PrintrBot Metal Plus – Heated Bed

Posted: 12th April 2015 by Quinn Dunki in Hacks

Stand back- I’m going to try SCIENCE!

 

My PrintrBot has been chugging merrily away for me, and while it works, the preparation of the bed before each print is a real hassle. The solution is a heated bed, and it was high time to install mine.

 

..

Once again, the PrintBot kit is nicely packed.

 

PrintrBot’s instructions for installing this kit are pretty good, but as you might expect, I ran into a few gotchas.

 

..

Overall, the process is pretty straightforward. Unscrew the bed from the machine, screw the heater to the underside, and reassemble.

 

..

The kit includes this insulation layer, which is held in place by the heater mounting screws. I found it valuable to get this insulation taut. Otherwise it can drag on the bed runners once reinstalled. Pull the foil sheet tight as you tighten the screws. Also hold the sheet tight while making the screw holes (by poking the foil with a sharp knife).

 

..

This shot shows how great the Metal Plus’ design is. The bed comes off without disturbing the drive-rail system or belts, and the bottom comes off without disturbing any other sheet metal.

 

..

Now it’s time for wiring.

 

It’s important to note that the heated bed requires a lot of juice, and as such the PrintrBot’s commodity laptop power supply isn’t enough. It requires an ATX power supply. Any ATX supply with PCIE support will work. PrintrBot sells one, if you want to guarantee getting the right thing. No offense to PrintrBot, but Amazon sells equivalent or better ATX supplies for about a third of the price with free shipping. I opted for the latter.

With all the parts in hand, refer to PrintrBot’s document on how to wire up the power supply. It can be tricky to identify the correct connector amidst the bewildering array of them on an ATX supply, but a voltmeter verified it. The PrintrBot wants 12V, so whichever connector fits that and delivers 12V is the one. Note that the connectors are modular, so you may have to separate a few to find the right one. The PrintrBot’s control board has a keyed molex connector on it just like the power supply, so there’s no hackery or guessing involved. There’s only one connector that will fit. Still- verify with a voltmeter to protect your investment. I was never a build-your-own-PC person, so ATX power supplies are pretty new to me. Apparently the yellow wires are +12V and the red are +5V, but I never trust wire colors.

 

..

Next, the bed needs to be prepped for the Kapton tape. Mine had dried up wood glue gunk on it from my previous system of “wiping glue on painter’s tape” to achieve adhesion.

 

The final step is to apply the large square of Kapton tape to the bed. Here’s where the instructions let me down, and where I wish I had done a bit more homework. The instructions fail to warn how tricky it is to get this tape down without bubbles. If you get bubbles, they are very hard to remove. There are good techniques out there which I wish I had known about. If, like me, you get bubbles, they can be worked out with a plastic scraper and a lot of patience. A few small bubbles don’t seem to hurt too much, but it’s certainly better not to have them.

 

..

Here’s my final result, before bubble removal attempts. I did get most of the worst of them out, but it wasn’t fun.

 

Now that the machine is running off an ATX, it seems silly for my Raspberry Pi print server (yay Octoprint!) to be running off its own USB power supply in a separate outlet. I poked around on the ATX with the multimeter, and found a convenient connector that delivers 5V for the Pi. All I needed was an adapter cable.

 

..

I found this weird Type-A-Female-To-Micro-B-Male adapter on the junk pile. No idea what it’s for, but it’s about to get a new lease on life as an ATX-to-Raspberry-Pi power cable.

 

Once I had verified all the pins and wires lined up the way I expected (say it with me: never assume the wire colors are right), I grabbed a 0.1″ header and some heat shrink to make a little cable.

 

..

The white connector is a small one on the ATX supply that was conveniently compatible with 0.1″ headers, which I have lots of. I do love any excuse to use heat shrink tubing.

 

..

Presto, one ridiculously over-powered Raspberry Pi. Again, never having been exposed much to PC guts, I have no idea what in Homer’s name all those other cables and connectors are for. I expected two or three, but an ATX power supply has closer to 30, if you include all the ways the modular ones separate.

 

One oddity about the PrintrBot Metal Plus is that is has no power switch. I don’t like plugging things in and unplugging them to turn them off. Luckily, the ATX supply makes this easy. There’s a particular jumper wire required to make the supply turn on (as shown in the PrintrBot instructions), so it was easy enough to put in a switch there instead. Maybe that’s what the jumpered pins are intended for? No doubt all the custom PC people are screaming at their web browsers right now, enraged by my ignorance. Anyhoo, the junk pile produced a rocker switch from who-knows-where (an 80’s coffee pot, maybe?), and I wired it up lickety split. One caveat to this- since the Raspberry Pi is now powered from the ATX, turning off the switch kills the power to it as well. Technically, of course, you’re not suppose to kill power to a Un*x machine without shutting it down in software first. In practice, for something as simple as this dedicated Octoprint setup, it doesn’t hurt anything. If somehow a file gets corrupted from improper shutdown, it’s a simple matter to image the SD card again. There’s no critical data on the Pi.

The heated bed is pretty awesome, I have to say. It immediately and noticeably improved my prints, and not having to do any bed prep is awesome. I did find that I needed to tweak my calibration a little bit. Well, more than a little if I’m being honest. This is likely a combination of the heated bed and the fact that I hadn’t yet tried to print anything really challenging. My spool holder was pretty substantial, but it did end up with some warping from poor adhesion to the blue tape.

At first, my heated bed prints seemed to be coming out great. The small test box looked fantastic, with good precision, smooth sides, and sharp corners.

..

Overall precision improved right out of the gate. This is about 0.2mm closer to spec than before the heated bed.

 

I then tried to print something genuinely large for the first time, and here’s where things got interesting. First things first, however. I decided to try making a case to house the blob of ATX-Raspberry-Pi-Switches-Etc that was now hanging off the side of my machine.

I designed a box that would sit next to the printer, and hold everything nice and tidy. Finding a model of the Raspberry Pi Model B (the one I happen to have) to use for designing the mounts wasn’t easy. Other Raspberry Pi models are plentiful. You can’t throw a rock at the internet without hitting a model of the B+, for example. Not so for the lowly B.  I did finally find one, and unfortunately I’ve misplaced the link to it, so I’m unable to give credit where it is due. Thank you, anonymous donor! Allow me to pay it forward here.

 

..

There’s holes for the fan and ventilation slots on the power supply, a compartment with brackets to hold the Pi, a hole for the switch, room for routing wires, and the whole thing prints with minimal support (the fan hole is against the print bed).

 

A print this large (it occupies most of the 10x10x10″ Metal Plus build volume) takes serious time to print. About 2-3 days, in fact. For that kind of commitment, I needed to know for sure the dimensions were right. To that end, I started by chopping up the model to isolate the areas where fitment was critical. Meshmixer is a great tool for this. Each of these little pieces could be printed in 30-60 minutes, so tweaking dimensions was easy.

 

..

The large print volume is a big help here as well. Here I’m printing the Pi cradle, the switch hole, and the slot/tab for the side cover all at once. These are all tricky fitment areas, so it was helpful to validate my dimensions.

 

..

Here’s a section of the base, which validates that the ATX will fit, and the power connector hole will line up.

 

With all my dimensions validated and tweaked as needed, I pulled the trigger on the massive multi-day print. The large round hole for the fan is face-down on the print bed, so it starts by printing a very large flat surface. It was clear right away that this was not going to succeed. In a print this large, I could see my quality was not where it needed to be. The surface was stringy, clumpy, and pulling apart in different areas. No good. I hadn’t done any serious calibration of my print settings, but it was clear I now needed to. How would I know what to tweak though? There are a lot of variables, and I didn’t know what effect they might all have. Clearly, this is a job for science!

 

..

Here’s a macro shot of the problems. The strands are not sticking to the bed consistently, and not sticking to each other in many areas. That causes loops, bubbles, and other chunkiness. The extruder head even snags on these bubbles sometimes, obliterating the entire print in short order.

 

I set out to create a series of small tests, controlling one variable at a time, and carefully logging all the results. I experimentally controlled Z-home height, head temperature, bed temperature, and flow rate. These seem to be the variables that most affect print quality and bed adhesion, based on googling various kinds of 3D printing problems.

 

..

Here are nine test prints of a single layer, carefully tweaking one or two things at a time. The bed was cleaned with acetone in between each, to eliminate that as a variable. After each test, I pulled up the print and assigned a rating for how well adhered it was.

 

I didn’t fully explore the N-space matrix of combinations for all these options, because I started to see patterns pretty quickly. Quality and adhesion were clearly improved by going up a bit on my head temperature, and starting the first layer a little low (effectively “squishing” it into the bed a bit). The latter is a tip I had read in a few places, and was suggested to me by a Blondihacks reader in an earlier post. Thanks for that!

 

..

Here’s the result of printing at 220°C (10°C above recommended), with a 70°C bed, and a Z home value of -0.05. The strands are tight, well adhered to each other, and there are no bubbles. The splits you see are damage caused when the sample was pulled off the bed.

 

Now that my first layer was looking great, it was time to print a small test box again.

 

..

On real prints, the results were less acceptable. The box on the left, though hard to see in the photo, is basically a gooey mess. The box on the right is after making the adjustment described below.

 

What I found was that my 220°C head temperature was too high for the later parts of the print, where the plastic needs to bridge gaps and cool quicker to stay in place. I also found that my prints were cemented to the bed so firmly that I had great difficulty removing them without damage when finished. My tuning had worked too well, in a sense.

A little research revealed that a common trick in 3D printing is to use different settings for the first layer or two, then change them as the print head gets higher. Cura comes with a plugin for just this purpose, in fact. I adjusted it so that the head would start at 220°C, and the bed at 70°C, then drop to 210°C and 60°C respectively after the first couple of layers. This worked really well, and resulted in a very high quality test box.

 

..

These settings produced my highest quality and highest precision test box yet. Just 0.07mm off spec!

 

All this tuning has had good results, but has also reminded me that these machines are a hobby, not an appliance. Stuff like this is going to have to get sorted before they become something a person can buy at Walmart, plug in, and start making widgets. They are roughly at the point home computers were in the 1980s- useful, but requiring a level of dedication and knowledge from the user that most people won’t have.

I felt like finally I might be ready to print this ATX box in one go, so I spooled up the black filament (to match the printer, of course), and kicked it off. It was tough to sleep that night, both because of the noise in the other room, and also because I had never really left the machine this substantially unattended before. I checked it before bed, and it was cruising along. I checked at 2am (I’m a restless sleeper) and it was still chugging along. Then I woke up to this…

 

..

The technical term for this is “Spectacular Fail”.

 

At some point during the night, something went awry, and bedlam resulted. It took some time to clean up the machine after that, I can tell you. From the aftermath, it appears as though a layer half-way up delaminated, causing the upper portion to warp, and impact the head. After that, all bets were off. It’s hard to say what caused this- possibly the colder room at night. I also realized that my bed was about 1mm out of level. Not enough to affect small prints, but as they get taller, this matters more and more. This might have caused an upper layer to get poorly deposited, causing it to separate and warp. Who knows.

I debated adjusting my bed and trying again, but in the interest of preserving my sanity (and getting some sleep), I opted instead to break the model up into pieces were are easier to print and can be assembled after the fact with cyanoacrylate glue. The other nice thing about breaking it up like this is that I can print the pieces in batches, as I have time, instead of having to commit to a massive two or three day job.

 

..

I used 123D Design’s Solid Split tool to break up the model, and then added tabs to help glue together the sections. The sections are small enough to print two or three at a time on the Metal Plus.

 

..

Big flat pieces are easy to print, so this dry-fit went well.

 

..

A quick glue-up with cyanoacrylate, and we’re ready to go.

 

..

Thanks to all my test prints, I was confident everything would fit. Still, it’s exciting and a bit of relief when everything actually does. You can see the Raspberry Pi securely tucked in up above, and the power switch on top, towards the back.

 

..

As hoped, the case is a nice fit against the side of the printer.

 

The final power supply/server box turned out pretty nice, if I do say so myself.  The only thing I might do differently is that it would be nice for the power outlet to be on the back. However with the way the vents and cabling are arranged on an ATX supply, there wasn’t any way to orient it that also worked with where it needs to sit on the printer. A minor quibble.

Overall, I’m very happy with this heated bed upgrade. It has made the printer so much easier and more pleasant to use that, frankly, I wouldn’t recommend anyone buy a printer without one. It has also enabled larger projects such as this ATX box. My printer is now, I think(?) finally at a point where I can start using it to solve real problems (for certain values of “real”).

 

UPDATE 4/26/15: I’ve posted all the files for this Print Server project here on Thingiverse. Enjoy!