Today I tried some settings in Slic3r. Doing so I printed a U-shaped testcube of PLA on a heated mirror. I printed the first layer at 200°C and the subsequent layers, too. It looked quite well, but at one corner it hadn't stuck to the bed as perfectly as I wished. Actually I wouldn't call it warping already, but maybe the very very beginning of it.
So I cleaned the mirror with alcohol. Then I printed a second cube with 200°C at the first layer and the other layers at 170°C. My expectation was a reduction of the warping, as a layer with a lower temperature wouldn't shrink that much like a hotter one.
The result was puzzling. The second cube showed strong warping.
To be sure I repeated the printing and printed a third cube with 200°C/200°C and a fourth with 200°C/170°C. While the third one again showed no warping, the fourth cube warped so much that it got off the heated bed and I had to abort the print.
Obviously the reduction of temperature at the second layer weakens the bond of the first layer to the bed. Maybe the "cold" second layer cools down the first one and causes it to shrink. Once the first layer got off the bed the whole part is an easy prey to warping.
Possibly this temperature difference explains a lot of cases of mysterious warping. A little option in the slicing software turns out to be a trap ...
I guess I will repeat this with ABS soon.