Voltage Control Current Source (car's blower motor speed controller)

[pplaninsky]

Hello,

recently I gto 3 heater's blower motors from Volksvagen.
There are pictures and all in the second post.
I did some reverse engineering - these motors are really useful for projects that require a motor with a turbine for various ventilation (or other purposes).
And they are dead cheap, mechanically sound and ready to use after cleaning.
I got interested how the thing works, so I reverse engineered them - which was quite a feat as the PCB is tightly potted.
After I mostly destroyed the first one and stabbed myself in the finger with a scredriver,
I did the second one better after I discovered that the potting compound becomes very soft after hot air treatment.
Then I burned my finger  on my hot air station, which was about 400 C.
It felt more like electric shock rather then the usual burning sensation. And it sizzled...

With the price of the aforementioned personal injuries, I present to you the schematic of the thing,
which I believe is accurate. Checked it 10 times, found mistakes and corrected them.
But there might be some that slipped my attention.

I have some question though - I am not very good in analogue stuff and power supplies.

* why there are two MOSFETs in parallel?
* Why there are two op-amps? It seems like U1B is just for sinking current? Is it a protection from short circuit and over heating?
* in general how the whole thing works - it seems the op-amps are comparators, I does not seem that there is any feedback. Just capactitors - I guess for stability.


Some info from my tests:
-> At about 3.8V at the CTL pin the motor has sufficient current and starts to spin
-> At about 4.2V it is at max and consumes about 4 amps. (It actually triggers the over
current protection on my supply, so I can't say for sure if it is max and how much current it will take.)
-> The ramp up in speed from is pretty fast, though it feele linear.

[pplaninsky]

And here are some pictures.
VW part number is: 740 226 033 F02

[pplaninsky]

And here is the PCB analysis for those playing along at home ...

[NiHaoMike]

Seems like a very weird circuit. It's like it was designed by a drunk intern who was given a breadboard and a box of parts with the instruction to "just make it work".

If I were tasked to design it, I most likely would have just implemented a PWM controller based on a 555 and called it done.

[Whales]

My head hurts trying to decode that.

Opamp U1B: Indeed it's some sort of shutoff, probably related to the NTC.  EDIT: Or maybe it oscillates for current control or something similar, as you suggest.

Why there are so many resistors and diodes however... was this circuit designed by a group that is only allowed to add parts to their prototypes, never change existing ones?  Maybe some weird safety/compliance/approval framework is leaking in nonsensical ways.  Diodes Means Protection, How Dare You Suggest We Remove A Diode.

I wonder if this is some evolution/adaptation of relay circuits complete with thermal time delay components.

Two mosfets:  My first thought was staggered startup to avoid large transients.  Perhaps U1A and C3 are used to turn on Q2 with a delay compared to Q1?

D5A/B: Probably some sort of TVS package for protecting the mosfet gates.  Alas the mosfet gate wires are not externally accessible, so does this actually help?  Maybe if we assume the whole circuit gets given large transients on the 12V rail and instead of filtering the power supply for the whole circuit they just TVS the most vulnerable parts (mosfet gates).

EDIT: Are you sure this is current controlling the motor?  The current sense resistor being used is the mosfets themselves, whose RDSon will drift with battery voltage and temperature.  Maybe they're just using the opamps as digital logic gates with arbitrary thresholds?

[Zero999]

I think you should double check the reverse engineering.

Where's the freewheeling diode across the motor?

[pplaninsky]

Hey Everyone,

Thank you for all your replies.

I did check the schematic and my reverse engineerin 10x times.
I wouldn't claim there couldn't be any mistake, but the chances are very minimal. But I will re-check once again.

I agree that the desing is weird. It was designed before 1997 with probably 105 German regulations in mind, i.e. fire-safety and tons of other safety, which I cannot guess.
Imagine turning on the heating of your car and it catches fire. Big news - even in 1997 

On the other hand - it is an auto part. I think there will be a lot of transients - I suppose - you know ignition coils and other things all connected to a common chasis
where the only main voltage regulator is the car's battery.

Meanwhile, with the second depotting, the PCB came out undamaged enough, so I was able to hook it up and probe around a bit.
In fact Q1 starts a bit earlier than Q2 and always has a bit more current on it.

@Whales - the voltage on CTL is controlling the current via Q1 and Q2. I did test that multiple times.
There is no current flowing in from the CTL - it is high impedence path.
However, about 3.8V on the CTL the MOSFETs start conductin and the motor starts turning. At 4.2V it is fully on.

"Diodes Means Protection, How Dare You Suggest We Remove A Diode." Laughed my a** off.
I imagine that is more true than we think in Germany. Everything is over-engineered. Or at least it was.
The other option is the intern with  bread-board 

[magic]

I think U1A is simply a comparator with limited slew rate due to C3.
Some sort of delayed-something, but what's the point if the FETs are in parallel

[inse]

The metal strip is a shunt for current sensing. So R15 is connected to the source of the MOSFET.
Also check the diode network D5/D2 as it does not make sense in this configuration, maybe D2 is a zener diode?

[pplaninsky]

The metal strip is a shunt for current sensing. So R15 is connected to the source of the MOSFET.
Also check the diode network D5/D2 as it does not make sense in this configuration, maybe D2 is a zener diode?

The metal strip has the number 10 on it. It does not register any resistance, not a miliohm. At least with my multimiters (3 of them).
I thought it could be a 10A fuse ...

It makes sense to be a zenner, so I did check that. However, up to 15V no conduction in reverse direction. My power supply does not go beyond, so if it is higher voltage Zener - I would not know. However, I deduced it should be a regular diode, because that is supposed to be a 12V system - but who knows. Voltage drop in the forward direction is the usual 0.6V.

[pplaninsky]

I think U1A is simply a comparator with limited slew rate due to C3.
Some sort of delayed-something, but what's the point if the FETs are in parallel

Yep the most puzzling thing is the parallel MOSFETs.
One guess I will have is: that is hard to replace part - you need a lot of dashboard dismantaling to change it.
Back in 1997 engineers thought serviceability as well. So, my buest guess is redudancy. The circuit should work perfectly well with just 1 MOSFET.
These are BUZ102 from Siemens, which have Id=42A - more than enough for that motor.
If it happens one of the FETs to go bad - everything still works ...

I will check that this evening and will post an update.

[pplaninsky]

One thing I forgot to mention is the I designated the 'DRAIN' pin on the connector as an output pin.
I guess it goes to the car control unit, which measures voltage drop or something, senses short circuit protections or something.
I would not imagine any reason that to be an input pin. Also the wire going to that pin on the connector is very thing, same as the wire going to the CTL pin.
The wires going to the power pins +12V and GND are thick, i.e. something like 8 AWG.
Whereas the wires going to the CTL and DRAIN pin are very thin 20 AWG or beyond, so they don't carry current for sure.

[inse]

Thinking again, D2 must be a zener, surely above 15V as it should not conduct in normal condition.
The voltage must be just below the rated D/S voltage of the MOSFET as the zener will provide a clamping function for back EMF.

[pplaninsky]

There you go ...
Disconnecting MOSFETs to see what happens.

[pplaninsky]

Made some further investigations.
Indeed some of the diodes are Zeners - tracked them down by the SMD markings.
However this does not change much the whole pciture.
D2 is a 39V zener protecting the gates of the MOSFETs
D4 is a 12V zener clamping the control voltage to max 12V
D1 is a 15V zenere regulating the power-in voltage.
D7 is not a zener (most probably). Just reverse polarity protection.
D3 (MELF) is not a zener, I think it prevents the op-amp to source current. So, this op-amp can only sink current.
D6 (MELF) is not a zener and it's function is a mystery to me.

U1B would indeed drops the voltage of its output in over current situations and will start sinking current - measured that in practice.

Checked the rest of the circuit once again and it is waht it is. Did not find further mistakes in my reverse engineering.