Pulsed DC can loosely be defined as PWM although, I think in this case it has distinctions.
If PWM was used in a system such as this, on would expect that the 24 VDC peak would be a constant value. i.e. a 0-100% square wave havig an amplitude of 24V.
So let's say measurements were made and that the CTI throttle reverted to PWM when the throttle was 10% or less. Further lets assume that the peak was 10% of 24V or 2.4V. If we have a 2.4 V pulse at low throttle, probably the trainwill laugh at the PS, but IF the amplitude was 24 V with a duty cycle of 10% to 0% from 10% to 0% throttle, then the train would move.
Hence, intuitively, the mechanism used for wimpy, low inertia motors is appropriate. For BIG, high inertia motors isn't.
PWM preserves torque.
For instance, as an intuative approach, take a bunch of people trying to push an MG and a mid-side car. In the MG case it would be a case of push, roll, push, roll (like full blown PWM on a small train). With the larger car, full blown PWM would move the car slowly.
Thus, I'd be more likely to advocate PWM over the entire range of control for BIG motors. This is the cheapest but it relies on some intimate details.
Their design wouldact a little better than a traditional power pack in any case. The only way to get true low speed performance is to use PWM at the motor's voltage. There will be the case of setting the power pack at 5% to get the engine moving and then reduce to 3% to keep it moving slowly when starting from a full stop.
It would be in CTI's best interest to allow setting of that point (% throttle), voltage to use for 100% starting at that point and PWM to start at that point. Now you've added complexity.