Print layers drifting towards rear

[DDLLC]

As we are clamping a sphere, the perimeter of the inverse conical nose of the OEM set screws will want to do one or two, or both, of two bad things; a) they will dig into the sphere and make irreparable divots that will become the new favourite seating/setting positon for the sphere and screws, or b) they will cause the sphere to rotate along the screw nose's path as the screw is tightened screwing up (no pun intended) up any positioning of the platen just accomplished.

I also put a dab of high pressure grease on the screw's new ball nose to minimise any tendency of he screw's nose to rotate the sphere.

That's what I originally thought but actually the set screws impinge on a split cylinder that is between the sphere and the screws. The split is between the two set screws so in theory, the sleeve is pushed in and squeezes the ball.

I did adjust z bearings and can say that stepper motor stalled when it tried to drive the arm off the top of the ball screw. I had to pull up on the arm while simultaneously hitting the up button repeatedly. Not a graceful process. The rods upon which the bearing ride seemed to be narrower at the very top. The rest of the travel up the mast didn't cause any audible lost steps. It needed quite a bit of adjustment to be able to reinsert the z stage. From all the things I've seen on forum, I'd suggest that every new owner just make this adjustment before getting started.

Of course that didn't fix my problem. Yesterday I made another run with standard exposure conditions with a 100 micron sheet for zero and the slant persists and the back is a bit thicker than the front.  I print a small open rectangle inside a larger rectangle and have had the smaller rectangle separate from from the FEP about 2/3 of the way through the print. Today I increased the base layers exposure to 75 s from 60 seconds and other layers to 10 s from 8 second. Zeroed off a 100 micron sheet as before. The rectangle pictured is the result. The thick end is about 0.7mm thicker than the thin end. The step backwards results in a 3-4 degree slope. The Z mast tilt backwards maybe 0.1 degree (tilt meter read 0.1 degree off normal when base has zero tilt but meter is supposedly only accurate to +/- 0.3 degree) so it's not the cause of the problem.

I was going to make another run with a zero on 180 micron sheet but during tightening of the front set screw, the thread in the aluminum housing was stripped. I should admit that I replaced the socket screws with a hex head since I thought maybe the original screws might be limited in the amount of force they could apply to the sleeve since it looked like it's shoulder might bottom out on the recess. So I'm out of commission for awhile until I replace this part. Maybe the problem I'm seeing is related to something on that head or the ball lock system, so maybe a new head might give superior results.

[cliffyk]

Yup--you are 110% correct--I opened up one of mine last night. It's not badly designed.

Good work and good plans to track down the problem. If anything else occurs to me I'll chime in...

[DDLLC]

Some updates and interesting observations. I had a replacement build plate clamp made out of STEEL and didn't counter sink around the set screws (to reduce risk of thread stripping. You can see it in the pictures. The set screws need to be longer ( 6 mm x 16 mm) to reach the split clamp sleeve. During zeroing (on a 180 micron paper) after tightening both set screws I notice the paper was actually only still clamped by the right front portion of the build plate. I had held the plate in position while snugging up both set screws with what I judge to be reasonable force. I re-zeroed being even more careful tightening and it still was held by the front left. Again I re-zeroed and tightened the front screw. The plate seemed to be holding the paper all along the perimeter. Once I tightened the right side screw the paper was only held by the right front corner.

My feeler gauges showed that gap being in 0.007-0.008 inches range and extends as far as I can slide the gauges in without damaging the blades. I played with various techniques but the best I can achieve is a gap of between 0.0025 to 0.003. I basically applied as much force on the rear edge as I can with my fingers while I tightened the right hand set screw.

I don't really have a good idea as to why this happens.

Right now I have a test print running and will know later today if this new head has fixed my "drift" problem.

Greg

[DDLLC]

Short story is that the layers are still moving rearward, but something new occurred. First the data.

On the large rectangle the thicknesses starting from the rear left corner and going clockwise to the left front corner the thicknesses (mm) are: 6.43, 6.31, 7.11, 7.20. Note that in the pictured called "10-19 front last print layer side" the image is now curved away from the front side. Hard to believe that is due to mechanical misalignment! That layer of curved printing is about 0.71 to 0.79 mm thick as best I can measure it. My Z step is 20 microns so that is many steps.

The small rectangle stuck to the FEP sheet again. It's dimensions clockwise from left rear to left front are: 4.7, 4.68, 5.06, 5.08. It's also shifting towards the rear.

The exposure time on base layer is 75s and other layers are 10s and this is with the elegoo grey resin. Also Zero was set with 180 micron paper.

I will run another trial but increase the number of base layers to 10 and will increase the normal layer exposure time to 12 seconds

I'm open to all suggestions as to what might be happening.

[cliffyk]

DDLLC--Great work !!!

Got me to thinking and I dug out my OEM Mars mast, and a dial indicator--set up as shown below:

"zeroed out"


The applied force is just the weight of the watch tool's handle with its tip placed 'tween the trolley carriage and rail. It produced a 0.005" (0.127 mm) deflection! Just very light force of my fingernail was required to create a 0.002" (0.0508 mm) deflection. Assuming the pivot point for this deflection to be the center-line of the topmost (leftmost in the photo) of the three guide pulleys makes the moment arm 36 mm. This calculates to a angular deflection of 0.08° to 0.20°

"pried upward lightly"


This is at the mounting surface of the build-plate support arm. That arm is mounted perpendicular the the trolley and is 95 mm in length from its mounting surface to the center-line of the build-plate post, allowing deflection at the build-plate post (and by extension the build-plate) of 0.134 to 0.332 mm--134 to 332 microns.

Deflection of this magnitude, brought about by application of modest external force, could easily manifest itself during printing as inconsistent positioning of the build-plate

The trolley guide wheels have a radiused groove, meaning that in practicality the radius of that groove in the wheels has to be a bit larger than the radius of the guide rods (t'would bind up ferociously were they not)--I believe that is where the measured deflection is originating. It would be better if they had a "V" profile groove, relieved at the apex.

Elegoo has acknowledged this weakness by changing to recirculating ball linear bearing and V-grooved hsrdened steel rail on the Mars 2 Pro mast.

Mars 2 Pro mast:

[DDLLC]

Clearly there is a good bit of slop in the design, but I'm not sure it's the source of my problem. I've just gone through more tests on zeroing and if the set screws are tightened enough (as in either one) some corner/edge of the build plate will rise above above the paper. I also checked the flatness of the build plate on my granite flat. The build plate itself seems to be very flat and even sticks a bit when one tries to remove from the granite. The newer mast/ z stage design is much better as is Jackson's design.

My best results seemed to be by just lightly tightening the set screws. I wasn't able to slide a 0.0015 inch gauge under the perimeter at any point.

I also use an electronic angle meter on the arm and it is 0.00 degrees in orthogonal directions both before and after zeroing. This meter is only accurate to 0.3 degrees, so it could show the deviations you measured.

I'm a bit worried as to if the plate will tilt during the ongoing test print. On the prior prints there must have been a lot of force to move the build plate to cause it to tilt and I think this print will be easier move out of level. Time will tell later today.

Greg

[cliffyk]

Clearly there is a good bit of slop in the design, but I'm not sure it's the source of my problem. I've just gone through more tests on zeroing and if the set screws are tightened enough (as in either one) some corner/edge of the build plate will rise above above the paper. I also checked the flatness of the build plate on my granite flat. The build plate itself seems to be very flat and even sticks a bit when one tries to remove from the granite. The newer mast/ z stage design is much better as is Jackson's design.

I found the same with my two plates, flat as that proverbial pancake we so often hear of...

My best results seemed to be by just lightly tightening the set screws. I wasn't able to slide a 0.0015 inch gauge under the perimeter at any point.

I have found that over-tightening those lock screws is common; I saw a video wherein a fellow twisted away on tyhe wrench as though he were tightening the log nuts on the rear of his F-250 "dually"...

I also use an electronic angle meter on the arm and it is 0.00 degrees in orthogonal directions both before and after zeroing. This meter is only accurate to 0.3 degrees, so it could show the deviations you measured.

The angles I presented were all calculated from linear measurements of the moment arm and deflection. I too do not have a protractor (digital or otherwise) capable of measuring such small angles..

I'm a bit worried as to if the plate will tilt during the ongoing test print. On the prior prints there must have been a lot of force to move the build plate to cause it to tilt and I think this print will be easier move out of level. Time will tell later today.

Greg

Keep up the good work, something has to be shifting dynamically. Have you tried different layer thicknesses? It is not inconceivable to me that what you are experiencing is being caused by some combination of machine precision, resin density cghanges during curing, and/or other antagonistic relationship of the system settings and components.

Often it is helpful when changing something changes an outcome--regardless of what was changed. Otherwise we are doing just what dear Albert warned against...

[DDLLC]

Well the light tightening with no perimeter gap didn't work out as can be seen in the picture. One improvement is that the center rectangle adhered to the build plate. The layer thicknesses are 20 micron and there are 25 base layers with 75 s exposure. Also the following layers have 12s exposure instead of 10s. Note also that Z arm was still level in two directions to 0.00 degrees.

This time the base layer count is also 25 and exposures are 75 s and 12 s. The screws are tightened as best I can. I did ultimately end up with the rear left corner having a 0.006 inch gap. The others seemed to be a couple of mils or less. Not sure why one of them isn't closer to 0.006.

The real question is what is forcing the plate to ultimately tilt. When I've really tightened the screws is near impossible to force the plate to shift unless I put a huge amount of force on the part. I don't see why that should happen. Something is systematically applying force on one side.

One other item I may not have pointed out is that the "design" part thickness is 5mm and yet ends up closer to 7mm on the thicker side. On this one it was more like 8.7 mm and the thinner side is 1.75 mm.

[DDLLC]

The second run from yesterday with maximum tightening on the screws also showed a tilting of the build plate and resulting in a print shift. The angle is more or less the same ranging from 3-5 degrees off normal. Clearly tighter set screws is better but it was unavoidable to have up to 7 mils gap between the zeroing paper (150 micron) on one side. The level meter on the clamp arm is still showing 0.00 deg in both directions. The printed  thick edge is around 7.25 mm and the other is around 6.6 mm. The small rectangle at the center also separated this time unlike the previous run. It's thickness ranged from about 4.7-5.1 mm.

 
This morning I disassembled the clamping head and cleaned /examined it. There is nothing obvious as to why the tightening of the right hand set screw causes the print plate to tilt.

Today I've zeroed on 120 um paper and still have a 0.004-0.006 inch gap on the left side. I'll see if that makes a difference although I really don't expect it to.

[cliffyk]

I wonder if altering the strength of the pre-load spring in the head would change anything, or lubricating the ball within the sleeve? The latter is counter-intuitive, however practical solutions often are...