I recall, but can not substantiate, reading that three pole switches and contactors used with all 3 poles parralled  can be loaded to TWICE the single pole loading, and not to three times as might be supposed.

There is no reason why each pole can not carry the full load current, the problem is that the load current may not divide equally between all 3 poles. 

Make certain that the switch selected is rated for DC, or alternatively use a switch rated for AC, but at TEN TIMES the voltage, that is a 480 volt AC switch.

I agree the problem with paralleling low impedance devices like switches is ensuring equal current division - whatever you use to link the poles together is likely to have a resistance in the same order of magnitude as the switch itself - so it would be easy to have one pole having proportionately significantly more resistance overall than another - even if it 0.0002Ω vs 0.0001Ω the current will still divide 1:2 rather than 1:1. Probably the least worst approach is to link each terminal to its neighbour daisy-chain fashion and then connect the main conductors to opposite ends (e.g. extreme left a the top, extreme right at the bottom) so the link resistances tend to add up the same overall. But still there'd be some imbalance due to variation in the resistances of individual links and indeed of the switching elements themselves.

In most other situations a tiny traction of an Ohm makes far less difference - e.g. when connecting long cables in parallel small differences in link/termination resistance would be swamped by the overall resistance of each cable core, or when paralleling the supply side of a 3P DB to be used single phase, the linking is only done on the supply side of the incomer - the current division is controlled by the connected loads on each of the three bus-bars.

    - Andy.

When paralleling up electronics, to ensure good sharing, we would use 'ballasting' resistors that are a few times the circuit impedance, so that the variations in wiring layout are less serious. If this cannot be done then a traveling wave layout is used, where the input and output are located on diagonals of the group of devices sharing the load, such that the one nearest the input with lowest input resistance, is furthest from the  output, so that the total path lengths of all devices are more or less matched lengths. Such things work on a PCB but are not so easy to arrange in a contactor - so instead  some de-rating is needed

Mike

Thanks all. That makes a lot of sense. The minor difference in impedances would clearly have a significant effect I've subsequently found the de-rating factors from various manufacturers which seem range from around 0.7 to 0.9 for 2 poles in parallel.

The following is for Siemens contactors, but I'd assume follows the same principles. I know it says AC-1, but the same document states that they are suitable for DC-!

- parallel connection of 2 main contacts: Ie AC-1 x 1,8
- parallel connection of 3 main contacts: Ie AC-1 x 2,5
- parallel connection of 4 main contacts: Ie AC-1 x 3,0

Also, similar derating information from Schneider shows different factors need to be applied when considering durability, breaking current, voltage etc.

In summary, I'd say that the principles pointed out about not assuming that all poles will carry the same current are sound, but trying to get 'universal' derating factors is a forlorn hope as they really need to be taken from the manufacturer.

Anyway, in my case, I've now found those and gained a little understanding as to why they need to be applied in the first place. 

thanks all