MOSFET Automotive cabin fan speed controller

Thread Starter

teliocide

Joined Sep 26, 2013
100
The cabin fan speed controller in my Mrs's Subaru Impreza only works on max or not at all.
Assumption is the MOSFET cabin fan speed controller is dead.
(I replaced the motor 12 months ago.)
This is the variable type not the preset 1 2 3 max type, so it will most likely be using a MOSFET as a power resistor
Subaru want $US299 for the part but I think it includes half the dashboard.
The Mrs will not let me pull it to pieces unless I can guarantee it will be fixed with in hours.
So I need to do some homework.
This is what it looks like: (Internet photo)

SUBARU MOSFET.jpg

This is my best guess as what the circuit is
AAC.jpgSo I need to stock up on a few components before I start the physical work
Any Guidance on circuit and/or component values would be very much appreciated.


Thanks
Greg
 

LowQCab

Joined Nov 6, 2012
1,572
A Full ~12-Volts is getting to the Blower-Motor on "High",
so the FET is not smoked.
In order to determine were You need to start looking,
You need an accurate Wiring-Schematic for the Car.

Something else is going on, it could even be as simple as a worn-out Pot in the Dash.

It's not likely that they are directly driving the Gate of the FET with a Pot, ( as per your drawing ),
if for no other reason than that would be too simple, ( and not very Linear ).

The first thing I would do without a Schematic is to measure the Voltages directly at the Pot.
.
.
.
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
A Full ~12-Volts is getting to the Blower-Motor on "High",
so the FET is not smoked.
In order to determine were You need to start looking,
You need an accurate Wiring-Schematic for the Car.

Something else is going on, it could even be as simple as a worn-out Pot in the Dash.

It's not likely that they are directly driving the Gate of the FET with a Pot, ( as per your drawing ),
if for no other reason than that would be too simple, ( and not very Linear ).

The first thing I would do without a Schematic is to measure the Voltages directly at the Pot.
.
.
.
The FET is usually by-passed on max setting, usually with a relay
Is there "too simple" when it means more profit?
If it does the job, or at least appears to, and lasts a reasonable time, simple is good
I remember the instrument gauges on French cars of the 60's and 70's, they used a wire and a spring attached to the needle.
As the voltage across the wire increased so did its length and the needle moved accordingly. What they lacked in precision they made with being robust and cheap.
The system in the Subaru is very not linear with nearly all of the adjustment restricted to the last 50% of movement>
I simply may have over-ride the authority of she who must be obeyed.
 

MrAl

Joined Jun 17, 2014
8,604
I have found this Honda circuit View attachment 253862which is a start
So it is a turn on and turn off circuit.

That gate capacitor bothers me in both circuits, but it appears that they may be using that to "soft" turn off the mosfet, which helps to prevent drain voltage overshoot. So it's a controlled turn off and a little controlled turn on also although that part of it may not be needed.

Now using the mosfet to control the turn off means that the mosfet will be in the linear mode for a short time, and during that time unlike when it is full 'on' or full 'off' it dissipates more power. The energy during that time will heat up the mosfet more than if it was just fully on or off.
The trade off is the 'slowness' of the turn off versus the power dissipation during the time when it is turning off. Since the time to heat up is probably faster than the time to cool down, that means the mosfet can not be turned on and off too frequently unless another means to clamp that potential drain terminal spike. Since this kind of calculation requires a lot of extra data like the specific heat capacity of the materials in the mosfet or similar, it's probably simpler to make some measurements.
Turn the mosfet on and off using a test circuit and with the actual load and see if the mosfet gets hot. You have to turn it on and off at about the same rate as it will see in the actual circuit. If it turns on and off once per second, then test like that. It would be best if it turned on and then turned off, and then wait 10 seconds or something like that.
What you look for is a leveling off of the mosfet temperature after several hundred on and off cycles. It may rise fast up by 40 degrees C above ambient but then level off at 45 degrees C or something like that, or it may only reach 10 degrees higher than ambient. The key is to make sure that it does not rise too high.
Another factor of course is the heat sink. The larger the heat sink the better.
If we calculate the average power and it turns out to be 1/2 watt or something like that, then it may work without a heatsink. As the power rises it is wise to use a heat sink though. 2 watts for example better have a heat sink at least 2 square inches or better. There are some cheap heat sinks for sale that are small but can work up to 2 watts or more, just look around the web sites that sell them if you need one.

BTW, do you know what the turn on and turn off timing is going to be, worst case?
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
Maybe that's not exactly your part but my point is that there's no way it should cost that much.
I agree but check the part number 73533FE000
Subaru prices, outside of profit gouging, make no sense
When the motor wore out last year subaru wanted nearly $US500 for a new one. But you get most of the plastic housing and ducting that not required. After some hunting I found the fan motor is standard to most Japanese vehicles and I got one for $US20. Very funny but the fan motor rotation is different for left and right hand drive vehicles.
 

MrAl

Joined Jun 17, 2014
8,604
Apparently not.
The mosfet is used as a power resistor>
There are several videos on YouTube that demonstrate this with no ambiguity.
OH wow, that's the worst way to run a transistor. That means it gets heated all the time at least to some degree. The power curve would max at 1/2 voltage so if it was a 12 volt system the max power curve would top out at 6 volts. If the run current at that voltage was 4 amps, then the mosfet would be heating up having to dissipate 24 watts.
Do you happen to know the run voltage and max current for the load?
We can calculate the max power and by the size of your heatsink see if it is good enough or not.
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
It is not about clever it is about PROFIT
The MOSFET is not run at max>
When the fan speed is set to max the Mosfet is by-passed using a relay
The fan motor would draw 10 amps max
 

MrAl

Joined Jun 17, 2014
8,604
It is not about clever it is about PROFIT
The MOSFET is not run at max>
When the fan speed is set to max the Mosfet is by-passed using a relay
The fan motor would draw 10 amps max
Ha ha, yeah isnt that the truth :)

But it is not the power in the motor we are worried about, it is the power in the mosfet, so we concentrate on the power in the mosfet not the motor.
When the speed is set to max that probably means the motor is getting max power, but that is not when the mosfet is dissipating max power even if there was no relay and it was just turned on fully.
The speed setting that causes max power in the mosfet is not at full setting it is somewhere around half setting or maybe closer to three quarters of the max speed setting. That is when the mosfet would get the hottest. That could be called the max power point of the mosfet.
That occurs when the voltage is 1/2 of full and the current is lower than full. The power curve would look like an upside down parabola with the peak in the center of course and that is at 1/2 voltage for a linear load (resistive) and for a DC motor is is probably around there too although may vary a bit depending on the motor.

If you really want to understand this in full, then set up a current and voltage meter. Measure the voltage across the mosfet and the current through the mosfet and take a reading once the motor speed levels off after changing it (it may be quick or take a few seconds depending on the motor).
If you do this for several speed settings we can figure out what the maximum power in the mosfet is and then figure out if the heat sinking is good enough. But remember we are concentrating on the power in the mosfet not the motor.
It would be great if you could take 10 readings from low speed to full speed, but at least 4 readings one at 1/4 speed one at 1/2 speed and another at 3/4 full speed and the last at full speed. This should tell us a lot about the situation.
If you really want to get more exact on this, measure the voltage and current as above, but this time multiply the voltage times the current, and as you do that increase the speed and look for the point where the voltage times the current is maximum. For example, in a 10 volt 10 amp system you might see:
Lowest speed: 9 volts, 1 amps (9 watts)
low speed: 8 volts, 2 amps (16 watts)
mid speed: 5 volts, 5 amps (25 watts)
higher speed: 2 volts, 8 amps (16 watts)
higher speed: 1 volt, 9 amps (9 watts).
Notice how the power came out to be a maximum of 25 watts at 1/2 of the full voltage. That's for a perfectly linear load but the motor max point may be slightly different so check for that if you care too.
That's the best way to tell the maximum power the mosfet will be dissipating. It just takes a little more work with a hand calculator to calculate the power that's all. It will in fact show definitively what the max power in the mosfet is, provided the motor has the real life load connected also (could be gears and pulleys and max lift load or similar for example).
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
A Full ~12-Volts is getting to the Blower-Motor on "High",
so the FET is not smoked.
In order to determine were You need to start looking,
You need an accurate Wiring-Schematic for the Car.

Something else is going on, it could even be as simple as a worn-out Pot in the Dash.

It's not likely that they are directly driving the Gate of the FET with a Pot, ( as per your drawing ),
if for no other reason than that would be too simple, ( and not very Linear ).

The first thing I would do without a Schematic is to measure the Voltages directly at the Pot.
.
.
.
I now have the unit out and apart and tested
The MOSFET is a NEC K2500.
There is a short between the Drain and the Source = Dead
I cannot locate any NEC K2500 in my part of the Globe.
China and the US appear the main Sources but I will not pay $US46 to post a $3 part from the US>
So are there equivalents?
 

LowQCab

Joined Nov 6, 2012
1,572
That's an outdated and barely adequate part.
The specifications are not likely to be critical.
I will assume that it's in a TO-220 Package.

This means that about ~500 to ~1000 different common part numbers will work.
What country are You in, or I should say, what supplier do You want to use ?
DigiKey is World-wide.

IRF2805PBF
PHP191NQ06LT,127
IRLB8743PBF
IRF2807ZPBF
IRLB4132PBF
IRF1405ZPBF
AUIRF3205

DigiKey has over ~1000 of these numbers in stock, but maybe not in your Country.
.
.
.
 

MrAl

Joined Jun 17, 2014
8,604
I now have the unit out and apart and tested
The MOSFET is a NEC K2500.
There is a short between the Drain and the Source = Dead
I cannot locate any NEC K2500 in my part of the Globe.
China and the US appear the main Sources but I will not pay $US46 to post a $3 part from the US>
So are there equivalents?
You have to match or exceed the specs of the original part except for the gate voltage and drain to source voltage. For the DS voltage you can go higher but have to pay stricter attention to the max on resistance.
So with that in mind, you have to match the max drain to source voltage, the max turn on resistance, the current handling ability (like the max drain current), the power rating (probably the same for a similar package), but very important also is the gate turn on voltage. I am not sure if you quoted the operating voltage of the system but if it is 12v or under it may be a low gate turn on voltage device.
I'll try to look it up as soon as i can and see what i can find.
Where are you ordering from?
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
MrAI,
Thanks for your generous input.
LowQCab (see above) has supplied me with a list of substitutes
Now to decide which one will be the closest to plug and play
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
[QUOTE="LowQCab,
[/QUOTE]

Thank you for the substitutes list
I have checked them out and all are available in Australia at humble prices.
What annoys me is the part which I posted an image of when I started this thread is, as a Subaru part $299 in the US $AUD400-$900 in Australia BUT if it is purchased as Nissan part it is $US20 or $AUD30
I will still repair the dead unit ..... Its in my Scottish genes to do so>
Thank you for your assistance
 

LowQCab

Joined Nov 6, 2012
1,572
All of the FETs I listed are compatible,
pick the one that is easiest for You to obtain.

In the Automotive-Parts-Industry the standard practice is that the
price gets DOUBLED every time the part changes hands.
Yepp, 100% mark-up from manufacturer to Warehouse,
then ~10% from Warehouse to Dealer,
then 100% from Dealer to Customer.
This generally applies to ALL parts that start at under around ~$200.oo,
parts that start-out at a higher-price may have a lesser mark-up.
.
.
.
 

Thread Starter

teliocide

Joined Sep 26, 2013
100
Time to put this to rest
Recap original symptoms
First the fan would only run at max speed.
Next........ no fan at all.
Next... fuse dead.
The motor speed MOSFET (Resistor :)) was shorted ie there is a short between the Drain and the Source = Dead
The motor was tested with an independent power source. It rotated and moved air.
Thinking back the power supply wires were a bit warm.
Replaced the motor speed controller and fuses.
All worked for 5 secs until the fuse went POOF!
The motor speed controller was too hot to touch.
Conclusion, partial short through the motor
Pull motor out
The motor could be turned easily. There was slight drag but no obvious cogging.
These motors are not designed to be repaired, but if you carefully unbend enough steel they come apart.
Bearings seemed OK but the commutator looked like it had been chewed my a giant feral rat, like wise the brushes.
Chucked the armature in the lathe and tidied up the commutator and cleaned between every contact.
Sanded the brushes back to a normal shape.
Lubricated and reassemble with gentle hammering
Now easier to turn and could feel cogging.
Put power on it and it purred like a kitten and did not get hot.
Put everything back together (motor speed controller was not cooked) and hallelujah it all lives again.
But for how long
Thanks for all the help.
Blower motor com rooted.jpgBlower motor com cleaned.jpg
 
Last edited:
Top