Hardware Hacking the T20 CPU Fan to Always-On

[HoldMyBeerAndWatchThis]

Anyone done this?

Since it's a variable-speed fan there is either:

• A signal input to the fan to indicate to a controller chip in the fan what speed to assume.
  
  Circuitry on the T20 Mother board senses signals output from the fan and controls the fan voltage/current from there.

It likely is not so simple that you could hard-wire a couple of connections to +12 VDC and call it a day.

I'm starting this thread to pick up the trail from anyone here who has started down that path. Likely this will involve hooking a 'scope to the fan leads to see what sort of signals are there (I'm an EE).

I want that fan on full all the time. If I have to sneak in a little 8-pin microcontroller to generate the proper signals, then I'll do it.

[BillMorrow]

i assume that you have visited the fan control utility confrence..?

here: http://forum.thinkpads.com/viewforum.php?f=27

[HoldMyBeerAndWatchThis]

Yes, there are two threads that mention the T20. Only one has a reply. No useful information.

The Fan Control Utility is software based and is not intended for and does not function on the T20, correct?

I want to modify hardware and have the fan on at all times. Depending on the level of difficulty in the hack, it may be possible for T20 owners to simply drop in a new fan assembly with the mod already done.

I have a load of other work to do today but I'll pull out the 'scope this weekend and have a peek at the signals on the fan cable.

[Timoti]

Just an observation to keep you from being bored:
The fan of the T20 has 4 cables: red, blue, orange, black.
The T22 fan is compatible with this, but the black wire is missing.
I would assume that the black wire is the ground, and the red is VCC. So maybe the T22 fan uses the casing to ground itself? I'd check it with a DMM for myself, but the [censored] thing went AWOL on me...

On the other hand, apart from professional interest, why would you want to hotwire it? If the CPU runs too hot, just throw in a T21 or T22 fan: it has a far better heatsink, and passive cooling spares your battery. My old 800MHz CPU did run about 10C cooler with it.

[HoldMyBeerAndWatchThis]

Thats for the information.

Black doesn't have to be "ground." If it is ground in the T20/T21 it could be an EMC ground, not necessarily a current-carrying ground. As far as using the case as a power return, my intuition bets against it. I haven't seen this in any consumer electronic or scientific instrument I can recall at the moment. It is generally the rule that power supply currents should run as close together, going out and coming back, as possible. This reduces impedance but most importantly greatly reduces EMI emmisions.

On the other hand, apart from professional interest, why would you want to hotwire it?

1. The fan cycles on and off based on the temperature of the CPU. The temperature at which the fan kicks on is ridiculously high. A CPU should run as close to ambient as practical to prolong its life.

2. When under varying loads, the fan kicks on and off every couple of minutes. The CPU core soars up somewhere not too far under the boiling point of water. When the fan kicks on the temperature drops tens of degrees in a short period of time. If there is anything worse than running a CPU too hot, its cycling the temperature up and down. This induces thermal stress on the CPU package and die. A CPU, any CPU, should run as close as practical to ambient without cycling.

3. If the fan stays off, heat escapes the CPU through conduction through the heat sink. This heats the air inside the laptop case and therefore contributes to the inside case temperature significantly. If the fan were on all the time, cool room air would constantly be drawn in to keep the heat sink at a low temperature, thus minimizing the contribution of the heat sink to internal case heating. (Ever feel how hot a PCMCIA card is after you pull it out? It's basically a "dipstick" for internal case temperature. Granted, PCMCIA cards are designed to survive this heat, nevertheless, heat is not good for electronics components, especially electrolytic capacitors.)

4. I probably will install a T21/T22 fan unit with the fancy heat sink compound. BUT, this won't improve my internal case temperature or average CPU temperature much. Remember the fan won't be turned on until the CPU temperature is "too [censored] hot" already. The new fan and compound will lower my maximum temperature and increase the rate of temperature change. The latter not really a good thing.

There are usually drawbacks to any solution and they will probably be increased dust accumulation and shorter fan life. As you know, fans are cheap. I'll stash one or two known-working fans away just in case. As far as the dust, I've owned my T20 for four years and just recently cleaned the fan. I use it a lot to rip DVD movies. It often runs 24/7. There was minimal dust. I run high-efficiency filters on my central heating system though and this could influence dust accumulation. I have no pets so this eliminates pet hair accumulating. T20 owners with cats and large dogs may not want to hard-wire their fan on permanently.

Any other questions?

[Timoti]

Okay, i see your point.

I agree that thremal stress is the natural enemy off any IC, but given that we're talking about portable devices, everyday wear-and-tear are worse, i assume. I walk around with my T20 in my bag during winter on the streets and in heated rooms. So maybe my processor won't live more than 15 years, but who uses more than 15 year old systems anyway? Wait, that would be me... Anyways, swapping CPUs every 10 or 15 years is fine with me.

Regarding the PCMCIA cards, i don't think they give you the ambient temperature of the notebook, especially in a T2x. They contain active components that dissipate heat. USB2.0 cards run kind of cool, my 10/100 ethernet card is warmer, and of course the wireless is a little oven. And in the T2x the graphic chipset and the north bridge is directly under the PCMCIA assembly.

Here are the measurements of the sensors in my T20 with a 1Gig processor with 27C room temp:

balthazar:/home/tamas# /root/ibmtemp.sh
CPU: 45 C
RAM: 49 C
T3: 38 C
T4: 29 C
balthazar:/home/tamas# hddtemp /dev/hda
/dev/hda: SAMSUNG MP0804H: 49°C

To my annoyance, i don't know where T3 and T4 is located in the system. Any ideas about it?

Anyways, i'm still very interested in your findings.

[jamerslong]

Speaking of PCMCIA and Hardwiring i thought of doing this a couple of days ago but my T22 Heatsink arrived so i didnt do it.

i was going to try running a cable from pin 17 of a PCMCIA slot to the fan, i am not sure if that is the right pin but going by memory, you can look it up at pcmcia.com it outputs 5+ volts which should run the fan, if it dosent also wire pin 37 (again unsure but going by memory)


i hope that helps.

[BillP]

Speaking of PCMCIA and Hardwiring i thought of doing this a couple of days ago but my T22 Heatsink arrived so i didnt do it.

i was going to try running a cable from pin 17 of a PCMCIA slot to the fan, i am not sure if that is the right pin but going by memory, you can look it up at pcmcia.com it outputs 5+ volts which should run the fan, if it dosent also wire pin 37 (again unsure but going by memory).

Pin 37 is a data line. Pins 17 and 51 are 5 volts but the source may be current limited to PCMCIA requirements. Also the fan could put electrical noise on the 5V line right at the PCMCIA socket.

There are a number of ways to drive a fan and some of the possible circuits are detailed in the Maxim MAX6640 datasheet, http://datasheets.maxim-ic.com/en/ds/MAX6640.pdf

It is possible that the black wire in the T20 fan is driven by a MOSFET to control speed. That is one of the circuits shown in the 6640 datasheet. If that is the case, simply grounding the black wire will run the fan at full speed all the time. A few measurements and some circuit tracing should reveal how the fan is driven.

[SMA]

Not that I have done any particular research on thinkpad fans, but I have no reason to belive that they should be different from other fans. That is, 2 input wires supplying power to rotate the fan and 1 output wire supplying a mechanically generated pulse for every rotation of the fan - useful for counting the rotational speed of the fan.

I would not expect to find a chip within the fan.

And yes, I seriusly belive it would be as simple as just hard-wiring a connection.

BillP:
I do not think that such chips as the MAX6640 existed when the T20 was designed. The datasheet for the MAX6640 says 2004. But nice reading anyway, thanks.
But there must be a MOSFET or something similar to drive the fan. In my opinion, doing a modification to that would be better, if it has to be a hardware solution.

Timoty:
What is the name of the program you are using to read the temperatures ?

[Timoti]

I use the Linux kernel IBM Thinkpad ACPI module. It has an interface in /proc/acpi/ibm/thermal and i wrote a little script to format it:

Code: Select all

#!/bin/bash

TEMPS=$(cat /proc/acpi/ibm/thermal | cut --fields=2)
CPU=$(echo $TEMPS | cut --delimiter=" " --fields=1)
RAM=$(echo $TEMPS | cut --delimiter=" " --fields=5)
HDD=$(echo $TEMPS | cut --delimiter=" " --fields=6)
TEMP4=$(echo $TEMPS | cut --delimiter=" " --fields=7)

echo "CPU: "$CPU" C"
echo "RAM: "$RAM" C"
echo "T3: "$HDD" C"
echo "T4:  "$TEMP4" C"

The hard drive sensors (if your drive has any) can be accessed through S.M.A.R.T., either with smartctl or hddtemp.