PC Fan RPM sensors

[Sunny]

I needed a case fan for a PC I was assembling today, and decided to
recondition one I had on the shelf and put it back into service. It
happened to be one with a third wire which provides RPM sense so the fan
speed can be monitored by software.

After cleaning and oiling, I tested the fan on my bench PC - worked
nicely, but the fan monitoring software said it wasn't spinning. I
checked the RPM sense output on the scope and found a 12V p-p square
wave when it's supposed to be at TTL levels. This damaged the monitoring
chip on the motherboard as the fan sense inputs are rated 5.5v maximum
and there's no protection on the board. Fixing the motherboard involved
replacing the monitoring chip, which was *not* easy as it's a 48pin LQFP
very close to a PCI slot - not worth the effort, but I did it anyway for
SMD rework practice. (actually I got more practice than expected as the
first replacement chip I installed was so dead the board wouldn't even
boot and I had to try another - to be expected when you re-use chips
from dead boards. The second one was good)

I took the offending fan apart and found the RPM sense output is nothing
more than a 2N2222 connected between ground and the 12v input to one of
the coils - the motherboard is expected to pull the collector up to 5v,
but the fan designer made no attempt to prevent the output going to 12v
when the transistor fails.

Am I correct in thinking an internal short is a common transistor
failure mode, and all it would take to protect the motherboard is a 5.4v
Zener between RPM sense output and ground?

TIA

Sunny

 

[Marco]

Sunny wrote:

I needed a case fan for a PC I was assembling today, and decided to
recondition one I had on the shelf and put it back into service. It
happened to be one with a third wire which provides RPM sense so the fan
speed can be monitored by software.

After cleaning and oiling, I tested the fan on my bench PC - worked
nicely, but the fan monitoring software said it wasn't spinning. I
checked the RPM sense output on the scope and found a 12V p-p square
wave when it's supposed to be at TTL levels. This damaged the monitoring
chip on the motherboard as the fan sense inputs are rated 5.5v maximum
and there's no protection on the board. Fixing the motherboard involved
replacing the monitoring chip, which was *not* easy as it's a 48pin LQFP
very close to a PCI slot - not worth the effort, but I did it anyway for
SMD rework practice. (actually I got more practice than expected as the
first replacement chip I installed was so dead the board wouldn't even
boot and I had to try another - to be expected when you re-use chips
from dead boards. The second one was good)

I took the offending fan apart and found the RPM sense output is nothing
more than a 2N2222 connected between ground and the 12v input to one of
the coils - the motherboard is expected to pull the collector up to 5v,
but the fan designer made no attempt to prevent the output going to 12v
when the transistor fails.

Am I correct in thinking an internal short is a common transistor
failure mode, and all it would take to protect the motherboard is a 5.4v
Zener between RPM sense output and ground?

TIA

Sunny

Hi

Transistors die in all possible modes, open, short circuited. Normally
however, a transistor in the situation you describe won't die at all. I
guess it had died because something nasty happened on the RPM output,
probably in a previous life of the fan.

Then for the zener... The transistor must have a base resistor. When the
transistor fails, this one will be in series with the zener, so a zener
would help. Although I would use a 4V7 one.

Finally... I think it is a bad (cheap) mainboard / chip design if
something like this fries a chip...

Marco