Originally posted by jpzk Is there a way to figure out what would be the combo, lens+TC, theoritical "sweet spot" where you'd get as little loss of IQ possible?
Not without actually measuring the combo.
But a perfect TC would maintain the sweet stop, i.e., if at f/5.0, the 1.7x TC would shift it to f/8.5. That's already close to diffraction territory for modern high MP APSC cameras.
A real-world TC may make it f/11 which is not as sharp one may wish but can still be sharpened up at least.
As a general note, I've found images from TC combos to sharpen well.
Photozone has tested one of the best tele lenses ever made (the Nikkor 200/2G) with 1.4x, 1.7x and 2.0x TCs.
Nikkor AF-S 200mm f/2 G ED VR (FX) - Review / Test Report - Analysis
In the center, the sweet spots shifts as follows (effective f-stops):
200: f/2.8!
280: f/5.6 (rather than f/3.9)
340: f/8.0 (rather than f/4.8)
400: f/8.0 (rather than f/5.6, with degrading max. resolution)
So, it seems the TC is eating about ~2 stops on its own for reaching the sweet spot. With even worse results in the corners. However, this isn't a shift as the 200/2G is so good the optical performance is almost only determined by the TCs.
I would have to run a numerical analysis to come up with a model which predicts the sweet spot of any combo. To simplify, it is SQRT(N_TC^2 + N_eff^2) with N_TC=5.6 or 8 for 1.4x or larger. And N_eff = sweet spot F-stop times TC-factor.
Something like SQRT(8^2+(1.7*5)^2)= f/12 for the DA*300/4 x 1.7TC combo. No idea if this sounds plausible but it certainly is close to the diffraction boundary for K-5 pixels. Feasible but needs lot of sharpening.
P.S.
Just noticing we're in the Q forum here: The Q's pixels are so small that the DA* at f/4 (f/4.5 maybe) is the only way to use this lens with the Q. Forget about stopping down or using a TC.