Total solution for absolutely ZERO oscillatory noise in t4x

[melberi]

I managed to fix the original constant growl by drilling the hole, but now I have a worse, rattling noise (of random nature instead of constant). However, flipping the fan upside down silences the rattle. I suspect the bottom part of the area around the axel got damaged or some small bits of metal got stuck in between. This is the fan on the long heatsink. I'll see if I can still fix it further to really silence my T40.

I wonder if the fan was originally designed to be used upside down? Possibly the unmodified fan would be dead quiet when flipped with gravity pulling the blade unit to the correct position for the magnetic bearing to work correctly.

[madkat]

not in the topic i've originally posted but - you've answered one question i've asked long ago
if there is a difference in wattage between the short and long fan

[dozer]

AdaSch.....a big "thanks!" for posting your solution, and the pics and diagram.

It indeed works well on the fans which use the magnetic-bearing.

(note to netzspannung: the mag-bearing isn't something you see when you drill the hole....it is inside the fan....see the .png diagram Adasch provided)

That "blipping" every 5 seconds was FAR more annoying than any steady 'whining' could ever be...

Rather than drilling the hole, I used a .093" carbide end-mill and plunge-milled the hole.

This technique keeps 99% of the chips out of the fan (assuming you're using a vacuum or air-blow while milling to sweep them off the surface as they're cut) because it doesn't 'open' the hole at all until 99% of the material is already gone away as chips (endmill has a flat face instead of a piercing point).

netzspannung: you may be able to salvage your 'ruined' fan simpy by inserting a nylon thrust bearing into the hole you drilled.

I say 'nylon' because it is an inherently 'lubricative' low-friction plastic.....and one that's very commonly used for exactly such bearings.

PVC, Polycarbonate, etc. are poor choices.

Nylon, HDPE, and Delrin (best of all) are good choices.

Cut a small piece of nylon rod of the right diameter to "press fit" into the hole (i.e. just a few thousands larger than the hole).

If you don't have any appropriate nylon rod in your junkbox, then look at "weed-whipper" line. A lot of that line is made from nylon.

You'll likely need to counterbore a recess into the 'inside' end of the rod-insert....so that it doesn't push the shaft up too high.

You may need to experiment with the depth of this recess until you get the optimum running position for the shaft.

Another possible technique would be to take a small piece of sheet nylon....say, .050" thick, heat it to near melting, then simply press it against the surface over the hole. It will naturally 'flow' around the edges of the hole on the inner surface; thus 'locking' it in place.

AdAsch, thanks again.

[dcouzin]

BTW, the better idea is to get a longfan assy with the 0.4 ampere fan (blue sticker on bottom) (shortfan is 0.2 amp) It covers the GPU, but it is also better because the fan itself is less thick.

I just replaced the FRU 26R7957 (with bad bearings) which had found its way into my T43 with a brand new FRU 26R9074 (available from a Chinese eBay seller). Both fans are marked "Fan Long M24", both have practically identical thickness, and both motors are marked 0.3 A. In which FRU did netzspannung find the souped-up 0.4 A motor, and was that find the source of his noise relief?

There are so many fan FRUs that the user community may never figure out what exactly is behind the T43 fan noise problem. I should report that my replacement resolved a horrible clattering noise. That was a bearing problem -- I could shake the old rotor to and fro. But also the 5 second pulsations are gone. [ERRONEOUS OBSERVATION!] This must have a different cause. There is a wee circuit board located at the housing end of the rotor shaft. What does this do? Is it a tachometer circuit? The old 26R7957 fan had motor MCF-208AM05. The new 26R9074 fan has motor MCF-208AM05-1. Does that that "-1" denote a difference in the (tach?) circuit such that the T43's Embedded Controller communicates differently with the 26R7957 and the 26R9074, producing the 5 sec pulsations only with the former? I'm sorry to add yet another speculation to the "total solution" strand but laptop fan noise can affect us.

[madkat]

thought about the circuit board - as the culprit for the noise

especially when last week - replaced the shortfan with a T43 0.3A longfan (details in the Intel Mobo Frankenpad topic) in my TR451 - it worked differently at startup:
- the shortfan would start at max speed and in 2 sex max shut off
- the long fan starts at a medium speed, than during the following 1,5 - 2 secs goes to max speed, stays there for about another second and then drops the speed and shuts off
mainboard is obviously the same, so it must be a difference either in what the mobo "senses" from the fan (speed info etc) or different behavior of the fan electronics with the same current provided by the mobo...

in conclusion - something DOES differ in the little board of the fan

[hmh]

The 30s rev-up noise in a T43 is magnified by a firmware bug, where it uses a non-initialized register to set the target fan RPM. Since it is uninitialized, changing something (like short->long fan, or BIOS option, etc) can change that behavior, and maybe make it less noisy.

Or you can install a bug-fixed unofficial firmware. It doesn't remove the 5s/30s recalibration speed changes, but it does make sure the 30s one is sane.

The -1 revision of the motor might have a smoothing circuit (just a capacitor, really) added and that would make it a lot more silent too. Or it just sounds like it should and you notice such a difference because you had a really bad one before

[madkat]

my frankenpad has a R51 mobo
so it doesnt have the T43 bug...
and with the T41 shortfan - it doesnt have the "t43 startup behaviour"

[dcouzin]

The 30s rev-up noise in a T43 is magnified by a firmware bug, where it uses a non-initialized register to set the target fan RPM. ... Or it just sounds like it should and you notice such a difference because you had a really bad one before

There's an ambitious thinkwiki page on the subject: http://www.thinkwiki.org/wiki/Embedded_ ... r_Firmware.
(My one complaint is that the page lacks dates -- the wiki presumption of perfect updating.)
According to that page the non-initialization bug has low annoyance level. It's the pulsation bug that has severe annoyance level.
Yes, my old fan's bad bearing amplified the pulsing sound. But my new fan isn't silent. It makes a good whoosh. Yet there's no hint of pulsation. [ERRONEOUS OBSERVATION!] So the same old EC firmware version 1YHT29WW (1.06) released 6 June 2006 is interacting differently with the new fan's electronics. The pulsation problem isn't a firmware bug per se but a miscombination of firmware and fan. When was the 26R7957 fan discontinued, and when was the 26R9074 fan introduced? Dates will help for understanding Lenovo's efforts.

[madkat]

Are you saying that pulsation isn't only a mechanical issue?
I'm thinking that at a newer fan seems gone due to different bearing structure (not modifiend especially for this, by lenovo)

At the T43 long fan (from a very little used machine) the noise is almost gone - but if you are paying attention its still there, just much quiter (different and newer bearing)

[hmh]

Fan pulse bug: http://www.thinkwiki.org/wiki/Embedded_ ... ng_pattern

Read http://forum.thinkpads.com/viewtopic.php?t=20958 if you're interested in the fixed firmware.

[dcouzin]

The pulsation often discussed (such as at http://www.thinkwiki.org/wiki/Problem_with_fan_noise ) can't be simply mechanical since it follows an exactly 5 second cycle, regardless of the fan speed. The fan goes surge-relax-surge-relax-etc. The Embedded Controller is telling the fan something every 5 seconds. One fan surges at this refreshed information while another fan simply continues at speed. [ERRONEOUS OBSERVATION!]

[hmh]

"simply continues at speed"... are you sure? Likely it just varies the speed a lot less, and therefore you don't notice it without resorting to measurement equipment (or someone with really good hearing).

[dcouzin]

"simply continues at speed"... are you sure? Likely it just varies the speed a lot less, and therefore you don't notice it without resorting to measurement equipment (or someone with really good hearing).

What we're trying to determine is whether firmware version 1YHT29WW communicates differently with the two fans 26R7957 and 26R9074. With the first I heard pronounced 5 second pulsation. With the second there I noticed no pulsation. [ERRONEOUS OBSERVATION!] It's irrelevant to the question at hand whether the second is producing 0 pulsation or extremely tiny pulsation.
The one trouble with my example is that fan 26R7957 had a bearing problem making it at least 10 times as loud as fan 26R9074. Someone might question whether scaling bad fan 26R7957's sound output way down to fan 26R9074's level would leave the pulsation noticeable. I'm pretty sure it would. The widely annoying pulsation involves both pitch and volume.

[hmh]

Well, you can go to the thread I linked a few posts back, get the firmware source code, and check it. I guess it you take a few hundred-man hours if you are not skilled at reverse engineering software state machines, or about 20 man-hours or so if you are. Maybe ask mg about it, he might know since he did study that code enough to fix the 30s annoyance bug.

Or you could get an osciloscope, tie probes to the three fan leads and examine the waveforms. A bit more hands-on, but probably a lot easier and faster than trying to make heads-and-tails from IBM H8S firmware, especially if you can simply compare before-and-after using the old fan.

I very much doubt the firmware is treating the fans differently. It is far more likely that a low-pass filter was added to the fan, especially if it is not doing the same amount of ruckus at boot as the old fan did.

[dcouzin]

hmh, I should have thanked you for that wonderful link to mg's work on the firmware. It was an extraordinary effort. As mg said (27 Feb 27 2006) Lenovo had included the fan speed fix the T43 firmware but "looking at sub_13b74, there seems to be a typo: er0 is loaded from RAM but then er1 is incremented and dereferenced ..." blew it. Did no one at Lenovo hear about this before new firmware was released on 6 June 2006?

I should clarify that my words "firmware...communicates differently...with...fans" is not in proper engineer-speak. I don't mean that the firmware sends a different signal to the two fans, only that the fans "understand" the same signal differently. Due to the fans' electronic difference they respond differently to whatever the Embedded Controller does every five seconds. This is consistent with your explanation.

[hmh]

Well, then we're talking about the same thing.

As for Lenovo/IBM, well, maybe someone in there did know, but I don't think so. What matters is that it is too late for that now, and it is best for everyone if there is no major noise about it (for one, I don't think Lenovo would appreciate that firmware being on the network).

Those who want a fixed firmware, can get it from that forum post. Most make do with the software fan control apps, that can do a lot more than just stop the pulsing, and don't have any real reasons to want the unofficial firmware.

[Senser]

Hi,

I'm having the annoying pulse sound on my T43 as well. I'm trying to find out if the long fan (type 26R9074) is compatible with my T43 (type 2668) and if the long fan will solve the pulse noise as well (I'm not very eager to drill a hole in my current fan actually).

So; will replacing my current (3 year old) fan by 26R9074 solve the noise problem? And will it fit?

Thanks!

[dcouzin]

senser, the FRU 26R9074 fan will fit the type 2668 Thinkpad. It is the fan specified in Lenovo's online partslist: http://www-307.ibm.com/pc/support/site. ... MIGR-58420
I must note that the nastiest part of installation is the need to wiggle and push the fan all the way into place.
This fan eliminated the 5 second pulsations in my type 2668 Thinkpad. [ERRONEOUS OBSERVATION!] It is probably important that the Embedded Controller be the latest version 1.06, 1YHT29WW. I'm also using the latest BIOS version 1.29.

[Senser]

Thanks for the information. If I'm right the 26R9074 is the same fan that's already in my T43 (3 years old). If so, I don't understand why a new fan (of the same type) will solve the fan issues.

Besides noise I dislike my T43 for the amount of heat it produces. Of course, TPFancontrol kan solve most of the noise of the fan, but the fan on fanspeed 1 or 2 results in a lot of heat (even with RMClock installed). Can this be solved somehow too?

Thanks!

Regards,
Senser

[hmh]

Yes. You have to use a better heat-transfer compound than what IBM used on the heatsink, maybe add a secondary passive heat-sink to the southbridge (hardware modification, requires a lot of hability and some tools) , make sure that evertything you don't use is turned off (serial ports, IR port, modem, etc), keep the WiFi card with the radio blocked when not in use, and use it in BATTERY power save level while in use (decreases throughput a lot. No, I have no idea how to tell the Windows driver to do this), and finally undervolt the CPU.

That will give you longer battery times, to boot.

But you probably want to start a new thread if you want to ask anything about the above.

[dcouzin]

Senser: "I dislike my T43 for the amount of heat it produces"
hmh: 'You have to use a better heat-transfer compound"

I disagree with hmh. Using different heat sink compound between a chip and the heatsink doesn't reduce the amount of heat produced by the chip. It just increases the fraction of the heat that is blown out by the fan (and reducing the fraction that increases the chip's temperature and also heats other parts). I agree with Senser. The T43 electronics simply produce too much heat. IBM/Lenovo can't expect a ventilation system like that of earlier T4x's to handle that. The heat production seems out of whack with the performance. My T43 peforms some tasks 25% faster than my T42, but the T43 runs about 15 C hotter even with the fan running about 10 times louder. Where were the engineers?

[AdaSch]

Result = new fan, cost me some money

Mega LOL

if someone has two left hands should not take for modifications. I did have (40?) many modifications of the 100 percent success

cheers

[dcouzin]

yes, under sponge is rivet. Get battery drill, and do hole in center about 2,2mm (diameter)
cheers

What sponge? What rivet? The drawing doesn't show this.

[dcouzin]

I examined a T43 fan marked 26R7849 and 26R7957. Under the motor label is the "rivet". This looks like a press fit aluminum part 6.5 mm diameter which would support the pcb and the shaft bearings. I've taken the liberty of modifying AdaSch's first drawing to show the "rivet". With apologies for MediaFire, here it is.

I found the rotor to be pulled strongly magnetically toward the "rivet". This is by permanent, not electro-magnets. I think that drilling a relief hole in the "rivet" of this fan would cause the rotor, instead of levitating, to simply rub against the pcb. (Clearance between rotor and pcb is just 0.6 mm.)

Toshiba, who makes my hard disk drive as well as this fan motor, can't be such idiots as AdaSch would have us believe. I think there is an thrust bearing (of some sort) between the shaft and the "rivet". This thrust bearing might contribute some noise.

Different T4x fan motors might have different bearing constructions, and some might levitate the rotor and benefit from the added relief suggested by AdaSch. Probably not the 26R7849 / 26R7957.

[netzspannung]

Mega LOL

if someone has two left hands should not take for modifications. I did have (40?) many modifications of the 100 percent success

cheers

Now, I surely will not go as far as blaming AdaSch for something I would do with my own hands, but as much as I can tell from my experience and also my correspondence with dcouzin, the drilling idea is not a solution. This advice is bordering on insane and I'm warning other forum members.

As always, you pays yer money, you taketh yer choice.

[AdaSch]

Ok, if you don't want - don't do it simple

netzspannung - PM

[sojourner]

I examined a T43 fan marked 26R7849 and 26R7957. Under the motor label is the "rivet". This looks like a press fit aluminum part 6.5 mm diameter which would support the pcb and the shaft bearings. I've taken the liberty of modifying AdaSch's first drawing to show the "rivet". With apologies for MediaFire, here it is.

I found the rotor to be pulled strongly magnetically toward the "rivet". This is by permanent, not electro-magnets. I think that drilling a relief hole in the "rivet" of this fan would cause the rotor, instead of levitating, to simply rub against the pcb. (Clearance between rotor and pcb is just 0.6 mm.)

Toshiba, who makes my hard disk drive as well as this fan motor, can't be such idiots as AdaSch would have us believe. I think there is an thrust bearing (of some sort) between the shaft and the "rivet". This thrust bearing might contribute some noise.

Different T4x fan motors might have different bearing constructions, and some might levitate the rotor and benefit from the added relief suggested by AdaSch. Probably not the 26R7849 / 26R7957.

I agree with most of your statements. The metal we're being told to drill is actually a metal PLUG (not a rivet), it is pressed in and peened. The fan motor shaft is hardened metal with a rounded end and is DESIGNED to ride against this plug. The fan is positioned in the magnetic field SPECIFICALLY so LIGHT PRESSURE ensures the fan shaft rides against this plug.

So far as I can tell what is happening with these old fans is two things.
1. lubrication dries up
2. (which causes) WEAR

Although the bronze fan shaft bearing encurs some wear, the bigger portion occurs on the plug. This PLUG is actually soft metal. When the fan lubricant dries up the hard shaft turning constantly against this soft plug (which acts as a THRUST BEARING) creates excessive wear and creates a dimple in the thrust bearing. It is this DRY RUBBING of the fan shaft against a dry worn thrust bearing which is causing bothersome noise.

Not knowing all this beforehand, and wanting a quieter fan, also did the MOD only to find a WORSE situation. Now, having done the MOD, because the fan shaft did not have a thrust bearing surface to ride against, the fan moved to where the magnets would have it position in a natural place between the poles. Unfortunately the fan moved SOOO much the fan HUB rubbed against the PCB. An even WORSE noise was made

That's when I knew this situation had to really be thought through, that's when I discovered what's mentioned above.

SOLUTION (for me, yours may be different).
Knew an appropriate thrust washer needed to be installed so I drilled an even bigger hole (don't recall the size). The intent was to make a new thrust washer and press it into the newly drilled hole. I chose to use the end of a drill bit for the new thrust washer. Reason is, the solid end of a drill bit is harder than soft metal (CRS, Cold Rolled Steel) but not as hard as the fan shaft. Knew this would made a good combination.

Found just the right drill bit so it would press into the new hole. Ground the end of the drill bit FLAT with a pedistal grinder, put a very slight chamfer around the end to remove any burr an provide easier entrance into the hole. I then cut the end off the driil bit so it was just a short piece of metal. After lubing the fan shaft with a liberial amount of oil (used Mobile Synthetic 30w for autos) the plug was inserted and ground flat (Dremel) to the fan housing. Finished off with good cleanup and placed a piece of aluminum air-conditioning duct tape where the paper label was.

BTW, found you need to press the new bearing in just far enough so the plastic fan hub does not rub on the PCB. If you press too far in the hub will rub against the alum. housing on top! There is a 'sweet spot' one must hit when pressing in the new bearing.

It's a lot of work I know, not sure I'd do it again either but the reward is the quietest fan I've ever heard on a laptop!

[AdaSch]

simply,

my method work or not?

i have right where is sound source?

cheers

[sojourner]

simply,

my method work or not?

Regretably, no! Created even more noise, but don't take it wrong. It worked for you and you wanted to make a contribution, which you did and I encourage you to continue to do.

I've too have posted things which another member had contrary remarks about ... just the way it goes here. And as long as we always keep our remarks (replies) 'civil' (respectful) it works out well for the group!

i have right where is sound source?

You said, "noise source is between chassis and axle". So, if one considers the 'plug' (actually a thrust bearing) 'the chassis' then you are partly correct.

Noise can come from either (or both) of two worn bearings. These fans have TWO bearings, an axle bearing (sintered bronze, fan shaft goes through it) and a thrust bearing (end of fan shaft rides against it). The prevailing cause of noise usually comes from the axle rubbing against a dry, worn thrust bearing.

[dcouzin]

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. Whatever is or isn't there we'll call the "thrust bearing".

You diagnose "dry worn thrust bearing" as the noise source. I doubt this because I've just replaced my T43 fan with a brand new 26R9074 fan and it was immediately noisy. Also I've just had my T42 fan replaced (on warranty, so I don't know what they put in) and it is silent. The T43 is famously noisier than the T42. Of course the T43 is hotter, so the fan blows more, but it's noisier as it blows. The T43 fan noise is a whiny mechanical noise, heard even when the fan is brand new. This nasty noise comes especially from the underside of the T43.

My guess is as follows: the T43 fan's thrust bearing isn't different from the T42 fan's thrust bearing; the T43 fan isn't noisier itself -- running outside the machine -- than the T42 fan; the thrust bearings make or convey a small vibration, but for this vibration to make a significant sound requires mechanical coupling to the bottom of the machine; this coupling is the T43's problem. This guess agrees with netzspannung's appraisal from 7 July. The guess is that the T43 has a dimensional/constructional problem under the fan.

I should explain the "make or convey" phrase. Even a perfect thrust bearing conveys vibration if the shaft is pounding against it. The sources of thrust in this fan are the magnet and the fan blade. The fan blade vanes are shaped such that they make some axial forces while blowing, and these forces can be a rapid pounding. AdaSch's proposal to eliminate the thrust bearing eliminates all the made and conveyed vibration. Unfortunately this fan does not magnetically levitate the blade so it needs a thrust bearing.

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?