Root cause insight into the common BMW blower motor resistorfailures

[Jamie]

Any evidence it was checked with a scope?

Yes.

This quote below is verbatim from this location:
http://www.bimmerfest.com/forums/showpost.php?p=6536514&postcount=131

Hi, After soldering the lost/refound component,
remaking the joints of the 2 mosfet and testing the FSU alone with an
oscilloscope, here are my observations:

- the FSU works again

- there is no PWM ,

You said you tested the FSU alone. If so, how can you
say there is no PWM signal between the car and FSU?


the gates signals are continuous voltage only , this is the reason
why it heat so much its aluminium box... In fact there is no point on
the board where square signals are present. Can somebody check its own
FSU if it's the same ?

- the 2 bridges are in fact 2 resistors 10 milliohm used to balance the currents between the 2 MOSFET and balance the power also. The mesure of the DC voltage on those resistors can be used to evaluate the current of the blower and its worn state.

Say what? 10 milliohms is .01 ohms. How could that
possibly balance the power to a motor in a 40 amp circuit?

Resistors are used in the emitter path when combining two or more
to the same circuit. This is needed to insure both transistors
share in the load when it comes to biasing... Otherwise, you'll
get one that favors beta and the other will sit back and snooze.
Call it a ballast R if you wish.. These R values are generally
low.

Jamie

 

[Jamie]

I 'think' (but I'm not sure) that these are the resistors in series:
http://www5.picturepush.com/photo/a/12516343/img/12516343.jpg

Yup, and I see the legs of the transistors (two). So those R's are in
line with the emitters where they joint.

Most likely thermo stress cracks due to the potting restraint.

Jamie

 

[Jamie]

If they are resistors, I've never seen any that look like that.
Also, given that you want to thermally bond any components
that generate major heat, why are they not heat sinked?
With any power design I've seen, the key components, eg
the transistors are directly bonded to the heat sink.

THey make power resistors in T0-220 style cases, for heat sink mounting
however, I doubt those are that.

Jamie

 

[jim beam]

Yawn. I worked with 16 layer boards at a factory troubleshooting
defective, new boards that cost over $8,000 to stuff.

fantastic. could you condescend to help these guys wit their project
then? or are you just here to whang your donger around?

The internal
routing of a simple low frequency board is irrelevant for drawing a
schematic. You can X-ray a board or mill it one layer at a time if you
want to duplicate the routing. All you need to do for bais reverse
engineering is to trace each with an ohm meter by probing every pad and
termination to identify the signal path and draw a schematic from your
notes. I've done this with four layer boards for 30+ years.

then you're a complete genius because other people find it hard/impossible.

"Specs on
the chips" makes things easier but knowing who made it, and how it's
connected will tell you if it is a custom part or just house numbered.

yeah, a schematic doesn't exactly mean much if you don't even know what
the components actually are. [see above]

There are industrial solvents to remove any potting compound, but
they aren't cheap or easy to buy. Failure analysis is a specialty in
electronics manufacturing. It is expensive but gives you the answers
you need so you can design out the problems.

expensive? you're not kidding. and failure analysis is a whole lot
easier if you know what you're looking at in the first place.

A jig to hold the module in a CNC machine would allow you to cut
accurate holes down to the right points to see if the IC was bad,
without unpotting a failed unit.

yeah, if you know precisely where you're drilling, in three dimensions.
and you have dil/soic [etc] type chips. anything bga or similar, and
you're sol.

and you're not doing that with a black box and getting meaningful data
out of it.

Then a test fixture with 'Pogo Pins'
would allow you to see what was bad. If there were enough bad units, it
would be worth designing and programming a computerized fixture.

ok, but you're missing one simple thing - this is a bunch of guys with a
dremel and a dental pick. money/time are limiting factors, even if
there were logic to reverse engineering, which i don't think there is in
this case.

 

[jim beam]

Like I said before, don't drive one then. It's kind of like going on a
date with that unbelievably attractive female type who is also smart,
witty, fun to be around, actually seems to like you, and oh by the way
is completely mentally unhinged.

no, what you're experiencing is her "disappointment" at discovering that
you are an anosognosic retard.

 

[jim beam]

jim beam wrote:

On 03/26/2013 05:58 AM, Scott Dorsey wrote:

On 03/25/2013 05:57 PM, Jamie wrote:

jim beam wrote:

Another way to do this, is to have an inductor on board with the
speed control circuit. You would PWM that inductor in series to a
filter
cap
on the output which will then give you a clean variable DC. THe
inductor
will be doing all variable voltages.



ok, as i understand it, and as i said to scott earlier, this is a
problem because it mungs low speed motor start and low speed torque.



Not really, it's feeding the motor with variable DC from that
integrator
stage.



i understand that - and variable voltage is the problem. the
secondary [bordering on primary in some applications] advantage of pwm
is low speed start and torque. if a motor starts at low dc voltage,
not only is the start speed inconsistent, it has little torque. pwm
can start a motor slower and at much higher torque. it's a big deal.


A properly working blower motor does not need extra torque to start at
low RPMs.

Torque is only needed when RPMs increase and mass air flow is
increased, thereby, putting a strain on the motor.

you're right, except that there are more variables. very cold days,
very windy days, blown snow powder, leaves, all kinds of things can
mess with the motor starting at a low speed.

Basic resistor
systems will vary in speed if air pressure isn't constant, and in
most cases it isn't..

When there is no air flow or the flow has been restricted somehow,
there is little to no torque demand, other than mechanical of the blower
blades and those should turn easy, sine bearings and balance permits
this.

PWM is just a cheap way of speed control, it does not mean it's
better, in fact in some ways it's not, due to over head in noise..

from what i can see, the /only/ drawback is noise. power efficiency,
controllability, speed consistency, and yes, sometimes price, all are
wins for pwm.

A linear control with feed back will provide the needed torque but
they do tend to run hot when throttled back, because of the resistance
being present between the 12V and the motor terminals. THis is where
PWM comes in a winner but then you need the added cost of noise
reduction engineering.

I can only assume the linear module at least uses a feed back to
maintain output voltage, if it is so cheap that it does not even
do that, then maybe they are trying to emulate a real resistor or
they are just shitty engineers or tightwads.

well, they're clearly failing at something if they're trying to
provide an engineering solution. if however they're providing a
financial solution with a per-determined failure rate, then they're
right on target.

If that being the case, I guess we now know why the module keeps burning
out

But there is factor that maybe you have forgotten or didn't know, and
that is, the resistance of the DC motor. stall torque can be limited to
what the DC R value is,

didn't know.

in otherwords, this value forms a voltage
divider and thus low voltage at stall current could seem like no voltage
and not turn.

which would be another factor in favor of pwm...

DC PM/SHUNT motors attempt to compensate when load is dragging it
down in speed, that also includes a slow start. So as long as the speed
control can maintain a low voltage set point even when the motor is
calling for high amps - the motor R, it'll still start. However, there
is another factor, the speed control may not be performing armature feed
back and simply supplying current only..

that would be my guess.

If this is the case, then the
motor will stall at low speed demands in conditions that make it hard
for the motor to start.

I've also seen them allow the motor to run in torque mode to adjust
for air density. It'll simply self adjust naturally, and in those cases
you do not want armature feed back but torque feed back. Of course,
this will cause heating of the module when torque demand is low..

i guess that's another part of what we're looking at here.

thanks for the feedback!

 

[Jamie]

On 03/26/2013 05:58 AM, Scott Dorsey wrote:

On 03/25/2013 05:57 PM, Jamie wrote:

jim beam wrote:

Another way to do this, is to have an inductor on board with the
speed control circuit. You would PWM that inductor in series to a
filter
cap
on the output which will then give you a clean variable DC. THe
inductor
will be doing all variable voltages.



ok, as i understand it, and as i said to scott earlier, this is a
problem because it mungs low speed motor start and low speed torque.



Not really, it's feeding the motor with variable DC from that
integrator
stage.



i understand that - and variable voltage is the problem. the
secondary [bordering on primary in some applications] advantage of pwm
is low speed start and torque. if a motor starts at low dc voltage,
not only is the start speed inconsistent, it has little torque. pwm
can start a motor slower and at much higher torque. it's a big deal.

A properly working blower motor does not need extra torque to start at
low RPMs.

Torque is only needed when RPMs increase and mass air flow is
increased, thereby, putting a strain on the motor.

you're right, except that there are more variables. very cold days,
very windy days, blown snow powder, leaves, all kinds of things can mess
with the motor starting at a low speed.

Basic resistor
systems will vary in speed if air pressure isn't constant, and in
most cases it isn't..

When there is no air flow or the flow has been restricted somehow,
there is little to no torque demand, other than mechanical of the blower
blades and those should turn easy, sine bearings and balance permits
this.

PWM is just a cheap way of speed control, it does not mean it's
better, in fact in some ways it's not, due to over head in noise..

from what i can see, the /only/ drawback is noise. power efficiency,
controllability, speed consistency, and yes, sometimes price, all are
wins for pwm.

A linear control with feed back will provide the needed torque but
they do tend to run hot when throttled back, because of the resistance
being present between the 12V and the motor terminals. THis is where
PWM comes in a winner but then you need the added cost of noise
reduction engineering.

I can only assume the linear module at least uses a feed back to
maintain output voltage, if it is so cheap that it does not even
do that, then maybe they are trying to emulate a real resistor or
they are just shitty engineers or tightwads.

well, they're clearly failing at something if they're trying to provide
an engineering solution. if however they're providing a financial
solution with a per-determined failure rate, then they're right on target.

If that being the case, I guess we now know why the module keeps burning
out

But there is factor that maybe you have forgotten or didn't know, and
that is, the resistance of the DC motor. stall torque can be limited to
what the DC R value is, in otherwords, this value forms a voltage
divider and thus low voltage at stall current could seem like no voltage
and not turn.

DC PM/SHUNT motors attempt to compensate when load is dragging it
down in speed, that also includes a slow start. So as long as the speed
control can maintain a low voltage set point even when the motor is
calling for high amps - the motor R, it'll still start. However, there
is another factor, the speed control may not be performing armature feed
back and simply supplying current only.. If this is the case, then the
motor will stall at low speed demands in conditions that make it hard
for the motor to start.

I've also seen them allow the motor to run in torque mode to adjust
for air density. It'll simply self adjust naturally, and in those cases
you do not want armature feed back but torque feed back. Of course,
this will cause heating of the module when torque demand is low..

Jamie

 

[[email protected]]

This quote below is verbatim from this location:
http://www.bimmerfest.com/forums/showpost.php?p=6536514&postcount=131

Hi, After soldering the lost/refound component,
remaking the joints of the 2 mosfet and testing the FSU alone with an
oscilloscope, here are my observations:

- the FSU works again

- there is no PWM ,

/You said you tested the FSU alone.  If so, how can you
/say there is no PWM signal between the car and FSU?

/ the gates signals are continuous voltage only , this is the reason
/why it heat so much its aluminium box... In fact there is no point on
/the board where square signals are present. Can somebody check its own
/FSU if it's the same ?

- the 2 bridges are in fact 2 resistors 10 milliohm used to balance the
currents between the 2 MOSFET and balance the power also. The mesure of
the DC voltage on those resistors can be used to evaluate the current of
the blower and its worn state.

/Say what?  10 milliohms is .01 ohms.  How could that
/possibly balance the power to a motor in a 40 amp circuit?

Not to the motor idiot. To the transistors.

- I guess the principal duty of the computer on the other side is
switching off the power transistors if the control voltage goes under 1V.

/Which makes no sense at all.

Certainly not to you.

I put the FSU back in the car and it still work, I don't know if it will
last long. because of the heat...

http://www.bimmerfest.com/forums/attachment.php?attachmentid=306784&d....

/It seems they last at least a few years.  Maybe it's like
/gas.  You buy it and you go so far.   I think you're in way
/over your head here;

But not yours?

No, because I'm not the one trying to reverse engineer
an electronics module in a car, that contains among other
things, an unknown 16 pin chip, without benefit of any of the
necessary tools. You guys don't even know what the interface to the
car is, whether it's analog or digital, etc. And you don't even have
an oscilloscope to look at
anything with.

Oh, BTW, if you're all so smart, how come I was the
first one to find out for you that the 16 pin chip number
you had is a vaild one for a real chip?

 

[Scott Dorsey]

You've never seen strips of nichrome in a space heater?

Bosch has done the nichrome insert power resistor for many years; I know
that they used them in the turn signal flasher in the late seventies when
they first went electronic. They aren't really very good resistors but
they are very cheap.
--scott

 

[Scott Dorsey]

Actually, it's the same transistor, which broke in half while
I was attempting to get the black rubber eraser stuff off of it
to read the numbers.

It's really going to be HARD to read those numbers now...
http://www4.picturepush.com/photo/a/12516062/img/12516062.jpg

Okay, there is a TO-220 package that is split open there, with the
backplate on the right and the cover on the left. Can you chip off
enough of the araldite from the cover to be able to read the numbers
on it?

Or, could you get a good picture of the die which is left on the piece
on the right? We might be able to identify it with a sharp photo of the die...
although from what I can see from the fuzzy photo it does not look like a
very happy die.

If it is actually a MOSFET it will look like this:
http://www.panix.com/~kludge/fet1.jpg

(That's a package that is a little bigger than the TO-220, but you can still
see the channel down the middle of the FET and the overheating damage to the
source.... the three leads have been torn off in the unpotting process though.)
--scott

 

[jim beam]

No, I'm not like you. I have worked in Electronics for 52 years. I
could draw the schematics, but I would need a handful of bad modules and
the time to do it right. Just like the kU band microwave audio, video,
and data terminal hardware I worked on that's in orbit aboard the ISS.
It takes hands on effort to reverse engineer a design.

so what you're saying is that you're just here to piss and moan because
you know how to help, but won't.

and credentials don't work on usenet - they're completely uncheckable
and many are bogus. what matters is whether you can walk the talk.

as for having stuff in orbit, i don't have anything, but two of my best
friends do. the difference between them and you is that they're not
jaded and they're actually helpful.

Bomarc has reverse engineered a lot of automotive modules:

<http://www.bomarc.org/basement/viewtopic.php?f=1&t=6&sid=8ee707756ef37b24ff5aa633f1b4548a>

The one you want might be on their lists.

that at least is vaguely useful, but you could have come up with that 20
posts ago.

 

[jim beam]

Yawn. I don't have access to any defective modules. I don't know
anyone at the local BMW dealership to ask for failed units and I'm not
going to buy a new module to destroy for a whiny piss ant like you.

i don't know who you think you're talking to, but i have consistently
advocated /not/ deconstructing this unit. it's not worth it when cost
of replacement isn't that high or you can build an alternate controller
that will be more reliable.

You can't even crawl.

at least i can follow a thread without being a crotchety old fart.

Then tell them to reverse engineer it for you. Or can't they 'walk
the talk', either?

i don't /want/ to reverse engineer it any more than i want to repair
broken light bulbs. you were the one bragging about how easy it was. i
said it wasn't. and when it comes down to walking the talk, you won't.

And you couldn't so you piss and moan. I couldn't remember their
name, since I hadn't seen one of their print ads for over 20 years. I
looked them up, when I did.

i guess we should be grateful...

 

[Leif Neland]

Følgende er skrevet af Nate Nagel:

Restriction of Hazardous Substances; that is, no lead (among other things.)

http://en.wikipedia.org/wiki/Restriction_of_Hazardous_Substances_Directive

This does not count the hazardous issue of the soldering failing,
causing hazards to the users of the equipment :-(

 

[Scott Dorsey]

Følgende er skrevet af Nate Nagel:
This does not count the hazardous issue of the soldering failing,
causing hazards to the users of the equipment :-(

That's why there are exemptions for the military and telecom industry,
where it's actually important that stuff work properly.

What I find ironic is that the shorter lifespan of consumer gear caused by
the RoHS manufacturing has actually increased the amount of electronics
going into landfills, making worse the problem that it was intended to reduce.
--scott

 

[Bimmer Owner]

I can only assume the linear module at least uses a feed back to
maintain output voltage, if it is so cheap that it does not even
do that, then maybe they are trying to emulate a real resistor or
they are just shitty engineers or tightwads.

Most of the people on the BMW forums think it's one of these.

 

[Bimmer Owner]

Most likely thermo stress cracks due to the potting restraint.

Other people have suggested this also.

Some say the potting is what is causing the stress cracks.

Re-insert without potting, is the "said to be" solution.

One question:
If the FSU works without potting, what was the purpose of the potting?

 

[Bimmer Owner]

It only does AC amps..

Good catch!

 

[Bimmer Owner]

Can you chip off enough of the araldite from the cover to be able to
read the numbers on it?

I will try this morning. (I was away on a trip for the past two days).

PS: Had to look up araldite:
http://en.wikipedia.org/wiki/Araldite

 

[Bimmer Owner]

Does the red jumper connect to the 40 amp fuse?

The red jumper was a hack added by one user to fix the solder cracks,
I think.

 

[Bimmer Owner]

In your other pictures, the spring clips just hold the transistors against
the heat sink.

This implies two transistors (although I only found one).

I will dig through the mess again - but I think I was too fat thumbed
when I cut it open, and may have destroyed the evidence.

I do have a second FSU (since two failed on me) though ... but I want
to try to FIX that one (instead of destroying it).