Here's the first iteration of 'Scout'. A simple printable frame for a robot based on two continuous rotation servos and an Arduino.
The idea was to limit the number of printable parts and 'vitamins' to the bare minimum and keep the total cost for the bot below $50. I'll use this to teach the robotics merit badge in December. Here's how the costs breaks down:
- 2 Futaba S3003 servos or equivalent (modified for continuous rotation). $4.50 each. I found deals on ebay to buy sets of 4 for $16 with no shipping.
- 1 arduino diecimila. $20 - 25. Also an ebay deal. A nano may work better. I'm still looking into this.
- 3 o-rings for tires. $3.00. Might be possible to replace these with rubber bands.
- 9V battery
It's taken a while to get used to the interface and there's still plenty of bugs to work around but the FreeCAD project is really moving along nicely. The sketcher makes designing parts like the ones below really simple. I learn things better with an actual project rather than just trying to copy tutorials so I launched into this silly thing:
I've been using kapton as the build surface but I accidently gouged it after my Z-axis home switch got bumped out of position. 200mm sheets with adhesive aren't cheap so I started looking for alternatives. It occurred to me that window tinting film is cheap, easy to find, and pretty durable. It also has an adhesive. Surprisingly, I didn't find any references to tinting film on the reprap wiki or forum.
Car windows can get pretty hot parked in the sun, but I wasn't sure the material would handle 100+ celsius. A little research on Google tells me the stuff is vinyl but that doesn't help much. Kind of like calling Bigfoot a 'critter' - Might be true, but ain't really saying much.
3M seems to be the big player and their site did have an MSDS. Their film is PET with a metal oxide coating. Score! - PET has a melting point around 250C.
I called a local tinting retailer and they gave me enough scraps to easily cover 8 or 10 build plates, way more than I need. I got samples of two different thicknesses.
My first attempt was just to clean the glass well, peal and stick. What looks like water droplets are actually air bubbles. But, my first test prints were very positive. ABS stuck as well as it did to kapton and separated easily as the bed cooled.
My next attempt, after googling a bit, was to clean the class and then spray with a mild soap solution. Then peal the film and lay it onto the wet glass. Squeegee the water out. Much better. But when I heated it, I got some large bubbles. I assume trapped water is boiling. Perhaps I should lay it in the sun for a few hours/days before heating it.....
Sprinter (formerly Klimentkip formerly Tonokip) and Teacup are the two leading firmware options right now. They both support retraction, acceleration, heated beds, thermocouples, etc. They're both in active development and things change fast. As I've worked to get my mendel reliable, I've bounced back and forth. Hence, I've learned a few lessons about each and the differences I need to make in my config to make each work. All this is just from my experience. If I got anything wrong, shoot me a comment and I'll fix it up.
- M103 / M101 are deprecated codes but Teacup uses them to trigger the firmware based retraction. Don't remove them through replace.csv
- Firmware retraction is enabled by entering a non-zero number for E_STARTSTOP_STEPS. I use 16X microstepping and about 700 steps = 1mm of feedstock. I wanted to retract about 3mm so 2000 is a reasonable number.
- Teacup Firmware retraction and Skeinforge retraction don't play well together. Disable SF by entering '0' in the mm to retract setting in the Dimension plugin.
- Sprinter doesn't do retraction in firmware but handles the Skeinforge retraction well.
- You must have your maximum feedrate for the E axis quite high. In fact, you have to set it much higher than you can actually extrude.
- Teacup uses G161 and G162 for homing. The firmware doesn't pay attention to the switches at all during any other move so manually seeking the switches with a G1 or G0 move will fail.
- When searching, Teacup will run at the axis maximum feedrate until it hits the switch. Then it backs off, and searches a second time at the SEARCH_FEEDRATE. This can be a little unnerving.
- To home a single axis you try to move the axis in the direction of the switch farther than it could possibly go. Something like G1 X-200 F1000 It will stop at the switch. Then you do a G92 to set the position.
- You can also use G28 to home all the axis, X, then Y, Then Z. G28 doesn't take an axis parameter and must home all axis. This doesn't work for me be cause I like to home my Z slightly below the bed height and then do a G92 with the offset to set the correct position. The dump location where I can move the nozzle below the bed height isn't at X=0 Y=0, so I need to home X and Y, then move to the dump location and home Z.
- Teacup wants distances in relative terms. Set this on both the 'preface' and 'dimension' pages.
- Sprinter seems to want to work in absolute numbers.
I haven't tried all these but noted them when I read them. Might be useful to someone
- A 10 uF capacitor across the thermistor contacts may help smooth out any noise.
- Settings in the configuration.h file are intentionally conservative. VERY conservative. The expectation is that a person will change them and not that some base config will ALMOST work for everyone. For example, the acceleration default value is 10. I ended up using 6000. Like I said, VERY conservative.
- If the result of STEPS_PER_MM_Z*MICROSTEPPING_Z is greater than 1000 then Teacup will ignore the MAXIMUM_FEEDRATE_Z setting. This can Z to do things you don't want or to just stall.
- you can change the pid factors stored in the eeprom using M130-M134, read source for details
- M200 prints the status of the available endstops.
I started seeing missed steps in my prints. I haven't had this problem for a long time so I went through the basics of tuning up, lubricating moving parts, etc. The problem remained. I started printing test blocks and noticed the problem looked very regular.
This made me think it must be in the software tool chain and I spent a couple days trying to narrow it down. Nothing.
Finally, I took another look at hardware, even going so far as to disassemble things that have been working reliably for a while. Lo and behold, I found this:
These were some old gears that I printed when my quality was still pretty rough. I should have reprinted them a long time ago, but I forgot when they seemed to be reliable.
After you figure out a problem, it always seems so obvious.
I finally broke down and replaced the Gen3 electronics with a RAMPS board and after working through some issues, it finally seems to be working.
I usually print in ABS so I also invested in a Prusa PCB heat bed from Ultimachine. There isn't much information out there on mounting the thing and most of the pictures I saw had people mounting them on a piece of plywood. This is probably perfectly safe but there's something that bothers me about mounting something designed to get hot on top of something flammable.
I had a piece of 12" smooth Hardiboard. This is siding material made from cement infused fiber. It's very flat, very hard and heat resistant. As a test, I held the torch on a scrap for a while to see what the result would be. It scorched the topside but barely penetrated at all.
It's a little heavier than I'd like, but the combination of the hardiboard and PCB is actually lighter than the old acrylic and removable MDF that I was using before.
The downside is this stuff is damn hard to cut and throws off a lot of dust. I milled a couple pockets for the thermistor and the wires so the bed would lay flat. I also cut a few notches to remove material and lighten it further. It seems to be working well.
The big things I wanted out of this build were a stepper driven extruder, 5D firmware, a Makergear hot-end and PLA printing.
Next steps involve a better build platform, debugging the z axis sticking and a bunch of testing.
I listen to a lot of audio books from my Sony Walkman (NWZ-E436F). I have an RF modulator so I can play them through the audio system in the truck. The only problem is the thing slides around and gets lost on the floor.
This is really the perfect kind of problem for a makerbot to solve. The pieces are simple and tolerances are forgiving. I could improve the design of the stem but recessing the bolt that holds the two parts of the stem together.
It seems like every magazine and blog I read lately is filled with reprap stories. Make magazine had a Makerbot Cupcake CNC on the cover of the previous issue and CBS even did a short profile. It's time to find out what all the fuss is about.
I ordered a Cupcake a couple weeks ago and it arrived on Friday. My truck was on the fritz and the family was away so I didn't want to stray too far from the homestead. This was a perfect distraction.
So far the build has gone together very well. It goes together like something from IKEA but the tolerances are really very good considering it's made of wood. The T-nut design and the fingers of the laser cut panels make it surprisingly rigid.
I haven't gotten past the mechanical construction yet, but here's a couple pictures.
Without a doubt, I can already tell that the "hot end" is going to be the weak link. It's the white and yellow bit in the middle of the extruder assembly below. Compared to the rest of the build, this felt very DIY and kinda kludgy. I've seen a few blogs and posts of people coming up with improvements.