Dane.Kouttron

[01.10.22]  Replicator 2 Revival

Did you inherit a crusty old maker-bot replicator 2 from a lab-clean out and are frustrated with how finicky it is to reliably print?
 
Get things working again and bring the machine back to life, with some details below.


What?

Hardware Repair

Conclusion Image Directory

Project Background
What is a replicator 2? 
The Rep 2 is a nearly 10 year old model printer at the time of this writing. Brand new this thing retailed for $2500 [1]. There were a few variants, a dual extruder model, and a heated bed model. The design became the state of the art for consumer-level printers and sparked a number of nearly direct clones.

Did you know you can literally buy a 2012 makerbot replicator clone in 2022? The Flashforge creator [link] is very close to being a modern duplicate, I believe the control electronics and main board are different but the mechanicals are so close that its hard not to notice. There are some improvements that I really like, namely replacing the membrane push buttons with tactile push buttons. Even the cut-outs for the filament spool holders are in the same spot. A colleague purchased one of these and I used it for a few weeks, its surprisingly good.
Keep in mind, at one point the replicator 2 was a reasonable printer, while there are a number of improvements, add-ons and gadgets to augment it, lets focus on getting the stock contraption running reliably. Its actually possible to have a makerbot that prints and isn't miserable, some of the keys to doing that are setting realistic expectations and not using the godforsaken makerbot printer slicer. Lets aim for a reliable 50mm/s print speed at 0.02mm layer height and work from there.
Dane why are you spending time waking up an old machine, you can get a replicator clone for $300.
I'm interested in printing out some parts using odd filaments and having a printer just for non-normal filaments lets you keep your useful printers working. Specifically some of the curious 'bronze' FDM filaments from the virtual-foundry folk are really interesting. Coupled with a tiny kiln and some associated hardware you can get some reasonable metal prints, which is incredibly exciting. Yes you can CNC parts but, weird elaborate shapes can get opened up with multi-material kiln-able prints

Step 0: Gather tools, Determine what works & what doesn't work
You have a pile of questionable hardware lets sort out what works and what doesn't
Make your life easier, get an old SD card

The Replicator 2's for whatever reason work better with old 2GB SD cards. Check that old derpy HP camera that ran on AA batteries, or that dubious digital picture frame. Yes that one. You can even use old MMC cards. We're going for simple 'almost reliable' functionality to get there we're starting from first principles. Do you know what the read-rate is for a 3d printer? 2MB/hour, yes an old card will work totally fine. For testing I'm using a 64MB MMC SD Card.
3D Printer Maintenance kit for <40$
Interestingly most printers topologically have similar parts internally, so for maintenance there's a lot of crossover on tools, here's my goto kit:
  • 0.012" stainless steel wire [mcmaster][amazon] [$8.90]
  • 0.4mm replacement brass nozzles [mcmaster][amazon]
  • 3d printer extruder cleaning filament 1' sections [mcmaster] [amazon]
  • Grease for lead-screws and linear rails [mcmaster][amazon]
  • Digital meat thermometer rated to 300C [amazon]
  • A good directional flashlight
    • This is remarkably important for watching the first layer go down, as well as sorting out bed leveling.

Fixing the push-buttons with conductive paint

Its almost a guarantee that the front-panel push-button is showing its age. This is a membrane-style push-button where the rubbery bit has a conductive paint on the back and mushes against a PCB to simulate push-button actuation. The conductive paint wears down and the front panels become useless. I found this stuff [link] and it works phenomenally well. What does work really well is to fondle the canister before use, it somewhat helps the paint mix if it has separated. After fondling in a warm environment, set aside a dollop on a piece of paper, use a matchstick end to transfer some of the paint from the dollop to the membrane push-buttons. `

A look at the push-button backside

Here's the push-button membrane thing removed from the chassis, to remove it you can just squeeze it and pull it out, its only really held in by its geometry. You don't actually need a lot of conductive paint to cover the button pads and for whatever reason the back of a matchstick works really well for applying. They do recommend making multiple small coating
A little dab will do


I just wanted to be clear, you really do not need much conductive paint to fix these pads. The dry time lists ~45 minutes, but I found its reasonable to let it cure for 2+ hrs before installing. If you do get some of the paint outside of the circular area, just wait until dry and scrape it off. A small scalpel knife works well for this.
Better than conductive paint, actual push-buttons

Conductive paint is great, but it is a wear item. You will likely re-paint those pads in a year or so. I was thinking 'wait there are small SMT buttons', so I found a relatively short height momentary push-button with the goal of soldering it on-board and having it sit in between the membrane thing and the board. Given that these are only 1.5mm tall this actually works reasonably well.
Alternate-Alternate Push-button fix

If your membrane push-button thing no longer exists, don't fret, somewhat larger adorable surface-mount push-buttons are a thing, these are incredibly small but do a reasonably good job at replacing the membrane thing completely. These particular surface mount push-buttons are from here [link] but the only important dimension is the pin spacing. These do have a bit of a tall 'stance' so try and provide enough solder for these to be supported structurally.
Getting some basic baseline filament

It is super tempting to grab the dusty 3/4 used up spool from 2 yrs ago, or some ABS you never got around to using, but we're trying to get some bench-marking here. I highly recommend the Dremel 1.75mm PLA. Its not anything particularly spectacular its just incredibly identical batch to batch and prints remarkably reliably. Its even available on mcmaster-carr for $30. All the benchmarking here is with their 1.75mm blue PLA.

Is the temperature sensor reporting real temperature?

One issue I've observed was the reported extruder temperature sensor and real life do not match at all. They can be comically wrong. This is a pretty big hurdle as plastic extrusion is incredibly temperature dependent. Fortunately even off-brand multi-meters feature thermocouple inputs, and in a pinch, a digital meat thermometer does go to 300C. Actually, a digital meat thermometer should be excellent at this, time to add it to the '3d printer maintenance kit'. I wouldn't use this to verify absolute temperature but '+/-5C' should be reasonable. With the tip of the thermometer in the top of a cup of boiling water it reads 99C, so its likely this is reasonably accurate.
Yeah the temperature is incredibly off

So the first test of the meat-thermometer and the makerbot reported temperature is really off. Shown in this poorly lit picture, a set-point of 210C and a read temperature of 168c. Whats properly bizarre is thermocouples shouldn't have offsets. Thermocouples topologically are just two wires of dissimilar metal welded together, they produce an offset voltage predictably with temperature. Its possible the thermocouple amp is failing? Unclear. As a result of this, an extrusion temperature of 200C now requires a print temperature of 240 C to compensate for this offset. Long term the actual issue should be resolved.



The Four Tenets of 3Dprinting: Filament, Temperature, Adhesion and Speed
Spark, Air, Fuel, Compression are the four tenets of small engine repair. If you have all four the engine should spin, it may not spin stellarly but it will indeed function. We're going for function first, so lets check each of these to make sure we're not missing something basic: Filament, Temperature, Adhesion and Speed.
Filament
If you have a questionable gasoline powered lawnmower, remove variables and test it on new fuel, just as such use new filament when testing a questionable printer. PLA is hands-down the easiest printing experience, it is incredibly forgiving. Good PLA is also a good experience.
Temperature
As mentioned earlier its not uncommon for temperature measurement to be incredibly off on surplus / for parts printers. You do not need elaborate test gear to verify extrusion temperature, you can get away with a simple digital meat thermometer for a first order approximation. Again PLA is fairly forgiving, having +/- 4C is not the end of the world. The easiest route to take here is to use the built in menu system.
Adhesion

Bed adhesion is somewhat a topic of contention, fortunately that arena has evolved so many times over the past few years that there are some great options. PLA on a glass heated bed is just so frigging good. I cant believe it took me so long to try this but it's great if you mostly print PLA. Glass has a fairly high coefficient of thermal expansion / contraction, which normally wouldn't be great but in this application is wonderful, you warm up the bed to 60C, print then let the bed cool down. While the bed is at 60C and clean of dust / oils, PLA adheres wonderfully, when the bed cools down to ambient the parts reliably dislodge. No chiseling with a screwdriver or messing about with bed misalignment every time you need to whack a part off the build surface.

I'm going to assume the bed's are made of plate glass, with a coef. of expansion of 9 x10^-6  (m/moC)  but just the linear expansion is shown as:
dl = L0 α (t1 - t0
where
dl = change in object length (m)

L0 = initial length of object (m)
α =  linear expansion coefficient (m/moC) 
t0 = initial temperature (oC)
t1 = final temperature (oC)
So on a 1ft (0.3m) square print surface: 0.3m * (9x10^-6)*(60C-20C) = 0.000108 meters, or 0.0108mm in either direction when the plate warms up / cools down. This is not huge but its enough for the part to 'pop off' when the base cools down. 60C is also not hot enough for the PLA to warp or mechanically change too much.

For the makerbot replicator 2, there are some heated bed options, but the ones I inherited are non-heated so I'm opting for the fairly tried and true 'blue painters tape' approach. I've found that using a wire brush on the blue painters tape helps to add some surface area and increase its bonding.

Speed

Doubling your print speed does not halve your print time. This may seem counter-intuitive but the velocity set-point is still governed by the system's maximum acceleration. We can roughly confirm this with just 'simulation'. Take a simple model (a calibration cube) and get slicer time estimations for a variety of speeds, then plot. While slicer time estimation vs reality has its errors, we can take it for a reasonable approximation. This plot shows a lot of information, and somewhat curiously shows why most FDM 3d printers end up with a realistic max operating speed of 100mm/s.  You simply run into decreasing marginal utility, somewhat bounded by machine acceleration, but more likely bounded by how much the extruder can extrude. Print at 100mm/s and it takes 40 minutes, print at 400mm/s and it takes... 24minutes, a lot of extra work for not a huge benefit.
Plastic is also just a poor thermal conductor and the faster you attempt to print, the shorter the available time there is to warm up the filament. This plot assumes your extruder can keep up, but generally for extrusions > 120mm/s you're pushing it on a pinch-gear extruder.

This is somewhat backed up by the default print speeds on popular printers:
  • Prusa I3 MK3S+:         85 mm/s
  • Creality Ender:          70mm/s
  • Flashforge Creator 2:  80mm/s







Step 1: Trying some pre-made test prints

I made you some test prints that you can copy directly onto the SD card, no slicer no nothing, just put on an SD card and go.

 Actually doubling speed does not halve print time

Its fairly easy to miss this one, but increasing your print speed does not linearly decrease print time. Here's a plot of data collected on the makerbot replicator printing fairly straightforward 35mm cube calibration cubes. You can see between 25 and 35mm/s there's actually a linear relationship, but after 35mm/s the selected velocity is bounded by the machine's maximum acceleration. As you increase print speed you also increase the extruder's back pressure, this is a potential cause for jamming. The makerbot extruder is not known for being all that great. It does work, but it has its limitations, if you can print slow you are literally trading time for likely hood of print completing successfully.


Step 2: Slicing with CURA
Don't you dare use makerbot studio, it somehow does a fairly bad job at generating reasonable print paths and causes more headaches than it solves.




Slicer time is very enthusiastic

During the test prints I logged actual print time versus slicer estimated time, and like anything there's some deviation. Interesting the slicer is incredibly accurate at estimating print time at ~35mm/s printing. This however falls off pretty hard as we hit 100mm/s. Part of this could be due to a number of things, the acceleration that the machine

Low Effort Upgrades
A UPS

Now that things work, whats the low hanging fruit to get more reliable results?  The makerbot uses a switch-mode supply generating 24v feeding everything on-board. That switch-mode supply can survive ~2 cycles of downtime before the printer starts to misbehave, as such you actually don't need a fancy UPS to keep things working. Almost all basic office UPS's are a dumb relay, they normally have mains connected directly from the wall to the output, on loss of mains (generally <90vac in the states) they wake up a very basic square-wave 60hz inverter and switch the relay to use that as the power source. I have a recycled APC XS 1500 that works in exactly this fashion. A UPS gets you out of the issue where your 9 hour print was affected by a momentary power blip as everyone turned on their stoves to go make thanksgiving dinner, etc.

The peak power of a single extruded replicator 2 without a heated bed (the type I'm testing here) is 150W, so even the super basic single 12V7 battery style UPS can handle that.  I haven't observed any issues using a dumb ups (square wave output) or modified sine wave output, you do not need a sinusoidal output ups.


Did you notice the lack of advertisements or Facebook nonsense? Consider sending a cup of coffee :]




(There's other photos in the photo gallery)
Concluding Remarks:

If you have questions or comments, ask below or send over an email.
Comments:
HTML Comment Box is loading comments...

Pack your homemade drone batteries safely, less they become inadvertent lithium hand-warmers.

Dane.Kouttron
Rensselaer Polytechnic Institute 
Electrical & Electrical Power
631.978.1650