Really printing printers this time

July 14th, 2013 by

The title of my last post was intended to be figurative. Yet here I am with a second printer on my desk. No instructions or a bill of materials this time around, just some photos to work backwards from and a Google Group.

It's not something that I would necessarily recommend doing unless you're slightly nuts and have some spare time on the weekends. But now I have about 6x the build volume and can print 3x as fast. And Deltas just look incredibly cool - My wife calls it the Spiderbot.

It's all open source and on GitHub, with the main development here and my fork, with some custom parts and modifications, here.

Printing PLA on frosted acrylic

May 3rd, 2013 by

(1) The adhesion problem described in this post caused by the filament. You get what you pay for. I only use Diamond Age (high quality, every color under the sun, expensive) or Printrbot (local, cheap, good) filament now. Skip the random vendors on Amazon.

(2) Printing on unheated, frosted acrylic is a bad idea. The roughness combined with the lack of thermal expansion makes over-adhesion a very real problem.

(3) If you have trouble printing on heated glass, something else is wrong in your system. If no heater is available, use blue tape from Amazon or Printrbot.


The adhesion between my PLA and heated glass bed suddenly degraded about a week ago. Before then, adhesion had been fantastic running the extruder and bed at 185C and 60C, respectively. I was using borosilicate glass from McMaster-Carr mounted on top of the heater PCB with clips. Two other things changed in the same time frame that might have led to the adhesion problem: I started using colored PLA filament from new sources (Amazon and 3D Printer Hub) and I started running the air conditioner.

The sudden degradation in PLA-glass adhesion has been reported by other people (e.g. here, here and here). I'm not sure exactly what caused it in my situation (build up on the glass? moisture in the PLA? black magic?) but I found a better alternative to heated glass in the process of debugging. I tried a few things initially that didn't solve the problem that are worth mentioning.

First I tried modifying the bed temperature (60C -> 90C) and obtained a slight improvement in adhesion but prints would still fail quite frequently. Next I tried cleaning experiments. For the first month I had been using nail polish remover (acetone + lots of other crap) and was concerned that contaminants might be building up on the glass. So I consulted the forums and tried cleaning the glass with hot water and detergent, vinegar, lemon juice, isopropyl alcohol and acetone (separate experiments). None of them solved the problem and I was starting to get frustrated with glass.

So I started looking around at alternative bed materials. Many people use kapton tape or blue painters tape with good results, but I like having a continuous and smooth printing surface. Polycarbonate has also been reported as working, but it scratches more easily than acrylic and is more expensive. The main problem with acrylic seems to be that it bonds too strongly to PLA even when unheated, but I remember seeing that sandblasted or frosted acrylic can give the right level of adhesion. So I picked up a 6" x 6" x 1/8" piece of acrylic from Tap Plastics and frosted it by hand with 400 grit sandpaper.

The frosted acrylic solved my issues and has a few significant advantages over a heated glass bed with just a few downsides. The advantages are that acrylic is incredibly cheap ($1 for my piece), it can be refinished using sandpaper in case it gets dirty or scratched, it can be easily drilled rather than held in place with clips, and it doesn't require any heating (faster prints, one less variable to control).

The first disadvantage is that you can't print ABS anymore, but I have no interest dealing with warping and rafts so this wasn't an issue. The second disadvantage, perhaps more significant, is that you no longer have separate control of first layer temperature and adhesion. So far this hasn't been an issue, but I can imagine a scenario where good first layer adhesion required a higher than ideal temperature which would cause other issues.

The main risk seems to be that the adhesion might be *too* good, although this hasn't been a problem with my first few prints on the acrylic. I also bought a sheet of HDPE while I was at the store; it is more chemically stable than acrylic and I haven't seen anyone write about trying PLA + HDPE so it's something that I'll try out in the next week or two as an experiment.

Printrbot insulator sleeve

May 3rd, 2013 by

A few weeks ago I started a print and took the dogs out on a walk. When I came back, there was a huge blob of PLA attached to the tip of the extruder because the first layer had not adhered to the bed. The filament gunked up the red insulation boot and I cut off the contaminated portion. Unfortunately this made it slightly too short to cover all of the hot bits, so I started looking for a replacement.

Printrbot does not sell individual replacement boots and finding a generic source seemed like a much better solution than contacting them for this one random part. After a bit of searching I found a 1/2" silicone coated fiberglass insulation sleeve that is similar (although not identical) to the sleeve on my printrbot LC. It is slightly looser than the stock sleeve but works great and 5 feet of it is about $20. I'd consider ordering the 3/8" sleeve in the future to have a slightly tighter fit if I ever run out of my 5 feet, although there is some risk that it'd be too small. It would be nice if printrbot sold their insulation boot separately but this will do.

(update: I contacted printrbot and they included a replacement sleeve with another order that I placed. Fantastic customer service!)

Printing printers

April 29th, 2013 by

I bought a printrbot LC kit about a month ago. I had been toying with the idea for a long time, but was spurred on by seeing a 3d printed cat toy that a friend made and realizing that there are relatively inexpensive printer options ($500-600) out there today. At that price it is relatively easy to justify; if my only options were the $2k monstrosities from MakerBot then I would have never purchased one.

So far I've been having a lot of fun and am very happy with my printrbot LC, which I would highly recommend. I occasionally wish that I had purchased the LC+ rather than the LC (200 mm vs. 150 mm of travel along each axis) but it hasn't been a significant issue so far.

You need a reasonable amount of mechanical, electrical and computer skills for assembly and setup. CAD experience is also helpful for designing things. My wife thinks I'm crazy and would never build one herself. You can buy preassembled printers, but there's something to be said for knowing how it all fits together so that you can tear it apart to make modifications and debug issues.

There are a few absolutely essential upgrades:

  • fan, fan mount and three-point bed leveling system. The fan addresses PLA oozing and makes bridging work incredibly well. The leveling system is necessary to get good consistent adhesion. All three of these things are included in the LC v2 (currently shipping) whereas I had to purchase them after the fact.
  • glass bed and clips. I separately purchased the 0.125" x 6" x 6" piece of borosilicate glass from McMaster-Carr to use as my bed. The glass is rigid, hard and chemically inert so easy to clean, and provides good adhesion to PLA. The first thing that I printed (after Mr. Jaws of course!) was a set of clips for holding the glass in place. I taped the glass to the PCB heater for this print, and there was a noticeably amount of sliding around.
  • wall-mounted filament spool holder. I mounted a toilet paper holder onto the wall in my room and it is absolutely perfect for holding a 1 kg spool of filament. It is a much better option that the lasercut spool coaster and spool options from printrbot in that the spool can never slide off yet rotates easily.
  • filament guide. I'd also highly recommend replacing the built-in wooden idler latch with one that has a built-in filament guide. My extruder used to occasionally skip if the incoming filament angle was off-axis; the built-in guide solves this problem.

The glass clips have worked quite well, although the combination of the clips and the leveling bed design effectively reduce my y-range from 150 mm to 110 mm. I will probably design modified clips in the near future.

Adhesion between PLA and the glass bed has generally not been an issue. I've printed with natural, red, black and white PLA to date. Only the black PLA has sub-par adhesion, requiring that the glass is squeaky clean and that both the extruder and bed are hotter than normal (e.g. 185/60C typically, 210C/65C for black filament). 1 kg of PLA costs about $30. I've purchased spools from from printrbot, Amazon and 3D Printer Hub and haven't had any issues.

Regarding extruder temperature, I'd recommend running M303 to automatically tune your PID settings if you have a printrboard with Marlin firmware. I use a set of 5 mm calibration cubes and a frog to check print quality whenever I've modified anything.

One other tip - rather than trying to fine tune the z-limit screw, I've found that using the "Z offset" parameter in Slic3r is much easier way to adjust the degree of first layer squishing. This does not affect the need for the bed to be level, just removes the need to mechanically fine tune any constant offsets. For leveling, I follow the standard procedure of putting a sheet of paper between the extruder tip and bed and adjusting the thumbscrews on the bed to yield a constant and light amount of rubbing between the paper and tip across the bed.

I've come across a lot of useful links. Here are some of the better ones that I'd recommend checking out: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10.

Here's a video of my printer chugging away:

On biking to work

April 14th, 2013 by

I've been biking to work for almost a year, so wanted to share some of the things that I've learned.

Route: I live about 5 miles from work. About half of the route is on an off-street paved trail, a quarter is on nice roads (residential, business parks) and the last quarter is on a busy 35 mph road with a marked bike lane. The biking layer on Google Maps is a key tool for scouting out a route, although I'd recommend using Street View to confirm that all of the roads are bike-friendly before trying them (e.g. Central Expressway is marked as bike friendly).

Showers: I sweat a lot. In the summer after a run it looks like I also hopped in a pool. So having a shower at work was absolutely critical for me. Your mileage will vary depending on how easily you sweat and if you wear a backpack, but having a shower and changing room at work is an important consideration.

Panniers: I bought a pair of Ortlieb panniers a few years ago, which was a fantastic investment. Compared with a backpack they are waterproof, can carry much more weight and volume, and keep your back dry (see showers above).

Lighting: I have four lights: two attached to my helmet using velcro and two attached to my bike. The front light on my helmet is a 2W LED beast so that I can see on the pitch black trail at night while the other three lights are generic bike lights. The front light on my helmet also serves as a headlamp when we go backpacking. Maybe the most important part of the lighting setup is a good set of rechargeable batteries. Eneloop is the best option here by far and puts NiMH to shame. Any charger will work; I have this one and it works great.

Clothes: I usually wear the same clothes I would go on a run in; shorts and a synthetic short or long-sleeve t-shirt. If it's cold, I have a dayglo windbreaker and gloves in addition to a headband that keeps my ears warm under the helmet. If it's really cold then I pull out the balaclava and as a bonus get to scare people.

Other gear: I have a set of plastic fenders for winter, although it's about time to take them off for the summer. They aren't my favorite things in the world, but they do their job. I also don't have them mounted perfectly so they interfere slightly with the operation of my back brake, which is why I'm eager to take them off for the year. Also, clipless pedals are a must and make biking in general so much more enjoyable.

Safety: I've never had any issues with cars even though I bike home in the dark about six months out of the year. This is probably due to the combination of my dayglo jacket, four lights and I'm on a car-free trail for about half of the trip. My main issue has been biking with slippery conditions when it's rainy and I've had three relatively minor solo accidents. All three accidents were on one region of the trail and caused by slippery road markings (two accidents) or biking too fast down a wooden bridge (my bad).

That's about it. So remember - lots of lights, Eneloop batteries and panniers.

Movies made before their time

December 23rd, 2012 by

I saw The Hobbit last weekend and thought that it compared very favorably with the 1977 cartoon version that is so near and dear to my heart. Along those lines, two surprising things:

1) An animated/rotoscoped version of the Lord of the Rings was made in 1978.
2) Somebody uploaded the entire thing to YouTube and redubbed it MST3K- style.

What you say Hammer? Proper

October 4th, 2012 by

Wow, I just came across rapgenius. For example, you can find where The Nothing from the Neverending Story is referenced. Almost as good as reading about how spiders walk on Wikipedia.

What is the thesis?

September 27th, 2012 by

I'm not sure, but it's online in its entirety here. Are you prepared to experience some advances in nanomechanical force sensors with integrated actuation?

Such a good quote

September 27th, 2012 by

I just came across this comment to a New York Times article. It is just so well written and perfectly conveys my experiences.

I've been a vegetarian for a long time. For both health and animal cruelty issues. But I don't preach; I just eat. Still, it's remarkable how vegetarianism angers folks. Really angers them. Especially the older ones. Somehow it's un-American. Many carnivores think I'm accusing them of something. Try telling people you're a vegan sometime. It's hilarious. Imagine a vegetarian running for president. A communist transexual would be elected before a vegetarian.

Speaking of which, I have explained what I eat and why so many times over the last few years that I've lost count. I never, ever bring it up but always need to be ready for the standard array of questions when I do not eat something with cheese on it.

No, I do not live in a tree with a spotted owl named Moonsparkle. Yes, I can eat mushrooms although they are not technically denizens of the plant kingdom. No, my shoes and belts are not leather due to the magic of synthetics. Yes, lions do hunt gazelle and all of that but they don't exactly have any other options. No, my cats and dogs do not eat animals and don't need to in order to be healthy and happy. Yes, I am physically fit with off the charts B12 and calcium levels, half the cholesterol level I had a few years ago and twenty pounds lighter. All of these, except for one, are real questions I've been asked. The vast majority of people are nice and polite, but some people just open up with a barrage of questions, usually seeking out some fallacious chink in the vegan armor.

If you're a question barrager, please take the time to read this extremely reasonable book written by a former Bush speechwriter or this short and sweet video (or any other Mercy for Animals video) and your questions will be answered. I'm happy to answer non-confrontational questions just to dispel the notion that vegans are the black-bloc wing of the vegetarians. But please, no more gotcha- or owl-questions. And I never plan to run for elected office.

Triggering traffic lights on a bike

September 1st, 2012 by

Today I learned how inductive loop sensors at traffic lights work.

The motivation was a traffic light on my route to work that never detects me. Until now I have been carefully positioning my bike into the middle of the loop. I had been assuming that it was based upon the movement of a small piece of steel somewhere on my aluminum bike inducing current in the loop. This concept fits with the fact that cars (large with lots of steel) reliably trigger loop sensors while bikes (small and usually not much steel) often do not. The solution then would be to stick a magnet on your bike. Problem solved right?

This is not how inductive loop sensors work, thanks to this fantastic writeup. They are resonant systems. The loop forms an inductor in an LC circuit that is driven on resonance (at 20-30 kHz usually). Any change in the inductance shifts the resonant frequency. What's great about this is that your car/bike is part of an oscillator. A car triggers the loop by being large and conductive, not large and magnetic. Your vehicle behaves like a shortened turn in the roadway inductor, increasing the resonant frequency of the oscillator.

Similarly, a bicycle triggers the loop sensor by decreasing the inductance seen by the circuit. The best way to do this is by bringing a large ring of metal to the region with the largest magnetic field. Your tire rims, unless you spent a small fortune on carbon fiber ones, provide the loops. But where should you position your bike? It depends on the way the loop is configured. If it's a simple square then you should position it directly over one of the long edges, not in the middle of the loop. If you can see two side-by-side squares in the pavement then you should position yourself over the center line where there are two wires carrying current in the same direction. If there are diagonal cuts in the pavement, just make sure that one of your wheels is over one of the diagonal lines. If the first link is too daunting, read this link (or either of these) for short and sweet diagrams.

It is impressive how many people think that run of the mill permanent magnets can trigger inductive loop sensors (myself included this morning). Loop sensors are normally designed to detect an increase in resonant frequency (i.e. inductance decrease). However, some fraction of loop sensors are configured to also respond to an increase in inductance and you can buy magnets on the internet that purportedly trigger loop sensors. So could you just strap a chunk of iron onto your bike frame? Probably not. If you look at the patent application filed by the Green Light Trigger company you'll see that the magnet needs to be ceramic, i.e. high electrical resistivity. As the excitation frequency increases the apparent inductance of a magnet decreases due to eddy currents which short it out. By increasing the resistivity of the magnet you increase the corner frequency up to which it looks like an inductor to the circuit. Some people on the internet claim that magnets don't work because only inductance decreases are registered. Others, who are trying to sell you stuff, claim that inductance increases are also detected and their magnets work. Either way, someone on the internet is wrong.

The frequency change required to trigger the loop sensor depends on a threshold that is programmed into the system. There is a tradeoff in setting the threshold - too low and cars in adjacent lanes might trigger a left turn light. Too high and bicycles and motorcycles will not be detected. This means that if the loop sensor is relatively small (< 10 feet long) then it's just a matter of adjusting the sensitivity. If you have positioned your bike properly and it still doesn't trigger then you should let your city know so that they can reprogram it. I haven't had a chance to ride through my problem intersection yet, but hopefully the power of physics will help me next time and if not, the magic of complaint forms on the internet will.

TLDR: Your bicycle rims are primarily responsible for triggering loop sensors. Learn where you should place your bike. If the loop sensor still doesn't work you should contact your city.