In the sience paper they used up to 90μMol/s/m² and measured an increased PS. The effect however was measurable even at lower intensities. I gave this time ~20μmol/s/m² in addition and will give them next time around 40μMol/s/m²(exchange the driver for one with 1050mA).
If I should make a recommendation I would say go for ~40-50μMol/s per squaremeter and use a driver which allows you to add more diodes in case you want to try it with even more far-red or with a mix of deep and far-red later.
To get ~40-50μMol/s you could use 6-10 top bin XP-Es in series or Osram Oslons and run them with 700 or 1050mA. A Meanwell LCM-40 is a driver with selectable drive current(350-1050mA perfect for secret diodes) and is usable from 2-100v and offers endless possibilities. It even offers 0-10v and PWM dimming which makes it compatible with the most reef controllers.
The driver has up to 42w with is enough to power more diodes if you want.
The shortening effect should be well feelable at 12/12 and the by a few percent optimized PS should largely compensate for losses in the end. As it looks like I get the same 50g/sft as usual but this time only from 240w LED, 40w UVA / B and 5w far-red (285w total) above a 2x 4'. That's only 42-45.000lx or 600-650μMol/s/m². I'll let how know how much it is when dry..
Here is a shot from day 48..
View attachment 4201393
The strange yellowing is a genetic defect of the SLH, btw..