Makes perfect sense! Thanks for the advice. One can think of the arms like cog wheels. A smaller arm attached to the motor not only means longer sweep of the motor for same effect at the end, but smaller resisting force or torque on the motor. I chose this setup with equal length of all arms only for simplicity. I knew there would have been a lot to do to optimize the lengths.
I made an Excel chart to simulate the four arms, but I used it only for checking my formulas for the arm movements. I need to get back to it. What I need is a chart with x-y-points showing a cloud of all reachable points for all angle combinations of the two servos, when the servos could only turn in say 5 degree steps. The density of the dots would show which regions of the drawable area would have the highest accuracy.
I guess the only thing almost sure is that the primary arms must be of equal length and the secondary arms must be of equal length. This gives a symmetrical setup. But the primary arms could be a third of the secondary arm length. And the distance from one servo axis to the other could be considerably less than a third to avoid the dead point where the two secondary arms would be stretched to a straight line. I know it will complicate the trigonometric formulas a lot, but I've already stated that I'm not afraid of floats :). The Arduino processor is doing great on that.
The effective sectors of the servos could also be adjusted. Right now both have their 90 degree direction straight up just because our standard way of looking at the coordinate system. But the left secor could be like 60 to 240 degrees and the right -60 to 120 degrees.
I see I've hijacked my own thread now.