voltage-torque curve

[kamocat]

I'm interested in how the torque of a brushed DC motor changes as its supply voltage decreases. I have seen a small amount of data that implies it is not linear, but I'm hoping for some good data on the CIM, Fisher Price, Globe, or RS-545, controlled with a Jaguar.
Currently I'm only interested in the data at the motor's max efficiency, but apparently that can be calculated from its stall torque and the RPM at no load.

My assumption is that, because it's PWM, the ratio between speed and torque will remain constant as long as the actual supply voltage remains constant. (I've heard the output of our motor controllers described as voltage, due to the lack of inexpensive oscilloscopes, and I realize this is confusing and most likely incorrect terminology. At the same time, I'd prefer not to limit my resources to teams that have both an oscilloscope and a dynomometer.)


Any links or first-hand data? Public libraries are notoriously poor at returning results for "brushed DC motor".