@Thomas_Judy : that is exactly the same kind of effect that I had had on lots of bigger models, it was also not totally predictable (I think due to the approx. 0.2mm shift accompanying removing the build platform, removing the model, and reassembling the build platform) - with me, it seems to be totally gone once the front-to-back non-parallel angle between platform and resin tank as well as lateral skew is compensated for… the bow in the platform can probably only be removed by milling it flat, which might loosen the 8 press-fit threaded sockets, but this approx. 0.2mm higher up in the centre seems not to be too critical if the other tolerances are compensated.
I know the method I used is not terribly high-tech; I also don’t know how long it’ll hold… compression of the paper shim may lead to mid-/long-term instability; really, the adjustable screws for the resin tank I read about should be able to give a first and perhaps in itself big enough improvement, but whoever wrote about that said FormLabs doesn’t allow you to touch these, so I stuck to modifying the build platform, where voiding the warranty isn’t a problem. Additionally, there is a variance from platform to platform anyway, so I doubt whether those screws would really solve it.
At any rate, for the moment, specifically coupled with @BrentONeill’s hint of using the hex screws to do some fine adjusting, with my method I really have got a) much more reliable and even adhesion to the build platform, plus b) much easier detachment of the models from the platform… no breaking tabs, no excessive sticking. Do be aware, however, that simply removing the build platform, detaching the model, and plugging the build platform back in and printing again can cause 0.2mm shift, thus different results, which in itself can make the difference between it-just-worked-that-first-time to oh-shit-what-went-wrong-NOW?!? before the compensation method is used.
Try printing my test file 2015_07_27-2cm_base_thickness_plate-x8_numbered.form (2.2 MB) - it prints in 14 mins on clear at 200µ / 18 mins in white at 100µ /24 mins in black at 100µ, using only 4ml of resin; adjusting the build platform using the results is a little trial and error, but I always got significant improvements for all 4 build platforms already after the first iteration; a maximum of 3 iterations requiring platform disassembly plus one further one without disassembly (using Brent’s fine adjustment) leads to around a max. difference of ~0.3mm across the build platform - this is compensated for very well by the simple but clever method FormLabs used… it’s only when you get a big tolerance problem that their method doesn’t seem to be enough, but it seems fairly common for this to happen. With 4 print runs, 3 platform disassemblies you are probably looking at 2-3 hours time frame per platform, with about 1 hour of actual work occuring. For me, a small price to pay for predictable printing.
The down side is having to keep a list of build platform/resin tank combinations with accompanying z-axis offset… AND having to remember each time you change platform and/or resin tank to readjust the z-axis offset… but I’d rather have that one-off effort than the constant frustration, money/resin wasting and resin tank ‘burning’ that accompanies the failed prints - maaan, do I HATE coming in in the morning to find a 12 hour failed print!!!
Give it a go with the tough resin, if you get a chance, and let us know if it helps.