I also get great results with Bungard boards. I used to use 1:9 metasilicate but have now switched back to 1:19 as that works fine. It can take a while to get perfect results but when you do you can then get really consistent results.
Some tips from me (not that you need them now but others might):
It's almost impossible to overdevelop using metasilicate. Try leaving it for 10 minutes to rule out underdeveloping. (3 minutes should be plenty though.) Rinse thoroughly.
I found a good way to check whether it's exposure / developing / etching that's causing the problem. Have one part of your test PCB completely wiped clean of the etch resist using meths or acetone. Problems here and it's got to be the etching. Have another part that you really expose by giving it 10 minutes under UV with no transparency. The rest of the board should still have the protective layer on at this point. Then do the usual exposure test by exposing the rest of the PCB for different times. Problems with the super-exposed area mean it could be the developer but it won't be the exposure.