Yeah, unfortunately it's only a patent, so it's vague, with no specific practical application data. It has been clear for a while from the Delphi website, that the actuator can be controlled on both PWM and CAN, but on there there was no datasheet, only PR stuff. The question is, whether they use the same wires, or are they different wires, but Holset excluded the PWM option...
I think it is technically doable, that they share. On Microship uC-s, special functions (RS323, A/D, USB+-) always share pins with general purpose I/O, maybe ST does too, with the CAN wires. It is just a question of flipping bits to switch which function of the pin is used. So the PWM and CAN wires could be the same. Also, the wire may be split, and go into both the CAN, and some GPIO pin. Looking at the flowchart, it listens to only on one "interface" at a time anyway, prioritizing CAN, so it won't mistake CAN frames for PWM. And there's no way a PWM signal will make up a valid CAN frame.
Mistaken, I don't know, what your plans are exactly with the unit, but if you decide to get an inside look at the actuator, look for the part no. of the ST, so that we can retrieve the datasheet. With a datasheet, we can expect to be a lot smarter:- it will list the pins I/O voltage levels, that should give us the voltage levels needed for the generated PWM. As for the frequency, it's mostly irrelevant, as the control information is carried in the duty cycle. Probably around 100kHz-1MHz should be fine, but the frequency limits for input pins should also be listed there- just generally looking at the pinout, and where the CAN wires are connected (keep a beeping multimeter handy! ) should give an idea about whether PWM and CAN share the wires, and is it one wire, or two wire differential, etc. If you're really lucky, the ST even has a PWM type input pin, and CAN lines are connected to it, too (if not, A/D pins can be suspects for a PWM input, too).
If you don't want to take the unit apart just yet, but feed it PWM on the CAN wires, I'd try generating a PWM signal, that slowly (few seconds) cycles from, say 10% to 90%, that should definitely get it out of its default powerup state. I'd try it with 2,5V peak first, then 3,3V, then 5V, and also make it a differential output, too (normal signal on CANhi, negated signal on CANlo, in sync as close as possible). CAN is differential, too, because this way it's much more insensitive to noise and EM radiaton/interference, I wouldn't be surprised, if the PWM signal would be expected in differential form, too. These actuators are intended for heavy duty industrial environments after all.
Something about J1939 voltage levels:
http://www.can-expo.ru/files/J1939.pdf, although without knowing whether CAN and PWM go into the same pins on the ST, not much conclusion can be drawn from this...
WhatsADSM, do you have this turbo also? Have you maybe pulled it apart already?