thats a good point about the undercuts… thanks… ill probably try it tomorrow. at this point im more interested in eliminating the possibility of any printer issues.
High temp resin printing problems FORM 2
feyr47
#74
hey there, did you take this file off drop box? I got the split master file, but I cant seem to get the corrects d20.
last night i re tried printing my “corrected” file set to .05mm res…it was very close to what I wanted “numeral issues aside” the print itself was ok but not what i would call functional.
In an attpemt to determine if there are any further issues with the printer, i wanted to prnt your corrected d20 file in the interum im goin to try the split master file.
its really amazing how unseen spurious data creeps into a file like this. When i fixed the floating anamolies which I caused during a cut uperation and flat out missed in the final model I had hoped thisgs would work out. but I guess that other flaw you pointed out is still having an effect., from what I can see it looks perfect now. but clearly something else is wrong… the split master should be finished by late morning… will update with results.
feyr47
#76
ok so here is the split master print! looks pretty much perfect…thanks allot for fixing the model!. this reassures me that its NOT my printer. I also attached what I printed last night to for comparrison purposes.
Now ill need to figure out how to detect those flaws you found “not the ones i left by mistake”
The print i did last night of the mold i made where i removed the floaters DID make a difference, but clearly there are still problems.
Any further suggestion on how to find these flaws would be appreciated.
thanks
Sculptingman
#77
The flaws can be impossible to find without a proper diagnostic software analysis.
The floaters were not the primary flaw- they were the render-able geometry left behind by a boolean or trim/stitch operation that resulted in non-solid geometry.
This might have resulted in duplicate points or polygons or line segments that perfectly coincide with other points, polygons or line segments.
Some of these bits of unconnected geometry or topology might be internal to the model… And most of these kinds of error will occur when you convert the file from a surface model to an STL model on export.
Something as silly as modeling in a space with a resolution of 0.0001 mm - but converting to or importing into a model space with a resolution of 0.001 mm can result in points in the modeling space being “Averaged” to the exact same coordinates… giving you two points in the same location, connected by line segments of zero length- either of which can give printers conniptions trying to figure out the shape of the object.
Just one polygon in a model that has its defining points connected in reverse direction creates a “hole” in a model that renders as a solid, but as far as the printer is concerned is a soap bubble with a wall thickness of zero.
Most of these errors are due to bad procedures or set up of a modeling space by modelers who do not understand solidity in 3D modeling, and therefore do not realize the kinds of things they might do that will violate solidity.
People who model in surface modeling apps like Solidworks, Maya, Rhino, or SubD apps like zBrush and Modo are far more prone to bad modeling technique, because the Apps themselves are designed to represent surfaces only and do not warn you when your model is not a solid. And so there is no pressure to learn what solid modeling is.
I can’t tell you how many jobs I’ve gotten where some really brilliant artist modeled a ZBrush figure with oodles of delicious detail… and the file gets handed to me because the artist did not make a model that could be realized as a solid object. ( a LOT of digital modeling if for CGI animation… or games… and does not NEED to be a solid- it just has to render nicely.)
Apps Like Meshmixer were largely originated as a tool for resolving the many different kinds of meshes produced by the many different types of modeling apps out there.
Preform itself has an import function meant to detect and correct many kinds of non-solid geometry- but it can not handle every possible combination of errors that modeling apps can produce…
And even though I run an app that ONLY can produce a perfect STL solid… Preform often claims that a model I import is broken… even though it is a perfect qualified solid.
The easiest solution is to spend the money to buy an app that models in voxels and can not export an STL that is not perfect… Like Freeform or its less costly cousin Sculpt from 3D Systems.
Failing that, the second best option is to use a modeling app that has solid modeling features, diagnostics and repair features such as Form Z.
And, of course, its better to read up on how Topology differs from Geometry and how the two have to be configured to achieve solid geometry.
For example- Imagine 8 points in space in the configuration of the apexes of a Cube.
That is the geometry of the model.
But HOW are those 8 points connected to each other? If each is connected in straight line segments defining the vertices of a cube- then you have a solid model of a cube.
However, if one of those corner points is connected to the diagonal opposite corner point thru the center of the cube… that can not describe a cube… can it? Those specific order of connections are the Topology of the model, and they must be in a very specific order for the computer to be able to understand the geometry as a solid cube.
In subD modelers you use 4 point geometry- meaning every polygon is supposed to have four points.
However- if One of those 4 points gets moved out of plane to the other three… then that polygon is NOT a flat plane anymore… and it will often render with a triangular ‘hole’ in the polygon. This violates solidity…
and, of course, STL conversion seeks to bisect every 4 point polygon into two triangular polygons because 3 points ALWAYS define a flat plane.
Yet- if one of those triangles is connect in the reverse direction to all the others, then its a hole into the model and it can not be a solid.
Some of these simple things Preform can correct for. Sometimes, in trying to correct bad geometry, Preform itself can create unprintable geometry… without being able to ‘recognize’ that the model its is displaying is a Ghost the Form 2 can not print.
3D modeling has become popularized… lots of folks get into it with no real training or background in how it works- they rely on software engineers to come out with apps that correct their mistakes without their even having to be aware there was one. And so those modelers do not improve in their technique.
There is no substitute for competence. And in the world of 3D printing… for every failed print that is the fault of the printer or its materials… there are 50 that are the fault of bad geometry in the model.
Sculptingman
#78
PS- you will note that in that split master print one of your separation planes is really ripply.
That is an artifact of that plane being very close to parallel to the build platform.
You will get a much better result if you angle the model such that the separation plane is inclined to the platform by at least 45 degrees. ( in the direction the angled faces are running so that ALL the separation planes are at around 45 degrees )
Because this is a split master- you do not need to use the costly high temp resin, and would be better served by an opaque resin like the grey- or the dental model resin so that you can better ‘read’ the surface of the model for finishing prior to pulling silicone molds from it.
Ideally- you want to orient the split master to maximize accuracy of the separation plane- you are less concerned with printing artifacts on the facets of the die.
the reason is because the separation planes of both sides of a split master need to Match and so you want to avoid at all costs having to retouch or alter the separation planes in any way. The surface of the model, however, you intend to polish and refine anyway…
so the printing angle that gets you the most accurate separation plane may well leave ripples or other artifacts on some of the angled planes of the die- but those are surface you plan to refinish anyway.
Ideally- You should create the opposite split master for the other half- and create some additional registration feature on your existing separation plane because the plane you modeled registers to two halves in ONE direction… but they can still slide relative to each other in the direction those angles run.
You will need to keep the CAVITY lined up in both X and Y and the existing planes only give you X.
And one last thing. I made that model as a demonstration from the file you provided. In addition to adding another registration feature… You can eliminate the Ejector port and have that face in the model.
And you should re-position your Gates ( the entry point of the resin you will be casting ) so that you introduce resin at the lowest point in the cavity-( As the mold will stand during casting ) with vents at the highest points… when you are not using an injection press, you need to consider that you are Not filling an empty mold with resin, you are pouring resin into a mold that is Already Filled, with Air.
Plan for the resin you pour or syringe into the mold to flow into the cavity at the bottom, and sweep the air up and out of the cavity ahead of the resin. This will result in better and more consistent castings.
feyr47
#79
Thanks for all the advise, I’ll need to reread this afew times to pick out all the details. I mainly work in solid works “because it’s what I have for machine design, solid modeling is clearly a side function” you make a strong case for some solid modeling software. Incidentally, for fun I ran that corrupted stl file through meshlab and freecad just to see what it would find. Both did find correctable errors, but they didn’t dlfind that “scratch” flaw in the face. I’m printing that corrected geometry mold.file just to see the results, I’ll have to redraw the whole thing to fix the numeral spacing. Thanks again. I’ll update when I have more results
feyr47
#80
Hi
Had some interesting results today, this pic is of the print I did last night from corrected geometry model you sent me.
It definatly came out better that anything Ive made, but still isnt useable, perhaps there are still some issues with the printer?
Sculptingman
#81
well that is not good. It should have printed perfectly… and again- the flaw is confined to the interior of the faceted cavity.
This is clear evidence that something is awry with your printer- Or it might be positional… that is, you appear to have the narrow hole at the top- closest to the build platform- so the facets are angling out as each layer is added. This might be resin clinging to the underside of these angled facets and getting cured by a laser spot that is not tightly focused due to some issue with optics.
If you haven’t opened a ticket with support, Do so and send them this and other pictures of the flaw.
I would tend to orient a model like this at a steeper angle… but this is a really strange and persistent flaw.
feyr47
#82
It sure is persistent, I’m in a bit of a pickle as far as angle orientation. If I set it too steep I end up with supports either in the mold cavity or the pin channel “even if I abandon this design, I think it should be printable”. So it’s a balance. I did open a ticket a while back and NEVER heard anything from form labs, we never purchased the pro service plan, I guess you get what you pay for. I had allot of success printing this design in flexible resin, the only real difference is the hole in the mold. I gave some thought about workspace setup and it occurred to me that I did all my designs at home on an older version of solid works than what I use at work where I print the molds. I did some work on the design at work “on my lunch break” using the new version of SW on the same file, AND I never checked to see if the files tolerance settings were the same… I would think they should be but I don’t know for sure. I’ll have to look into that as well. Do you have any opinions on meshmixer and it’s stl/ solid model analysis tools? I plan on getting sculpt or freeform but meshmixer is an interim step.
JohnHue
#83
Same here, but the few times I did contact them I got swift and useful advice. I think you should try again and point to this thread as well as the fact that Ike didn’t get the same result with the same file.
Ike
#85
For what it’s worth, here’s the orientation I used:
new d20 straight numberstop-hightemp.form (758.5 KB)
I did manually remove some of the supports from the edge of the triangular hole on the bottom face before printing.
feyr47
#86
thanks, I also remove, move the supprots near the hole. did you make any corrections to the model outside of preform? if so what did you use?
thanks
I also openeda new ticket on this.
feyr47
#87
Hey there any opinions be on why that splitmaster mold printed, but not the fixed mold model? I’m struggling to explain that besides an unseen flaw in the model.
Sculptingman
#88
that is super bizarre.
I think you should discuss this with Formlabs support…
The only thing I can think of is that as you print it in that orientation - each time the platform comes down, the resin inside the die cavity is being funneled up and into a narrow dome shape of the angled facets, and literally splashing up onto those faces and the laser is being diffracted into that wet surface and partially curing it.
the real Mystery is that IKE was able to print the orginal file without issue other than the deformed 1.
OR- its possible you did not print the corrected file.
The Form 2 keeps a Queue in memory of recently printed files. I have more than once accidentally sent a PRIOR version of a file in the queue to print, because the Image of the model on the Form2 screen LOOKED the same as the file I meant to print.
For example- the last time this happened, I uploaded a revised file to the printer, and when I went to look at the printer to confirm the print… I did not realize that I had not hit the Back Arrow to see the most recent imported file… but saw the image and name of the LAST printed file- which was a prior version of the same model… I pressed the Print button, and the machine started to print. Luckily, when it could not read the cartridge, I looked at the screen and realized that the name of the file ended in V2 instead of V3 and was able to abort the print.
So you can double check to ensure you printed the corrected file… if the numeral 1 in the 12 does not look full height, then you printed the uncorrected version. ( assuming its partially visible despite the flaw )
feyr47
#89
OK so it just got weirder! over the weekend I took the file I originally sent you in an attempt to replicate the flaw “not the floaters,i removed them” you showed me in the screenshots from before. I ran the file successivley through meshlab, freecad and meshmixer. I used the analysis tools provided to fix any errors found… I then tried printing this file off in corner of the buld plate i dont normally use and it turned out! weired numerals and all.
I had already contaced formlabs support and they suggested dirty galvos, i also wonder if maybe the window is at fault somehow? it SEEMS perfectly clean but perhaps im missing something… i used a strong flashlight while cleaing with pec pads according to their instructions. the pic dosent show everthing but its pretty much spot on.
Sculptingman
#90
the galvos are down inside- you can’t really assess them without opening the optics box.
If this error keeps happening in the middle of the platform, but not at the edges, then there are two possible problems. One is that the optics have a slight spot flaw. Just a particle of dust on one of the galvos can cause a diffraction flaw that repeats in one are, but not others.
The other is that if this is the first time you have moved the model any significant distance on the build platform then your repeated attempts might have burned in the middle of your tank floor.
Check the burn map in Dashboard for that tank to see if it shows significant shadowing where you are tending to place the model.
It is good practice to MOVE models all over the build platform from one print run to the next. keep mixing it up.
feyr47
#91
thanks, im leaning twards a galvo issue.
- the tank im using is brand new with only 6 or 7 prints, im aware of the practice of moving the prints and I do this, this current print is just in an extreme position.
- I have seen this error crop up once on one other print before on a totally different model.
I suspect something is up with either the galvos or the possibly the main window… i can see some faint “wear” marks on the surface that is otherwise very clean, it almost looks like a coating on the window is worn off somehow.
Id like to get a replacement if possible.
i am re-printng this in the center of the tank to see what happens now.
Ike
#92
Galvo issues usually affect the whole build surface, so if your window is clean but it’s still bad in most places and good in one place, that might point to a problem on the main mirror, not the galvos or windows. Cleaning the main mirror is slightly less difficult than cleaning the galvos, but if you end up following Support’s instructions to open up the printer and clean things, I think you should probably clean all 3 mirrors.