Not to hijack this , but it might matter to the OP.
I modded a Nokia charger to deliver about 13.4V into a PIC hydroponic controller (3.5" x 1.8" of surface mount parts, dbl sided PCB) I built. There is a Lead acid 7AH gel cell being trickle charged as battery backup and a 12V low amperage pump.
Ok, without the pump attached the PIC is having 'lockup' probs sometimes when sampling the adcs (Supply volts, CDS LDR, Thermistor etc.) once the charger is connected. It is ungrounded. I note with a grounded (much noiser) PC 12V supply all seems well. In fact simply clipping the PC supply's gnd terminal (not the +ve) to the system powered by the nokia the 'lockups' go away. Also attaching a grounded O'scope also causes the prob to disappear!
Clearly an apparent 'noise' prob. I thought. After much efforts with ferrite cores, ferrite beads, assorted caps ranging from .01uf to 100uF across the supply I can't figure it. The O'scoped 'noise' is under 20 mV (abt 200Khz) on the 13 V line and almost nothing after the 7805 Vreg. Yes I have a 0.1 uf cap on the PIC Vdd/Vss pins and a 1 uf on the 7805 output & input. The nokia charger has a 47uF 16V on its output line.
I setup a watchdog and a counter....I am getting about 10 resets per hour as measured over a 12hour cycle with the charger or charger AND battery as the supply. No resets with the battery alone.
Any suggestions?