Total solution for absolutely ZERO oscillatory noise in t4x

[AdaSch]

sojourner, I'm glad you agreed with my revision, as far as it went, to AdaSch's sketch. I assumed Toshiba put something, nylon or something, between the shaft end and the aluminum.

first i hear bout this

my all fans don;t have any isolator between chassis and axle

maybe in short fan?

[sojourner]

sojourner,
You diagnose "dry worn thrust bearing" as the noise source.
...

Who has examined the mechanical coupling below the T43 fan? Is there room for a millimeter or two of isolating material? What is the best material?

Well, you are bring out more and more detail which I avoided because my posts are often lengthy.

I knew when writing the source of noise, the description was not complete but chose to leave out some details; they're mentioned below:

Talking the T41 fan only now, the noise source (excluding WIND) IS the fan running with dry worn bearings, but as you bring out, we have to consider the environment of the fan. Unfortunately the fan in my T41 did touch the bottom of the base. Noise set up in the fan was then coupled with and AMPLIFIED in the base! To resolve this I did a little extra bending on the fan housing to lift the fan a bit, then installed either a very thin piece of foam (or electrical tape) in the base between it and the fan housing. The reason, wanted to isolate the fan and stop direct metal to metal contact! Made a big difference in quieting the fan too. Been running quietly for over 6 or 7 months.

BTW, also found fans can be designed quite differently between TP's. Don't have a T43 fan so not sure about it. But I do have an R51 fan which incorporates an ALL PLASTIC housing and some type of metallic/plastic insert for a thrust bearing. And yet another fan I've seen actually relies on the sweet spot in a magnetic field to rotate (no thrust bearing). So fans between different models can be quite different, all of which of course effects noise levels.

[sojourner]

sojourner, I'm glad you agreed with my revision, as far as it went, to AdaSch's sketch. I assumed Toshiba put something, nylon or something, between the shaft end and the aluminum.

first i hear bout this

my all fans don;t have any isolator between chassis and axle

maybe in short fan?

I work with metal for a living. Have also worked with a variety of plastics, including teflons, acrylics, delrins etc. When I cut something it is just my nature to really examine what I see and do not recall any plastic (or non-metallic) when I initially drilled through the thrust bearing. If any intermediate was used between the metal thrust bearing and axle, I didn't see it; and presume none was used. And BTW, the fan here is the short fan.

[AdaSch]

ok, short fans don't have oscillatory BUG!!!

please hear sound sample and replay

[milo]

Hello todo el mundo,

I just want to say I made the little hole under the fan chassis to stop the oscillatory noise problem on my T43P 2668 WLN and it works.
What is call the "oscillatory noise" is a distinctive noise, like the noise of a -very mini- "demolition hammer"
Don't merge this typical noise with the noise of the air blown. If you don't have this noise, don't make a hole under your fan!
I listened the noise samples posted by AdaSch but it's very difficult to heard the difference "before and after".
Now, with this mod and the install of TP Fan Control, it's a smooth song due to the air blown and disk access. A delight.
Last remark : I don't have two left hands but it not so easy to make a little hole. You should be patient, attentive and meticulous, not nervous

Thanx AdaSch !!

Milo

[LPC]

OK. I have had enough with the fan noise on my T43P (2668-H2U).

After reading the full thread it appears the drill fix by AdaSch works only on certain fan designs. Mine is a long fan M24 26R9074. Can someone confirm what fans work with this fix? The sticker over the rivet states MCF-208AM05-1.

My sound is very loud and def not a wind noise, but it is not a pattern noise. Once it starts it is constant until fan shuts off.

Thanks

[sojourner]

Perhaps someone will have performed the fan mod on your exact model and will get back. In the meantime, sounds like you've done your homework and come to a right conclusion regarding this 'fix'. I believe oiling your fan will remove much noise and improve performance (because your symptoms sound like a dry bearing) BUT, whether drilling the hole will help facilate that end or worsen the problem is anyones guess at this point. I'll just leave you with this tip:

If you drill the hole and the fan axil drops too far, be prepared to install an insert in your drilled hole to act as a thrust bearing (after oil is added). You'll recall some of us had to do this and is discussed in this thread. I don't think I mentioned this in my responses but if you have to use a thrust bearing you can keep it from falling out by using a very good tape. I did a light press fit of the bearing but also used air-conditioning duct tape (aluminum) to insure the bearing stayed there and the oil bath didn't leak out. That stuff is great, it sticks and holds for a LONG time.

All the best!

[milo]

Hello LPC,
The sticker over my rivet states :

Toshiba Home Technology Corporation
DC Brushless
MCF-208AM05??
D 5V 0.3A
Made in China

See the photo of my fan :
http://goodbox.free.fr/misc/fan-t43p-26 ... shless.jpg

(I began to scratch the sticker before the idea to do a picture, so I don't know the end of the serial number...)

Milo

[milo]

Tip to make the hole on the fan chassis.

You have to drill until to discover the axle. Before to discover it, it's possible (it's my experience) the axle don't turn correctly (with a big noise when you turn it manually). In this case, be patient, so continue to drill. When you see the axle, if the dimension of the hole is good, the axle will fall a little automatically : you win!

After, I put a drop of kitchen oil and closed whith tape.

Milo

PS : sorry for the bad english...

[LPC]

SWEET! Thanks Milo! I am doing a little more checking before I do the fix, but your confirmation of the motor gives me hope

[LPC]

Hello LPC,
The sticker over my rivet states :

Toshiba Home Technology Corporation
DC Brushless
MCF-208AM05??
D 5V 0.3A
Made in China

See the photo of my fan :
http://goodbox.free.fr/misc/fan-t43p-26 ... shless.jpg

(I began to scratch the sticker before the idea to do a picture, so I don't know the end of the serial number...)

Milo

Thanks again Milo for taking the time to help.

Was your fan fru 26R9074? the sticker is on the opposite of the motor sicker. It on part of the copper. Also when you look sideways into the fan is there is a small circuit board located at the housing end of the rotor shaft?

Thanks

Leon

[dcouzin]

These pictures show a partially disassembled Toshiba MCF-208AM05 brushless DC motor: http://www.mediafire.com/imageview.php? ... tloenhnnm2

The fan blade includes a cylindrical ring shaped permanent magnet. This is the rotor of the motor. The circuit board bears metal plates and windings which comprise the stator of the motor. Read about brushless DC motors at http://en.wikipedia.org/wiki/Brushless_ ... tric_motor.

The permanent magnet ring is 5 mm high and (as AdaSch observed) the shaft length does not allow this ring to match the height of the stator. This assures that the thrust bearing will feel thrust from the shaft. The motor was clearly designed for this. The end of the shaft is spherically polished and greased.

AdaSch's idea to simply eliminate the thrust bearing makes some sense after examining and measuring the motor. The distance from the center of the metal plates to the circuit board is 2.5 mm. Therefore if magnetic levitation takes place and keeps the rotor exactly centered heightwise on the stator the rotor will just graze the circuitboard. I doubt that levitation is so simple. The axial thrusts on the fan blade (rotor) depend on the airflow. If more air enters the fan from below than from above the net axial thrust is downward toward the missing thrust bearing, now the circuit board. Also the fan blades have a slight twist causing axial thrust downward toward the missing thrust bearing, now the circuit board. So AdaSch's modification amounts to replacing Toshiba's shaft end thrust bearing with the circuit board as thrust bearing. The modification might reduce the net axial thrust, but it will be born by an unlubricated plastic-to-plastic contact. This might be quiet. It might cause trouble later, as some have reported. It is a dicey cute solution.

[milo]

LPC, I don't know the FRU number. Sorry.
Dcouzin, interesting your analysis. If I will have some trouble later, I will have to say here. At the moment, no trouble.

Milo

[sojourner]

The permanent magnet ring is 5 mm high and (as AdaSch observed) the shaft length does not allow this ring to match the height of the stator. This assures that the thrust bearing will feel thrust from the shaft. The motor was clearly designed for this. The end of the shaft is spherically polished and greased.

AdaSch's idea to simply eliminate the thrust bearing makes some sense after examining and measuring the motor. The distance from the center of the metal plates to the circuit board is 2.5 mm. Therefore if magnetic levitation takes place and keeps the rotor exactly centered heightwise on the stator the rotor will just graze the circuitboard. ... So AdaSch's modification amounts to replacing Toshiba's shaft end thrust bearing with the circuit board as thrust bearing. The modification might reduce the net axial thrust, but it will be born by an unlubricated plastic-to-plastic contact. This might be quiet. It might cause trouble later, as some have reported. It is a dicey cute solution.

Great post (nice photo's too). It's rather interesting some people perform the mod and it works great, some such as myself, hear more noise. After doing the mod and seeing the results I came to the same conclusion as you, i.e. the thrust bearing is supposed to be there. This leads me to conclude what is really needed is, oil for the bearings.

Also, I've wondered if the 'cage' surrounding the fan can be easily removed or if doing so will break the the little metal tabs holding it to the copper plate? Your pictures show these disassembled. Do you know?

If the cage can be removed and easily replaced, I think the way to go is to remove the cage, remove the fan (pull up gently, may want to use bent paperclips hooked close to the hub then gently pull - don't want to break the fan) then OIL using a good light weight motor oil (SAE 5 or 10) mixed with graphite powder. Apply oil solution into the bronze bearing hole and to the axil. This oil/graphite solution is a very good lubricant and a long lasting solution. If you can get a couple of drops into the bronze bearing to create an oil-bath for the axil/thrust bearing, the better (that would be my aim).

BTW, I've also used the oil/graphite solution to re-lub expensive irreplaceable cooling fans on expensive computerized machine tools ... works great! For those in the USA, your local Walmart should carry tubes of graphite powder.

[dcouzin]

I ground off the outermost parts of the four metal tabs allowing the aluminum case to be easily pried from the copper plate. The remainder tabs are still good for locating the case but some glue should be added for holding it tight. There aren't large forces between the case and the plate.

I found undried grease on the tip of the shaft of this MCF-208AM05 motor from my old 26R7957 fan. As mentioned in November, a brand new 26R9074 (with MCF-208AM05-1 motor) made the whining noise in my T43. These two facts argue against the dried lubricant hypothesis.

[sojourner]

As mentioned in November, a brand new 26R9074 (with MCF-208AM05-1 motor) made the whining noise in my T43. These two facts argue against the dried lubricant hypothesis.

Interesting, so what do you think was causing the noise?

[dcouzin]

sojourner, we're going in circles. I made my guess in my 26 Nov post and you somewhat agreed in your 27 Nov reply. However I must retract the suggestion to listen to fans while they're unmounted, since the airflow influences the axial thrust and the axial thrust influences the thrust bearing noise, if any.

I think the clue to the puzzle is that fan whine is much more reported for T43s than for T42s. Granted there is an embedded controller firmware problem with the T43 which (a) causes failures to fall back to slow speed after heatups and (b) causes 5 sec pulsation. (a) can be solved with ThinkPadFanControl. (b) doesn't cause whine, it just makes it more gruesome. Granted most T43s run hotter than most T42s. Still as owner of both, I find a whine problem with the T43 fan and not with the T42 fan.

Toshiba made the rotor seat about 0.6 mm above levitation height (height where it matches the stator). They had room to raise the stator by 0.6 mm, but I think they chose to make the action certain: the blade always presses down onto the thrust bearing regardless of airflow; it never wanders up into the aluminum case. There is considerable magnetic force against the thrust bearing but therefore no bouncing on that bearing, which could be a noise source.

And there is a thrust bearing. The shaft is 1.5 mm diameter so I drilled a well-centered 1 mm exploratory hole through the aluminum plug. After passing through about 0.4 mm of aluminum the drill went through about 0.15 mm of bronze. Indeed the sintered bronze bearing inserted into the brass center of the stator dead ends as a thrust bearing -- one piece sleeve and thrust bearing. Reports of the shaft riding on the "chassis" or on the "aluminum rivet" were false.

Yet I don't know the source of the whine. The fan noise might be tiny and much amplified in the T43. Next time I'm inside the T43 I'll measure the fan assembly and maybe trim and isolate it.

[Darkness Master]

Openinig topic for explanation what cause this anoying fan oscilation. In fact dcouzin was RIGHT.
There is NO mechanical problem. I hope that nobody will be drilling fans any more . Problem is purely in PWM driver from EC on board.
There are Scope waveforms from my measurement of fan driver which shows GAPS in PWM pulses. these have 5s periode and different length ( +-10ms) in each GAP is voltage on HI level so fan try to fully spin for this short time . And from this there are stupid oscilations.

So how to fix this?
One way is to disassembly EC code and try to make better algorithm. Ok there is some group of people which try to do this but its long way trip.

Second way is make new fan driver. I am working on this, in week I will try to post results.

Another way. Go and buy a new model of laptop

Scope waveform:
http://pictureup.cz/picture/verejny/18951895ibm t43p fan problem scope.png

After finish , I will try to make guide to thinkwiki.

Darkness Master

[micl9]

wow am glad I stumbled on this thread AFTER D'Master chimed in!
With my penchant for tinkering and hacking I would have torn into my toys drill in hand!

As I read thru this thread I kept thinking wait a minute the fan designers had to have thought of this - glad we have proof. Seems its always blamed on the Mech Engineers when its really the EEs fault!
[censored] I feel like I am at work - oh wait I am

[hmh]

Interesting. Such a gap would cause a lot of noise, indeed.

Can someone measure a ThinkPad that doesn't have this problem? One of the newer ThinkPads from Lenovo, or at least the T42? It would also be nice to get a measurement of another T43...

Talks about how the EC operates and possible fixes to this issue are best carried out in this thread, where at least one person who knows a bit about the firmware internals hangs out:

http://forum.thinkpads.com/viewtopic.php?t=20958

And BTW, it doesn't have to be the ThinkPad fan controller EE's fault. It could be software/firmware people's fault, or shoddy fan sensor design. Given how simple the fan driver in the planar card was reported to be, and how it is connected directly to an EC pin, it is likely either shoddy fan sensor design, a crappy PWM generator inside the EC, or a firmware (likely design) bug.

The EC firmware people could easily be doing something like generating the PWM on software through soft-realtime, sometimes missing interrupts (we know for a fact that you can screw the EC up if you pester it too much over the LPC bus, e.g. doing ACPI EC transactions), or maybe they use an EC-embedded PWM generator that is half-baked and stalls while it is being reprogrammed, and reprogram it often.

It is also possible that they did it on purpose (feed full power to the fan for a few moments) to be able to properly read the tachometer every once in a while (i.e. shoddy fan sensor design. Pretty much describes every 3-pin PC fan), instead of doing the really hard work required to avoid that and still do closed-loop speed control (maybe the issue is not the hard work, but one of the border conditions it requires was deemed unacceptable).

[micl9]

*****Expletives removed by Moderator***** I feel like I am at work - oh wait I am

Sorry Can't even remember which one I used.

@hmh I was going to throw the SW/firmware guys under the bus too, but as an ME I depend on the SW guys being late so the mech stuff doesn't appear late!

Ok enough off topic...

Isn't a pulsed power feed like that also used to conserve power. I know we did a lot of that in cell phones.

[hmh]

Well, the PWM signal has a very superior instant torque profile at the expense of an inferior acoustics profile. That said, it is worth remembering that what is extremely annoying on a T43 is not the noise made by the PWM itself, but the noise caused by the changes to the PWM duty cycle (we can easily ignore constant, low power noise).

Direct voltage control without PWM gives you a much better acoustic profile, but you will have less instant torque than you would have with PWM at any point but the maximum voltage (which is the same as 100% duty-cycle PWM), so you will stall the fan at the lower speeds much sooner than you would with PWM, and it will stop spinning much easier if there is any increased mechanical resistance (due to lint in the rotor, blocked air outlet, and so on).

I don't recall if PWM uses less energy, but it just might if it causes the motor to operate at a better performance point throughout the torque/speed range you require...

[Cepheus]

Hello everyone!
Does anybody know how to remove the thermal grease? I opened my T43 to make that hole in the bottom center of the fan, but I realised I couldn`t take the fan off because the thermal grease was not allowing it. Well..the paste is not really hard, but enough hard to be very risky to take try taking the fan off. Do i have to heat it?
So, those who opened up your thinkpads: please tell me how did you proceed.

[Navck]

Um are you talking about the paste over the CPU or the "paste" over the GPU? The stuff on the GPU will tend to solidify and stick, the paste under the CPU shouldn't... Weld to the CPU...

[Cepheus]

I mean the paste wich is applied on the zone within the red rectangle.

[Cepheus]

So, does anybody know how to take the fan off when dried thermal paste is keeping it sticked with the GPU?

[ajkula66]

A very thin blade (such as a thin box-cutter) applied carefully and with a lot of patience, or dental floss would be my advice.

Good luck.

[Cepheus]

Ok. I did the hole, the noise level is much more lower(in fact, I can hear only the blowing air..) but I have another problem now. the laptop crashed several times untill now. I think it may be an overheating problem. Well, when I just browse or do things that do not solicitate the CPU or the GPU, the temperatures are normal. But when I play a game(e.g. Freelancer) the laptop is suddenly shuting down after a while(10-15 minutes).
I checked the temperatures after a little bit of Freelancer playing. The CPU was at 80(and a few) degrees Celsius and the GPU somewhere about 60.
Are these temps to high? Do they cause the crash?

[Cepheus]

I just found out that the max temp for the T43 CPU is 92. I checked again the CPU temp while playing Freelancer, and it was at 89 wich is very close to the edge. So, I guess the answer to my question is "yes, overheating causes the crash".
But now, here comes another question: what should I do to prevent overheating of the CPU?
From what have I seen when I opened up the laptop, the CPU does not have direct contact with the fan. Why is it heating so much now(after I made those modifications to the fan) if it does not have anything to do with the fan?

[Cepheus]

Sorry. Big mistake. CPU is right under the fan so I guess I did not applyed weel the paste over the CPU.