[s-cars] HVAC diagnosis/repair - not for the faint of heart

[Mike Fitton]

After several months of diagnosing and slow repair of my HVAC system, I 
thought I'd do a full writeup and share my experiences with the list and 
make a contribution to the archives for posterity.  All of the following 
assumes you're prepared to rip your car apart in search of these hidden 
servos, hotwire little motors, see and tolerate the occasional puff of 
smoke, and that you're proficient with a multimeter and soldering iron 
in addition to basic hand tools.  If this is you, you may save a small 
fortune in overpriced Bosch electrics.

I had the relatively common problems with the servos not working 
correctly or at all.  Early in my ownership of the car, I used the HVAC 
headunit to pull the codes out.  The procedure's floating around 
somewhere...been a long time so I don't remember exactly.  I remember 
being scared and confused by the results of that, so I just sort of 
tolerated it for a while longer.  Later I finally got a VAG-COM which 
made things easier.  VAG-COM reported 12 errors in total.  Didn't write 
down all of them, but they were mostly servo-related.  No shock there.

So here's the skinny.  All four servos threw codes.  All four 
potentiometers threw codes.  That means every single servo box in the 
car was malfunctioning in some undefined way.  Under dealer/warranty 
situations, the driver notices the system misbehaving, takes the car in 
for service, it gets plugged into the dealer's scan tool, the problem 
servo is identified and replaced.  However, with warranties on our car 
long since expired, that service would now run about $500 in parts plus 
labor.  Totally unnecessary, I thought.

Okay, electric motors just don't "wear out" per se.  Yeah, there are 
moving parts and friction so they do wear out eventually, but these 
servos just don't face that kind of duty cycle.  They should last 
through the normal lifespan of the car in general.  Like any common 
electric motor, the servos use tension fit copper brushes on the rotor's 
shaft.  The friction at this interface eventually wears off enough 
copper dust to get in the commutators and short the motor out.  Motor 
stops and throws a code.  Four of these incidents do not add up to you 
needing to spend $500.  Just pull the freakin motors out, clean them out 
well, and reinstall.  Once reassembled, I recommend running the bare 
motor on a bench with like a 5-9vdc power supply.  On one of my motors, 
I found a little residual contact cleaner in there that had to be burned 
off as well as some burring on the copper brushes that caused some 
arcing.  Just run the motor both directions until that stuff stops so it 
doesn't happen in your car.  If you're at this point it's already 
broken, so you're not going to break it further, even if you melt the 
whole thing into a puddle of motor juice on your bench.

The motors themselves are small and a little intricate in their 
assembly.  Work slow and note how it all goes together.  Try not to lose 
any little pieces.  Also, I didn't have them all out next to each other 
at the same time, but I believe all four of the actual motors are 
identical.  In fact the only electrical difference among the four servos 
appears to be the resistance readings for the potentiometers.  Those are 
integrated into the servo unit and are used by the HVAC headunit to 
determine the location of the actuator.  It also appears that the 
headunit continuously learns the high and low values, so there's at 
least some measure of compensation for potentiometer wear.  There are 
five wires going to each servo.  Two control the motor; three are for 
the potentiometer.

But here comes the real curve.  Once all the motors were thoroughly 
cleaned, tested, and reinstalled, none of them worked.  After some 
initial swearing on my part, I enhanced my calm and got the voltmeter 
back out.  I used the output test thingy in the VAG-COM to operate the 
servos.  Their electrical designations are V68, V70, V71, and V85, 
though I can't remember which one's which offhand.  So anyway, I 
activated the servos directly with that and then measured the voltage at 
the harness.  They should all read in the neighborhood of 8.5v.  All of 
my harnesses instead read in the neighborhood of 0v.  Okay then.  Just 
to satisfy my curiousity, I checked for continuity in the wiring between 
the headunit and one of the harness...it checked out.  I didn't think 
that was it, but it was easy to rule out and I was going to feel stupid 
if that was it.  But it wasn't.

It was the voltage regulators on the HVAC headunit logic board itself.  
It goes down like this.  Two voltage regulators.  Each one supplies two 
dual-power op amps.  Each op amp handles one servo and is reversible so 
that it can drive the servo in either direction.  My voltage regulators 
were bad.  They're STMicro VRs, model L4940V85.  Mouser (.com) sells 
them for $1.50 each.  So.  Aside from the cost of however much Deoxit I 
hosed in the motors, I spent $3.00 plus shipping repairing my HVAC 
system.  A dealer would have replaced all four servos (~$500), found no 
change in the system, and then replaced the HVAC headunit at god only 
knows what cost.  That's fine under warranty.  It's not fine if I'm 
paying for it out of pocket.

Enormous credit goes to a UK man whose name I don't even know.  My work 
on this was enormously influenced and illuminated by his.  He provided 
me with all the particulars to put this together, including basic 
explanations of how it works and why it breaks along with the necessary 
part numbers.  His site is required further reading for this job and is 
located at:

http://www.eveshamsquash.co.uk/audi/climate.html

Happy soldering!

-Cheers!
Mike