If I have this correct the Form 2 only has a discrete resolution in Z, meanwhile X and Y had “infinite intermediate” resolution due to being handled with galvos.
Here is what I find confusing I get that steppers have discrete steps but only X has been switched to use a stepper Y is still on a galvo so shouldn’t it still have “infinite” resolution shouldn’t you be able to cure a straight line along the Y axis instead of one with overlapping dots. I get that the difference will be imperceptible but it seems it should be doable. Also my understanding of steppers used in FDM machines is they produce similar “infinite” resolution in X and Y.
So all that being said, why has switching to a stepper on the X axis caused the printer to have a discrete resolution in both the X and Y axes? Or was that an independent decision unrelated to the design change?
it is a very interesting question, it asks another:
on the form 2 we can see the laser from the top fill the surfaces and then pass on the outer edges for finishing …
on the form 3 that no longer be the case? since there is a line-by-line sweep, the edge will be as beautiful?
It might be that due to the new optics to be able to get the laser to shine perpendicular to the build platform that it makes the steps between the Y axis more apparent. The galvos don’t get rid of stepping entirely, it’s just that the difference between movements was small enough where you wouldn’t be able to notice it previously.
The reason is because galvo aimed laser on only a single axis has zero effect on perceived resolution.
Picture it… Place a pencil tip in the straight slot of a drafting template… That you can move it up and down with infinite precision makes no difference in drawing a cross sectional profile… because you move the template over 1 mm and draw another straight line… and then move it a millimeter, and draw another straight line… The round tip of the pencil Still leaves two round END POINTS to each and every straight line. And all the nuancible points in between are NOT going to be visible as they are just a solid.
In 3D- the axis on which the galvonometer sweeps the laser makes the INTERIOR smooth… but all YOU can see is the end points of each straight line that define the exterior surface of the model.- and that endpoint is GOING to be defined by the diameter of the laser spot.
So the machine has a finer laser spot than the Form 2 - but it CAN’T use that laser spot to draw a continuous line at the SURFACE of the model where you can see it, like the Form 2 does.
As a result, the exterior surface of each layer will be the stepover of the lightbox in one axis, and the diameter of the laser spot in the other. ( less the cure overlap between straight lines - so maybe 70% of the diameter of the laser spot )
This is why Formlabs stated that the Form 3’s resolution is better than the Form 2 in some ways, and worse in others.
I don’t believe the Form 3 actually prints with a series of dots like a 2d printer (or an inkjet type 3d printer for that matter,) but it’s always a series of parallel lines. The reason for applying a discrete resolution to both X and Y would come down to 1 of two reasons.
1: It’s really only actually discrete in one of the X,Y axes, but it makes so little difference compared to if it was discrete in both, surface finish wise (as well as customer perception,) that they just list the resolution for both.
2: Something was simplified in the software by discretizing both axes. There are certainly advantages to having “square” units, smaller data, less processing, etc. Combined with the above (not much advantage to keeping the higher resolution of the second axis,) they may have chosen this route.
The previous Formlabs printers still had discrete steps for the galvonometers, it’s just that they were very small, and they were not constant across the build area (smaller steps in the middle, larger towards the edges due to the geometry.) The Form 3 has optics (the parabolic mirror) that I think should eliminate the nonlinearity, so there at least is a resolution number that is consistent for each axis.
So perhaps the reason they never published a resolution before was because the size of the discrete steps was not constant and now that they have made them constant they are publishing them. But that leads me to my next question why 25 microns. With a stepper motor and a lead screw fairly inexpensive FDM printers can achieve ~0.4 microns per step, and I am virtually certain that the galvo can do better as well but don’t know the details well enough to argue it. In the Z axis I am also certain the hardware is more precise than 25 microns just like in an FDM printer but there are practical limits to layer heights and you get diminishing returns and longer print times as you go smaller not to mention more precise requirements on your materials. So I am guessing that they chose to make it 25 micron steps in X and Y to simplify things. However if I where them I still would only have forced 25 microns on the X axis and left the galvo it’s full capability. I can’t see any reason to constrain it other than it makes the numbers consistent and gives a constant value for that axis.
One- it makes no difference. The ability to print finer than 25 microns does not confer a perceivable improvement in the smoothness of the part.
I almost never print at 25 microns. At 50 I get just as good a result when I need fine detailing. At 100, I get better dimensional accuracy and the detail is good enough.
Two- resin bleed. Just as the machine has some difficulty printing layers much thicker than 150 microns, because the laser does not penetrate very far into the various resins, there is also a certain amount of light scattering that occurs when the laser hits a liquid. This is why the machine has difficulty printing a HOLE smaller than 0.3mm
An Three- time. The laser takes X amount of time to cure the resin… the finer the laser spot, the more linear meters of line the laser has to draw to make each layer.
The biggest complaint about the Form 2 already is the TIME it takes to print.
Most users, most of the time, print at 100 microns just for the TIME factor. And formlabs just managed to come out with firmware that enables it to print at 130 micron layers and draft resin that prints at 300 microns, to help address this issue of glacial print speed.
What question are you attempting to answer? I am failing to see the relevance. The Z axis I explained was constrained for a reason which you have done a good job explaining here. I also said if I where them I would still constrain the X axis. I then said I couldn’t see a reason to also constrain the Y axis. None of your points seem particularly relevant to constraining the Y axis if X and Z are already constrained except the one where you claim it doesn’t confer a perceivable improvement, which clearly is conjecture since you have not seen the difference of Y constrained vs Y unconstrained, but is probably true.
you said you didn’t see why the limitation of 25 microns…
I addressed 3 different contributing factors for why resolutions below 25 microns might be avoided, even though technically possible.
There is no limitation on the Y axis in terms of laser positioning… but that is irrelevant for reasons I outlined in a prior response. A single axis being unconstrained in Lengthonly can not contribute any refinement to the overall model.
While the laser in Y can still start and stop any straight line at a fractional laser spot’s width relative to the layer below or above… it still is going to move only in the Y direction from there… so the surface of the model will always be the endpoints of a line that is at leaser the laser spot size wide. Because it can not really draw a continuous line outlining the exterior surface, like the Form 2 can, this means that every layer’s exterior surface will be composed of the starting and stopping points of the laser spot. lending a very fine, but still granular appearance in the x and y plane.
The only instance in which any other point on a Y axis is going to be visible on those extremely rare instances where one portion of a given model’s surface happens to be in perfect alignment with the y axis for some distance… and even then, any deviations from that line in the x axis will have the resolution of the stepover of the print head.
As a result, the resolution in the Y axis will always appear to be the minimum diameter of the laser spot, slightly overlapped.
I said I didn’t see why the limitation of 25 microns on the Y axis if the other two were already limited.
Most of which were not relevant to the narrow topic of limiting Y if the other two where already limited.
Then what does it mean to say it has a resolution of 25 microns on the Y axis?
Are you suggesting they mean to say the line can start at any position but must be a length evenly divisible by 25 microns. That wouldn’t save them any effort doesn’t simplify the engineering and is not what X and Y resolution means in any other context I know.
Either way this is demonstrably false, however I will give you that the improvement may be negligible, but the cost of not constraining the Y axis to start and stop each line at points relative to a 25 micron grid or at lengths of multiples of 25 microns is negligible as well. When the cost is negligible I would go with more accurate.
As for your three points:
This is all based off your experience with Z resolution you do not know whether it would be perceivable in the case of constraining the Y axis to a resolution or not. However more precise positioning of the start and end of lines drawn in the Y axis isn’t going to lead to the loss of dimensional accuracy you see from finer Z resolutions.
So what is the problem that will occur due to resin bleed if a line drawn by the galvo on the Y axis starts and stops in a position more precise than 25 microns?
This is definitely relevant for the Z axis which is the slowest axis since each layer requires raising and lowering the bed which is very time consuming. It will also be likely to be somewhat relevant for the X axis as the LPU is on a lead screw so probably is fairly slow too. However in the Y axis the laser moves quickly. Also in the worst case for a given line will a finer resolution will add less than 50 microns of travel distance which takes an extremely small amount of time. But more interestingly some lines will also be shorter thus taking less time. Since the source data is arbitrary you will probably overall get a fairly even distribution of shorter and longer lines resulting in the time differentials canceling each other out, so in most cases it probably won’t add any noticeable time to a print and occasionally it may take slightly less or slightly more time.
I said I didn’t see why the limitation of 25 microns on the Y axis if the other two were already limited.
I answered that- Take a marker with a round tip and a straight edge ruler- try to draw a solid circle shape by strictly drawing straight lines that are parallel and spaced as far apart as the marker tip is wide.
What you will get is a roughly circular cross section that APPEARS to have a specific resolution in BOTH X and Y.
It will not matter that your straight lines can have any length- the combination of your marker having a round point, and that you have to make each successive line exactly the same width apart will compound to create what looks like resolution in both X and Y.
now Draw a square the same way, Aligned with X and Y.
On the Y axis- either side of the Square will appear perfectly straight and smooth- But its positional accuracy can not be any finer than the stepover of your ruler- You can NOT draw a smooth line HALF a step over. So Y resolution, though smooth, can be OFF by as much as the stepover of the ruler.
IN the X axis, the other two parallel sides of the square, the Y axis can draw a line of very precise length- however, either end of that line MUST be a round spot the diameter of your marker tip. Since you move the ruler over one marker tip width for each line- the result will be that that entire side of the square will look scalloped- like a bunch of DOTS in a row. A row perfectly aligned with the dimension of the “square” but NOT smooth.
Thus you have a limitation of stepover resolution in the smooth Y axis, and the resolution of the Laser spot size in the X axis. Unless your square happens to be a dimension equally divisible by the stepover, You will always have a dimensional variance in X- and an exposed raster in Y on every layer.
SInce the Form 3 has a stepover smaller than its laser spot size- the exposed raster in the Y dimension is the same as the stepover of the X dimension- so the visible surface resolution in X and Y are always the same.