Gas station without pumps

2021 August 23

Dragon key holder

Filed under: Uncategorized — gasstationwithoutpumps @ 15:18
Tags: , , , ,

In 3D printed names, I showed a key holder for my home and office keys, which finally failed after 3 years.  My family has been bugging me to replace it anyway—pointing out that it is stupid to have one’s name on a key holder, as it makes it easy for anyone to figure out what doors the keys are for if they get lost.


The key holder cracked at one of the bolt holes. It was still sort of usable (the keys were still constrained by the bolt and the cover plate was not yet coming loose), but it looked bad and would sometimes snag when I was trying to get it out.

I decided to make a new key holder with a dragon motif, based on a sweater from Past Times that has gotten too moth-eaten to wear, but that has a pattern I like.


I took only the dragon part of the pattern, which is 18 pixels by 109 pixels (20 by 111, if you include a 1-pixel white border).

My first job was to transcribe the knitted pattern onto squared paper, then type it into the computer.  I had hoped that OpenSCAD’s “surface” function (which I used successfully for the quantum dot pendant) would let me convert this image easily to a 3D relief, but it did a terrible job, as each black dot became a sharp peak.

I ended up using the Pillow fork of the PIL Python package to manipulate the image and export it in PNG format.

    im_5=im.resize([5*(num_rows+2), 5*(num_cols+2)], Image.NEAREST)

I resized it by a factor of 5, ran it through a “max” filter to spread out the black, then smoothed it.


The PNG file is the negative of original image. Unfortunately, the checkerboard grid does not print well at the size I needed to make the image.


I edited the pixels manually to get a somewhat more printable shape, then did the same sort of spreading and smoothing.

I initially printed just the dragon on a little oval, to see how it would come out, before adding the rest of the key-holder design from the original key holder.  Because OpenSCAD produces huge STL files when the “surface” function is used, I simplified the STL files with


Here are the 3 samples. The top one tried just expanding the pixels to 5×5, the second one did the smoothing, and the third one got rid of the checkerboard patterning.


The first two prints were failures. The top print was done face-down on the glass bad (as I had done the previous key holder), but there was too much spreading on the first layer. The second print failed because the glass plate rotated during the print—I had to add hot glue to it again, to keep it in place.


The final key holder is ok, but it would really look better if I had a smaller nozzle (the 0.4mm nozzle limits the horizontal resolution).


Here is the finished key holder with keys—I expect that it will last another 3 years (though I did use 50% fill this time, to make it a little stronger).

I’ve uploaded the design to

Final 3D-printed “quantum dot”

Filed under: Uncategorized — gasstationwithoutpumps @ 14:44
Tags: , , , ,

In 3D-printed “quantum dot” and 3D-printed “quantum dot” revisited, I wrote about my attempts to 3D-print the image from

I finally got good prints from the resin printer at work (they had to clean the optics on the printer) and a decent print of the “stage jewelry” version on my Monoprice Delta Mini printer. I gave away all the prints (including the failed ones) to the physicists who provided the data, except for the one best print in each size.

The STL files from OpenSCAD are ridiculously large (17.8MB and 19.1MB), but they can be reduced using without much loss of detail to under 1MB.

The OpenSCAD program, scaled data file, and two STL files are available at


Here is the resin print, which is 50mm in diameter. The peaks come out clean and sharp, but my only color choices were black and clear (the only two resins BELS had).


The back of the print has the scaling information, but even with sanding the spots from the supports are annoyingly visible.


The stage jewelry version is twice as big, with a diameter of 100mm (I measured it at 104mm—I think my printer calibration may be a bit off).


Again, the back has the scaling information. Using “concentric” for the bottom layers made for some interesting patterning.


Here are the two quantum-dot pendants side-by-side, to show the relative sizes.

2021 July 26

3D-printed “quantum dot” revisited

Filed under: Uncategorized — gasstationwithoutpumps @ 14:55
Tags: , , , ,

In 3D-printed “quantum dot”, I wrote about my attempts to 3D-print the image from, especially how I was unable to get a good print using my Monoprice Delta Mini printer.


Here are two not-very-successful prints using silk-gold PLA filament. There was a lot of stringing and the peaks were too fragile and snapped off.

I decided to try again, but at a bigger scale: 70,000×, rather than 32000× in the xy dimensions, making a 10cm diameter pendant.  My first attempt, using a layer height of 0.14mm was OK for the peaks, but the hanging ring did not fare so well.  Part of the problem was that the ring was too thin, and part was that horizontal circular holes do not print well—the flat top at the inside of the circle is insufficiently supported.

Update 2021 July 28:  I was looking at the original data file today, and it looks like I dropped one of the zeros in the xy scaling (I now think the scaling is 700,000×, not 70,000×).  I need to check the z-axis scaling also.


The image on the right shows the collapsed circular ring on the version printed at 0.14mm layer height. The image on the left shows the redesigned hanging ring and printing at 0.1mm layer height.


Here is the whole medallion at 0.1mm layer height in CC3D silk gold PLA with 20% infill. There was a little stringing and a few “zits” on the surface, but not too bad. I tried printing at 70micron layer height, but pronterface complained about not being able to allocate enough memory, so I gave up on that.

I’ll probably do one more post on these medallions, once I get the resin-printed ones that are printed without support.  The 10cm diameter is a bit too large for ordinary jewelry, but could work as stage jewelry.

2021 July 21

3D-printed “quantum dot”

Filed under: Uncategorized — gasstationwithoutpumps @ 10:40
Tags: , ,

Earlier this year, I saw an article at that included a very pretty picture of the scanning-tunnelling micrograph of the trapped electron.  I asked the author (Jairo Velasco Jr) for a copy of the data, so that I could 3D print it.  It took me a few attempts to get the scale and clipping right, but I was unable to get a good print using my Monoprice Delta Mini printer.


Here are two not-very-successful prints using silk-gold PLA filament. There was a lot of stringing and the peaks were too fragile and snapped off.

I finally got around to asking BELS to print one for me on their resin printer.  I had two choices of resin (clear or black), so I picked the opaque one.  The results are much better:


Top view of the black resin-printed electron density.


Somewhat more side view. I’ve played with the darkness here, to make the shape and layering more visible.

Unfortunately, they printed the part with supports, which rather spoiled the back, so I’ll probably order another one without supports.


The back gives the scale: 32,000× in the horizontal direction and 2,000,000× in the vertical dimension. The ripples and dots are from the support structure, which was really not needed.

Correction 2021 July 28: The xy scale factor is incorrectly printed—it should be 320,000×, not 32,000×.  I had an error in my OpenSCAD code in copying the step size from the original data files.  The z-axis scaling is ok.


This is what the support structure looked like after I cut it away from the medallion.

The prints on the resin printer cost me an $11 setup fee, plus $3.64 for resin per print.  I think the tank is big enough to print 6 or 7 at a time, which would reduce the cost from $14.64 to about $5 each.  I’ll want to print a couple as gifts for the physicists who gave me the data. Each medallion is about 4.6 cm in diameter.

2019 October 15

Extruded clamp

Filed under: Uncategorized — gasstationwithoutpumps @ 16:51
Tags: , ,

A few years ago, I found a little (6.5cm long) aluminum clamp on the street, probably fallen out of someone’s truck. I liked the design of the clamp(though the original rubber band has died):

Although the “Taylor” brand name is clearly stamped on the extrusion, I have been unable to find clamps by Taylor for sale online.

Extrusions are particularly easy to copy with a 3D printer, and so I decided to make my own clamps. I took off the rubber band, put the extruded aluminum pieces on my flatbed scanner and scanned them at 600dpi. I then used Inkscape to manually draw a Bézier-curve outline of each piece. I used the circle tool in Inkscape behind the curve to tweak the hinge contacts to be very close to circular.

I imported the SVG files into OpenSCAD and extruded the pieces to 12.5mm (the thickness of the original clamp). The first printing was not entirely successful:

The first few layers warped on the end of the handle—apparently the thin end of the handle did not adhere well to glass plate and warped up.

I fixed the problem in two ways: I made the handle a little beefier (it seemed a little thin on the original anyway) and I printed with a brim. While I was at it, I increased the extrusion from 12.5mm to 15mm to make a slightly fatter clamp. The resulting clamp was successful:

Here are the two pieces of the clamp separately.

The clamp assembled. By using three wraps of the rubber band, twists in the rubber band can be avoided.

The difference in the handles between the first print (on the left) and the second print (on the right) is clear in this image.

Removing the brim was a bit of a hassle—I need to think about applying a brim only to the handle part. I don’t think that the Cura slicer makes that easy, so I’d probably have to design it into the model (union with a 0.14mm thick plate around each handle).

Now that I have the basic model working, I can play around with different jaw shapes and different sizes of clamps.  With the jaw all the way open the clamping force seems to be about 15N, though that obviously depends on the rubber band used.

Update 2019 Oct 21:  I have released the design as  I was reluctant to release the design, as it was reverse-engineered and not my original work.  I contacted the company whose logo was on the clamp for permission, and they responded:

I have looked through our old brochures and have not seen a clamp like that.
We have not built anything like that in my 41 years here.
Good luck with your project.

I took that as permission to release the model on Thingiverse.

Update 2019 Oct 28: After some image searches for the logo, it looked to me like the logo might be Taylor Guitars, so I wrote to them and asked them about the clamp.  They said

Thanks for writing!
That is a clamp we made, it’s called a kerfing clamp. We use them in our factory and there was a time, many years ago, that we sold these clamps to guitar builders. We no longer sell them but we did give this guitar supply store, Stewart MacDonald, permission to copy the design.

I wrote back asking for permission to put the design on Thingiverse, and if it is denied, I’ll take down the Thingiverse post.

Next Page »

%d bloggers like this: