486 DataExpert 4045 Motherboard - Stuck in Reset

[auxie22]

I posted this to Vogons but it's probably not the best place for it. I have started looking at repairing some old 486 motherboards and am a bit stuck. I am not sure if anyone here is good with these kinds of boards

I have a post card that shows the reset LED always on and all reset pins on all chips are high across the entire board

I have checked the clocks and they are all OK.

I have checked all the voltages and they are OK too (voltage regulator also changes based on jumper settings).

My understanding of most of the boards around this era is that they use the powergood signal for reset and that reset should go high for a second and then go low so it can then initialise the BIOS code

I have traced what I can and found the following:

powergood signal (5.1v) goes to pin 5 (input) of the 4069 Hex Inverter

pin 6 (low inverted output) of the 4069 appears to go nowhere unless it's broken somewhere inside the board as I cannot see any traces coming to/from that pin. (I'd expect this to go to a schmitt trigger buffer / hex inverter somewhere or 2 before getting to the chipset)

powergood signal splits off before getting to pin 5 of the 4069 and goes to pin 8 (input) of an SN7407N schmitt trigger buffer

pin 9 (output) of the SN7407N then goes to a diode which goes to ground

positive diode pin splits off to a 0.04 nf capacitor which goes to ground

positive pin of capacitor splits off and goes to the chipset (W83C491) - I am unsure what pin this is as I can't find the datasheet for this chipset, all I know is that it's a renamed Winbond / Symphony Labs

I have also traced back the RESETDRV from the ISA Slots

this goes to pin 10 (output) of SN74F04 Hex Inverter

then pin 11 (input) goes to pin 3 (input) of the 4069 Hex inverter (very odd)

pin 4 (output) of the 4069 Hex Inverter goes to the chip select of the RTC clock chip

I traced the reset switch jumper pins and they go straight back to the chipset (W83C491) - It doesn't seem to go through a buffer at all which is odd

I think I traced some of the other pins on the Hex buffers and inverters which also go back to the chipset but without the datasheet it's really hard to know what they are doing

I assume there is a RST In pin and Powergood / PWR OK pin on the chipset but don't know

If you have a working board and can trace where pin 6 of the 4069 goes that would help a lot or have datsheet for the chipset that would help too

But any help would be appreciated

Links to the various chips etc

https://theretroweb.com/motherboards/s/dataexpert-exp4045

https://pdf1.alldatasheet.com/datasheet-pdf/v … TI/SN7407N.html

https://www.ti.com/lit/ds/symlink/sn74f04.pdf … uct%252FSN74F04

https://pdf1.alldatasheet.com/datasheet-pdf/v … C14069UBCP.html

Thanks,

[fzabkar]

powergood signal splits off before getting to pin 5 of the 4069 and goes to pin 8 (input) of an SN7407N schmitt trigger buffer

pin 9 (output) of the SN7407N then goes to a diode which goes to ground

positive diode pin splits off to a 0.04 nf capacitor which goes to ground

positive pin of capacitor splits off and goes to the chipset (W83C491)

SN7407N-9 appears to be a buffered reset input to the chipset. I expect that there would be a pullup resistor, either discrete or internal to the chipset.

[auxie22]

Thanks fzabkar,

I will have another look but I didn't see a resistor when I traced it out so as you say it may be in the chipset itself.

What would be the point of pin 6 on the 4069?

Why go to an inverter at all if it's not being used?

[m k]

PowerGood is a pre-requirement, but the actual timeout for reset can be elsewhere, simplest is an external capacitor that is loaded, but it can be something inside a chip also.
Nasty thing is that there can be what ever other pre-requirements.
Sometimes RTC chip or similar was the actual resetting thing.

Keyboard controller is independent and primitive, so it can also act independently.

[fzabkar]

PowerGood is a pre-requirement, but the actual timeout for reset can be elsewhere, simplest is an external capacitor that is loaded, but it can be something inside a chip also.
Nasty thing is that there can be what ever other pre-requirements.
Sometimes RTC chip or similar was the actual resetting thing.

Keyboard controller is independent and primitive, so it can also act independently.

The RTC doesn't have a reset output (M5818 is a clone of MC146818):

https://www.nxp.com/docs/en/data-sheet/MC146818.pdf

Bit 0 of Port 2 of the keyboard controller is "RCB: System Reset":

http://www.bitsavers.org/components/winbond/W83C42_Keyboard_Controller_199407.pdf

http://www.bitsavers.org/components/winbond/

[auxie22]

Thanks guys,

I will have a closer look at the keyboard controller and surrounding components... Hopefully it's a faulty cap and/or resistor... and will let you know

[m k]

PowerGood is an active high open collector circuit.
It also means that "active PG" can only be not good.
Open collector also means that the same signal can be pulled down from multiple positions.
Since 4069-6 is low the power is good, at that point.

RTC CE low usually means that normal power is present.
(the chip generates some interrupts)

Chipset reset can have a feedback.
It can also be a chain, usually not long, but obviously spreading.

Post card is "later" in boot sequence.
The machine can be in a loop that never goes so far that the card is activated.
If that is the case then ROM must have some activity.
It's also possible that the system is halted after a serious error.

W83C42 and MC146818 RESETs are active low.

[auxie22]

Thanks m k,

4069-6 is behaving exactly like a reset signal when turning on the power, this makes me think that there is a broken trace between it and another device.

pin 21 (RCB: System Reset) of the Keyboard Controller is an output and goes straight to the chipset when turned on and is always high

I will check the RTC and Keyboard Controller again to confirm what their reset signals are doing

I might trace out the other pins of the Hex Inverters / Buffers and make a rough circuit diagram and then post it here

[auxie22]

Hey guys,

The board is now out of reset on start-up (cold solder joints on chipset) but I still have some issues as reset isn't working because when I use a switch to reset it, it goes back into reset as soon as I press it rather than pulsing.

Post card still shows no codes

Also every now and then the board will start up in reset again with no clock.

The reset switch is directly connected to pin 140 on the 83c492

I noticed that the 83c491 was getting warm when it was stuck in reset and it still is getting warm now that reset is working but 83c492 is cold

cache chips are also getting quite warm

CPU is getting hot 

I will check the clocks to make sure they are still OK but is there anything else worth checking?   

[m k]

Maybe there are still more bad connections in reset subsystem.

Maybe one of chipset chips is a memory controller and one is not.
Cache is also in place when memory is used.

From where you've got that reset must be pulsing?
Reset is usually a level and watchdog is pulsing, but watchdog can also be completely internal, so invisible.

When boot start there are some CPU internals happening before anything else.
Very soon after that at least part of low memory is checked.
(there pretty much must be a program stack)

What ever functionalities the machine is doing it must have a controlling software.
So ROM must be accessed.

[auxie22]

Thanks m k,

I think you are right about 491 possibly being a memory controller as it does connect to the cache but I am still in the process of mapping out the traces. I have been trying to find a datasheet but still no joy.

With regard to the reset, i meant when the button is pressed it will go high and at startup when it "pulses"  once. Probably a bad choice of words.

I have mapped out pins from a few of the chips, mainly the inverters and buffers and have found that 8042 gets a reset signal from pin 82 of 491 and that sends a signal to CE of the RTC and RSTDRV on the ISA slot.

I haven't reflowed 492 yet but I might do that and see if it resolves the issue.

[m k]

Maybe the initial reset pulse is from somewhere else.
Or maybe it's a multipurpose pin.