I've come to this late so I don't think that I have anything practical to offer, but from the title I guessed that it was an EMI issue.
When controlling a motor by PWM, the voltage is switching rapidly between 12 and 0. That can cause radiated emission from the PWM wire, which is in this case the fan negative. However the larger issue is the variable current taken by the system. When the switching device is on, the current is taken from the 12 V supply. When the device is off, the current freewheels back through ground and no current is taken from the supply. If there is any inductance in the wire feeding the PWM controller, there will be large voltage spikes on that wire when the current stops, unless enough capacitance is added at that point to absorb the charge.
Also, the ESR of the capacitors is important. The ESR can often cause a larger voltage ripple on the capacitors than the impedance of the capacitance.
I guess that the voltage spikes were causing the alternator to shut down. I don't know why spikes had a larger effect than the average voltage.
When controlling a motor by PWM, the voltage is switching rapidly between 12 and 0. That can cause radiated emission from the PWM wire, which is in this case the fan negative. However the larger issue is the variable current taken by the system. When the switching device is on, the current is taken from the 12 V supply. When the device is off, the current freewheels back through ground and no current is taken from the supply. If there is any inductance in the wire feeding the PWM controller, there will be large voltage spikes on that wire when the current stops, unless enough capacitance is added at that point to absorb the charge.
Also, the ESR of the capacitors is important. The ESR can often cause a larger voltage ripple on the capacitors than the impedance of the capacitance.
I guess that the voltage spikes were causing the alternator to shut down. I don't know why spikes had a larger effect than the average voltage.