Hang on, if all your resist dissolves away in the developer then you've over exposed it. At least that's the way the stuff I use works - exposed parts dissolve away and then get etched away, non exposed parts ... leave copper.
Have you tried doing a test strip ?
Chimed in to say the same: in **positive** boards light *destroys* emulsio, so you are overexposing.
Much worse, you already expose them too much *before* even exposing

Peel them only inside a room, use only a 25W yellow lamp; incandescent!! , not CFL nor Fuorescent tubes.
Tape your transparency over the PCB or the glass so you don't move it and get a piece of cardboard.
expose 1 minute, cover 1/4 of the image, then complete 3 minutes, cover an extra 1/4; then 9 minutes and finally 27 minutes, so you have 4 widely different exposed areas.
Develop.
Then repeat, splitting time between the best 2.
In a couple tests you'll have the proper time.