I tried a ghetto liquid culture technique
http://basementbiotech.org/penicillium-isolation-from-citrus-fruit/
with inconclusive results.
Isolating a strain of pencillium is pretty easy, just let some citrus
fruit rot for a little and look for the olive green colored mould that
arises. You can make a simple microscope slide as shown, and
pencillium has very unique fan-like conidia. To make isolation of
the secondary metabolites (of which penicillin should be one) easier,
culturing the strain in a liquid culture is advised. I just boiled
some lemon chunks for a while and threw them in coffee cups and seeded
the liquid cultures with mycelium / spore swabs off the rotting citrus
fruit.
You could try some different media, there are tons of papers out there
about pencillium cultivation. I have some growing in potato dextrose
broth at the moment and it seems to be growing fine.
The reason the experiment ended in inconclusive results is that I
waited 3-4 days to check the disc diffusion assays, enough time for
the gram positive bacteria to grow back over its fallen comrades into
any area once made clear by the diffused penicillin.
I generally do DCM workups of fungal fermentation liquid media, save
and concentrate both the organic and aqueous partition, and will
sometimes macerate the mycelial mat and do an MeOH extraction. Ethyl
acetate is also a good organic solvent choice, probably better than
DCM.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1258127/pdf/biochemj00965-0101.pdf
That's an old paper but, sometimes old papers can use more simple
equipment, since fancy stuff wasn't available yet, though the catch-22
is using solvents and reagents most people would try to avoid now a
days.
I'd say, isolate some pencillium and start some pure cultures on a few
different liquid media types, then do some more reading while it grows
(around 10-12 days is when I believe optimal penicillin conc. is
reached).
Liquid liquid extraction with a separatory funnel would be pretty
easy, then you could do minimum inhibitory concentration tests by eye,
or with a spec at OD600, or disc diffusion.