Well, no, 20% of efficency in conversion is considered good, for DPSS lasers ..... i'm not sure if the maximum actually obtained from ambient temperature assemblies is still 25%, or if someone have finally reached 30% ..... anyway, as basical principle, if you double the energy of the outcoming photons, you can never expect an efficency more high than 50% ..... then ofcourse you must considerate the loss from coupling, alignment, polarity rotation, faces reflection, the part of the energy dispersed from the crystal in form of heat, and so on .....
The maximum theorical efficency is obtained using a collimated IR diode with FAC lens, that can collimate all (or the maximum possible) of the IR in a point inside the crystal, typically from the middle to 3/4 of the lenght of it, with the crystal perfectly aligned and the polarization planes from the diode and the crystal axis in the range of one degree from the optimal one .....
Usually, instead, the normal green modules, uses an open can diode, with the face of the emitting chip the more close possible (half millimeter, usually) to the input face of the crystal ..... less cost, but also less efficency .....
If you're planning to use the crystal of a 50mW unit, probably you can get from it 100 or 150mW without too much problem, but over this, the heat generated start to be too much, and the crystal can become damaged ..... considering the better alignment and polarization that you can get from these assemblies, can be possible to obtain 100mW from a 600/650mW of IR, and maybe 150 from a 900mW of IR, but can be difficult to find 5,6mm IR diodes that emits 900/1000mW ..... open can units, ofcourse ..... and also, you need to heatsink your unit very well, or your crystal will become soon a cooked ex-crystal
..... (also, all the modules rated 200mW that i've seen til now, that effectively emits over 170/180mW, uses 9mm IR diodes, not the 5,6mm ones)
