FuST wrote on 2023-06-16, 20:26:

The first T1000XE wasn't that hard to fix, replacing a couple of capacitors and some minor PCB repair did the trick. I started w […]
Show full quote

The first T1000XE wasn't that hard to fix, replacing a couple of capacitors and some minor PCB repair did the trick.
I started work on the second machine I got but haven't got around to finishing it, real life kind of got in the way.

I did start work on a T1000 which confirmed my thoughts on all the T-series machines having a similar architecture for the PSU section.

If the PCB looks good you're probably going to be okay with just replacing the caps in the buck/boost converters (i.e. the larger ones). The switching MOSFETs/transistors can take quite a beating on these.

Well, I have replaced all the electrolytics. Many were leaking. The pcb looks great though.

I am actively scoping the various rails to try and see why it won't fully power up.

What I observe is that, on pressing power button, all the converters activate but after a second or 2 the controller shuts the supply down via logic on pin 33 of the processor.

I can see the 6xAA kluged battery pack struggling under the load which I think is more than an Amp because the AA cells are <1V during power up. With full
Nimh Batts that's a lot of current.

All the major transistors seem ok.
I found a damaged diode that appears to stll work...curious.

I feel like the 5V rail is where the current is going. I have removed all 5 soldered jumpers which I think isolates the supply from the board. Same failure. So that makes me suspect the 5v converter is somehow Sinking a lot of current . I don't have a good way of measuring the transient current through the psu to really know though.

So my thesis is that the 5v converter struggles and pulls a lot of current, which the controller senses, and them shuts down.

Tough slogging. Your T1600 schematic is helpful as is your repair guide.

The controller can't sense current draw directly, it doesn't have the circuitry to do that.
A possibility is that one of the rails gets loaded down a lot which causes the voltage to not reach the appropriate level in time causing the controller to shut things down.
A 4-channel scope is handy in these situations as you can monitor all voltage rails simultaneously.
The damaged diode seems suspect to me, the switching regulators use shottky diodes because of their low forward resistance. I'd advise to check (out of circuit) that they are within spec and are actually still diodes.
Scoping both sides of the diodes when the PSU is running (or trying to) should also validate this: one side should show the switching of the transistor and the other should be smooth which is a nice property of how buck/boost converters work

IIRC the problem with my second T1000XE is that it doesn't power up when the LCD is connected, which is where I was stuck at at the time.
Disconnecting the LCD and removing all the jumper links that connect the PSU section to the rest of the board is how I started troubleshooting, it isolates any other problems.

Also note that the battery pack must (for some reason unbeknownst to me) be present on the 'XE and have at least some charge in it so it is detected by the controller IC.
I've replaced the cells in one of my T1000LE/XE batterypacks with a couple of NiCd AA cells from Amazon, those work like a charm and don't require any changes to the charging circuitry.

Gut check on output voltages..

W504 W504 = 5V ie VCC
W501 = 12V
W502 = -9v

VEE for LCD =??? What's this rail supposed to be? Probably not -25v I would guess. That's what I have.

Edit. Vee for LCD is 22V. So that's good. I found a table of supply ratings for T12oo in the maintenance manual. Very helpful. Lists out what subsystems get what power.

Well thank you for your hard work on that T1600 schematic. T1000XE is close enough for that doc to be quite helpful. I got my machine working today. Quite a bit of effort to find the blown parts. For me, I had a blown LM358 that provided sense signals to the controller. Also I had a blown transceiver which was loading -9V way too much. Also some small diodes were blown. Lots of time to find the issues!

That's good to hear, glad the T1600 schematic helped.
It's unfortunately not uncommon for more parts (like diodes and the comparator) to fail as people try to power the machine up without replacing the caps. It can cause some pretty large voltage spikes when they're not doing their job correctly and it's a nightmare to track down without knowing how it all works together.

Anyway, glad another T1000XE was saved from becoming scrap!

Hey, reviving this thread and I'm reviving a toshiba 286 laptop as well

In my case it's a Toshiba T2000/40 -- which seem to be rare and information about it is hard to find.

I have two issues
1) I've got the freaking PSU problems "red blink" everyone else is having with this generation of laptop, I won't go into that now.
2) my screen behaves in a weird way

As you can see, it's perfectly split in two. It does not seem to be a backlight problem as the separation is pixel perfect
In (reverse) text mode, you can see the white is just less white in the bottom half of the screen

In WIndows 3.1 you can see that actually colors (out of the 16 gray levels) are actually missing in the second half. Bottom of the window's corners are missing the white outline, pixels are also missing here and there.
I tried lowering the number of colors in the bios (16 or 😎 and the result is the same

Did anyone ever see this problem, I'm really open to suggestions.
Is it in the screen ? In the video RAM ? In a cable ?

It's really like a bit or two is/are missing out of the bottom half of the screen.

Any advice is greatly appreciated

I've got a similar problem on an AST Premium Exec 386 with mono screen... I've tried everything up to dismantling the LCD module, sometimes pressure on the edge of that seems to cure it temporarily. So I am assuming there are solder or other problems inside. I have not gone into it yet, due to having no info on how "reversible" the disassembly is. Probably the easy fix is to take model info off the panel assembly and attempt to get a good deal on a replacement.

Could be a number of things, really.
What springs to mind:
Dodgy capacitors on the LCD module
Cold (cracked) solder joints
Bad contact on the LCD zebra strip (I don't know if this LCD has those and I'm not sure that wpuld cause this type of fault)
Bad LCD driver ICs on the LCD module
Bad video cable
Bad video RAM / display driver circuitry.

Does this model have an RGB (CGA/EGA style) or composite video connector so you can rule out the video circuitry on the mainboard using an external monitor?

Okay, guys... next one 😀
I also got the T1600 with non-working PSU. "Troubleshooting manual" describes the procedure of PSU diagnostics as the batteries should not be connected. However, on the 1st page of this thread it is clearly indicated that they should.
So the question is: do I have to repair batteries (main right and left and ram-backup-battery) and connect them prior to trying start the PSU ? I have already checked the board, FETs, +5V and clock on both controllers - and got nothing. As soon as I press start button red LED starts blinking whithout any attempt to start PWM converters.

Batteries are not required for the T1600 PSU board to operate. The rest of the computer isn't even required, it will work stand-alone (i.e. only connected to the 12V DC brick).

Is there no PWM on the MOSFET gates at all? If it attempts to start but fails to reach the proper voltages in time there should at least be some PWM activity for a short (couple tens to hundreds of ms) while.
If there's no PWM at all I'd check if the internal 18V main bus comes up as this drives all the other converters.

Thanks a lot! I am reading through your notes and diagrams for last 2 weeks 😀
Well, "just on the table" the PSU behaves a bit erratically. It does not start and does not even try to start 18W buck-boost when I press the button. I attached LEDs to 5VA, 5VB and 18V rails - well, right after switch on 5VA and 5VB are on, couple seconds after that moment 5VB goes down. Push the button and 5VB goes up and immediately red light starts blinking. No sign of flash on 18V and no clicking sound, just silence. Actually, no PWM on IC13 outputs.
However, just at powerup (+12 from mains) PSU might (or might not) produce a couple of clicks and very short flashes on 18V rail meaning it tries to launch main converter. Then it reverts to usual state: steady red, +5VA and no luck when button pressed.

Had to have a bit of a think about this.
The behaviour you're seeing with the 5VA and 5VB bus makes sense. I had to wait a couple of seconds after 5VB went off before attempting another powerup.
A clicking sound would suggest that a capacitor or inductor is shorting, the PCB should be completely silent when running.

Let's start at the 18V bus to make sure it actually comes up to the full 18V.
The 18V bus is enabled by Q14.
The converter itself is controlled by IC13, the signal is buffered by IC5 and then sent to Q29. It further consists of inductor L6, shottky diode "7" and C7.
I'd start by checking these components (Q14, Q29, L6, "7" and C7) are good.
A bit further down the signal path:
The 18V zener diode "9" controls Q27 which can in turn pull the PWM signal to Q29 to ground.
This works as a failsafe, when the 18V rail rises too high the PWM is shut off.
C6 is a tank/smoothing capacitor.

The voltage on the 18V bus is sensed through a voltage divider formed by R57 & R58 and fed to IC13 pin 2 through current-limiting resistor R96.
Check continuity on those current paths.

Do you have access to a 'scope? It's pretty much required to do further troubleshooting.
Even one of those cheap Chinese pocket-type ones can be helpful as the signals that need to be checked are way too fast/short for LEDs.
For example: The 18V bus should come up within a couple of 100 _micro_seconds. Too slow and the processor will give up.

Thanks a lot for your support!
By "clicking sound" I mean normal sound of working inductors, especially they are switching on. They arent completely silent 😀
I will re-check the sequence you suggested, but as far as I remember, PWM do not appear at all on pin34 of IC13.
My scope is Tektronix TDS1012, old, but seems to be working good enough.

One more question: can you measure signals on pins 1-3 of IC12, coming from what you named "current sensing"? In Multisim model I got they should be around 0 volts during normal work. In reality I got approx. +1.1V there.

My PSUs don't make any sound when running. At least nothing I'm able to hear which is generally anything below ~18KHz.
When the board is first switched on there is a bit of sound from the inductors physically deforming a bit under the magnetic field, I guess that could qualify as a "clicking sound".

That Tektronix 'scope should be more than enough.
When trying to see if there is any PWM signals make sure to set it to single-shot trigger mode.

The voltage on those pins seems about right.
This is one of the problems with circuit simulators: they assume perfect components, zero-resistance wires, etc.
Pin 1 measures the current consumed by the entire board (and entire PC when installed) which can never be zero.
Because the MPU uses a 0 - 5V reference voltage it has to be a positive value.

I've checked on one of my T1600's, I get between 0.8V and 1.2V on pin 1, and 1V on pin 2 and 3 when powered off.
Powered on pin 1 hovers between 1.5V and ~2V, pins 2 and 3 are unchanged.
This makes sense as the board is running, thus consuming more power which is reflected by the higher voltage on pin 1.
Pin 2&3 are the same as no batteries are connected.

Oh, thank you so much and, of course, Merry Christmas and Happy New Year!
I was quite busy with a bit bigger machine (particle accelerator, which doesnt want to work too) but never forgot about the laptop's PSU.
Now I spent couple hours looking for some patterns in how does it start.
Well, I still do not understand for what reason it tries (or not - 50/50 chance) to produce PWM at the very power-on (before ptressing the button), but at least in some cases the signal at pin34 IC 13 looks like pic.1 Good enough?
Then in couple milliseconds it comes to DC 5V and stays so for 3-4 seconds (pic.2 ) until IC12 removes 5v_B from IC13. If the button pressed within these seconds (while 5V_B still high) - no PWM and LED starts blinking.
If the button is pressed after 5v_B is down, then 100% chance of getting one more attempt to start 18V PWM as on pic.3.
Looks like normal logic for CPU? I guess I should trace this PWM signal further thru the gate and FETs (however, I've checked all the FETs out of board)

Those PWM signals look normal to me.
The blinking LED when trying to re-power the board is something I encountered as well. You have to wait for the 5V_B rail to drop below TTL logic high (around 2V) but preferably 0V.
Makes testing and debugging a bit tedious as the capacitance on that rail means it can take up to 5 seconds before another attempt can be done.

Next check I would do is if the PWM makes it to Q29 and if any 18V comes up and regulates properly.
If so it's time to start checking the other PWM driver pins on the ICs, check the MOSFET gates, etc.

Hi all,

I am also trying to repair 2 T1600 - one is in a very bad condition, so I started from the less rusty 😀. I did not try to switch it on and I directly substituted all caps, luckly there are no significant corrosions...now I am verifying if it works properly before reassemble it. When connected to its power adaptor, the offline voltages are fine, I get 12v on the Mosfets (Q14 Source, Q1 Drain, Q5 & Q6 Source, Q2 & Q3 Drain).

When switched on, the Led remains solid red, no blinking at all...so I checked the PWM signal at the Mosfets' gates. Q7, Q5 and Q6 look fine to me...but on Q4 gate I have no PWM, just a small fluctuation, almost 0v. I can measure 12v on Q4 gate when the PSU is off, but when I switch it on it goes to zero volts. I checked also all outputs of the Mosfets, again Q7, Q5 and Q6 look good, and on Q4 Drain I have 12v volts stable (the same on C3 positive) - does it make sense as I have no PWM on its Gate?

Considering this behaviour I checked Q23 base, also here I have 12v when the PSU is off and 0v when it's on. Then I checked IC5, UPA1600, here I have the same situation on the output pin 15 (PWM 12V): 12v when psu is off and almost 0v when it's on. I also checked the input (pin 6) and here the situation is different, I have 0v when the PSU is off and a PWM at 5v when it's on.

Could be the mosfet inside the IC5 UPA1600 broken? Any other options to consider? I add some picture to better explain the situation...

Thank you!

Q7 Gate - PSU ON

Q5 Gate - PSU ON

Q4 Gate - PSU ON

Q4 Drain - PSU ON

Q7 Source - PSU ON

Q5 Drain - PSU On

Q23 Base - PSU OFF

Q23 Base - PSU ON

IC5 UPA1600 - Pin15 P12V_PWM - PSU OFF

Ic5 UPA1600 - Pin15 P12V_PWM - PSU ON

IC5 UPA1600 - Pin6 P12V_PWM_DRV - PSU OFF

IC5 UPA1600 - Pin6 P12V_PWM_DRV - PSU ON

This seems like a straight forward issue with these for once.

The mosfet in IC5 sounds like the most likely culprit.
The UPA1600 is a pull-down only inverter (see datasheet). The base of Q22 and Q23 (totem pole transistor drivers for Q4) is pulled up to 12V through R35 when the board is powered off.

Since both IC5 and IC11 have at least one spare channel (see schematics) you could try to lift or cut the leg on IC5 pin 15 to isolate that channel and use an unused one instead to see if that improves things.

Edit: on second thought, that doesn't make any sense. The only failure mode would be that specific channel on IC5 being marginal and not switching fast enough.
That thought lead me to another one: are you sure your oscilloscope is fast enough to register the PWM signal? In your last picture it looks to just be able to detect it but I'm not convinced it's within the supported bandwidth.
I unfortunately don't have a T1600 handy right now to see what you should be getting.

Is the PSU running stable or is it switching off immediatly?

My fault - very sorry for this, it was a matter of scale of the signal, I increased the sensitivity and the PWM showed up.
I was expecting a different voltage on Q4 Gate, on Q7 Gate I have 5.8 Vmax and 6.8 Vpp, while on Q5 Gate I have 11.5 Vmax and 5.2 Vpp, also Pin6 P12V_PWM_DRV on UPA1600 has 4.4 Vmax and 2.6 Vpp ... but for some reason on Q4 Gate the PWM I measure has just 0.58 Vmax and 0.25 Vpp...anyway, it's still a PWM, with 78 Khz freq as expected I think (please see picture). I don't know if these voltages make sense, anyway the PSU is stable, it remains "on" and I have never seen the led blinking. I will assume it works, and continue cleaning the rest of the machine... luckly it seems this PSU was saved "in time", the caps were just starting to leak...

Q4 Gate - PSU ON

I also have another T1600, in this case the PSU was in bad conditions, really soaked in the liquid leaking from the caps...on this I am still working, this is what I did so far:
- switched on without checking nothing as soon as I got it -> very bad idea indeed...I smelt immediately a strange odor and the PSU led started to blink...
- substituted all caps
- switched on -> smoke from Q2
- substituted Q2, switched on -> smoke from Q29
- substituted Q29, switched on, I have again 18v -> but Q4 becames very very hot...
So i started to dissolder components around Q4: Q4 itself, Q23, Q22 and D12 seems fine according to my chinese component tester. I also dissoldered the small inductors and they are also fine, measure are as expected.
Now I don't know what to do next, it seems there is something wrong around Q4 but I have no idea to test next...

Thank you!

Another question: any suggestion on how to substitute the 2 battery packs? I have not yet search for their specifications, the yellow 4 battery pack perhaps it's easy to replace, for the brown one that seems make up by 3 cells I have no idea...

If the board is running normally I wouldn't worry about the level of the PWM signals. As long as the MPU's are happy with how it's running so should you 😉

As for the second one, the only things I can think of would be either Q4 not being switched correctly (i.e. turned ON all the time) causing a dead short through Q4, Q2 and Q3 or a short on the +12V output rail.
Some missing traces might also cause this but I can't point to a specific one.

Electrolyte can do a lot of damage, cause shorts and corrose traces (especially underneath IC's) and is very hard to clean off. Have you thoroughly cleaned the board? Inspect it closely and measure every suspect trace start to finish. Some trace damage is invisible tk the naked eye and even hard to spot under a microscope.

Edit:
About the batteries: I wouldn't bother to be honest. I used to replace them but stopped doing it a long time ago for multiple reasons.

The brown 3-coincell pack is just for keeping the RTC running and persisting the CMOS settings. I used a battery pack for DECT phones in the past. The yellow pack is for the suspend to RAM (i.e. stabdby mode) functionality, it consists of 4 or 5 AAAA cells. Both aren't that useful especially since the default CMOS values are all you need.

On top of that, both should be NiCD cells which are getting increasingly difficult to find. You might get away with using NiMH but I haven't had any luck with those, they tend to overheat when used with a charge circuit built for NiCD.
Another problem with those battery technologies is that they self-discharge rather quickly, damaging the cell when it approaches 0V. Not using the machine for a couple of months can be enough and would require disassembly and replacement of the cells again. Not worth it IMO.

In any circumstance DO NOT just replace them with lithium batteries! Can't stress that enough as that poses a major fire and/or explosion hazard 100% of the time. Lithium battery technology requires a lot of monitoring and control logic like overcharge and overdischarge protection, charge/discharge management, etc. for it to be used in a safe manner.