316 Photos - Feb 5, 2013
Photo: The KAOS Lab's new Makergear M2 3D printer arrived (actually, a couple days ago, but it didn't go to the office we expected).  Ordered fully assembled in early November.Photo: It is _incredibly_ well packed.  Everything is bagged, then locked into place with foam sheets, then infilled with peanunts.  I (ecology and mess) hate that kind of peanut, but no harm was going to come to the packed machine.Photo: Comes with _everything_ needed to get started (except a 7mm wrench for fiddling with the Z stop).
Spool of PLA, big roll of Kapton, very nice US-made ball-end Bondhus hex drivers and a ton of spare parts.  Also includes solid documentation.Photo: Photo: Photo: Two calibration prints (sasquach head and bracelet)  in the box from the factory test.  Can still see the red drip on the nozzle from production.Photo: Photo: The M2.  
Hardened steel rails and GT2 belts on X and Y, trapezoidal thread (I didn't gauge it, it's 4-start but I don't know if it's metric trapezoidal or Acme) and 2 10mm rods with linear bearings on Z.  Steel frame, machined aluminum brackets, borosilicate glass bed.  
Very, very well made. 

I have to chuckle that with all the custom machining, binder clips are still the best bed hold-down.Photo: Photo: That is some serious linear motion hardware.  Not entirely sure what was being pointed at, it is near the homing switches (which are only on one end of each axis, not end-stops) and the awesome LED-on-contact indicator integrated into the bracket.Photo: The parts bag.  Big roll of Kapton tape, feed tube, spare sleeves for the hot-end, bag of spare hardware, and the mysterious "Glider print HW" bag.  Turns out, there is a many-hour pre-sliced glider model (http://www.thingiverse.com/thing:22268 ) on the SD card that ships in the RAMBO board, and those are the nose weight.Photo: Wiggling it around from my laptop.  The brick peeking out behind the screen is one of the two PSUs, a massive 12V 18A Dell power brick.
Around this time, we learned that the geometry and drive means that if you cut motor power, the bed gently glides away to safety under gravity.Photo: One layer of the glider after some mucking about.  The first pass crashed the hot-end into the clamps, sprayed some plastic past the end of the bed (feature for priming the head?), and revealed that the bed was seriously un-level.  Also, that was _supposed_ to be a small ~1-hr test instead of a ~14-hour marathon, but jobs printed from SD apparently like to resume themselves.Photo: At some point in our explorations, the bed heater cable snagged on the Y bracket while near maximum Y+, pulled the power connector apart, and damaged (pictured) the thermistor connector.  Fixed with soldering iron, hot glue, and a great deal of fucking around to extract the female 43645 crimp terminal from the header.Photo: Photo: Printing the test companion cube.  My tramming job leaves a bit to be desired.Photo: Our first object, a 1" Companion Cube ( http://www.thingiverse.com/thing:13588 ) printed in MakerGear provided Black PLA.  

Slic3r tells me it took 2430.0mm (5.8cm^3) of filament, which are obvious lies and not consistent when multipled by the cross-section of the filament.  The actual time was around 45 minutes (~10 less than estimated, not sure why. 

Came out pretty well, bottom was a tad smushed (my inadequate tramming), the skirt was lightly fused to the print, and the top was a little rough (known to be hard with that model).Photo: Photo: Photo: Photo: Photo: Photo: Photo: Another round of fiddling with the Z stop produced much better results.Photo: Makerslide Endstop Mount ( http://www.thingiverse.com/thing:42469 ), as it came off the printer.  Was set to .2 honeycomb fill, but definately used the alternating-45° linear pattern.Photo: A pair of finished (ish, I did a little more razor work after) Makerslide Endstop Mounts.Photo: The position of the little Z-home adjust screw makes all the difference in the world.  Those are the same sprue, the bad one was stopped because it was terrible, a nice part was cut out of the good one.Photo: Pretty much all the objects printed on our M2 so far - the two factory samples, my endstops, the core for one of those cute square gear things, a scaled-down threaded rocket nose cone ( thread test, mediocre), and a companion cube.Photo: My printed endstop holders fit!  Useful printed parts!Photo: We've been having some problems with doubly oblique corners along an axis distorting, snagging, and pulling the part off the bed. Higher temperatures and only having obliques on diagonals seems to help.Photo: First attempt at making a camera part - lens board for an old 4x5.  
Model made in Openscad.  Pretty proof of concept, but unusable -  The bed heaters came online _after_ the part started printing, so there is some corner lifting, the smaller bore was .2mm too small (error in model), and an offset introduced by the minkowski function for corner radiusing has the hole not centered.Photo: Back of lens board, the seam down the middle is a seam in the kapton tape currently on the bed.  Smooth, but corner lifting due to failing to heat the bed is evident.Photo: Printing with a 10% hexagonal infill.  The yellow tint to the bed is because we decided to try a layer of kapton.Photo: Failed part with 10% hexagonal infill.  I set the Z stop juuuust too low, so it squished the layers and dragged a bit.Photo: TOTORO!
http://www.thingiverse.com/thing:12146

Just finished, with a bit of thread between the ears.

Technically, this is an experiment in radiused parts, relief features, hex infill, and small flanges, and it was effective as such, but really I just wanted one.Photo: Totoro, exactly as it came off the bed, lighted to show relif.Photo: Photo: Totoro's butt is a little rippled, and I don't think it was part of the model.

We've been having trouble with that sort of smearing on the Y- side of parts with increasing radius.Photo: Totoro, after quick post-processing with a hobby knife.  The web in the ears is tenacious stuff.Photo: Underside of a part printed with the Z alignment too low, notice the smear/ripple pattern.Photo: Some more retained examples from failed prints.Photo: The sprues and failures bag from our first week of ownership.  Much smaller than expected.Photo: Trying out some blue Repraper PLA. 
It needs much higher temperatures for bed and head (we're still not quite dialed dialed).
Prints are starting to look good as we get close.Photo: Several kilos of PLA from RepRaper. White, Blue, Glow-In-The-Dark Blue, and several spools of black.
It was cheap and seems to work.Photo: Photo: This is a little copy of the MOG Maze we use for demos.  Something about the slicer or model causes it to make a sloppy underlayer for the floors of the maze before it puts down a full layer.  Mushing the fill in that process has ruined a couple attempts, but it makes a good test for our temperature/cooling rate.Photo: Photo: Printing a smaller MOG maze.Photo: Photo: Photo: Misusing a soldering iron as a heat-sealing device.Photo: Heat-sealed a piece of transparency sheet (PET) to the edge of the maze for ball retention.  Slotted side was using the sharp edge of the iron, the good sides were "brushed."Photo: Balls are 1.5mm BBs stolen from the bearing in a dead 120mm fan in the dead parts pile.Photo: MOG Maze v3.  Intermediate size, thin walls, top facing text in the rim. The slicing was certainly more elegant than the large version, and the print looks clean.Photo: MOG Maze V3 samplePhoto: Several generations of maze.  There is one more the size of the smallest with an embossed "UK" on the back that prints in about 20 minutes one-off (no optimization attempts yet) that will likely be the E-Day giveaway.Photo: Debossing with near-resolution features in the bottom of a print.  Works surprisingly well.Photo: Photo: Photo: Daaaaang.  Look at that extruder grip.Photo: Repraper black is not _nearly_ as black as MakerGear  black.  Testing here with an LED flashlight.

Not sure if we're looking at different dye density, or different dye, or what.Photo: Otherwise identical rim from MakerGear black and RepRaper black. 
Both prints were under temperature for their respective filament, but look at the transmittance. difference.Photo: Trying a two-part recessed lens board for an old camera.Photo: Lens board + barrel assembly, and mount as a separate piece.Photo: Came out pretty well.  Adhesion is not perfect, but not bad.
50% rectilinear fill to try to inhibit light transmission through the not-as-opaque black we had loaded.  We are seeing some transmission through the disc.Photo: Ripple pattern on one side.  I'm guessing bed temperature variation.Photo: Photo: rough assembled with a Graflex Trioptar f/4.5 103mm lens installed.  The shutter hits the wall of the tube (Expected, necessary to get it into the camera body).  The flash sync pins BARELY clear and touch in some positions.Photo: Rear view of previous.Photo: Recessed lensboard installed in a camera - there was some light leakage at the assembly point (fixed with aluminum tape), and it is a _bear_ to get in and out, but it seems to work.Photo: Using honeycomb support to try an a-mount pinhole "lens."Photo: The KAOS 3D Printing table as it stands.Photo: Finished the a-mount print.Photo: Photo: Support material nightmare.Photo: The hexes are _supposed_ to pop off when gripped with pliers.  Sometimes they do.Photo: Overly-attached outside support removed.Photo: The flanges look dimensionally plausible  might need some filing to fit.Photo: Photo: Inside support removed.  This part _didn't_ bond to the walls, but also did bond quite well to the top.  I see the appeal of printing in to materials and dissolving one off.Photo: This is the base of a Raspberry Pi case, but the picture is to note the weird ripple in the surface fill.Photo: A completed Rich Rap Raspberry Super-Pi case with mounting ears ( http://www.thingiverse.com/thing:25363 ) 
It was run at like 205/75 with RepRaper black PLA.Photo: Attempt at the perforated top for the same Raspberry Pi case.  It immediately turned into a gooey mess.

I lowered the bed temperature to 70(first)/55(rest) for the second try, and it didn't bond as well.  Still hunting the goldilocks point for this stock.Photo: Not a failure mode we've seen before.  It _could_ have been another Z homing problem, but I didn't adjust before the next print, and it worked.Photo: Fairly successful perforated top for the Rich Rap Raspberry Super-Pi case with mounting ears ( http://www.thingiverse.com/thing:25363 ) 
It was run on a cooler (55°C) bed and didn't bond as well as we've been seeing.
Maybe we just need to tent the whole damn machine to get better temperature control?Photo: First attempt at popping the case together.  It went and held, but the tab length was a hair off, causing the arching toward the far end.Photo: Other end of case.  My phone's focus is super sloppy.  Note that the surface of the top piece has a lot of inadequate-bonding striation.Photo: Trying to put my B1 RPi into the case, after adjusting the tabs a bit with a hobby knife to fix the arching.Photo: It went.  not beautiful, but functional. The top looks worse than it is in this light.Photo: Something doesn't sit quite right with the board installed.  It latches, but I get a bit of gap on the closure.  I touched it up with a hobby knife, but can't figure out what is binding, it seems to be soft interference.Photo: This is squeezed so the gap is at its worst to the visible side.  Not bad, but something is juuust wrong.Photo: Slightly better picture with my B1 RPi installed after a bit of fitting. 
Not perfect, but not bad.  I am increasingly concerned that it isn't a matter of tuning, as we've tried a wide variety of settings now, the black Repraper filament just sucks.Photo: Printing a one-layer thick  hilbert curve halftoned gray scale image.Photo: The hilbert halftoned image in pronterfacee.Photo: Hilbert halftoned print.Photo: Photo: Manged to lift it from the bed except for a weak spot in the first corner.Photo: Leaving the nozzle sitting hot is not recommended.Photo: Photo: Two differently-seeded layers of hilbert curve halftoning.Photo: Assembly-Line cleaning and populating the little mazes for our E-Day giveaway.Photo: A tray of E-Day giveaway mazes, with a big sibling.Photo: The printer set up to draw attention on E-Day.Photo: Ran off another RPi case during EDay.  We've definitely tuned our settings for the RepRaper filament better.Photo: Apparently hotter bed = better surface.Photo: Easy test for dimensional accuracy: print off some calipers, compare to real ones.  It came out quite accurate, but the scale is smudgy because our layer clearance was too low.Photo: This is an obnoxiously hard print that we've been using for testing.  Huge overhangs (which have never worked for us), print-in-place hinges (which have) and an embedded pivot (which has always fused).Photo: Totoros. On companion Cubes.  With most of a tetrahedral gear set in the back.Photo: Ironing on the bits of transparency to complete mazes.Photo: Scrwless Cube Gear with Pins, core, pins, first two gears installed.Photo: Photo: Photo: Our second screweless cube gear.  This one uses separate pin connectors instead of integrated pins, they seem to work much better.
http://www.thingiverse.com/thing:10483Photo: Trying blue painters tape and lower temperatures for superior grippyness.  Part is a lead tube for a respiration sensor for another faculty member's research.Photo: Tube sitting on harness.Photo: Photo: Glass (Left) vs. Tape (Right).
Glass is smoother, tape has fewer aberrations.
Also, it had to be brought down under 40°C to get the taped one off the bed, and that involved prying, while they just pop off the glass at like 45°C.Photo: And _this_ is what happens when the nozzle picks up the part and starts waving around a molten mass.Photo: Fixing the M2's head clearance and bed level properly.  

.020"=.508mm feeler gauge blade, two 7mm wrenches, a 2.5mm hex wrench, and and hour and a half of my time.  

Still end up making a mess with the layer height way down, but behavior in general seems to be better.Photo: Film carrier for an old medium format.  Printed-in-place springs. Snazzy. (also, printed with 5% fill and hence transmitting a little too much light).Photo: Same film carrier.  Trouble where bridging over the fill didn't work on the inside edge.  Could probably be fixed with a more clever slicer...Photo: Film carrier goes here.  The idea is that we can make backs to fit digital sensors.Photo: Turns out the attachment holes aren't quite aligned (turns out the ones one the camera aren't centered), and there is O(0.1mm) shrinkage in all dimensions, but not bad.Photo: Bed cooling system, made entirely from conference schwag.Photo: Printed a webcam mount for the printer.Photo: New webcam-enabled printer view.Photo: Photo: 1/4 scale totoro!Photo: Not very detailed, but 1/4 scale Totoro actually came out OK.Photo: Thinking with 3D Printers:
"The door stop keeps slipping, so let's just print one with _fucking cleats_"

The cleat size/spacing is even designed to bite well in our carpet.Photo: Seems to work, getting a tiny bit of deformation along the top surface.Photo: Trying one of the little TC logo badges with a .1mm layer height  and the bed zeroed with a 0.25mm gap.Photo: Top surfaces look good, just a bit of threading...Photo: Bottom side is... huh.  I think something in slic3r didn't compensate for the changes caused by the smaller layer height?Photo: I found an exciting model on the Internets...Photo: In principle, I am testing the new .9.9 Slic3r release.Photo: Photo: Photo: Photo: Fire Nation Headpiece!

http://www.thingiverse.com/thing:38369Photo: It looks like it was running a tad hot, the top flowed a bit where the head was lingering.Photo: Print complete.Photo: Fire Nation hairpiece!Photo: Fire Nation Hairpiece!Photo: My hair _might_ be long enough, but I totally don't understand how to put it on.  Going to have to interrogate one of the long-haired friends about the mechanics.Photo: Complete Hairpiece.  Worked incredibly well, except a little bit of rippling on the bottom of the ring, and a few dropped loops at the bottom of the flame where the climb angle was higher than it could handle.Photo: Had our first clog.  Feed gear stripped the hell out of the filament.Photo: With the cooling assembly off, it is obvious _why_ this drive/cold end works so well most of the time.Photo: Dat feed gear.Photo: The thermistor and heater are coded by gender, the two fans on the hot end are not keyed in any way.Photo: Hot end and mount, removed.Photo: Hybrid barrel.  Pretty.Photo: Piece of filament sticking out of the brass parts.Photo: Trying a half scale standard test owl since we've had really good performance after the slic3r upgrade and head clog settled.

http://www.thingiverse.com/thing:18218Photo: In this experiment you will require
One sterilized, standardized, scientificized owl.Photo: That is a fine-ass owl.Photo: Few dropped loops on the bottom of the beak and such, but the nostrils are clear, and it looks good.Photo: Smooth owl backside.Photo: Printed out a couple of the M2 Z-adjust wrenches from http://www.thingiverse.com/thing:53870 since we frequently need them

Blue is 100% and a hair too tight, Black is 103% and a hair loose.Photo: 49mm filter screw thread (.75mm pitch), printed well enough to screw in without a fight on the first try. 

Black RepRaper PLA, .1mm layer height, .2mm for the first layer.  Printed to heated glass bed.Photo: Photo: Printing a 49mm filter, just to test threading.Photo: Photo: Standard Owls, 1/4 scale, left printed at 0.25mm layer height, right printed at 0.05 layer height.  

The 0.05 is doable, but not at all helpful.Photo: Hank got his foster kid a little Nabi Jr. Android tablet.  The first party kick-stands cost $20.  We have a 3D printer. Hence, stand.Photo: Nabi Jr. Stand, alternate angle/position.  It has a 2x4 grid of little square mounting flanges on the back, which makes a single stand really flexible.Photo: Same stand, different position.Photo: And yet another orientation.Photo: Front view of device, and down the mounting groove on the stand.Photo: Both sides of the mounting flange.  There should be an instructable shortly.Photo: Photo: It _looks_ like a lozenge, but it might be... the futurePhoto: Our growing family of forest spirits (that never quite print right).Photo: Printing off a support frame for an imaging project.Photo: Full, Half, and Quarter totoro.Photo: Camera prototype, in stand.Photo: Rear of camera prototype stand.Photo: Friction-fit into slots.Photo: Photo: A-Mount flange, with support.Photo: Photo: Heart Parts! Gears look great, Slic3r botched the pins and core.  Don't really know what went down, it's a beautiful not-the-right-part.Photo: Heart parts! The gears look great, slic3r did something very wrong with the pins and core.Photo: KISSlicer doesn't seem to understand that pausing in place to change temperatures between layers is not OK.Photo: One of Hank's no-overhang, no-support print-in-place hinges.  
http://www.thingiverse.com/thing:83033Photo: Same hinge, stops at very close to 90°Photo: Photo: Same hinge, holding an arbitrary position under friction.Photo: First apparently successful KISSlicer print.  The infill speed is too high, and it had to be done at one temperature because changing is causing a pause-in-place and glob behavior, but it actually looks good.Photo: Correct core and pins for a gear heart.Photo: Glow-in-the-dark gear heart, in phase.

My camera phone had trouble with the low light behavior, but it looks _great_ in a dark room.
Picture/Comments from the adviser and his better camera here: http://www.thingiverse.com/make:38002Photo: Gear heart, out of phase.Photo: The raw filament is very transparent and distorts light as expected.Photo: We got a couple spools of MatterHackers' "Pro" PLA to expand our color palette/hopefully have more consistent filament.Photo: Clear PLA!Photo: Stressful 150% Secret Heart Box ( http://www.thingiverse.com/thing:44579 ).  The (58mm!) spans sagged a little, but had the slicer made better choices even that might have been avoidable.  All the joints went with only a little force.Photo: Heart box, "Locked" position.Photo: Reverse side of heart box.Photo: Worlds fanciest filament cleaner. 
1x Swag lens cloth
1x Long twist-tiePhoto: Hank printed a whole fucking birdhouse in two pieces.  Quarter for scale.Photo: Birdhouse parts.Photo: _Large_ print, partly to flush out contaminants after switching to clear filament.Photo: Birdhouse rear.  Many, many hours of printing.Photo: Print of a Dremel 3/4-12 attachment collar (http://www.thingiverse.com/thing:25062) in clear matterhackers PLA. 
Testing to see if it makes a mechanically usable part.Photo: Dremel Collar seems to work.  Threads are a little crispy going on, but if feels solid.Photo: Dremel collar seems to work from more than one angle.Photo: Penrose Tiles!Photo: They need a little bit of finishing to fit together nicely, but they are good enough  for desk toys.Photo: Canon FD to telescope lenspiece camera adapter, with tripod thread.Photo: Canon FD to telescope lenspiece camera adapter, with tripod thread.Photo: Bird house, 3D printed in three pieces, then fused and painted. 
Currently sitting outside the Marksbury building.Photo: Canon FD to telescope lenspiece camera adapter , with tripod mount. 
Yes, the sensor is much smaller than the lenses.Photo: Penrose tiles! (P2 ,just kites and darts)
Plated to the suggested standard ratio.Photo: One piece car.  The dam axles work (with a little prying/coaxing with an exacto knife)Photo: Underside: wheels and axles printed in place.Photo: The bed side is a little less pretty, but works.Photo: 2.2mm lump and impurity in our 1.75mm filament (this is the RepRaper stuff).  
First problem of that sort we've had, the M2's feed arm trapped it before it got to the head, which was confusing but helpful.Photo: Hank's hinge library is getting pretty impressive.  Folding puzzle box, printed in one piece, generated parametrically.Photo: Hinge box, folded.Photo: M2 printing replacement parts for itself.  The hex nut trap on the original wore out, but MakerGear is a good citizen and provides STLs.Photo: New and old, same orientation.Photo: Hank printing off an older birdhouse design, because his last one was too early for the Thingieverse contest.  Printed in 4 pieces: front, back, floor+unadorned walls, and the roof is a single piece printed on one of the outside flats.Photo: Deep debossing!Photo: Hank printing off an older birdhouse design, because his last one was too early for the Thingiverse contest.  Printed in 4 pieces: front, back, floor+unadorned walls, and the roof is a single piece printed on one of the outside flats.Photo: The Molex 3953-2002, melted out.Photo: Photo: So _that_ is why they used a right angle connector.  The 39532-2002 fits fine, but is pointed at an awkward angle.Photo: MatterHackers blue PRO PLA (opaque) vs. MatterHackers Blue PLA (Translucent).  
I think it is dye vs. pigment.Photo: Photo: Photo: ABS/PLA contrast parts.  Same G-Code except for the temperatures.  Parts are (surprisingly) interchangeable, and even move smoothly when exchanged.Photo: PLA is about 27.5GPhoto: ABS is about 21.8g on the same g-code with the temperatures changed.Photo: Fitted plate to retain the feed and prevent further wear to the cold end block. 

Our last attempt to print a replacement cold-end didn't end well (minor misalignment resulted in terrible jams), so we're trying to make sure the existing one stays healthy.Photo: A bit of weld build-up (Soldering iron + filament) to patch the worn spot in the original cold end block.Photo: Nothing says fault like vaporized metal (Notice that one of the leads on the black terminal is _gone_. 
Soldered in a spare screw terminal and our heated bed is back up and running, but that is a repeated and nasty failure mode.Photo: The quick-disconnect terminal for the heated bed is no longer quick-disconnect... because it is completely fused. 
I'm pretty sure there is a good reason the newer M2s use 24V supplies, the current to run a Heated Bed at 12V is excessive.Photo: Photo: The college of engineering *lost* their poster hanging boards, so we needed an alternative.  Thinking with 3D Printing = 4x8s of coated sound insulation board and brackets.
Works great (Except it will get tarry gunk on you if you press on it too hard)Photo: Stand brackets.Photo: 5 generations to successful upright 5x8 triangle clips.Photo: Photo: Southeastcon technical program awards.Photo: 1lb rolls of t-glase in clear, blue, and black, and a roll of higher quality GITD blue from Makergeeks to feed the filament supply.Photo: Fun with TGlase.  It looks amazing in person, and holds tolerances well, but is kind of twitchy to print.Photo: T-Glase heat creep jam.  Had this happen on two different printers.Photo: T-Glase heat creep jam.  Had this happen on two different printers.  Look at that snarl.Photo: Clear T-Glase on to your left (less yellow), cheap clear PLA on your right.  The T-Glase is not a huge win.Photo: New version of the Z-adjust wrench.Photo: New version of the Z-adjust wrench.Photo: Dat reach.Photo: The long-term damage to the bed ground from our earlier problems added up to an intermittent connection.  Replaced with another 6-position Molex Eurostyle compatible plug.Photo: Some time ago we added a second spool holder ( https://www.thingiverse.com/thing:159350 ).  It works better for most spool designs, but is a little ungangly.Photo: Second spool holder, rear view.Photo: BladeKey, printed version, paramaterized for my particular stack of keys.Photo: BladeKey, top view.  I haven't bedded the ziptie because that ziptie sucks and I don't want to "finesse" the hole until I have a permanant one on.Photo: Key upgrade v2: 3D Printed custom BladeKey, sex bolt edition.
Slimmer, and with better control over the deploy tension.
Did have to touch some things up with a drill because of dimensional lies by the bolt mfr.Photo: Photo: BladeKey, top view.  I haven't bedded the ziptie because that ziptie sucks and I don't want to "finesse" the hole until I have a permanant one on.Photo: BladeKey, printed version, paramaterized for my particular stack of keys.Photo: When I got to the lab,  I was told the can chute on the lab fridge was broken.Photo: Because too many of the inserts were missing.Photo: So I grabbed an existing one and a pair of calipers.Photo: Described the problem to OpenSCADPhoto: Printed out a batch (Full disclosure: there was a single test print from the iteration between which was slightly too tight on the ID)Photo: Inserted (Fits better than originals, that one has the ID bored because it was the first iteration test fit)Photo: OEM and PrintedPhoto: OEM and Printed, IDPhoto: Fixed! (Note also the snazzy made-out-of-a-chopsitck retention bar)Photo: Switched our slightly-melted RAMBo 1.0e for a 1.2h.

Note that the bed thermistor goes in T2, even though it looks like it should go in T3.

Microstepping change does make a slight movement improvement with no other changes.Photo: That print was on the bed when we left, what is it doing on the floor?Photo: ...Fuck.  Meltdown.Photo: Got hot enough to _crack_ and discolor the _ceramic_ around the nichrome.Photo: Photo: PEEK and PTFE both utterly destroyed.Photo: Looks like the thermistor (&co.) snagged on a lifted area of print, and the thermistor pulled away, causing the current to ramp until it hit >550°C.Photo: Melty and crackly.Photo: Photo: Photo: Kapton that was securing the thermistor to the head, now melted into the defunct part.Photo: Remnants from our old Nichrome Makergear heatcore.Photo: New Makergear V3b hotend (With 19v cartridge for older machine)Photo: Cleaner looking than the old wound-Nichrome design.Photo: New hot end is completely drop in.  Slide into groove mount, connect keyed connectors, go.
It's very close to the same length as the old one so Z home didn't even require more than routine adjustment.
I think the thermistor is different or we need to fiddle some PID parameters, because it seems to be reading below the setpoint.Photo: Some odd banding in one of our test prints with the new head... looks like missed steps?Photo: Missed steps because the motherboard fan died and the drivers were overheating.  One quick rooting around in the parts bin, and we're back in business.  
Thinking about a new top that takes a low profile 60mm...Photo: PVA glue on glass does indeed seem to improve adhesion/discourage curling for large flat parts.Photo: Photo: Trying PVA glue coating on the glass for bonus adhesion.Photo: Feeding the machine.  4 black, 3 blue, 2 white, one gold and one grey 1Kg spools of Rosewill 1.75mm PLA.  

We've run some of theirs before, it tends to run a little undersize, but that's fixable with extrusion parameters, it doens't have any other annoying properties, and is reasonably priced.Photo: Photo: Photo: Video: Makergeeks' "Raptor" High-Temperature/Dishwasher Safe PLA works exactly as advertised, and it's pretty amazing.

Blue double-wall band of High-Temperature PLA, "annealed" at 100C for 10 minutes.
White similar double-wall band in normal (IIRC Rosewill) PLA out of the giveaways bin.
Mug full of boiling water out of the microwave.
Observe.Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: Photo: