It's bad idea but may often not be fatal.
Rough BOTE calculation suggests that it would be eays enough to get say doubling of the maximum allowed charge rate of a cell if cells with say 0.3V imbalance are hard interconnected at the cells and then charged immediately after connection.
If batteries are not hard connected at the battery but instead have leads to a common power supply point then not charging them for about 10 minutes after interconnection *should* allow safe enough [tm] self balancing. Adding a very small resistor in each battery lead or ensuring leads of a minimum resistance would aid this process. See text.
The existing rule of thumb is probably an empirical one based on the practical reasons behind the guideline in the paragraph above - see text.
- Mutual balancing "on the bench" befor installation with a resistor or a purpose built bidirectional current limiter would be a good idea.
NOT an authorative statement. I've never hard paralleled LiIon cells.
But I do have lots of general battery experience and have thought about this specific issue before now.
Hard paralleling should be avoided if possible. With modern electronics it is extremely easy to make a switch which allows independent current paths when charging and discharging.
The "rule of thumb" MAY be based on experience and this in turn may be based on the happenstance of battery connection resistances - see below.
If you have cells which area rated at say 1C max rate and you charge two together at 2C, the charge may distribute unevenly and in addition you can get considerable intercell currents. The net result is that (it seems to me) that you could easily enough double a single cells rate charge rate.
Even simpler, if you can tolerate a small amount of voltage drop from the battery then adding a small amount of resistance in each lead such that it drops say 0.1V at full charge will allow quite substantial differences with minimal effect. If max charge is say 1C (common for many LiIon, some manufacturers allow up to 2C) the R ~= 0.1/C (C = Ah capacity in amps). So eg an 18650 cell (not a LiPo but same principle) may have a 2Ah capacity so R = 0.1/2 = 0.05 ohms. You can achieve something like that just by using two battery leads to wherever the cells connect to rather than hard connecting between cells and using a single lead. If 1C (2A of charge flows between imbalanced batteries the drop will be 0.2V - so 0.2V of imbalance at initial connection will be accommodated within spec if you have only battery-to-battery balancing to considerAs a very very rough rule of thumb LiIon capacity increases by about 6% per 0.1V across the constant current charge region. (That's based on a quick mental calculation of Vmin = 3.0V, Vmax = 1.2V, capacity at constant voltahge pedestal ~~ 80%, linear capacity change with voltage change). Capacity is NOT linear with voltage change but it gives us some idea. So a say 0.2V differential ~~~= 2 x 6% = 12% of C.If mac intercell balancing current = 1C then this will take ~~ 12% x 1 hour =~ 7 minutes. So if you parallel connect two cells with >= ( R/0.1C) lead resistance in each cell's leads and don't charge them for say 10 minutes after connection you'll "probably be OK" [tm]. Operation from battery immediately after interconnection is OK.
Effect on charge and discharge: As the above allows for about 2C of intercell transfer and as cells are not normally discharged at a 1C rate (laptop users usually value having more than one hour of battery operation) then enough resistance to provide Heath-Robinson interconnection protection would have minimal effect on cell discharge voltage. If charging at maximum capacity via these resistors te cell voltage will be reduced accordingly but as the system comes off constant current into constant voltage mode current will drop and battery potential will be made up. So the net effect is to slightly increase charge times.