds and upwards flexing x-and-y-rods as a result of the nozzle pressing downwards are not okay. 

 

I see that two people already posted the total height of their bed module (being 28.67mm). Mine is currently at 31.34mm. When I check if I am close to the 14mm you mentioned, I see a red layer on the bottom of the bed. Measured without that red layer, the distance between the bed and the base plate is 6.4mm and at no point that distance even approximates your 14mm. I must admit I am a bit lost ... Looking at the image you included in your post, I can only guess that you might have a different base plate?

 

Is this discrepancy something I can fix by turning the knobs the appropriate amount and redo the manual bed leveling procedure, which is available through the settings menu of the UMS5?

 

 

On 8/4/2020 at 1:49 PM, SandervG said:

 

I suppose the key lies in nuance. They should touch, and the bed should slightly push the nozzle up to perform the measurement. It is possible that it pushed the bed down all so slightly at the end of its measurement. But it is not supposed to push it down, which is what I understood from your initial message. 

 

I believe the space between the Z stage and the heated bed is suppose to be 14mm ideally. Perhaps you could check this too @neute?

 

 

We do not allow active leveling to be disabled on the Ultimaker S5 because the bed is significantly larger than an Ultimaker S3 or Ultimaker 3. This means that a minor offset can go by unnoticed on an Ultimaker (S)3, but can compromise bed adhesion on an Ultimaker S5. And we know that a successful first layer is key to a good print. 

 

It shouldn't be a pain and you shouldn't resort to Cura to work around it. Make sure the bed springs are at the correct tension and your print cores are clean. You can find more tips via this link, via our partner in the us FBRC8. 

 

 

Edited August 6, 2020 by paxpring

[SandervG 1,521]

13 minutes ago, paxpring said:

I see that two people already posted the total height of their bed module (being 28.67mm). Mine is currently at 31.34mm. When I check if I am close to the 14mm you mentioned, I see a red layer on the bottom of the bed. Measured without that red layer, the distance between the bed and the base plate is 6.4mm and at no point that distance even approximates your 14mm. I must admit I am a bit lost ... Looking at the image you included in your post, I can only guess that you might have a different base plate?

 

Hi, please ignore the image I've shared. This image belongs to an Ultimaker 3. Please use the values @Smithy and @UlrichC-DE kindly shared. That was a mistake on my end (confusing the 2 Z stages). So try to get closer to 28.67mm. 

[paxpring 4]

3 minutes ago, SandervG said:

 

Hi, please ignore the image I've shared. This image belongs to an Ultimaker 3. Please use the values @Smithy and @UlrichC-DE kindly shared. That was a mistake on my end (confusing the 2 Z stages). So try to get closer to 28.67mm. 

 

We're only human 🙂

Thank you for helping out, much appreciated!

 

Edit: and thanks to the others for the input

Edited August 6, 2020 by paxpring

[Smithy 1,146]

39 minutes ago, paxpring said:

However, downwards moving beds and upwards flexing x-and-y-rods as a result of the nozzle pressing downwards are not okay. 

In my opinion it is normal and ok, no need to worry about that. I have this behavior on the UM3, the S5 and the S3, it works and to be honest I don't care about it. It is not so, that the nozzle is pushing hard into the bed, just a little bit and then the bed, the head, rods, "bend" also, but not so much that I have to worry that something gets damaged.
 

The measured value is also not so important, so when you have something similar it should be ok, so round about 28mm and I am sure if you have 27 or 27,5 or 29 it also makes a not a difference. It is just important, that the springs have enough "room" to work. They should not too tighten and not too loose/soft.

 

When you adjust the screws, do a manual bed leveling afterwards, that you have a good leveled bed as starting point for the active leveling, otherwise the active leveling has to correct too much which is not necessary. 

[neute 16]

Hi all, 

 

thanks for the fruitful discussion. I've tested my print bed after your comments. When I tried to reach 28,5 mm, the screws bottomed out. I turned the backwards screw all the way in and then half a turn out to have a little play, and did the leveling. I had to adjust the both screws in the front so that the distance between print surface and bottom of the print bed is now about 29,5 mm. There hasn't been a significant change in the first layer of the part I printed before and after this procedure, but I'm willing to accept that the print bed is "optimal" for the active leveling to work best now.

 

I'll try to make a comparison of what I consider a "perfect/sucessful" first layer and what the machine achieves without touching the settings.

 

Best regards, Niklas

 

Edited August 6, 2020 by neute

[Smithy 1,146]

Just to clarify, the hight of the bed, manual levelling and so on, don't influence your first layer. It is just a preparation for the active levelling and the sensor. The actual distance is then given by the active levelling. 

 

I am happy with my first layer, but some have the feeling it is too close to the bed and they use the z-offset plugin.

 

 

[UlrichC-DE 115]

Calibration is recommended during initial setup. (Although I suspect the Ultimaker already does that at the factory) The automatic algorithms for adjusting the Z-values on an uneven printing plate (for example, only the pythagorean theorem) work more reliably on a reference surface. It is also logical that the fixed XYZ geometry of the printer cannot work on sloping surfaces.

The resistance of the table springs must of course fit. That only approximately pure distance values are calibrated.
It is important that the table springs are stronger than the print head springs. Otherwise only a convex surface or the resistance of the springs will be measured. Then the measured value is millimetre off by a few millimetres.

There is no 0 or 1 sensor. A BLTouch is also inaccurate. This is simply called measured value tolerance. Mathematically, this can be compensated by repeated measurement (avg() ) and fusion of measured values.

 

In other words. The fact that you can work with the Z-Height plug-in proves that the calibration and its sensors work. (If the calibration is successful)