Strange shrinking problem with a flatcar

[Lemosteam]

I am printing a very short flat car, the one in the printed mold thread.

The top surface is perfectly flat.

I have tried all manner of supports and I am insisting on printing it parallel to the build plate.  I have tried printing it ON the build plate (which yields a VERY nice surface by the way, but the bottom layers are exposed so long that the growth ruins the side of the car, rivets , etc.

What is happening is it prints the supports great, and we are out of the bottom layer thing, but when it prints that first large rectangular layer it shrinks proportionally below the horizontal yellow line in the image, next layer is a little larger, and when that large flat area is complete all the rest of the layers above that are near perfect.  The red lines show what happens to the perimeter vertical surface below the horizontal yellow line.

When I pull the print off the build plate, if you look from the side or the end the supports are angled as they go up, toward the very center of the car, like the yellow rays in the image below. the angle is less and less as your eye move closer to the center of the car.  Same is true when looking from either end of the car.

Note as well that the shrinkage is PLANAR, (TWSS) and that the print does not bow at all.

Resin is Siraya Fast, Layer thickness is 0.03, exposure time is 3.5 seconds  10 bottom layer at 30 seconds each.  It's like that first layer is shrinking, and then the next layer shrinks a little less.  I'll try to get some pics later.

My plan is to flat-sand the top anyway to rid the print of the pooling resin on top from printing horizontal, but it looks like hell when the top 1mm is tapered inward when it should not be.

Maybe I take the bottom layer limit all the way up through the supports to this first layer with the 30 second exposure, and sand it away like flash after?



The massive number of supports is there mainly to support around the edges- I was getting separation with that first layer earlier too.

Suggestions?  getting mighty frustrated.

[Chris333]

It is because there is such a dramatic change in surface tension right at that layer. I have no idea how to fix it.

[narrowminded]

I don't know this for sure but I have experienced a similar issue when printing flat.  What I suspect is happening is that the span between the supports when the first layer prints is so thin that it stretches when peeling from the film.  When it re-engages for the next layer it is flattened again, this time pushing the supports ever so slightly outward as the now stretched layer dictates, also not necessarily leaving the intended build gap but something less.  That sees the light and cures, not necessarily at the intended thickness, repeats the cycle but with the stretch error improving as the layers thicken in the span.  This causes a distortion that is gradually corrected to the proper position but carries the distortion where each layer found its "home" as the process progressed.  The supports hanging in the breeze and relatively soft at this point allow this flex and stretch to happen.  It's also why angled positioning can help but I know, it's not always desired for other reasons.

An effort to stop or hinder that action might include an additional support piece, unconventional by typical support styles, made as a sacrificial part of your design.  What I would envision would be a thin walled grid spaced just above the desired surface with attachment points at the spacing and cross section you will use for your normal slicer supports and then another attachment to the model surface about midpoint on your new grid between those typical support points.  The slicer attachments could be at the intersections of the grid squares, following through to the model surface, with the second span attachments to the model surface made at the middle of the span.  The grid piece with some vertical section (possibly 6mm/ 1/4") would give some vertical stability to the sag effect described above and the removal would be similar to the normal slicer attachments.  All of the span errors and stretch issues (if that's even the problem) could be contained/ corrected in the sacrificial grid before it reaches your model surface. 

This is all speculation but not without some experience with similar issues.  Good luck!

[SkipGear]

You need to support the middle as much as any other part. On a large flat surface, I make a grid work of supports trying to keep the spacing between them as even as possible.

[Lemosteam]

@SkipGear  that's what i thought so many more supports, see below. Seriously?   

The part is 6mm off the build plate, with a 1mm thick raft on the build plate leaving 5mm long @ 1mm dia tapering down to the tip for 1.5mm taper from the tip end.

This is the amount of supports I have under the image above, a veritable forest.  The car is ~19mm wide so at any one cross section I have 7 supports across so one every 2.7mm leaving 1,7mm between supports:



@narrowminded that kind of makes sense, but look above WTF?  Fortunately this one turned out the best in actuality, abs the slight taper appears like a small mold draft parting line.

[Chris333]

The whole floor is one big solid rectangle that creates more tension on the FEP as it pulls away. Then once the last floor layer is done and you just have the side walls the FEP tension is reduced all the sudden in one single layer by about 85%. The result is the ring around the model you see.

[Chris333]

As an example if I print a tender tank shell with the tender deck. When I get to the layers with the flat deck I will get a ripple down the out side of the tank right at those layers. So I print the tank without the deck and it come out perfect.

I know you need the floor, but print one without it and see how it turns out.

[narrowminded]

Without seeing the whole model I would consider flipping the whole thing, printing the floor last.  All of the supports underneath.  And probably with fewer supports.  Or, as Chris says, print in two parts and glue it together when complete.   

[Lemosteam]

The reason I am printing in this direction is so I can just sand that top flat and be done. I have a belt sander that will make quick work of that surface, and I can grind down to a certain depth. This way the underside is perfect for the truck bolster mounts.

There is always a method to my madness.

[SkipGear]

@Lemosteam - OK, you got me there, that's more supports than I would try.

I think Chris has the right idea. Print it without the floor. Print the floor separate and use styrene or scribed wood for the floor depending on what you need. It is probably the way the real thing was built anyhow.

[Lemosteam]

Rather than print without the floor, I may blow holes through the floor were the structure is not. Those could be replaced by strips of sheet lead, underneath the etched brass deck for mor mass.

Hmm.

[Chris333]

I just suggested printing without a floor to see if that makes the problem go away. Then you'd know. As for fixing it with the floor... I dunno.

[haasmarc]

I was thinking about using a piece of sheet brass for the subfloor.  3d print the sides, ends and undercarriage.  Tap the brass for the couplers and add a laser cut deck.

[SkipGear]

I just noticed your layer times....how much do you have the UV turned down? My "standard" exposure time anymore is 1.4-1.6 seconds with the UV lamp at full strength. I just printed a replacement flywheel coupling for an out of production Athearn Genesis Pacific and it came out dimensionally perfect the first time which sort of surprised me. The shaft fit was perfect and it is in the loco running now.

[Lemosteam]

I just noticed your layer times....how much do you have the UV turned down? My "standard" exposure time anymore is 1.4-1.6 seconds with the UV lamp at full strength. I just printed a replacement flywheel coupling for an out of production Athearn Genesis Pacific and it came out dimensionally perfect the first time which sort of surprised me. The shaft fit was perfect and it is in the loco running now.

Wait I can change the intensity of the light output?  no wonder I am seeing so much bloom. not knowing this the only thing I have been messing with is the standard layer exposure time of 3.6, and that was based on the RERF recommendations,

How do that from factory settings?