Friday, July 24, 2020

The Effect of Drawing Speed on a Sand Table Pattern

Most people run their sand tables at 100 mm/sec or less which helps keep noise down and detail in the drawing high, though it can be a little dull to watch. I run The Spice Must Flow at 500 mm/sec most of the time so that it will be entertaining to watch as the pattern is being drawn. I have always suspected that running at such speeds reduces the detail in the final drawing but didn't really know how much, so I tested it.

I ran one detailed pattern at 100 mm/sec and the same pattern at 500 mm/sec, and took some pictures so I can compare the resulting drawings.

Here You Go

100 mm/sec, acceleration 10k mm/sec^2, jerk 200 mm/sec, 190 minutes drawing time

500 mm/sec, acceleration 10k mm/sec^2, jerk 200 mm/sec, 38 minutes drawing time

100 mm/sec, acceleration 10k mm/sec^2, jerk 200 mm/sec

500 mm/sec, acceleration 10k mm/sec^2, jerk 200 mm/sec


Yes, loss of detail as expected. Is it acceptable? I guess it depends on which aspect of the table's performance interests you. If you're more interested in the final pattern, go slow and maximize detail. If you're more interested in the process of drawing the image, go fast!

Note: acceleration is high at 10k mm/sec^2. That ensures that regardless of speed, the drawing time will be minimized because most drawn segments will hit the target speed. That also tends to cause the ball to throw sand at the start of each new segment as the ball rapidly accelerates to the target speed. Throwing the sand (onto previously drawn lines) is one way that detail is lost when running at high speeds such as the example here.

One last thing, because who doesn't like cat videos?

Friday, July 17, 2020

A Car Stand for a OneWheel Pint

My son rides around on something called a OneWheel. It's like an electric skateboard but it has only one large tire and it sort of self-balances. I think he's nuts, but what are you going to do?

The OneWheel weighs about 9 kg. When he transports it his car, its odd shape allows it to tumble around every time he hits the bakes or gas or turns, so he needed some sort of stand for it that would hold it stable and keep it from damaging the back of the car. He showed me a couple designs for stands on Thingiverse, but as is often the case with models on Thingiverse, they are best used to illustrate how not to do something.

I started my design the usual way, by making a 3D model of the OneWheel with just the pertinent details modeled to the accurate dimensions, measured with a caliper and measuring tape. I used that model to design the stand. The design has a large, flat base, 286 mm in diameter by 6 mm thick, that is a silly waste of time to print. It would be more sensible to start with a wood, metal, or plastic base, either round or square, and then screw down some printed parts designed to hold the OneWheel, but I didn't have any suitable material on-hand so I went ahead and printed the stand with the large base.

CAD model of the OneWheel on the stand. The side of the tire and wheel rim rest on the stand.

The tabs just fit into the rim of the wheel but don't interfere with the valve stem.

Print in progress on UMMD- it took about 10 hours and 45 minutes, about half of which was spent printing the large circular base, and used 307 g of PETG filament.  It was printed in 0.3 mm layers at 50 mm/sec with 20% triangular infill, 4 perimeters, 0.5 mm line width, and 4 top and bottom solid layers, in one piece without supports.

Here is the OneWheel on the printed stand. Like our genius president, it's quite stable!

Wednesday, July 8, 2020

To a Hammer, Everything Looks Like a Nail

A couple years ago I installed a timer switch on my front porch light so it would turn on at dusk and off at dawn. It was one of those smart switches that knows what time it is and when the sun sets and rises and even keeps track of daylight savings time. The light uses two candelabra type bulbs and is about 12' (almost 4m) off the ground. At about the same time I installed the timer, I installed a couple LED type bulbs because they are supposed to last 20 years and I don't like climbing up on a ladder to change light bulbs that are 12' in the air.

Well, it's two years later and one of the so-called 20-year LED bulbs failed. I wanted to replace both of them with yellow bulbs because the white light attracts bugs, and the bugs attract spiders, and the spiders make webs and catch the bugs and leave things looking very messy. I found a couple candelabra base yellow LED bulbs at Home Depot, got out the ladder and went to install the new bulbs.  

The fixture hangs from a chain, way up in the air, and requires two hands to get it apart to install the bulbs. That means I'm up on a ladder, with two hands on the fixture and none on the ladder, not a good situation. I unscrewed the bottom part of the lamp fixture, and it and the glass then came down, granting access to the bulbs. I had forgotten that the glass doesn't actually attach to the bottom part of the fixture and ended up dropping it. Oops!

So now I had to replace the glass, an impossible task because that stuff is not standard and even if I could figure out who made the lamp, they've long since stopped making them or stocking the glass parts. So my choices were either buy a whole new fixture or make my own replacement for the broken glass. Being a cheapskate and 3D printing type person- you know the type, every problem can be solved with a 3D printer- and having recently made a nice lamp using some beautiful edge-glow glass PETG filament from Keene Village Plastics, I decided to try using the stuff to make a replacement for the broken glass. If it didn't work, I'd only be out about $1 worth of plastic and a couple hours of print time and I could always replace the whole fixture.

After I finished cleaning up the broken glass, I made a couple quick measurements of the fixture and went to work. The only dimensions that matter are the diameters of the top (222 mm) and bottom (102 mm) parts of the fixture, and the distance between the top and bottom (180 mm) when the fixture is screwed together. In Fusion360 I created a profile of 1/2 of the glass using a few straight lines and a spline curve, then revolved it 360 degrees to make a solid of the correct size and shape, then exported the STL file. I sliced it with 2 perimeters with 0.5 mm line width and 0.3 mm layers, no top, no bottom, and no infill. It literally took less than five minutes to design and slice.

The print took about 3 hours at 50mm/sec. Judging by the surface of the print, I'd say 50 mm/sec may be pushing the speed a little high for this filament. No matter- it will be 12' in the air and no one will ever get close enough to see any imperfections. The whole thing is pretty flexible, but rigid enough to maintain its shape as long as I don't screw the bottom of the fixture on too tightly.  

The print doesn't look particularly transparent, but it actually lets a lot of light through. 

And here it is installed, with the new yellow bug lights on:

Yes, those are bugs sitting on it! So much for the idea that yellow light bulbs don't attract bugs...

Next project: clean all the spider webs and dead bugs off the porch walls and ceiling.

Wednesday, July 1, 2020

YAFL: Yet Another Fractal Lamp

My brother, the distiller, built a bar in his living room, and I thought that he might like a bar-worthy lamp for it, so I got busy.

I recently bought a couple 5 lb spools of transparent PETG filament (edge glow glass from Keene Village Plastics) that has some bluing in it to make it look like the glass that old Coke bottles were made from- I thought it would be great for a lamp shade, especially for colored lights because the bluing in the filament emits a beautiful, pale blue fluorescence. I've used PETG filament before, but this is the best I've used. It printed without a lot of hairs or charred blobs ending up all over the prints. I will be buying more. I suspect that all the "edge glow" colors they sell are fluorescent.

In a single wall, the stuff really does look like glass:

I searched through my collection of fractal based designs and settled on one I had previously used. I made some changes to make it more printable and for esthetic purposes, and sliced it for the new filament.

The print failed twice with the controller complaining of a heater fault. The first time I thought it might be because of a power outage (that's not a heater fault, is it?) due to a local thunderstorm, so I examined the printer and restarted the print. It got to about 23 hours and failed again. I took the cover off the printer so I could access the controller board and found that the screw terminals on the PT1000 interface board were stripped and had let the pressure off the wire enough to make the connection intermittent. 

The nice thing about prints like this is that even the failures can be useful, and no one knows they're failures if you don't tell them!

I replaced the cable and switched to the other PT1000 input on the interface board, then restarted the print. It made it all the way this time- 564.5 mm. It used 759 g of filament.  I printed it with a 0.4mm nozzle at 50 mm/sec, 3 perimeters, 0.25 mm layers. Total print time was 36 hours, 13 minutes and 57 seconds. Print time was included about 2 seconds per layer in order to make the layer synchronized time lapse video. 564.5 mm x 4 layer/mm x 2 sec/layer = 4516 sec = 75.27 minutes.

The camera had a problem when shooting the photo sequence that would become a layer synchronized time lapse movie. Even though I locked white balance, focus, and exposure, the brightness of the images varied a lot resulting in some really annoying flickering in the video. This only seems to happen when the print is lit by UV. I couldn't see any brightness variation or flickering of the light in the printer, so I checked the EXIF data of some of the brighter and darker images and found that the camera was varying the ISO value. Maybe a bug in Open Camera? I also found that the Mooni button suddenly isn't reliably working with the camera. Sometimes it triggers the camera and sometimes it tries to change the speaker volume setting. Hmmm. More problems to solve...

Here are some still photos shot while the print was running and almost finished:

This lamp has a round base, printed with black PETG. Unfortunately, the filament pigmentation wasn't consistent so parts of the print look lighter than the deep black color it started with. I used concentric top solid infill that makes the base look like a stack of vinyl records.

The light source is the same type Feit Electric 1600 Lumen, multicolor, wifi controlled LED bulb that I used in the last lamp, treated the same way- the two PCBs were separated and laid flat. The LED board is clamped to the top of the base by three screws that go through the heatsink plate inside the base. Yeah, I know it isn't going to be very good at dissipating heat that way, but the original heatsink was smaller and covered with plastic, so how warm is it going to get?

Pink/purple light, which is a combo of light from red and and blue LEDs, looks particularly nice. The bright center of the lamp looks pink while the rest of the shade fluoresces a pale blue color because of the blue LEDs. This photo doesn't quite do it justice: