ORIGINAL: gaRCfield
I don't mean to be doubting the masses, but the report does have me thinking. Even if the load was 4 times what he tested, it still would be under the 88oz-in of a DS821.
Nothing wrong with thinking, it would definitely be worth looking into. Even if your thought on the torque requirement turned out to be wrong, something else useful might be discovered.
Last thing is I have a hard time wrapping my head around the 'more power to hold the servo' idea. Everything I've been learning in school for the past 3 years talks about forces being greater during acceleration. I haven't taken electrical classes yet but I'm not sure if it's an electrical thing.
If that is in fact the case, I think this has to do with the fact that a servo contains an electric motor. Any electric motor will use more current at zero or low RPM (with a load) than when they are rotating at faster speeds.
On a brushed motor the current passes from one brush, through the commutator then through the other brush. If that commutator is not turning, quite a bit of current can begin to flow. Once it is rotating, brush to commutator bar contact is shorter in time so less current flows. I'm sure RPM and brush to commutator contact frequency come into play as well.
Brushless induction motors act in a similar manner because, from a dead stop, more current has to be applied to them to develop a magnetic field strong enough to get the rotor and load attached to it rotating. Once it is turning, the current demand drops.
Ever listen to a big fan or other electric motor when you start it up? It kind of hums and strains at first until it gets going. If you graph the current, it has a sharp initial spike then drops in a smooth curve to a reasonable level as the motor begins rotating. Prevent that motor from rotating somehow and the current spikes back up quickly and stays there. I've seen this dozens of time testing electric lift trucks while driving them, climbing on ramps and then staling them on ramps.
Back to your original point though... I wonder if anyone has actually measured the force required to move the control surfaces on various airplanes or do we just automatically assume that we need the biggest, baddest servo we can find? It would be interesting to find out.