There's a few odd numbers and salesman-like claims being bandied around here that don't make much sense to me.
First off, the quiescent current of the Shenzen/Xiechang controllers is nowhere near 150mA normally, it has been between 55 and 60mA for every one of them I've looked at, so this means that this particular controller may not be that efficient to start with. The power consumption with the controller on, but the throttle closed, will normally vary from about 1.5 watts on 24 volts up to about 6 watts on 100 volts. Not exactly a lot, but enough to get the dropper resistors a bit hot, as has already been said.
A switched mode regulator, as used on one or two controllers already, makes a lot of sense, but only really to get a source of heat out of the case. Using internal heat reduction as the reason for making the change would be sensible, as heat does generally reduce reliability.
Controller losses, and hence controller efficiency, are governed by three factors:
Firstly, there are the I²R losses in the FETs, internal wiring, board tracks and shunt resistor. At 50 amps, these will typically be around 6 watts for the FETs, maybe 1 watt or so for the shunt and maybe another 1 watt or so for the tracks and wiring. Let's say that the I²R losses add up to 8 watts (this is probably on the low side).
Next, there are the FET switching losses. These are difficult to calculate accurately, but are likely to be at least 1 watt or so per device, so maybe 12 watts in total for a 12 FET controller, again on the low side.
Finally, there are the quiescent current losses mentioned above, which vary from maybe 1.5 to 6 watts, depending on supply voltage.
This makes the total controller losses around 26 watts for a 100V, 50A controller. To get efficiency, all we need to do is work out the percentage power lost in the controller and subtract it from 100% power. 100V, 50A is 5000 watts, so the controller efficiency for the optimistic example above is around 99.48%. In reality, the FET switching losses and the I²R losses will probably be a bit higher than I've quoted, so this figure is about as good as possible.
This "tens times more energy efficient" stuff is bullshit, at best the controller power consumption "might" be improved by maybe 5 watts, changing the optimistic example above from around 26 watts of losses to around 21 watts of losses. The impact on controller efficiency is to improve it from 99.48% to 99.58%, not exactly "ten times more energy efficient"...........
I detest sales bullshit like this, for that is what it is in my view. If you need to misrepresent one aspect of your product in order to try and sell it, then frankly it leads me to believe that other aspects may well not be stated correctly, either, which in turn would make me walk away.
Jeremy
That's what I get for trying to do math without a formula and a calculator and a pad of paper. 
