Interesting fix to T23; constant fan;no beeps, video or boot

[rkawakami]

I decided to work on a T23 that I received a couple of days ago. It came from eBay (naturally) and the seller indicated that it tries to power on but doesn't. He said that it also had a 1-3-3-1 beep code (memory) at one time. When I powered it up this is what happened:

- CPU fan constantly on
- no video; LCD or external
- no backlight
- no beeps
- no visible/audible attempts to boot from floppy, CDROM or hard drive

Okay, I thought that I had a power issue. First order of business was to remove everything removable but it still acted the same. I've seen systems like this before so I tried something that has worked for me in the past. I grabbed the front left corner and flexed it downward while pushing the power button. No difference in the power-up. Then I flexed the corner upward and turned the system on. 1-3-3-1 beep code! Life! I plugged the module back in and repeated the power-up, still torquing the front left corner. Again, 1-3-3-1 beeps. Pulled out the memory and tried it in another T23 (known good system) and it beeped there too. So, the seller had a bad memory module. Put a known good PC133 module in this new T23 and back to the same thing during power on (only the fan was on). As soon as I twisted the corner I was greeted with a splash screen!

I found that as long as I continued to grab the front left corner and twist it upwards (relative to the rest of the base), the system would get through the POST process and commence booting. In fact, if I let go of the corner, the process would suspend until I once again grabbed the corner. I put a copy of PC Doctor in the CDROM drive and after the system finished loading the diagnostic, I let go. I first ran the CPU tests and they all passed. I then ran the Systemboard tests and it only failed one portion: "RTC Clock", the Real Time Clock. Putting PC Doctor into a loop on that one test, I discovered that as long as I held onto the corner, the test would pass. Stop flexing the corner and the test would fail. Hmm, not a typical power issue after all.

I stripped down the system, pulled the motherboard out and went looking for the RTC chip. Knowing that Dallas Semiconductor makes many RTC chips for computer and laptop systems, I searched for that logo first. I never did find one, but I did see at least three crystals. Those crystals are used to generate the clock signals for the motherboard. Since the three were located on the top of the motherboard, I re-assembled the system enough so that I could mount it in a port replicator. The reason for that is so that I can turn on the system using the power button on the port replicator, keeping the keyboard removed, allowing almost full access to the motherboard. I then pushed the power button and then pushed on what I thought was the most likely part to case the problem, the crystal marked Y4 (located at I4 in my diagram from this thread: Documentation of some components on T23 motherboard). The system "woke up" and started to boot. AH! Close inspection of the way that it was soldered onto the motherboard (slightly off-center of the pads on the board) seemed to look good, but I hit both end of the part with a soldering iron anyway. The system would still not boot unless I was bending the corner or putting downward pressure around the Y4 crystal. I looked over to the Y1 crystal and it appeared to be well connected to the motherboard. I then checked the third crystal (Y6) located at J5/J6 and saw that two of the leads had hairline cracks in the solder joint.

Heated both of those legs with the soldering iron and turned the system on. It booted without any assistance. Six hours later, it still continues to boot after being completely re-assembled. I can't claim to have found a common failure mechanism. Another T23 I have that requires a similar twist to power up has what appears to be good solder connections on all three crystals. Perhaps there's another component that has worked itself loose from that board.

Cost of the 2647-9NU, 1.2Ghz, 0MB memory (256MB module was bad), 0GB hard drive, perfect SXGA+ screen, WiFi, DVD drive and AC adpater... $122.05.

Cost in time to strip, test and re-assemble the system.... 2 hours.

Satisfaction in getting another T23 system to live again.... priceless .

[RealBlackStuff]

Great find, congratulations!
Now that you have become such an expert on T23s, how about expanding into T30s?
Lots of people have issues with them, that may be comparable to your T23 experiences.
I have both a T23 and a T30, but I find the T30 a lot easier to work on.

[tom_k]

Thanks a lot Ray

Excellent Report (as usually;) and beside all those "Laptop broken down" stories really encouraging
think i`m going to check some t23 mobo`s :D

regards tom_k

[dorronto]

Tremendous Ray,

I will print your comments and solutions and save them. Most interesting. As always, the "Ray101" clinic never fails to arouse curiosity with solutions........

Thanks Ray.......appreciate it..........hope I pass "101"


Ron

[tfflivemb2]

Chalk another solution up for Ray!!! Great job!

Could this be the same reason for the Blink of Death on the T20-T22 series?

[phool@round]

Both are very nice finds. I'll put that in my bag of things that go wrong, lately I've been working on the X ultra lights.

I might hit you up off list for some advice on a device I've found that is causing grief for those boards. I have no idea what it is yet and nobody seems to know on that side of town.

Looks similar to I3 top, just right of dead center, six legs, 1 middle leg wider than all others. Power/battery circiut heavily influenced. Became a LET......

[rkawakami]

... how about expanding into T30s? <snip> ...I find the T30 a lot easier to work on.

I hadn't planned on taking on another line of systems to troubleshoot. Hopefully whatever I am finding in the T2x family can be applied elsewhere. After stripping down the same T21 to the motherboard four times in the last week and a half and the various T23s recently, I've forgotten what it takes to pull the motherboard out of a 600-series system. Right now, I find the T2x a piece of cake to work on .

Could this be the same reason for the Blink of Death on the T20-T22 series?

I had the same thought as I was re-soldering the crystal last night, but the BoD problem still seems to be a general power system failure. That's not to say that the root cause isn't the same; I still believe that there's simply a broken connection somewhere on the board. I also entertained the thought of borrowing one of the illuminated stereo inspection microscopes from work to aid in my search. Holding the T23 motherboard under a lamp, while looking through a 2" x 4" magnifying glass with one eye yesterday wasn't that fun.

<snip>... six legs, 1 middle leg wider than all others. Power/battery circiut heavily influenced. Became a LET......

Ouch! I know that feeling. In my case a power MOSFET turned into a photon emitter. From your description, the part located at I3 appears to be a voltage regulator as the designation "VR3" is nearby. I can't find anything matching the "AOH 252" that's on the package. Said package is normally called "SOT23-5", for small-outline transistor, 2mm x 3mm, 5-pins and I can find many references to voltage regulators or DC-to-DC converters that are housed in that. The sixth "leg" is actually a heat sink tab which may or may not be attached to anything. One would assume that the motherboard designers knew what they were doing if the heat sink tab is not connected anywhere. But seeing where this one is on the T23 (directly underneath the hard disk drive and covered by the HD guide rail insulator), I'm sure that the part is getting pretty warm.

[phool@round]

Thanks Ray, you've confirmed where I've been going.

[zmyx]

I have exactly the same problem but resolder all three crystals didn't help. All three inductors are also firm in place (did resolder too) and the two fuses are ok (checked with ohm-meter). CPU and RAM are ok. All cards removed.

still constant fan, no beep, no video. only one LED is green-- the "Z" one near power button.

Any other possible things to check? Some ideas?

Thanks in advance.

[rkawakami]

First order of business was to remove everything removable but it still acted the same.

Have you tried removing all of the memory?? You should then get the 1-3-3-1 beep code. I'd also make sure that the hard drive and Ultrabay drives are taken out too.

I grabbed the front left corner and flexed it downward while pushing the power button. No difference in the power-up. Then I flexed the corner upward and turned the system on. 1-3-3-1 beep code!

If applying pressure like this to the system helps it to boot, then I'd say that there's an intermittent disconnection on the motherboard. Given the "stop-and-start" nature of my original problem I thought that it had something to do with the timing circuitry. Since the crystals are one of the major components in the timing system I looked closer at those and I got lucky.

About the only other thing I can suggest is to connect an external monitor and see if you can get a video signal out the VGA port. You may have to press the Fn+F7 key combination when the system is powered up in order to activate the VGA port. Actually, make sure that you press those keys at least three times, pausing for a couple of seconds each time to give the system some time to change video modes.

[zmyx]

Thanks for your quick reply!

First I should mention that I got this T23 from ebay and I thought it might be just a loosen inductor but as I said in the previous post it's not. I was already frustrated. After checking everything I could think of the system still didn't boot. I put everything back and leave the system for about an hour. Then I remembered that the three LED near the bottom right of the screen are all off during the constant fan on. I mean when I press power button every LEDs on the panel flash which indicates that the LEDs are fine but then after few seconds only "Z" LED is on and the rest are off. Anyway right after I noticed this I put a battery in and also the AC and it won't charge the battery at all. No LED is on, not even orange, totally dark, all of the three on the bottom right of the screen.

Then after some playing around the system booted! It booted on battery and then I plugged the AC in and it was still working. Just like that! IBM splash screen came and went all the way to report that no media is found. I plug the HDD in and restart and it booted into Windows!!! I waited until Windows finish the boot up and, still curious about the battery charging, went to click the AC icon in the task bar to see more info about the battery and, just like when it booted, it just freeze there at the battery info window and I can't turn the system off even by holding power button for a minute long. I didn't get to see the battery info too because it freeze before I can click the info tab. I have to disconnect the AC and the battery pack to shut it down.

Since then it didn't boot again and the constant fan problem is back. In the hope that it might magically boot up again I load the CD-ROM drive with a bootable PC-Doctor so that I can run the diagnosis. But until now it still no go.

Anyway, when the power is on and the constant fan problem is there, I notice that the HDD LED would stay on when I plug in the CD-ROM drive and turn off when I take it out. Is this LED just represent a simple electronic switch or does it mean that the system is really trying to work on the CD drive? I don't know how notebook computers do POST but perhaps there is something there that freeze the POST from continuing.

Now the answers to your suggestions.

Have you tried removing all of the memory?? You should then get the 1-3-3-1 beep code. I'd also make sure that the hard drive and Ultrabay drives are taken out too.

Yes and there is no beep code at all. The HDD and the CD-ROM drive were taken out with no positive result and I put the HDD back in when it magically booted once as I described above.

I grabbed the front left corner and flexed it downward while pushing the power button. No difference in the power-up. Then I flexed the corner upward and turned the system on. 1-3-3-1 beep code!

I tried many twisting combinations and they were no positive results. I also tried to flex the bare motherboard when I took everything apart and also no good result.

I don't have an external monitor now but I will try to get one tomorrow and I will let you know once I try it.

[rkawakami]

From your description of how the system first didn't boot, then booted but froze, it sounds very much like what I experienced with the T23 from two years ago. I don't recall what the status of the hard drive LED was when the system was in the frozen state, but those LEDs are supposed to flash and then remain off, except for the "Z" power LED when booting. If the hard drive LED is constantly on, then I believe that indicates an actual attempt to read from the bootable drive (HD or Ultrabay). My best guess is that there is still some badly soldered device on your motherboard, particularly something associated with the timing circuitry, that is causing your problem. A good magnifying glass or microscope, along with a tiny tool to pick at or pry up on any component/connection is the one thing that I can suggest. The other thing to try is to have the system assembled, but with the keyboard bezel and keyboard screws removed. Power up the system, then remove the keyboard and try pressing on different areas and components with an insulated probe; a toothpick, chopstick or even a cotton swab will do. It's even easier to do this if you have a docking station or port replicator since they have a dedicated power button. You can mount the system in the dock with the keyboard already removed and use the power button on the dock/port replicator to turn the system on. Once you have identified any sensitive area(s) on the motherboard, then visually inspect those places for questionable solder joints.

[RealBlackStuff]

I'd forgotten about this thread, which popped up today!
Just this afternoon I was going through some spare T23 mobo's and one of them had exactly the symptoms as in Ray's first post.
Just plain resoldered all three quartzes Y1/Y4/Y6, put it in a base with RAM and keyboard, and it booted, HURRAY!
This mobo, that had never worked since I got it in a box-lot, came out of a T23 I never heard of until now, a 2647-7G1.
Thanks again Ray!

[rkawakami]

Ah, see what dredging up old threads can do!

[tfflivemb2]

Makes me want to dig out some of my old T23s...lol

[zmyx]

I tried the external monitor but it's all black. I will take apart the T23 again and try to find loosen parts.

Anyway, what if the charging IC (ADP3806) is broken? Since the battery doesn't charge so I think there should be something wrong about it. Although the system can be powered by AC or battery or both but only charging seems not to work (the battery LED doesn't indicate any activity for charge/status). Could a broken charging IC, power supply controller or IO controller, stop the system from POST?

[rkawakami]

Not sure, but if I have to guess, I'd say that a defective charging circuit would not affect POST from completing. Most likely, given my experience with T23 battery charging problems, you probably have a loose L7 inductor located between the S3 SuperSavage and Intel 82801CAM (I/O Controller Hub) chips, near the main battery terminals on the top side of the motherboard.

And to correct something that I said earlier since I'm now looking at a bare T23 motherboard... you should also remove the black plastic hard drive bay and the speaker assembly from the system in order to gain better access to the top side of the motherboard. A couple of the crystals are located under the hard drive bay and the Y6 crystal is covered by the speakers.

[zmyx]

ok, I made another "operation" to my T23 again yesterday. I plugged the mainboard to a port replicator and use it to power on the system. I hit all the crystals and all the 3 troublesome inductors again with my soldering iron and nothing happened. constant fan still there. Pressing various big/small components on the mainboard also didn't help. Then I randomly resoldered some pins of some chips on the mainboard in the hope that I might hit the problem but no luck.

I tried again to resolder the inductors, now every of them that I can see on the mainboard. I added some solder to the big inductor on the top left of the mainboard, the big flat one near the CPU. Then, while trying to power it on from my port replicator, I press that inductor and also press the CPU several times (I had the heat sink removed to work on the nearby inductor) and the system just spring back to live! I have to say that I have already reseated the CPU more than ten times before with no good result.I tested the CPU in another system before and it's working fine.

Now that the system works I'm not so sure that it's only the problem of the CPU and its socket. It could as well be the big inductor near the CPU. But a working system gives me hope now to continue working on it

So I have been trying several things with the T23. I can enter the bios and I saw that the installed bios is an old version (1.08a and 1.03a Embedded Controller). I booted into windows and try to upgrade the bios but it failed (DeviceIOControl return 24, please restart your operating system and execute the BIOS or Embedded controller update utility again). Then I tried to upgrade the Embedded Controller and it went well until the point that Windows try to shutdown I got a blue screen (STOP 0x0000007E - DOSBOOT.SYS ...). I repeated the whole process again and got the same errors.

Another problem is that the battery LED is still off. It doesn't show any battery activity and the battery info in Windows also indicates that no charging occurs. So the battery doesn't charge at all.

So, these are the update and some questions for now. The T23 is still running fine up to now except the problems stated above. I might start to dig up some other threads for these error but I would also appreciate any pointers form here.

[rkawakami]

The inductor next to the CPU, when it is loose from the motherboard, causes a totally "dead" system. At least that's what I've found on about three T23s. My recommendation when working on any of the inductors is to remove them from the board, clean the two contacts on the bottom edges of the inductor and then re-solder them back on. There's a problem with "black pad", a layer of oxidation, which prevents the inductor from bonding with the solder. You can clean the contacts by scratching them with a small knife or screwdriver until they are shiny-looking; actually there's usually a copper color to them. I would also suggest that the pads on the motherboard where the inductors are mounted be cleaned. This can be done with a de-soldering braid (i.e., solder wick). Apply a small amount of solder onto the inductor's contacts before soldering them to the board.

It certainly is possible that some of the CPU pins were (are) dirty. Removing it from the socket and re-seating it can sometimes help. What also can be done is partially locking the socket with the CPU sitting about 1mm above the socket (pins are inserted most of the way) and then gently pushing the CPU down. This will scrape the pins inside the socket. Unlock the socket, make sure that the CPU is sitting all of the way down and then lock it in.

Don't know about the errors you are getting when updating the BIOS and EC. It appears like you are trying to use the non-diskette update, correct? (running the update programs directly off of the hard drive instead of using a boot floppy).

The battery charging problem could be the inductor I previously mentioned, the ADP chip or a blown fuse. Check this picture:

http://www.rkawakami.net/ibm_t2x/main_fuses.jpg

for the F9 and F12 fuses. The inductor is at the top of the picture too, with "150 N21" printed on it.

[zmyx]

The T23 is still running

I was thinking about re-seating the CPU, too, but I would leave it for now since it's still working. The only problem now is the battery. Now that windows is booted I can install battery maximizer software and to my (not so) surprise it shows that no battery is installed. The battery LED is still off too and the system now only runs from AC. More about this below.

Don't know about the errors you are getting when updating the BIOS and EC. It appears like you are trying to use the non-diskette update, correct? (running the update programs directly off of the hard drive instead of using a boot floppy).

Yes, non-diskette update and it was running today so probably it's the fault from Windows. I went through the process and it stopped where it asked me to confirm that the AC is connected, then that the battery is connected. This is where it stopped because the system doesn't see my battery at all. I check again the inductor and the two fuses that you pointed out and they are all ok. So I checked the ADP3806 IC and I found out from the datasheet that the first pin is the VCC and I checked it with my voltmeter and it showed 16V which is ok I guess. But the datasheet indicates pin8 (REG) 6V and pin9 (REF) 2.5V but my measurements were just 0V from both pins. So I guess it's the ADP3806 that is broken here.

Does a broken ADP3806 make the battery unseen by the system? If I try to boot with battery it gives me critical battery error (0190 or something, will check again) and the system would shut down. It still boot from AC though. And as mentioned battery LED is not showing any activities and battery maximizer doesn't see the battery and the battery is not charging too.

[rkawakami]

It certainly sounds like something with the ADP3806 circuit is bad. Could be the IC itself or any of the surrounding components that support it (resistors, capacitors, transistors, etc.). The ADP datasheet should have a typical application schematic if I remember correctly. You could try doing a little bit more circuit probing to see if there's some voltage level that's not getting to the ADP. The problem is that there's no published schematic of the laptop as far as we know, so you'll have to figure out which component on the board represents which one in the datasheet reference circuit. The only other things I can suggest would be to measure the resistance of the inductor (between the two metal tabs on top of the inductor) to make sure it's near 0 ohms. I also measure between two spots on the board which the inductor is supposed to be connected to. This is done by looking at the where the copper traces are routed from each leg of the inductor and connecting the multimeter to some of the nearby components. There's usually a large capacitor on one side of the inductor and/or a ground plane. If your meter's probes are thin enough, you can even use some of the vias (the through holes in the motherboard) next to the inductor. This confirms that the inductor is actually making electrical contact through the soldered connections. Visually inspect everything in the area; both top and bottom of the board. Look for obvious signs of damage like in this picture:

http://img.photobucket.com/albums/v356/ ... smpart.jpg

(referenced in this post: http://forum.thinkpads.com/viewtopic.ph ... 75#p435675) and use a toothpick to see if you can pry up any loose parts.

[zmyx]

Thank you so much for all the helps. Since my T23 is now running I will continue my journey on the thread about SMD parts on T23 that Ray referred to.

Anyway, I will post further updates if something new about my constant fan problem is available.