FYI: T2010 Ways to reduce heat and noise

Discussion in 'Fujitsu' started by Frank, Jun 18, 2008.

Thread Status:
Not open for further replies.
  1. Frank

    Frank Scribbler - Standard Member Senior Member

    Messages:
    3,847
    Likes Received:
    3
    Trophy Points:
    116
    :)A lot of people said removing the pads at the bottom of the tablet which cover the vents will reduce heat and noise. Well I’ve done it, too, and tested it. It changed nothing regarding noise and CPU temperature, but it reduced the heat at the bottom about 6 degrees. So the CPU temperature will stay the same, but the temperature at the bottom will drop and get more comfortable. I also hope and think that it will reduce the system heat when the tablet is inside the convertible bump case.

    :mad: Next option is lifting the tablet. Well, to be honest, this is senseless and changes nothing. Again: nothing.
    This image shows the temperature at different places in different states: (I lifted the tablet about 1cm)
    [​IMG]

    :D The third and first real option is applying new and better thermal grease. And this helps. If you stress the machine you’ll notice a temperature drop of up to 5 degrees, and while it idles you’ll notice a temperature drop of about 1.5 degrees. You’ll also notice that the fan is quieter (while doing light work) and gets faster quiet again. It’s still loud but you’ll notice a difference while doing light work.

    A diagram which compares the different methods. I switched on WLAN and BT, connected a BT mouse and ran Super PI on one core and played Aquanox on the other core 20 minutes, to stress both cores, the IGP and consume a lot of power, that's why the temperatures are so high. After the 20 minutes I closed Super Pi and Aquanox, switched off the BT mouse (BT and WLAN still on) and let the PC idle 10 minutes.
    [​IMG]



    Other options would be
    • controlling fan speed via software, which does not exist.
    • replacing the fan with a better, quieter one.
    • building a small hardware fan controller which sits between the main board fan connector and the fan, which reduces fan speed and fan aggressiveness. (I think about building such a thing, I just don’t have enough time in the next time)
     
    Last edited by a moderator: May 18, 2015
  2. thrawn86

    thrawn86 Pen Pal - Newbie

    Messages:
    45
    Likes Received:
    0
    Trophy Points:
    15
    interesting results. Most of us seem to be running our cpus throttled pretty heavily. I'd like to see the same tests at 30 or 40% throttle and 50% load to simulate constant onenote usage.

    as for the hardware solution: It would be pretty simple I think with one of these http://ourworld.compuserve.com/homepages/Bill_Bowden/317.gif and a small pot. you could probably steal v from the battery terminals (you wouldn't want to use the fan header as its being controlled by whatever demonic pwm is in there) but then again the laptop would probably take serious issue with you trying to boot with no fan "connected". perhaps you could just leave the sensor wire connected...

    or maybe just throwing a couple of diodes inline with +v would be enough, but that would just reduce the max fan speed, when the main issue that is annoying people is the speed ramping.

    If I had a supply or an analog trainer or something I could mess with it but I don't have access to the school's labs over the summer, so its all theorhetical at this point.
     
  3. Frank

    Frank Scribbler - Standard Member Senior Member

    Messages:
    3,847
    Likes Received:
    3
    Trophy Points:
    116
    You mean connected to the AC adapter with a full loaded battery, one core stressed with Super PI while throttled at 37.5% or 50% with RMClock running about 20 minutes? Well I can do it, but you will have to run this test, too on your machine, so we can compare a T2010 with the old thermal grease with a T2010 with a better thermal grease.
    If you do this, than please use RMClock and activate the log. Run this 20 minutes and send me the log file, so I can create a nice chart.

    As you already said, I also believe that the fan is controlled via PWM. But your solution only adjusts the voltage and sets it to a fix value, independent from the temperature or CPU load. This is not a good idea in my opinion ;) It's just too risky.
    The diode is more easy but, as you also said, does not solve the main problem with the fan and it's also risky, because you also only lower the voltage which increases the chance that the rotor stalls.

    The best way in my opinion is to use a small microcontroller which reads the PWM signal from the mainboard and sends an optimized, new calculated PWM signal to the fan.
     
  4. thrawn86

    thrawn86 Pen Pal - Newbie

    Messages:
    45
    Likes Received:
    0
    Trophy Points:
    15
    right on both counts, however I think for an attentive user manual fan speed control would be just fine. my fan doesn't need to be at 100% when the cpu is <55C, thats a perfectly acceptable/stable temperature. There are certain situations where I *HAVE* to have quiet, such as during class. Besides, the problem was never the noise (=< 50% is fine), it was the ramping. I'd much rather the fan stay constant at 30-50%, rather than jump violenty between 0% and 100%.

    Feeding the temp sensors (somehow) into it and writing your own fan control would be an ambitious project, but probably the most effective. There would be more parts to cram into there though since you'd need an opamp or transistor/etc.

    I'll try and run those tests for you
     
  5. dmlerner

    dmlerner Scribbler - Standard Member

    Messages:
    109
    Likes Received:
    0
    Trophy Points:
    30
    Just so no one else has to make the same mistake as me, the only vents are under the long skinny piece on the T2010, the other's are fine. :(
     
Thread Status:
Not open for further replies.

Share This Page