On 03/25/2013 06:59 AM, Scott Dorsey wrote:
In article , jim beam wrote:

well, you live and learn. apparently the reason they use a linear
controller is because it allows the fan to run near silently at low
speed. with pwm control the fluctuating magnetic fields in the motor
coils cause it to vibrate and make a humming noise at the pwm control
frequency.

Yes, this is why you put an integrator stage after the pwm stage, so that
the motor sees nice filtered DC with very little of the PWM leftover.

Problem is that the integrator stage costs money and big electrolytics
tend to have limited life, so auto folks don't like doing that.

you don't want to integrate the output, merely rub the shoulders off the
square waves to get the harmonics down. the whole point and benefit of
pwm is that you have full voltage full power available in each pulse.
that's how you can start and control a motor with high torque at low
rpm. if you integrate or smooth out the motor's supply, you effectively
lose that and the motor won't start or torque the same way or even at all.



that doesn't of course get around the fact that the unit in question
here is apparently badly under rated, but the above does at least
explain why it's used.

It's a cheap, reliable way of doing the job, if it's done right.

in this day and age, that's no longer true. motor control is one of the
hot ticket items on the silicon fab agenda, and has been for some time.
there are some great pwm options out there, and for not a lot of money.


It's clear
that it wasn't done right, but I'm still waiting to hear what was done wrong.
Given all the RoHS-related failures and the report that touching up solder
joints on the transistors fixes the problem, I am suspicious that it's a
soldering issue made worse by the extreme temperature cycling.

you definitely have a point there, but given the size and shape of that
heat sink, i don't think there's any way that silicon is getting
sufficient cooling, and is clearly way up against its ceiling. whether
that's accident or design is another matter, but the bottom line is that
it's an issue that spans multiple different module manufacturers across
multiple continents with different internal designs - that reduces the
probability of it being rohs and slaps it firmly into the vehicle
manufacturer's lap.


--
fact check required