The following Instructable uses the same resolution metric and copies the test geometry from greener1's Instructable on Ember Printer's Horizontal Resolution (https://www.instructables.com/id/Ember-Printer-Measuring-Horizontal-Resolution/).
Ember and Form-1 are two SLA (stereolithography) 3d printers that print using photo-cured resin. In an attempt to compare Form-1's horizontal resolution with that of Ember, a test stl was created, printed on the Form-1 and Ember, and photographed under a microscope. While the resolution metric suggests that the horizontal resolution of both printers is equal, at 5 line-pairs per mm (lp/mm), there was considerable trouble printing, and other factors make paying attention to a single number an oversimplification.
Step 1: Creating the Test Geometry .stl File
The Form-1 only takes stl and obj files as input for printing. Because the final file for the resolution test in greener1's post was a png stack, the first step was to make an stl that would properly represent the png-stack. This was done by close inspection of the png's and the knowledge that the smallest pixel in the line-patterns were 50 microns (.05 mm) on a side. Using a lot of copy, scaling and rotating tricks, the geometry was replicated in Rhino3D using planar curves and extruding to solids. Note that angled lines in the image are unavoidably pixelated, with stepped boundaries. This feature was not copied in the stl. Instead the angled lines were made to be 50 microns wide and smooth (in the case of 1 pixel per line pattern) or 100 microns (in the case of the 2 pixels per line) and so on.
The final solid was booleaned together and exported to a massive poly-count stl. No wonder the original file was a png stack, it was likely much more efficient at storing the data than saving all those polygons!
The same stl pattern used for the Form-1 was given to the slicer in EmberPrinter.com. When EmberPrinter.com makes a png stack from the stl the resampling step creates artifacts which actually made the apparent resolution worse. Changing the position or rotation of the line patterns relative to the build area would undoubtedly cause different sampling patterns to occur, making the apparent resolution seem dependent on the position and orientation of the pattern (see the last three images in this step)
Step 2: Printing on Form-1
Printing such a large continuous cross-sectional area part was a challenge on the Form-1. The Form-1 has a peel action, where the resin tray rotates about one edge, in order to detach the most recently cured layer from the resin tray. This motion tends to suck layers with large area off their predecessors, which leads to the unsightly peeled edges seen in the images, and resin seeping in between layers.
After some observation it became clear that the peeling happened on the edge of the rectangle farther from the hinge. In an attempt to mitigate this a wedge was added to the rectangle, which at least seemed to act as a sacrificial region at the very least.
The support structure option in pre-form (the Form-1 printing software) was used once, but resulted in another fail, with only the support structure successfully printed (see image).
Step 3: Printing on Ember
Small Test Object
A smaller test object, focusing on only the 1 and 2 pixel width line patterns, was printed. After a considerable number of failed attempts, likely due to a cloudy PDMS window in the resin tray, a series of settings on emberprinter.com (on the print settings page, under 'show advanced settings') that resulted in consistent prints for the smaller test object was found (see attached file). I suspect that the aspect of these settings that was most critical was the 3.5 seconds exposure time for model layers. Changing this parameter from 2.5 to 3.5 seconds, all else constant, made the difference between failed and successful prints.
It is unlikely that these setting will work for different geometry, and much of the difficulty arose from a clouded or otherwise blemished PDMS window.
In addition to these settings, the smaller test object was placed as close to the central axis of ember as possible, so as to reduce the moment arm, and the relative velocity between the surface of the print and the window as it rotated to separate a cured layer from the window (see hand drawn image)
Full Size Test Object
The full size test object, without the taper, was also printed on ember. The only successful print occurred on a new PDMS window, with an anodized aluminum build head. The settings were identical to those of the small test object shown in the images and in the print setting file except that the model exposure time was increased to 4 seconds. The quality of this print is notably better than that of the Form-1 or the small test object printed with the damaged PDMS window.
Step 4: Analyzing Result
The most successful print, the one with the tapered edge, was photographed using a Stemi 2000-c Zeiss microscope at 1.0x magnification.
Prints on the form-1 with a layer height less than 50 microns were failing. The only layer height that produced this somewhat acceptable result was 100 microns. For this reason the one pixel line pattern is not seen, because its lines were only 50 microns high.
Form-1 X/Y Resolution
The Form-1’s minimum feature is reported on the form-1 website as being 300 microns wide, and the laser can “move in steps if fewer than 10 microns.” However, from the test geometry that was printed it appears that the form-1 can resolve horizontal features (x/y) of 100 microns in width, yielding an estimated 5 lp/mm (line pairs per millimeter).
Due to issues with printing large adhering-surface-area models it may be that smaller features are still possible.
Ember X/Y Resolution
Greener1's Ember print and the successful print of the larger test block suggest that Ember’s x/y resolution is also 5 lp/mm. The smaller print shown in this instructable suggests that this resolution is only truly possible with a clear PDMS window and if the test pattern happens to line up correctly with the pixels in the png stack. However, it is important to note that the small test geometry would not print successfully on the Form-1. With a brand new PDMS window the entire test geometry can print with great success.