Ideally, the higher voltage cell charges the lower voltage cell until they reach an equilibrium, but it must be such that Lipos would need to be paralleled for a long period of time to do so. TrueRC and MaxAmps both build packs with cells in parallel, but with the cells constantly connected, it should be much eaiser for them to maintain equilibrium.

Based on what you've personally experienced though, I can't help but recommend separate balancers for each pack.

If I was to parallel to seperate sets of packs:

• Charge and balance each pack seperately, but preferably at the same time on different chargers. If you charge one pack, then charge the other, the voltage on the first one might have fallen a little. Probably only ~.05v total. Measure with a voltmeter to be sure.
  
  
• Hook the main charge leads together + to + and - to -. If there's even a slight difference in pack voltages, you'll probably see a spark.
  
  
• At this point, each pack is the same voltage but the individual cells between each might be slightly off. So, then use a small value (~1ohm) resistor on the balance taps between each parallel cell to "soak up" any potential differences before hooking them permanently. Maybe even hook up a voltmeter across the resistor. When the voltage drops to 0, there is no current in the resistor then the two cells are equal. The voltmeter is simply to verify that there is no current from cell 1 trying to "charge" cell 2, or vice-versa.
  
  
• Once SURE each cell of pack 1 is exactly the same as the corresponding cell in pack 2, tie the balancer leads like MM has in his diagram.

Maybe I'm a bit over cautious, but that's what I'd do...

Once the cells arae permanently wired in parallel, each cell in the parallel pair will remain the same and never be different from each other. The balancer wire assures this.

The more I think of this, the more I don't like forcing the cells to equalize through a small balancer wire. They SHOULD stay equal, but if one is weaker than the other, there will most likely be current flow in the balancer wire. In a stock 2p pack, the cell tabs are tied together for a low resistance/high current connection.

So, to revise my previous statement, I would actually pull the packs apart, and make two seperate series-connected 2s2p packs instead. Or, balance each pack seperately as skyello said...

my packs are made up of cells from an A123 dewalt pack, brand new out the package, never charged, never run. This being the case is it safe to make up a harness like metalmans diagram or should I still go through the process above?

It's easy enough to check the voltage of each cell in the pack using a DVM (digital voltmeter). Of the maybe 6-8 Dewalt packs I've come across, I'd guess half of them have had a cell or cells of lower voltage than the rest. If that is the case, you can use the resistor charging method BrianG outlined above to charge up the low cell(s) using the other cells.