Fan-Hax Inc
For some reason, some people on a local electronics forum got the idea that i do reballs. I only reflow, i don't have the equipment to change the solder balls. I used to send them to a guy i know, but since i got burned with my own dv9000, i no longer recommend him.
Now, i got a message about a tx1000 with nvidia 6150. Goes black when heats up, and starts up again only if it's allowed to cool. Guy bought it from an auction or something, as the touchscreen needs to be replaced, and he told me he opened it up to find a coin (!!!) between the 6150 and heatsink. Now that's ugly... especially since our pennies are made from an alloy with a crap thermal transfer coefficient.
I told him i can only inject flux under the chip, reflow it, and make an aluminum shim for it, and that it's likely to need a fan mod as well to make it react to the GPU/NB temperature as well, not just the CPU. Now, i've had this idea for a while but never gotten around to trying it out.
As you know i've modded my dv9000 with an extra fan over the GPU, but it's been bugging me for a while that while the tiny 30mm fan i used there is whining away, the main fan stays lazy because it's only dependent on CPU temp. Since the fans in these things are 4-pin, accepting a PWM control signal, it's fairly trivial to design a circuit to take input from a couple thermistors placed on the heatsink, one near the GPU, one near the CPU, and make it into a PWM signal, hijacking the control line of the fan.
The beauty of it is that the fan is still powered from the same place, also the sense line remains there, so the computer won't know the difference, and the result is OS-independent and does not require ugly patches like fooling the temperatures in the DSDT and such, as seen elsewhere.
I will not make an attempt to linearize the thermistors as their resistance vs temperature curve is advantageous here - i need the fan to stay as quiet as possible below 50C, but do its best so 60C temperature is not exceeded. I've calculated the thermistors to provide maximum input voltage to the controller at 55C for the GPU, and 65C for the CPU. I will be using a MC34063 SMD for my PWM control, and by the looks of it, it won't need much extra parts. Two resistors for each input (just because i can't find exact values), another resistor for the output, two diodes for summing, a few ceramic caps and of course, the thermistors. That's all.
Told him to come on Monday, so that'll be plenty of time to tune the design.
Now, i got a message about a tx1000 with nvidia 6150. Goes black when heats up, and starts up again only if it's allowed to cool. Guy bought it from an auction or something, as the touchscreen needs to be replaced, and he told me he opened it up to find a coin (!!!) between the 6150 and heatsink. Now that's ugly... especially since our pennies are made from an alloy with a crap thermal transfer coefficient.
I told him i can only inject flux under the chip, reflow it, and make an aluminum shim for it, and that it's likely to need a fan mod as well to make it react to the GPU/NB temperature as well, not just the CPU. Now, i've had this idea for a while but never gotten around to trying it out.
As you know i've modded my dv9000 with an extra fan over the GPU, but it's been bugging me for a while that while the tiny 30mm fan i used there is whining away, the main fan stays lazy because it's only dependent on CPU temp. Since the fans in these things are 4-pin, accepting a PWM control signal, it's fairly trivial to design a circuit to take input from a couple thermistors placed on the heatsink, one near the GPU, one near the CPU, and make it into a PWM signal, hijacking the control line of the fan.
The beauty of it is that the fan is still powered from the same place, also the sense line remains there, so the computer won't know the difference, and the result is OS-independent and does not require ugly patches like fooling the temperatures in the DSDT and such, as seen elsewhere.
I will not make an attempt to linearize the thermistors as their resistance vs temperature curve is advantageous here - i need the fan to stay as quiet as possible below 50C, but do its best so 60C temperature is not exceeded. I've calculated the thermistors to provide maximum input voltage to the controller at 55C for the GPU, and 65C for the CPU. I will be using a MC34063 SMD for my PWM control, and by the looks of it, it won't need much extra parts. Two resistors for each input (just because i can't find exact values), another resistor for the output, two diodes for summing, a few ceramic caps and of course, the thermistors. That's all.
Told him to come on Monday, so that'll be plenty of time to tune the design.
There's one ugly jumper wire but i couldn't be bothered to shuffle the component placements again. For now, one thermistor input only, as that's all the tx1000's gonna need. PCB for 2 inputs soon to follow.
Upon inspecting the thing i noticed that the fan was 3-wire, so it was voltage controlled not PWM. I had to do some extra hax to the board, namely adding a MOSFET and lowering the PWM frequency from 25kHz to somewhere around 10Hz, didn't measure it exactly. Anyway, the PWM clicking is barely audible and that only if you put your ear close to the exhaust, and once the fan ramps up you can't hear it at all.
CAPs POOF
Comment