No, thanks to you (and all others) for helping me, im learning a lot!

I did the bleeding test before, but got the same behaviour, thats why i didnt say anything about it.

- About the battery, all test was done without battery in the slot. Now i did one test more WITH battery on, a CR2032 3V, got same results. Anyway i have an adjustable PSU if we need to do more tests on this. How many volts and current should i use? the original battery its 3v6. (tomorrow)

- Something interesting, after a couple of hours it was enough to "force" the boot by conecting C83 and R13, (see image), but only works 2-3 times, and then i have to let dry again...bleeding the caps does nothing.

- About the TC4069, i did a drawing of the corresponding conections:

I think your problems are caused by the battery leak (corroded traces or vias). Photos you posted yesterday show there's quite some damage. All the traces that have as much as a single point of corrosion must be checked for continuity. Removing the parts from the affected area is recommended, if you have the proper tools. If possible, snap another photo of the area but with the keyboard controller removed.

Well corrosion could still be the culprit but I think we should make sure the battery circuit is operating as it should. Personally I dislike desoldering parts from old PCBs like these, they can be very fragile. Trying to remove a chip can easily rip traces or pull vias, especially those already damaged by the spill. So that's why I ask to probe things first.

Anyway, that connection from pin 2 of the RTC/CMOS to pin 12 on 4069 (the pin count for 8-14 should be in opposite direction, they go counter-clockwise) is the clock signal. Datasheet for the HD146818 confirms it. Now, without oscilloscope you can't really see the signal, if it's correct or not, but even with volt meter you should be able to test it, more or less. When the PCB is powered there should be about 2.5V there (the actual value will depend on how the meter sees it because it's AC, not DC). If you see close to 0V or close to 5V then the clock is not running. Most likely because the 4069 did not start oscillating due to incorrect battery voltage. 3V battery might be too low for this board.
Keep in mind that the clock output will be lower when on battery. In general you should see about half the voltage that's supplied to the chip.

You can pretty safely connect anything from 3V (which you already have) up to 5V as the battery. 3xAA(A) pack will be 4.5V (3.6 if those are NiMH cells). An external DC PSU is OK too, anything will do really, the power draw is very small (way less than 1mA). So pretty much the voltage you see at the output when nothing is connected to that PSU is what it will feed to the motherboard.

So for this test:
- see if the RTC clock is running or not
- see if the 8042 keyboard controller is connected to the 4096 as well (It might be a different pin than the RTC though)
- see if both ends of R13 are connected to 4096

This will let us know where to look next.

Posting from work break...

quicknick wrote:

I think your problems are caused by the battery leak (corroded traces or vias). Photos you posted yesterday show there's quite some damage. All the traces that have as much as a single point of corrosion must be checked for continuity. Removing the parts from the affected area is recommended, if you have the proper tools. If possible, snap another photo of the area but with the keyboard controller removed.

Well already replaced the whole Keyboard rom socket, because it was very affected by corrosion, and even had couple of pin holes broken, also cleaned the corrosion in there, in the process as Deunan says, ripped off some pads/vias, and had to do some repairs with cables. Checked all visible traces for continuity btw, so in theory this is ok, but i can be wrong of course.

- About next tests...only need an excuse to buy an osciloscope (cheap 30€ amazon one max) always wanted to learn using that. Will this be enough for most basic operations?

https://www.amazon.co.uk/Kuman-Oscilloscope-S … rds=osciloscope

I am somewhat on the fence with these cheap "osciloscopes". On one hand, it is a scope and works (mostly) within the advertised parameters - you will not get anything better for 30 Euro. And because it's cheap there won't be much of a loss if you break it by accident.
On the other hand... it's very basic. 200kHz bandwidth, poor sampling rate, even worse triggering capabilities. It is basically good enough to test audio equipment but only the low frequency part of it (so no radios or modern PWM amplifiers). All scopes have some limitations but once you get familiar with one you can quickly tell if what you look at is a bad waveform or is it the scope not capturing it properly. This is even worse with cheap digital scopes like this one. It might be a good first experience as long as you understand its limitations and not trust it too much in other words.

So if you want to play with it, and 30 Euro is no big deal, get one. It's enough to see that 32768 Hz clock signal. But you won't be able to use it to test the CPU/ISA clocks as these are way in the MHz range. So if it's only for this job, maybe put that money towards a better oscilloscope.

Ok, ill forget about the osciloscope then for now. I did resolder all pins in 4069 and RTC, also R13 then Did the tests:

- There was no signal, 0v from 4069 pin 12, to RTC pin 2.
- This is the conetection between RTC and KB_ROM

- And finally R13 left end IS conected to 4069 pin 1, so it means it is also conected to C83. Im almost sure this connection was not in before.

So you did not find connection from R13 right side to the 4069? Than it's a problem you have to solve now 😀

The circuit should look something like this, based on the tests so far:

However I'm unsure about the first gate used, I guess it's 3-4 but could be a different one. So:
- You need to find the missing connection to R13. Perhaps the copper is eaten away.
- Pin 13 might actually be connected to the first gate output, that's a different variant of this circuit, but I'm assuming it's like I have drawn
- To confirm, the other pin of X4 crystal should also be connected to the 4069, if not it's another broken connection

So, all acording to the schematic. If the connections are broken then the circuit will not oscillate, and there will be no RTC clock.

Will test the circuit this weekend, i'm out of time!

Ok i've tested the circuit as the image, and its not working. Basically R13 Right end is not conected with any of the 4069 pins. Tomorrow / sunday will do some bypasses and try to boot all in video to see if there is any changes.

Hi, today completed the circuit using a cable bypass, sadly nothing changed, still no signal from the inverter to rtc, this scheme shows how it is now conected:

May be i need to change the suspicious capacitors now, its only a couple of them.

Well that's not good, I had hoped it would start working.
Okay, so one more thing, can you measure voltage on each pin of the 4096? It might turn out it has to be desoldered to check for any damage under it (hopefully the chip itself it OK, but if not a replacement would not be difficult to find).

Well i've measured the 4069:

Anyways, i've desoldered, 3 suspicious capacitors, especially the red one below R13 (C83), wich was snap open and burnt! that wasnt visible without extacting it:

Also desoldered, the suspicious 4,1 volts below isa 8 bit port 1, checkd it and it had 116nf its a common 104 one, so it seems normal, i will replace just in case.

And finally desoldered C74, ich has same problem of C83, but no visible damage, im replacing this too.

Im going to shop now, lets see if i can do this today.

2.5V so about half the VCC on pins 1,2,3,4,12,13 is good. It probably means the clock is now running (even if the system is not working still).
Without a scope you can confirm that with a LED indicator. Solder a probe like this:

wire -> 1k resistor -> LED kathode (-) -> LED anode (+) -> wire

So basically a series connection through the reistor and LED with 2 wires on the end. When you connect the wires to GND and +5V you will see the LED only lights up in one direction, not the other. Now, connect one wire to pin 12 and with the other touch to GND and then touch +5V. If you see the LED light up in both case (although it might seem dimmer) then you have a working 32k clock now for RTC. And that is fixed.

Confirm that it doesn't work with the test card, see if the RESET LED goes off or not now.

OK, i did the capacitor replacement, with no changes, even the new capacitor keeps some current on power off. But while i was tesiting the TC4069 voltages discovered this:

https://youtu.be/UgNND2ogCgI

Does it make any sense to you?

I'm sorry but I can't view that video, says it's not available. Wrong link maybe?

Deunan wrote:

I'm sorry but I can't view that video, says it's not available. Wrong link maybe?

Sorry, always forget to set it public, now you can view it.

Pin 6 is output from gate 5-6. So clearly it is driving something but shorting it out like that is not good for the chip. However according to the measurements you've made pin 5 is at about 0.2V, so the gate is working OK. The problem is why pin is low and what it is connected to. It can't be connected very far, it must be either the 8042 keyboard controller or the RTC.

So, take a close look at the mobo again and find that trace that goes to pin 5. If it's under the chip you might still need to desolder it...

Or you know, that might just be the connection to PWR_GOOD pin that I suspected before. So also try to figure out if that signal goes anywhere near that gate.

EDIT: One small typing mistake and a whole different meaning. Should be "not good".

Hi, new update!

You got it right, removed the power conector and found pin 5 was conected to POWER_GOOD pin, but the path was 100% corroded. So i've now repaired that path and the reset signal got back to normal! yay!

Still the error codes are the same.

and

https://youtu.be/84b4mSxHX8w

Hi, just like to say "thanks" for your discussion here which I am following with a lot of interest, learning a lot 😀

You can see also this circuit in my manual here
download/file.php?id=33356&mode=view

Whoa, schematics! What a find. Is there any more than this page, just in case?

I will have a closer look at this in the evening, but for now I'd suggest building the probe I mentioned earlier to test if the RTC is getting clock now. This will allow us to cross that out of the list. Actually with the clock now probably present, and RTC being quite far from the spill, I wonder if the error code is actually due to CMOS data issue or maybe rather the keyboard controller. So, since the 8042 is socketed now, remove it and see if the code is the same.

The next thing would be to check if the 8042 is getting correct reset and clock signals, and if the bus signals (/RD, /WR and /CS) are OK. For now measure the voltage on pins:
2, 3, 4, and also 5, 7, 20, 26, 40.