Short answer: about 1 g*cm per gram of weight.
I measured a BLDC fan motor from ADDA, model number AD1505LX-90, by taping an arm on the fan, and taping a US penny to it (cut in half, each half at 1 cm from the motor axis.) At full rated voltage of 5V, the motor would start the arm rotating.
A brief pulse of electricity (say one tenth of a second) would NOT start the motor. The arm would twitch, but fall back into detent (between permanent magnets of the rotor.)
A little longer pulse (say one third of a second) would rotate the motor say half a rotation (past one or more detents.)
The fan weighs about 2 grams. It moved a penny weighing 2.5 grams on a 1 cm moment arm (a torque of of 2.5g*cm.)
So the starting torque per unit of mass is about 2 g*cm per g.
I know this is fuzzy, but:
- the manufacturers don’t publish the starting torque for these fans
- there is a research paper about the torque of these fans but you must pay to see it (thanks, but no thanks, Prof. Fussell)
- most posts don’t quantify, just saying “not much torque.”
I believe this will extrapolate linearly to larger motors.
Look for another post about the design of BLDC motors.
Tested a ADDA0205MX-K53-LF weighing 5g. It has about 20g*cm of starting torque. So it has about 4 g*cm of starting torque per gram of weight.
These motors are very thin and wide (like what is called a capstan motor in VCR’s and floppy disk players.) This flat and wide configuration probably has much to do with its high starting torque: the permanent magnets are at a long moment arm from the center of the motor.