Weird curling with PLA on heated bed of glass

Hi, I've finally took the dive with my new build and was really excited to up my game a little into a true open hardware Reprap.

First thing I wanted to do is print Nophead's Z couplers and I seem to having a strange issue with curling. Pictures below after I cancelled the print, but a few notes first.

I'm using a PCB heater MK2 with a 3.5mm thick pane of borosilicate glass on top, anchored with binder clips. I was having trouble getting it to stick when the bed was set any less than 80C. I liked the idea of the MK2 because it has a hole for a thermistor but I have to wonder how accurate it is because it's under the glass, and I'm assuming it's the top end of the glass that counts. I haven't seen to many thermistors glued to the topsides of their headed bed, though, so, when people say "heated bed 60C" does that mean 60C on the surface or 60C on the thermistor?

Anyway, at temperatures under 80C on the thermistor, the skirt prints fine but the first things it draws after that are some circles and they usually start lifting off the bed while they are printing. So that's how I arrived at that number.

Anyway, it starts printing ok but somewhere about 20% through I start noticing some parts are curling and lifting off the bed along the corners. I am spraying it down with an ammonia-based glass cleaner before starting so I don't think it's a finger oil thing.

I thought it might be that my extrusion steps are too high, but when I look at the bottom first layer, the coverage doesn't look thick at all, so I'm not sure that's the issue.

Previously, I've printed on an Ultimaker on masking tape but not only am I at a dead end with maintaining that hot end with it not working but I wanted to print larger single pieces hence the heated bed in the first place, but I'm puzzled why these small items are giving trouble. Only thing I can think of is that I don't have a fan like the Ultimaker has but I don't see a lot of designs on Thingiverse for them so I'm thinking they're not necessary, but designing and printing something air-tight might be problematic with this curling action.

I suppose I could just lay down some masking tape and leave the heat off, but I'm going to have to conquer this eventually.

Additional stats:
gcode by slic3r 0.6.0: as pictured, 2 copies along X
0.1mm layer height
extrusion temperature 185C
default speeds
hot end = Budaschnozzle v1.0 w/ 0.35mm nozzle
Ultimachine PLA 3mm (I measured an average diameter of 2.87 from 10 samples)
Printing with Pronterface
Running current Marlin on Sanguino 1.3a

The first layer is a little too high, which will reduce the adhesion.

80C is too hot for anything but the first layer. It needs to be lower than the glass transition after that. I measure on the underside of the bed and use 60C. I use a higher temperature for the first layer though to get it to stick well, but it is then too soft to hold the layer above, so the temperature has to drop before the object gets too high.

I don't use a fan.

---

[www.hydraraptor.blogspot.com]

Are there any hard numbers to how high it must be off the bed? I'm going through an iterative cycle where I turn loosen each of the M3 bed screws 1/8 of a turn to raise 0.0625mm and while the bottom is looking much more squished I'm still getting subtle curling.

I have feeler gauges but, aside from not being sure how exactly to use them (the naive "try to slide it under the homed nozzle" doesn't sound right since the bed is on springs), I'm not sure what values I'm looking for.

It should start exactly the layer height above the bed. Any other value gives you an object the wrong height. Some people go lower on the first layer to get more adhesion but that makes a ridge around the bottom of the object. When it is too high the first layer infill has gaps between it.

The easy way to set it is to extrude a rectangle around the bed, let it cool, peel it off and measure the thickness at several points and average it. That also shows how level the bed is.

---

[www.hydraraptor.blogspot.com]

Once I get the bed level I only adjust the Z end stop to change the height above the bed I want the nozzle to be. Then I only have to adjust one screw - I have opto stops and have it mounted on springs so I can adjust them just a small amount to change the height

grog Wrote:
-------------------------------------------------------
> Once I get the bed level I only adjust the Z end
> stop to change the height above the bed I want the
> nozzle to be. Then I only have to adjust one
> screw - I have opto stops and have it mounted on
> springs so I can adjust them just a small amount
> to change the height

Do you have a picture of that set up? For how low the tolerance seems to be for proper adhesion, I think I can use all the help I can get and fine adjustment on one screw sounds really helpful. I don't know if I'll be able to bootstrap myself if it requires pieces I can't print without warping, so figuring it out with modifications would work great.

I've got a fistful of curly plastic chips.

Ok, I kept dialing in the Z height and I think I'm at the point where I cannot print any lower and yet I'm still getting curling.

Attached is a picture of the outline, which is one thread of plastic that looks squished pretty good. The second shows a nice reflection that shows the curl towards x,y (0,0). Finally, the last picture shows all the pieces printed, each one has various degrees of curling and they all exhibit a sort of concave wall that I would blame on lower layers being squished too closely to the bed while extruding.

Can anyone offer any additional advice? These are now 0.2mm layer height, printed at 185C for the hot end and 60C for the heated bed. Also worth mentioning is that they look really secure while the first few layers are printing, it's when they reach a certain height that the older layers start to curl.

Edited 2 time(s). Last edit at 02/26/2012 01:58PM by Mazaw.

On my Mendel90 that last time I printed PLA the bed was 90C for the first layer, 55C for the rest. I use 2mm glass on top of a MK1 heater with the thermistor underneath the PCB, so the top of the glass would be somewhat cooler.

I would say that 60C measured at the glass might be too hot. The actual surface that meets the PLA needs to be below the glass transition temperature after the first layer. Tg is about 55C for PLA.

---

[www.hydraraptor.blogspot.com]

I'm gonna pick up an IR thermometer and see if I can't figure out what the temperature at the glass really is compared to what the thermistor is actually saying. Maybe I'm setting it to 60C and it's not even getting where it needs to be at the glass.

Be careful about measurements with the IR thermometer, the reflectivity of different materials will throw your measurement. You may want to put down a matte tape on the glass to make sure the reading from the thermometer is more accurate.

Mazaw Wrote:
-------------------------------------------------------
> grog Wrote:
> --------------------------------------------------
> -----
> > Once I get the bed level I only adjust the Z
> end
> > stop to change the height above the bed I want
> the
> > nozzle to be. Then I only have to adjust one
> > screw - I have opto stops and have it mounted
> on
> > springs so I can adjust them just a small
> amount
> > to change the height
>
> Do you have a picture of that set up? For how low
> the tolerance seems to be for proper adhesion, I
> think I can use all the help I can get and fine
> adjustment on one screw sounds really helpful. I
> don't know if I'll be able to bootstrap myself if
> it requires pieces I can't print without warping,
> so figuring it out with modifications would work
> great.
>
> I've got a fistful of curly plastic chips.

Sorry, I didn't see your response earlier. This is what I use for my opto stops now.
I really like it and it is easy to turn the screws just a little bit.

I don't use a heated bed for PLA, just have 3m blue tape on a piece of glass. Ultimachine PLA sticks fine on that. I got some PLA from the future is 3D and it doesn't stick as well, but when I clean the tape with isopropyl then it sticks.

Grog, I think I might have to derive from that, I'm printing on an Iteration 2 Prusa that doesn't get the X bars all the way through the X ends. Although I liked how small it is, I might have actually gotten away with printing it. I've printed a lot of masking tape in the past but I found, even with Ultimachine PLA, I was curling quite a bit, printing on an Ultimaker. I printed an X-end idler at the time and the bottom wasn't very flat at all but still stuck so well that I tore through the tape trying to get it off. I'll admit I haven't tried tape on this Prusa but that shiny glossy bottom is too lovely for me to give up on it so soon.

Thanks for the tip, Thav, I covered the glass with masking tape and let it "cook" for a while and I was reading pretty wild measurements around the surface. I don't know if these are you-get-what-you-pay-for but I was under the impression that the PCB heaters worked pretty evenly, so, I'm not sure how confident I am in this measurement device. I wonder if a thermocouple from my multimeter would work, but I'm concerned that the business end of that probe is a sphere and not flat.

Nophead, have you actually measured the heat on the glass? Or did you just dial it in with repeated printings at different settings? Is it borosilicate glass or window glass? I'm wondering, if 90C heats up to just under 55C for you, if I could calculate what the delta should be for my 3.5mm thick glass to have a better starting place than using up all my plastic before I can get a non-curly print.

I actually tried printing a vertex foot the other day and I let it unattended for a bit and, while I wasn't staring at it, there was a Y-axis issue that shifted all layers above a certain height a few centimeters towards +Y, like a torque issue on Y. That's not really that important (I just increased the current slightly) but what I found interesting is that the top layers also curled quite dramatically. I would have expect them to curl downwards towards gravity as there was no support under them. I wonder if that indicates something.

With my bed set to 110C an IR thermometer gives 104C on the top of the glass. As glass is transparent I don't know if the IR thermometer is reading the glass temperature or the PCB temperature.

To get a reading with a thermocouple I have to cover it with good insulator (ceramic cloth tape) and tape it down. Otherwise I get a low reading because there is only contact on one side of the thermocouple and convection cooling on the other side. However the insulation then probably creates a local hot spot so the reading may be a little high. There again, when an object is deposited on the glass it insulates it somewhat anyway so it will be locally hotter.

IIRC I calibrated the bed thermistor with a thermocouple on the top of the PCB.

I don't think the delta across the glass is much because it is a much better conductor than air.

90C will be well above the glass transition of PLA and that makes it stick well to the glass but it is still soft. I then drop it to 55C and that is definitely below the glass transition, so the bottom layer hardens and can then hold down the layers above.

I can't find a source of borosilicate glass in the UK for a reasonable price, so I use plain glass.

---

[www.hydraraptor.blogspot.com]

Dunno if maybe I should switch to regular glass, tried to do a plate of 2 x 2 LM8UU holders and one of the islands just consistently popped off during prints, even with the 95/55 profile for the bed. I love the part bottom, though, I can't even see the tool path of the first layer anymore. Downside is, well, popping off. lol

I think I'm going to have to just abandon it and deal with printing things no larger than 100x100mm on masking tape, and dealing with chiseling them off afterwords. Since I can't get this working, is it worth trying Kapton or PET tape or should I expect the same results? I don't want to buy any of that if it isn't going to significantly help.

Is your bed level?
Are you printing in a draft?

I can print PLA on Kapton and PET but there again I can print it on glass, YMMV! I prefer glass for PLA because the the parts don't pop of when cooled when using tape.

---

[www.hydraraptor.blogspot.com]

The only other thing I can think of is that I'm way down in Florida, US, near the shore, and it's always humid. But I don't think there's really anything I can do about that short of building a chamber for the printer and loading with desiccants. Another thing I could think of is that I'm using a Budaschnozzle 1.0 with a heater-block design hot end. That version featured a very short nozzle, so maybe the waving of the hot Al block so close to the printed piece encourages remelting the plastic?

Trying a quick print this morning over coffee, one of my connector crimps for the hot bed loosened up and inadvertently I tested out just how flame retardant they were, and how responsive my smoke detectors are, so I think my experimentation will be on hold for a little bit.

When I figure out what the first M in YMMV actually turned out to be, I'll be sure to share it!

Think I've got it. Issues with the lower layers spreading (going to experiment with lowering the temp after the first layer) but so far, 80C solid seems to work pretty well for me.

Posting in case someone with a similar set up has the same issue.



Edited 1 time(s). Last edit at 03/10/2012 06:22PM by Mazaw.

Your Milage May Vary.

I can't thank you enough, I was having real trouble with PLA curling on delicate parts but the 80C first layer temp seems to have solved the problem.

For people having difficulty "leveling" the build surface: I took a piece of roofer's tin and soldered a length of wire to it. I then hooked one lead of a multimeter to that plate and one lead to my hot end's tip. I simply zero my Z axis, then I go to each corner of my build plate and raise it up until I have continuity on the meter. Easy, fairly accurate leveling.