Disclaimer (October 2016):
I started writing this article, but never finished it. I got sidetracked, or distracted by something shiny, and never finished and published it. For that, I apologize. However, there is quite a bit of important information in this unfinished article. Unfinished or not, it needs to be seen. So here it is, in the RAW form.
Original draft (October 2013):
In a previous blog post, I talked about a chamber that I am making for my 3-D printer. This post is a follow-up, discussing what went right, and what went wrong.
One important thing that went right is the overall concept. I am super happy with how it ended up. It was a lot more work than I was hoping for, but oh well, it was worth it.
The spool holder looks nice, and it is, but it only holds 3 spools comfortably. With more than 3 spools, you can’t slide them back and forth enough to align any one of them with the hole for the filament. On the other hand, I switched to 5 pound spools and it is nice to have that much filament on hand.
One major problem was with the filament feed tube. The amount of force required to push/pull filament through that tube was much greater than I had anticipated. Part of this is due to how the feed tube is bent (vs. how it is done in a standard Replicator 2), but I think this is still a problem with the standard configuration. I ended up designing a system that uses a second stepper motor to push filament, on demand, down the tube to the extruder. There is now a sensor on the extruder that detects when filament is needed, and activates the stepper motor.
The filament feed stepper motor is controlled by an ARM Cortex-M4 microcontroller. Yes, it’s overkill. But the board costs US$13, so who am I to complain. Added to that is a stepper motor driver and some power supplies. Currently the programming is very simple, but it does detect if too much filament is fed in too short of a time. When that happens, it stops feeding filament and blinks some LEDs. This saved my rear end several times. I don’t want to see a 5 pound spool of filament filling the printer chamber.
I am 90% convinced that when people add a little oil to the filament it is for the feed tube and not the extruder itself. Adding oil to the tube greatly reduced the force required to get the filament through the tube. Before the extra filament motor, this was the only way to get reliable printing. Even after adding the motor, oil is still highly recommended.
The feed tube is made out of Polyethylene, or PE for short. PE is a difficult material to work with, since it is slippery enough that friction is not enough to hold it in place, and almost every glue does not stick to it. (But not slick enough to make the filament slide easily!) I was constantly having issues with the tube slipping out of the new feed motor block. I finally fixed this by heating a small section of tube with a hot-air solder station (like a tiny heat gun) and then pushing the tube together. This caused a nice ring-like bulge around the tube. The feed motor block was then designed to hold this bulge in place. The end result is a very secure tube attachment.
Update (Oct 2016): There are Teflon tubes available that might work better since it is more slippery than the PE. But the difficulty with Teflon is in attaching it. The heat-and-melt approach didn’t work, mostly because Teflon melts at a higher temp and gets somewhat toxic at those temps. I’m still using the PE tube, and add some oil about every hundred hours of printing.
Originally the new stepper motor was mounted on top of the chamber, outside of the soundproofing. That didn’t work so well, because you could hear the motor when it was feeding. I then mounted it inside the chamber, but screwed directly into the plywood top of the box. That was about the same loudness as before, since the plywood was being used as a sound board, magnifying the sounds. The third try had the stepper motor mounted to the already existing t-slots that was holding the lights and fans. These T-Slots were hanging from the top of the box with elastic cords. This is the key. The elastic does not allow the vibrations of the motor to transfer to the plywood. The result is very quiet. It’s not perfect, since some sound travels up the filament and out into the room. But the room has to be very quiet to actually hear it.
Update (Oct 2016): I removed the elastic cords and mounted the T-Slots to the plywood top with some 3-D Printed brackets. This is more secure and didn’t change sound leaking through. It’s not a lot of sound, so it doesn’t bother me.
Another source of noise was through vibrations of the printer itself working through the base of the chamber. Originally the printer was on a plywood platform that was itself supported using about 3 inches of shredded rubber from tires (called rubber mulch at the local hardware store). The result was that the rubber was both not elastic enough (vibrations were transferring through) and yet too elastic (printer was shaking too much as the extruder moved around).
The shredded rubber was in a tray made from 1×4 inch pine boards on the sides, and 1/8th inch thick hard-board on the bottom. I removed the rubber and replaced it with about 65 pounds of concrete. The wood tray was sealed with silicone caulk before adding the wet concrete. The concrete tray has some cut up mouse pad as feet. Above the concrete is some more cut up mouse pad feet and the plywood platform. This worked really well, although if the room is very quiet you can hear the printer.
Temperature control was another big issue. Before I did anything to specifically address it, the temperature would start out around 72F and over the span of 4 or 5 hours would rise to over 110F.
I first addressed the heat by getting an aluminum extruder upgrade. Then I used thermally conductive epoxy to attach heatsinks to the extruder, Y-Axis motors, and the new filament feed motor. There is a small 40 mm fan on the feed motor, as well. At some point I will add aluminum arms to the bot, but it seems to be working fine for now. The goal of these changes were to help the bot handle the high temperatures.
Next, I did changes to actually reduce the temperature inside the chamber. There were already two 80 mm fans inside to circulate air, but I added a third one that blows air across the glass door. Mounted on the outside of the door is a quiet 120 mm fan, also blowing across the glass. The two fans, inside and outside the glass, help to transfer the heat outside. The fan on the outside normally runs at 12v, but I am running it at 5v. This results in lower airflow, but it is much quieter.
With these changes, the printer can easily print with temps up to 110F, but the actual temp hovers closer to 102F.
The next temperature issue was the low temps. It takes this chamber a long time to warm up. Many hours. Adding 65 pounds of concrete makes this take even longer. My goal was to have a consistent but high-ish temperature to reduce warping and other things. Having the chamber start out at 72F and slowly rise was not the plan.
I fixed this by adding a heater and temperature controller. The heater is basically a heat sink with some power resistors and a fan attached. It dissipates about 76 watts, which is intentionally on the weak side. It takes a long time to heat up, but it doesn’t create hot spots inside the chamber. Plus, it is easier to just keep the heater on 24/7 since it keeps large temperature swings from throwing the platform out of level. I am using an industrial temperature controller and solid state relay to control the heater. This is able to keep the chamber within 1 deg F of the desired temperature.
The desired temperature is in the 95 to 100 deg F range. When the printer does it’s thing, the temp will rise to about 102F. The end result is a consistent temperature!
Update (Oct 2016): I keep it at 101F, 24 hours a day, 7 days a week. Because of the insulation, it doesn’t take that much power to keep it at temp. I don’t dare run it hotter because the printer electronics can’t take much more, and there are a lot of PLA parts that might start getting soft.
So far, everything is working nicely. I have done several 12+ hour long prints, and many 8 hour long prints. Obviously there were some failed prints along the way, but I feel comfortable doing large and long prints on this thing.
One nice side effect of everything is that I can spray the glass build platform with hairspray and place it inside the chamber to dry. Between the heat and the many fans, the platform dries in minutes! I have another 80mm fan mounted on the T-Slots on top of the chamber (just under the filament spools). If I need to dry a build platform while I have another print going then I can use that fan to help dry the plate– it just doesn’t get pre-heated at the same time.
Future changes that I see are to better address the temperature. Currently the lights add about 12 watts of heat, and are always turned on. These are RGB LEDs, so I wanted to add some color control anyway. But if they were turned off when not needed then the chamber temp could be reduced to below 100F (or work more reliably in the summer). Also, adding some temperature control to the door fans would reduce outside noise.
Update (Oct 2016): My printer has been running in this configuration for quite a while now; over 5,000 hours on the odometer. I get amazing and reliable results. I am currently working on more aluminum upgrades to the Replicator 2 printer, including aluminum carriage, aluminum arms, and a heated build platform. The heated platform is going to cause a problem with cooling, and so I am working on a new cooling system to augment the heater. The heated platform isn’t for ABS, but rather to print PETG, so I won’t have to run it that much hotter than I already do.