Fan Noise, Always On - Interesting Info (From HP!)

[Basker]

To everyone interested in the T43 fan issue: I happened upon a post over on the HP forums with regards to their HP nc6120 laptop, which is one of the T43's direct competitors. The folks there have the same complaints that people here have with the T43's fan. However, in this post, one of the HP engineers gets involved and tries to tackle the fan issue. I strongly think is identical to the problem with the the T43. As best I can tell, this is exactly what Lenovo should be (and maybe is) doing. I'd love to think they are trying some of the same solutions. It would be great if someone could bring this thread to Lenovo's attention, if they have a real contact inside.

Look for the posts by Andy Fisher in the HP business forums thread here: HP nc6120 fan always on.

The most interesting and pertinent quotes Andy makes are what he sees as the problem:

Here's what we know:

There is no "solution" to this issue, but before you go running to the hills yelling the world is coming to an end, let me say a few things.

The laptops you have are thin and light and built to the highest design standards. However, the trade off to a thin and light unit is heat dissapation. Even while idle the fan will cycle on and off. To often? I think so, we are evaluating our trip points.

For those people that have Celeron processors that don't have the power management features of the Pentium M processors then the fan will be on most of the time. The Celeron processors are lower speed processors and may run cooler than say the 2.1Ghz Pentiums but nevertheless the fan will run more. This may be a good thing since you probably don't notice it as much as if it turned on and off all the time.

On Pentium M systems that utilize speedstep technology the transition from low speed to max speed causes a significant temperature spike for about 1 sec which happens while browsing a web site for example. The spike puts us over the trip point and causes the fan to turn on. We are working to filter out the spike and prevent the fan from coming on so frequently. However, we need to do this without violating the ACPI specification, this is under investigation.

Another thing we have successfully accomplished is to ramp the fan speed up and down in small increments to prevent the suddenly on/suddenly off syndrome. This far improves perception in a quiet office environment.

Work is not finished on this issue, we are now investigating the features of the thermal sensor to see if we can improve upon the temperature reporting.

I am trying to provide you with all the information I have as we attempt to satisfy our customers.

and...

The solution which HP is actually implementing:

Time for an update and to set expectations.

The problems as we see it:

1. First trip point for the fan is set too low. Therefore fan off point is also set too low and may not turn off.

2. When running an adaptive power scheme such as Portable/Laptop a speedstep boost forcing the CPU to full performance causes a temperature spike which turns the fan on unnecessarily.

What we changed:

1. We have raised the first tripoint on and off temperatures slightly such that the fan is able to cool the system to a fan off point. Meaning that when idle the fan should turn off.

2. The fan on and off routines have been modified to turn the fan on and off very slowly. This should alleviate the fan coming on due to spikes in temperature caused by speedstep jumps. The fan will start to come on but since the temperature change is only a spike the fan will turn off before it realizes it was supposed to be on.

3. The F10 setup option to enable Fan on all time on AC has been changed to only run the fan at 20% while on AC. This keeps the system cool enough to prevent temperature spikes above the "fan on" trip point while running quieter. This is now defaulted on.

Celeron systems that don't have CPU power management may run the fan frequently although lower speed CPUs run cooler by nature.

nw82xx workstation units obviously run hotter since they have a more powerful video controller, probably more fan activity on these systems also.

Running the CPU full on using an "Always On" power scheme will cause fan activity.

Running on battery hasn't changed much other than the higher tripoints. Our testing shows that the slow fan ramping removes the audio shock of the fan suddenly coming on and turning off.

Our sample of a single outside tester still thought it sucked. This is the compromise with a high performance system that is still thin and light.

Can't give you a timeframe other than we're done in engineering. The rest is up to regression testing and schedules. Releases for 61xx, 62xx and 82xx do not run on the same schedule so releases will be staggered.

Try it, if you don't like it then you've got to do what you've got to do. If you think it's OK or you like it then post praises here, alot of work went into this.

Thanks for all your patience.

[sugo]

I wish ibm/lenovo engineers are reading this.

[Navck]

Same, and hope they release fan control software (Hey, some of us would like to have 100% fan during heavy load)

[BillMorrow]

i am impressed with the above h/p explanation..

/begin editorial

what i get from the above is that h/p seems to care more (officially) than ibm ever did IMO..
unofficially a lot of great ibm'ers would chime in here and on the old webboard forum..
who knows about lenovo or what they will do but since lenovo bought the who shebang, "lock stock and barrel" they got the good with the bad..

how they (lenovo) will filter out the bad (and who will choose what is bad and what is good and thus work keeping) is anyone's guess..

one would hope that lenovo will take more of an active role here and at least say SOMETHING rather than keep to the old ibm practice of official silence..
(IMO "official silence" = ostrich with head buried in the ground)

i believe that they (lenovo) would only gain by such participation..

/end editorial

[JaneL]

i am impressed with the above h/p explanation..

/begin editorial

what i get from the above is that h/p seems to care more (officially) than ibm ever did IMO.. <snipped>/end editorial

Yeah, they can talk a good game, but my experience was that they have poor follow-through. I've been through a similar situation with HP/Compaq where someone slipped on the forums and actually told the truth while the official support line was "we know nothing, no one else is having this problem, etc". I truly enjoyed cramming the messages from the forum down their throats (along with 600 notebooks). They immediately deleted the messages...