If you have a heated bed, you can dump your supply of blue tape. It will in all options be replaced by something that lives longer. The bottom of your printed objects will be very flat and shiny. The object will also detach easily from the bed once the bed has cooled down sufficiently. Some refer to this as "popping off". A twisting motion works easiest.
For the different types of heated beds available see: http://reprap.org/wiki/Heated_Bed
Connecting to Ultimaker
Do NOT try to install a heated bed if you have just one power supply and DONT connect it directly to the Ultimaker electronics. The power connector on the Ultimaker electronics is not rated for high current. Even though it might work, it’s not safe. Also the Ultimaker electronics should not be powered by a power supply which supplies more than 19 volt. This will drastically increase the chance of the 7812 voltage regulator to break, or worse. Always install a heated bed powered by its own power supply via a relay or with a seperate circuit to control power to the bed. Make sure you use a seperate power supply which can deliver enough power to the bed.
How to connect
If you solder a 4.7kOhm resistor to the Ultimaker PCB (R4 if you use TempSensor3 or R21 if you use TempSensor2) you can measure the temperature of the bed using a Thermistor more info. The thermistor sits near the center of the hot bed, and has thermal contact to the bed material. Its a 2-wire connection with interchangeable orientation. The firmware needs a thermistor table for exactly your thermistor. You can find them here. Thats why its best to use a thermistor thats already supported by the firmware.
In this PCB, you can use the spring-screws as for the acrylic bed. However, you cannot detach the bed anymore. Then, the actual bed material is screwed onto the PCB, and held by some grips. You obviously need electric insulation between an aluminium bed and the PCB, because the aluminium would short the PCB.
Glass vs. Aluminium bed
- window glass: 9€ if cut by glass company
- tempered glass: much more. 3-4 times normal glass
- alu: 2.5mm 10€ + 30€ invest for large width Kapton tape
Hoops to jump through:
- thermal stress
- holes/cutaway for joris-style startup is not possible DIY. The glass will break there easily.
- Upbending. See below
- needs Kapton tape on the surface, that can be scratched. Needs replacement every half year probably.
- Cuttin alu to shape wihtout deforming is difficult DIY.
- Insulation of the PCB necessary
Heated Bed Power considerations
A typical heated bed is approximately 20x20 cm. It is heated by e.g. a FR4 PCB, with some heating power P0, e.g. P0=100W. The PCB heats up he actual bed, either glass or aluminum. It is cooled by thermal radiation and natural cooling (no fan, but buoyancy of hot air bubbles going upwards) and the airflow by a fan (convective cooling).
Because all this can be expressed in formulas, its pretty easy to estimate the required heating power to reach a given target temperature:
What you see here is that at 80°C, radiative cooling is as strong as a constant airflow with ~0.35 m/s, which would blow however over the whole plate uniformly. The common fan I would guess at 0.3 m/s for25% of the area. This is why for the total heating power, one could estimate 0.1m/sec, where however the natural convection is stronger. You cannot just add the two, but for correction a 0.2 m/sec estimation seems fair.
Now one can create a coupled system, where the "insulation" of the glass above defines the target temperature at the print-side of the glass, while the PCB has to be hotter to compensate. Then one can see how much more power is needed for driving a glass bed.
You see that for heating a glass bed, you need ~5-10W more power than for an aluminum bed. Its not a big difference.
Now lets look at the velocity dependence. The plot shows the velocity by which the bed surface is cooled homogeneously, and the bed temperature that hence develops at 80W constant power.
So this exposes a possible flaw of heated beds in combination with fans. If we choose to heat at a power, so that the region under the fan (0.25m/sec) is at the target temperature, the outer side of the bed will overheat. This is not the whole story for an alu bed. It will conduct well in the sidewards direction, hence the temperature distribution will much more uniform, while glass is by a factor of 250 worse in conducting heat. One mm alu would have the same spreading effect as 25cm glass.
No this leads to major distinction in the battle glass vs. alu. Alu beds "compensate" for inhomogenous cooling.
From the previous section, its clear that if you cool the glass surface, it will become much cooler locally, even if its a thin glass sheet of 2mm. To make PLA and ABS stick to glass, it needs a minimum temperature, e.g. 70°C. If you are too hot, the printed object above is too liquid.
Now lets consider a printed object:
The printed object will be cooled by the fan, even better than the bed, as it is more exposed. Typically, it also has a lot of surface area, where it can radiate heat away. As heat conduction from the bed through object is possible, the object "steals" thermal energy from the glass surface, and hence cools it. And it cools more than the normal exposed plate, as the solid->air step is a bottleneck. This effect of the object cooling the bed is worse, if the object has a thin wall. Thats basically a cooling rib for the heated bed.
Now, if one considers a perfectly printed object with a thin wall, the thin wall will be cooled to a rather solid state by the fan and shrink a bit. The bottom of the object is also cooled, and cools the glass. It cools down to a degree, that the plastic does not stick well anymore. It is lifted from the surface due to the thermal tension in the wall above. Thereafter, its gets even more crazy, the glass underneath is not cooled anymore, it will get hotter.
All this would not happen, if the glass would spread heat better. And thats also why I think that a aluminium bed is superior.