My Toshiba T1100 Repair

[CedsRepairs]

I'm opening this to share my adventures in repairing a T1200 Toshiba, circa 1984



I wish the alreadi thanks Modem7 for his great Toshiba ressources on minuszerodegrees.net





I'm very happy with the exterior condition of the unit.
I also have what seems to be a VERY rare modem/RTC/rs232 optional card.


When all goes well, here's what you can see:






I'll first describe my battery replacement ( in the next messages ), then other repairs.

 

[Franchute13]

Very nice machine!!
I wait for future post about repairs!

 

[CedsRepairs]

This is obviously a T1100 (not plus) and not a T1200 as I put in the massage body but it seems I can't edit.

Step 1 )

Batteries !

What I got with it looked like a 1992 replacement, this was not a 1984 battery, but it's not something someone has built in hosue either. I guess they put tape on it to avoid rattling.




Batteries where in a sorry state to say the least (1992 at best)




I bought equivalent batteries in Poland, where NICD are still in stock (don't ask me why) despite beeing forbidden in europe since 2006.



Soldered them



I also recycled the original 2x5A fuse ( which would make it a 10A fuse).



Battery I created looks as good if not better then the original from toshiba

One thing that was really difficult was to recycle the original cables (which I did) because they were oxidized even inside the plastic, in the while length ! Never saw something like that. I really struggled to solder them back, after cutting 10cm out of them.

Strangely enough; 100% fully charged this delivers 14V (on an empty load) instead of 12V. It works anyway and is within the Toshiba specs (+/- 20%).

 

[CedsRepairs]

Issues i've got

- the freaking "B" key does not work. I tested with a multimeter and contact does not happen at all when you press it.
I plan to unsolder the $/pounds key and swap them, as i'll never type "uk pounds" on this laptop.

- I have to power it on at least 10 times to get it to boot. This is a real issue for me.
I could blindly "recap" all the electrolytics and hope for the best ; but they don't seem leaky, the one i've unsoldered "as a test" is perfect in an ESR meter, and i'd like to get a real diagnosis. I'm poking around with the oscilloscope.
So for all I can tell is that plugging in the screen provokes a 3A current spike at boot, which doesn't happen with cd screen unplugged, and I find this very weird. All voltages are correct (+12v,+5v,-5v,-12v) and all within tolerance. It's difficult to make measurement with the laptop running because it has to have the lid cloded to have keyboard & screen. This makes debugging a PitA.

 

[modem7]

I have to power it on at least 10 times to get it to boot. This is a real issue for me.

For readers, is this intermittent 'boot' failure: 'nothing appearing on screen' or 'will not boot from diskette' ?

I could blindly "recap" all the electrolytics and hope for the best ; but they don't seem leaky, ...

The vast majority of faulty aluminium electrolytic capacitors that I have encountered, showed no visible signs of failure.

... the one i've unsoldered "as a test" is perfect in an ESR meter,

The primary property of a capacitor is capacitance. If I was measuring an aluminium electrolytic capacitor, capacitance would be my first measurement (measured out-of-circuit), and then I would measure ESR (which is a crude measurement).

And for any in-circuit ESR measurements, I would be watching out for capacitors wired in parallel, something that would compromise an ESR measurement.

It's difficult to make measurement with the laptop running because it has to have the lid cloded to have keyboard & screen. This makes debugging a PitA.

Your "it has to have the lid cloded to have keyboard & screen" has me confused. If the lid is closed, the keyboard cannot be accessed, and the screen cannot be seen. Maybe you meant partially closed.

So for all I can tell is that plugging in the screen provokes a 3A current spike at boot, which doesn't happen with cd screen unplugged, and I find this very weird.

I do not see 3A peak at power on time of either of my two T1100's. It is about 500 mA peak (measured using the min/max mode of my Fluke multimeter).

My T1100's can power up and boot to DOS from a (720K) boot diskette when the display unit is unplugged from the mainboard. I can tell that by:

* About 35 seconds after turn on, I hear a beep from the speaker (if fitted), then I see the access LED on the front of the diskette drive turn on; then

* Once I am sure that the DOS boot from diskette has finished, I will hear the speaker beep whenever I enter the following DOS command: echo ^G con
(The '^G' bit is done by holding down the CTRL key and then pressing the G key.)
(Of course, the DOS command could optionally be put into an AUTOEXEC.BAT on the boot diskette.)


BTW: If your T1100's speaker is missing, a speaker gets connected to connector PJ1 on the mainboard.

BTW: On both of my T1100's, I am yet again seeing poor power connection to the floppy drive. The connector (shown [here]) is very loose, and the poor connection stops the drive from working. Need to permanently resolve.

 

[CedsRepairs]

The boot failure/success ratio is currently 9 misses for 1 boot.

On a full battery, I switch the power button, the screen "blinks" (you can see the pixels lighning up if you put contrats to 100%)
It often briefly displays two horizontal lines while I power cycle.

Then nothing happens.
No memory count/test, no beep; nothing. Also, no error code on LPT (I checked)
I can leave it like that 5 mins, nothing will ever happen.

With an oscilloscope but without the top cover ( so keyboard + screen removed) all I can see is :
- All voltage rails are OK ( good voltage + no ripple )
- The CPU gets a good 4,77 Mhz clock on it's clock pin.

I've got this behaviour both on "LCD" setting, but also on CRT setting. (And with or without the keyboard+screen.)

Basically something crucial fails to power up,
I suspect the "bus controller" which got a tiny little bit of corrosion out of the old battery, but the rest of the motherboard looks tidy.

I could (as many people do) "recap" the power stage but again, I see nothing wrong with the voltages. Could they have leaky/faulty caps while outputting the perfect voltage they're supposed to ?

Hope this clarifies and really thanks again for your time on this,
(forgive me if my english sounds weird, not a native speaker, also the fact the forum doesn't allow me to edit my previous messages does not help)

 

[modem7]

... also the fact the forum doesn't allow me to edit my previous messages does not help)

In a way, that is good. E.g. Since a particular thread started a month ago, I have been reading new posts as they have appeared. The post count in that thread is now at 50, and the last post that I read was #47. All I should need to do now is read posts #48, #49, and #50. I should not need to also re-read the previous 47 posts because of the possibility that someone may have added/altered information in one of those posts.

But there is some time-limited post editing ability. For example, 30 minutes ago, I made a post to a thread. Right now, I still see an 'Edit Post' option (to the left of the 'Reply' option) for that post, but I know that sometime later, my 'Edit Post' option for that post will disappear.

 

[modem7]

With an oscilloscope but without the top cover ( so keyboard + screen removed) all I can see is :
- All voltage rails are OK ( good voltage + no ripple )
...

So you have eliminated the possibility that your 3A (peak) drawing display unit is causing power instability at start-up time (the time period that it is drawing 3A).

... so keyboard + screen removed ...

In case the information is useful to someone later, the diskette drive is not required either. With no display unit, no keyboard, no diskette drive, and no option card plugged into my good T1100 mainboard, the POST issues a single beep about 35 seconds after I turn on the power switch.

 

[modem7]

... the POST issues a single beep about 35 seconds after I turn on the power switch.

About 18 seconds for 256 KB of RAM (i.e. that soldered to mainboard).
About 35 seconds if an additional 256 KB is added via the plug-in module.

 

[CedsRepairs]

In my case removing the floppy did not help.

So even without screen, keyboard, floppy : no post beep. (Well : once every now and then)

I'm really a bit out of options. Voltage rails seem fine, unless a very short transient at boot time causes an issue (that is not to be ruled out)

Battery is as charged as can be (it's new). Motherboard looks OK. I tested one of the big 470uF caps of the dc-dc converters, out of circuit, and they are ok on esr & capacitance.

I feel i'm so close to find the solution and yet, i'm not making much progress.

Really the issue here is that my problem doesn't seem to be a power problem.
And the 80c88 CPU get's it's clock. (that's as much as I can tell with an oscilloscope)

So what is the actual boot process of the thing would help investigate a bit.
I'm still feeling that this is related to the bus controller. (the chip in the middle of the motherboard)...
I cleaned it at least 5 times...

 

[modem7]

So what is the actual boot process of the thing would help investigate a bit.

The very crude sequence will be:
1. CPU held in reset state.
2. CPU released from reset state.
3. CPU starts execution execution at address FFFF0, which corresponds to an address in the BIOS ROM.
4. POST within the BIOS ROM executes.

With no known circuit diagrams available for the T1100, you may need to work out some of the circuitry.

An oscilloscope will show some things. For example. For the times your T1100 starts, your oscilloscope could show the RESET pin on the CPU being HIGH and then going LOW. For example, the presence of activity on the CHIP ENABLE pin (or equivalent) of the BIOS ROM.

I'm still feeling that this is related to the bus controller. (the chip in the middle of the motherboard)...

The chip affected by battery leakage.

I cleaned it at least 5 times...

Rhetorical question: Assuming an intermittent bus controller chip, might the leakage have somehow made its way into the chip ?

 

[CedsRepairs]

We're talking about minimal leakage here , I don't think it could harm the chip itself
I tried to check the connectivity of each trace but the freaking thing is so small that it makes it a pain

Also, the fact is you're experiencing the same issue, and I suppose you don't have that leakage problem ; so we both have the same problem and I think I should forget about the Bus Controller then.

Problem is you clearly have a power issue (you see wrong voltages) where I see perfect voltages.
We have the same symptoms, which leads me to believe we have the same problem.

I'll order what's necessary to properly recap the whole DC-DC boost converters section. I have no other idea at this stage ; all I can say for sure is that what's wrong is on the main board.

 

[modem7]

Also, the fact is you're experiencing the same issue, and I suppose you don't have that leakage problem ; so we both have the same problem and I think I should forget about the Bus Controller then.

Problem is you clearly have a power issue (you see wrong voltages) where I see perfect voltages.
We have the same symptoms, which leads me to believe we have the same problem.

I'll order what's necessary to properly recap the whole DC-DC boost converters section. I have no other idea at this stage ; all I can say for sure is that what's wrong is on the main board.

Like your T1100, the symptom in my T1100 is intermittent. It makes the problem more difficult to diagnose. After heading down the wrong path a few times, I am now convinced that the diskette drive is the cause (intermittently resulting in the +5V being 0.3V). I will do further investigation/diagnosis.

 

[CedsRepairs]

On my side I'll rate the ripple/noise on each power line

I'll also try to capture how the powerlines behave when powering the thing on.

 

[CedsRepairs]

As promised previously ( each message I post here takes 2 days to be approved...)

Test set up


A lot of hardware for just an old vintage PC

First measurements :

Ramping up of 5 volt rail, looks ok. Minor overshoot.



Levels (voltages) of all 4 voltages rails, simultaneously
All voltages are within range.



5V is a bit high, but this is without screen, floppy or keyboard.

Ripple and other noises.



Here it looks worse than it actually is.

The two negative rails (-5 : pink, -12V : cyan )
are somewhat ok. All signals have a clear "high frequency" 33Khz ripple due to the DC-DC boost converters.
The levels ( 20-30mV are way within specs )

The two positive rails ( +12V blue, +5V yellow : which is the most important one )
We're around 60mV ripplet, specs is 100mV according to maintenance manual.

However all these measures are using average mode.

At worst we're slightly out of specs for 5V ( a bit high, specs says +7% ok, we're at +10% )
Ripple is close to the max of the specs.

Wether or not they call "ripple" the high frequency switching of their DC-DC is unknown.

In short,
I've seen much worse voltage rails. If the thing is really not working because of such "slightly out of specs" it's really sensitive. But it could well be.

 

[CedsRepairs]

I posted an answer here a week ago, still waiting for approval ? Seriously ?

 

[modem7]

Refer to the information at [here], which provides some detail into the (initial) startup sequence of an IBM PC. Of course, the T1100 will be different in some respects (e.g. how RESET is generated, etc.), but in essence, the T1100 will be doing similar.

The oscilloscope screen capture at [here] shows that on my good T1100, the approximate delay between apparent +5V stability (top trace) and the CPU's RESET pin being taken low, is about 150 ms. Several measurements done; 150 ms approximate is consistent.

The oscilloscope screen capture at [here] shows that on my good T1100, activity on the BIOS ROM's /CE pin (pin 20) (bottom trace) starts about 2 us after RESET on the CPU is taken low.

 

[CedsRepairs]

Hey Modem7,

Well I've made some progress in some areas, and went back in others
Summary

- Now boot reliably (!)
- Now spits out a "NO KEYBOARD" error whatever I do. Yes, I'm close to toss it thru the window.

In details (mostly if it helps someone in the future than for me)

I've recapped the entire screen. The old caps where within specs, but as I had a transient speak of current that only appeared when plugging in the screen, I thought, hey why not.

Original caps



New caps (panasonic)




I've also recapped the entire powerstage (electrolytics) + a few tantalums

Original ;


New caps (smaller, technology advanced in 34 years)



All of this was... Quite useless :D
I've tested everything I desoldered and they were all within specs. Those Nippon Chemi-Con are just immortal.
I've tested the power supply after recap and it displays marginally better ripple.

Again I was pretty sure from my measurements caps where not the issue, but well I wanted to try anyway.

 

[CedsRepairs]

What seemed to be the root cause of my boot issue was, as I suspected, the bus controller.

It probably suffered more thna I thought from nicd battery spills. These were very small, but here's what I learned
- The liquid that spills it actually NaOH so highly corrosive.
- If you neutralize it with vinager, which I did, you end up with sticky salts residues that conduct electricity even more than the original spills

SO clearly I wouldn't recommend at all to neutralize with vinegar on a motherboard, as the resulting salts made it even worse.

I've got back my booting after washing the entire motherboard under the shower with soap + rinse with 99% IPA. Half a dozen of times, with little brushes.

Now i'm with

NO KEYBOARD

Whatever I do.

 

[CedsRepairs]

I've probed a bit with the oscilloscope on the keyboard controller
(or what appears to be a 8 Bit microcontroller used, probably, as a keyboard controller ; a 80c49AF-6) - lower right of the board

The thing gets a 4Mhz clock in its XTAL pins (around 33/34 iirc) - top left pins.
Gets a proper RESET (high). top left pins as well

Nothing seems to happen further than that.
Either this is the most silent microcontroller on earth, or it's dead.
Not sure where it gets it's code from. Not sure what could make it not work ( broken trace, broken pin ? )

But I'm a bit lost with my "no keyboard" error ( of course the flat cable is connected )
I checked the keyboard board and it seems fine. Switches are fine. Flat cable is fine and tested a dozen of times. Solder joint on flat cable & keybaord board are fine.

I believe either the 80c49 fails to start for some reason, or it's just broken.
WIthout a schematic it's really not easy.
I need a real hobby