So after the better part of a year trying to set up this experiment. I've finally managed to get a respectable amount of signal at my detector.
Basically the optical fibre that was meant to couple the signal into the detector was damaged or cracked somewhere. Which resulted in very poor signal to noise ratio.
I've since moved the detector onto the table and reconstructed the beam path so I now couple the fluorescence directly into the detector. Even with a fairly unoptimised alignment I'm getting massive amounts of signal. Which makes sense since I can visibly see the fluorescence coming out all over the shop.
To quantify the improvement. Previously using the damaged fibre, it took me about 5 minutes of collection at max power to get an intensity count of about 160,000 or so.
Coupling the light straight into the detector. It is taking me about half a second to get just as much. That represents an increasing in signal by about 1000 times.
Things are looking good for me
Also forgot to mention. With the fibre I needed all the room lights off and the sample and detector covered from as much ambient light as possible. With the detector on the table, I can do it with the room lights on! (obviously I wont when trying to collect publishable data on samples with much lower quantum yields)