Direct blowback is beautifully simple, but not always intuitive. With respect to the design of the CZ Scorpion Evo our testing suggests that the mass of the bolt is far more significant in the bolt velocity/speed than is the force provided by the recoil spring (or springs). One of the constraints in the Evo is physical receiver space; if there was room for "theoretically infinitely long" spring or distance to the ejector we could definitely tune the bolt velocity with spring rates, but in the space given by design the bolt mass seems to win.
We have tried several different designs and combinations including "conventional" and "non-conventional" spring types with marginal results.
As for flinging brass, the Evo extractor is a reloaders nightmare... but what we have found is that tuning loads to suppressed/unsuppressed in order to achieve a desired bolt velocity or extraction pattern is the most effective way to deal with some of the issues presented in this thread. .02 cents :-)