Page MenuHomeAleph Objects Inc

Bed corners bowing glass
Closed, DuplicatePublic

Description

While doing some diagnosis on quiver 1, I noticed a squish inconsistency that might be explained by the printed bed corners pushing up on the glass and making it bend upwards.

@matth has been doing some testing so here is a ticket to record the results and confirm if this is an issue or not, and if so, to work towards a solution.

Event Timeline

kent created this task.Feb 8 2019, 12:58 PM
kent triaged this task as Normal priority.
kent created this object with edit policy "LulzBot Hardware Products (Project)".
matth added a comment.Feb 8 2019, 2:19 PM

First I ran a calibration and didn't change them at all before doing this test.

I put in bed corners that we modified by just cutting off the triangle portion of them. I then did a print that printed a 2 layer square in the middle and four corners of the bed. The print seemed consistent for each location. Repeated this two more times with similar results.

Then I put in the default bed corners and did the same print with all the same settings. Just from the skirt it was evident that the center of the bed is higher up than the corners, as it would look more squished in the center and less so in the corners. Repeated the print again with the same results.

I then replaced the bed corners again with the modified ones and got the same results as first test run.

I can say, for quiver 1 at least, that the default print corners were bowing the glass a bit. I'm curious to see if other alphas would get the same results.

gcode for my test print:

DaniAO added a subscriber: DaniAO.Feb 13 2019, 8:45 AM

@matth the bed corners that were on the Alphas have been changed to help with this. I will get an MO put in to get some of the new ones printed up for you and will let you know when they are ready.

DaniAO closed this task as Resolved.Feb 21 2019, 3:20 PM

@matth I apologize for the delay but I do have those bed corners when ever you are at Mountain next.

matth reopened this task as Open.Feb 25 2019, 12:41 PM

I installed the corners and did a test print with them. Then I put in my older modified corners and did the same print.


With the new corners, I still see a noticeable difference in the bead height depending on where it's printed. In these pics, the inner skirt is with the older corners and the outer is with the newer corners.

In this picture, located in one of the corners of the heat bed, both skirts have a decent, similar bead height.

However, towards the middle of the bed, the new corners look different, it looks a little more squished/closer to the bed. The inner skirt, the older corners, appears to still have the same bead height as elsewhere in the print.

My older corners have the triangle parts removed, which doesn't seem to make the bed bow. If you have a depth gauge, you could more accurately measure this.

logan raised the priority of this task from Normal to High.Feb 27 2019, 7:31 AM
logan reassigned this task from matth to west.
logan added a subscriber: logan.

Seeing this issue with beta bed corners, the bow in some cases is so great you can see it from several feet away.

west added a comment.Feb 27 2019, 7:38 AM

We could just move to the TAZ 6 modular bed ones and deal with the gap another way

logan added a comment.Feb 27 2019, 7:40 AM

@west Except TAZ6 modular is the same as stock and used M3x8 fasteners, the M3x10 we're using there now would protrude from the bottom of the bed plate, which may not be a big deal.

logan added a comment.Feb 27 2019, 7:41 AM

But we use flush with the bottom of the bed plate still to specify how far to tighten those fasteners, fwiw.

west added a comment.Feb 27 2019, 7:41 AM

Then we could just remove the triangle section of the current bed corners and use that

logan added a comment.Feb 27 2019, 7:42 AM

@west That would be my best choice as well, and is how I had alleviated the issue before.

I received a set of these from @EricNugent and it brought my bed center Z offset from -0.3 to -1.05; much better.

@EricNugent would we be able to get a few more sets to test?

I would like @matth to at least get a set if possible.

@DaniAO I can drop off the extra parts from the print if you wanted to hand them out for testers. I installed a set onto Quiver 7 but haven't performed bed center offset test with them

@EricNugent thank you!

@matth I can leave some on my desk for you, when ever you head this way next.

DaniAO added a comment.Mar 8 2019, 8:13 AM

@matth any updates with the new bed corners?

matth added a comment.Mar 8 2019, 9:46 AM

I've been having good, consistent results with them.

kent closed this task as Resolved.Mar 20 2019, 9:43 AM
robert added a subscriber: robert.Apr 3 2019, 6:27 AM

Not sure if its the new bed corners(PP-GP0396) or not, but my heat pad is sagging by as much as 2mm from my glass. I even tried different heat pads, that start out touching the glass while at room temperature, but once they heat up they sag dramatically.

tutley added a comment.Apr 3 2019, 7:13 AM

@robert @logan @EricNugent @kent this is why the prototype/alpha corners had the little legs as they help support the plate and prevent it from sagging. With the right infill density they were laying flat and not bowing the beds upwards.
I would recommend we go back to those corners and revisit the slice and or model if need be.

logan added a comment.Apr 3 2019, 7:23 AM

The bow was less with the right infill, but not near as flat as the current corners. If we go back, we should look more closely at model dimensions so that there isn't any upward force in the middle. We need only to prevent excessive sag, I don't think we need as much pressure upwards as those corners previously had.

MikeR reopened this task as Open.Apr 4 2019, 10:26 AM
kent claimed this task.EditedApr 4 2019, 1:07 PM
kent added a subscriber: west.

The design that @west commited with https://code.alephobjects.com/rTAZQUIVER346990dde1862da593f5da21ae130a73b52eea06 is the same dimension as the one that we had bowing issues with.

It has those legs at 4.75mm. What is the actual clearance? My recommendation is to find out what the actual clearance is, then make the dimension of the model match that.

kent added a subscriber: paulette.Apr 4 2019, 1:36 PM

I took some measurements and it looks like the distance between the bed plate and the silicone heater pad is averaging about 4mm.

adam added a subscriber: adam.EditedApr 4 2019, 1:56 PM

didn't read all the way up - we do have sag issues. is this a matter of mixing and matching the existing corners to better account for the sag or would it be better to reslice the other design?

kent added a subscriber: david.hall.

The existing corners are dimensioned so that there would be about 750 micons interference between the corners and the heater, which results in upward force and bowing of the glass even at low infill densities. I just pushed 9b31fda49f07 to the T5668A branch which uses a 4mm leg height which more closely matches the actual clearance between the bed plate and the heater. I think will address both the bowing glass issue and the sagging heater issue.

To confirm this fix, we will need to print and test some.

logan added a comment.Apr 4 2019, 2:14 PM

tagging @bigmansas for slicing

I believe that the photo above was the last slice for Quiver. Haven't we removed the triangle? This is the PP-GP0396, correct? If we are going to change then we will need to test and remove backstock?

kent added a comment.Apr 4 2019, 5:30 PM

Using a flexy standoff like these would probably work too and we have tested this solution already in the past. Not sure which size will work best yet.

I sliced the file and set up test prints for 16 of them last night. We have them in hand now and are trying out the v2.4 of the bed corners.

kent added a comment.Apr 5 2019, 9:27 AM

I installed the 4.2mm standoff and it supports the bed without pushing up too much on the middle. That's what I recommend using. I did notice while testing that if the bed corners are installed without pushing them in towards the glass, then there is room for the glass to slide around on the top of the heater, but that's an issue for another ticket.

Here is the cluster gcode for the 4.2mm standoff's, just in case they work after testing.

west added a comment.Apr 9 2019, 7:21 AM

@kent the problem ive seen with adding a support in the center is the perimeter of the bed sags

west added a comment.Apr 9 2019, 7:39 AM

@kent i just pushed a revised bed corner with these dimensions

@west We need to move away from that type of bed corner. It takes far too long to print and we are already at requiring 3 days time to print enough printers for one day of assembly. The support in the middle should work well as the corners that are there should keep the perimeter from sagging. We have seen it work on the Taz 5 and measurements that MER has been doing have indicated that it will work. I would like to see the smaller bed corner with the center support used if the testing comes out okay.

west added a comment.Apr 9 2019, 9:37 AM

@david.hall the support in the center doesnt fix the perimeter, it only prevents the very center from sagging. @tutley and I have both seen this. A center post was tested when quiver was in prototype phase and ruled out for the perimeter staging

kent added a comment.Apr 9 2019, 10:09 AM

@west have you tested this new design on any of the betas you have?

I apologize, I thought that the issue that we were trying to fix was the center sagging. I didn't know that we had any perimeter sagging.

west added a comment.Apr 9 2019, 10:13 AM

@kent I havent, mostly because cluster slices are always different from my slices to resulting data would be invalid

logan added a comment.Apr 9 2019, 10:18 AM

I would really like to see the perimeter sagging with the center post installed. I haven't seen that and the only way it seems possible is to push the corners in to put a downward bow.
The claim was also that prototypes didn't see the bow in the glass from these corners, which doesn't seem at all possible given the geometry of the early parts was very clearly showing a significant bowl shape.

@logan while the center post support does reduce the sagging around the edges, there is still some sagging.
the prototype corners possibly sag in the middle a little as only the edges were supported, however there were no heating issues in the center of the bed (likely because the plate bows upwards when heated) and the edges were pushed up flush to the plate. This addressed the number 1 issue we had heard with the modular taz beds, a visibly sagging plate.

@west and i have both seen the plates sagging with center post support installed. I can try and get one installed and take a photo if you absolutely must see with your own eyes.

logan added a comment.Apr 9 2019, 10:43 AM

@tutley I am not he who makes these decisions, if I am to take your word for it, that will suffice...
This method was approved for TAZ 5 modular beds, FWIW.
I think users and calibration techs alike would agree that a slight perimeter sag is much more acceptable than an extremely bowed bed, which has been the persistent & easily reproduced problem with the proposed larger corners.
How large is the perimeter sag compared to what we see from an unsupported bed?
If the bed tends to bow upward when heated, is a slight perimeter sag going to be an issue?
Do we have thermal imaging comparing these multiple scenarios?

The benefits of using the center post seem to outweigh the negatives, that is, without any measurements of perimeter sag given.
The benefits include:
Not bowing the print surface
Supporting the middle
Less material/less print time = more production output
better fit - the large corners tend to creep out the front and back sides due to interference with wiper mounts
threaded hole in sheet metal already there

@logan this is with the latest rev of the corners and the latest center post. There is .5-.75mm gap at the edge. The bed is heated to 100C
This gap is smaller than i had previously seen (using taz6 corners and the original center post) but it is still there.. Dont get me wrong, I think it is probably fine, but you said that there is NO gap, so heres the proof that it is still there.
This is on the quiver at my desk, and is only a sample size of 1.

Overall, the center post definitely reduces the gap, but does not eliminate it.


logan added a comment.Apr 9 2019, 11:22 AM

I don't recall having claimed there wasn't a gap, only that I had not yet seen it with the center post installed. Thank you for the visual.

Looking at what is provided, I think that is acceptable personally. But we'll have to reach a consensus as a team as far as what we should proceed with.
The scenario in which the bed remains flat and evenly heated, while keeping in mind manufacturability, should be the end goal.

@logan no prob. sorry if i misinterpreted your comment. I think this amount of gap is acceptable as well