trace what its connecting to, hopefully its Winbond, then it will become clearer what its meant to be

It looks like this transistor is connected to +5V, LM78CCVF-J pin 40 (/SMI) and (I think) chipset (trace goes on back side down to the bios, jumps back to front and goes under ISA slot)

/SMI is a digital output (open drain), so I guess this transistor is supposed to pull it up.
I don't know electronics well enough to deduce anything...

Where is the second terminal of resistor R9 connected?

And the resistor values (10 kΩ and 2.7 kΩ) in the control circuit clearly indicate that this is a bipolar transistor, not a MOSFET.

shevalier wrote on 2026-03-22, 04:59:

Where is the second terminal of resistor R9 connected?

And the resistor values (10 kΩ and 2.7 kΩ) in the control circuit clearly indicate that this is a bipolar transistor, not a MOSFET.

Oh, I missed that. R9 is connected to +5v as well, so that makes it a pull-up resistor, right?..
BTW, R9 is 1 kΩ

Ah, I've got what you mean. R5 and an unmarked resistor create a divider:

But I'm not sure what this is supposed to do... 🤔

IIUIC, this transistor must be NPN and would act as a current sink when /SMI is pulled down by LM78 (assuming pinout is BEC).

imo looks like N-channel mosfet doing level conversion/buffering logic signal, but its only purpose would be buffering SMI interrupt, and that would be only used in case over voltage/temp alarms are enabled in bios. Might be safe to leave the spot empty.

rasz_pl wrote on 2026-03-22, 07:49:

imo looks like N-channel mosfet doing level conversion/buffering logic signal,

Given that it is not a signal but simply a resistor divider connected to the control pin, that is entirely possible.
But the pull-up resistor R9 is connected to +5V.
That is, +5V is the ‘low-level side’, so a higher (rather than lower) voltage should be taken from the drain, because of the diode built into the MOSFET.

It looks more like a level shifter using an NPN transistor.
5V from the monitor is converted to 3.3V for the chipset.
Although, why didn’t they just pull it up to 3.3V directly using R9?

deM-on wrote on 2026-03-22, 07:36:

IIUIC, this transistor must be NPN and would act as a current sink when /SMI is pulled down by LM78 (assuming pinout is BEC).

It looks like you need to find a pull-up/pull-down resistor connected to the drain/collector
Depending on the voltage this resistor is connected to, this will determine the type of transistor

I have a BX3 and may be able to trace the part later today. Mine has no PS/2 or USB functionality.

PcBytes wrote on 2026-03-22, 10:30:

I have a BX3 and may be able to trace the part later today. Mine has no PS/2 or USB functionality.

If you can, a highres picture of the SOT23 parts marking would be great.

Perhaps this gives some ideas:
https://github.com/raszpl/440BX/blob/main/ref … DEL%20440BX.pdf
Instead of the LM78 they're referencing the LM79 here, but they appear to basically the same. The signal there goes to the FW82371EB, without any transistors in between.

tehsiggi wrote on 2026-03-22, 10:37:

PcBytes wrote on 2026-03-22, 10:30:

I have a BX3 and may be able to trace the part later today. Mine has no PS/2 or USB functionality.

If you can, a highres picture of the SOT23 parts marking would be great.

Yes, please! 🙏🏼

tehsiggi wrote on 2026-03-22, 10:37:

Perhaps this gives some ideas:
https://github.com/raszpl/440BX/blob/main/ref … DEL%20440BX.pdf
Instead of the LM78 they're referencing the LM79 here, but they appear to basically the same. The signal there goes to the FW82371EB, without any transistors in between.

Oh, great staff! I checked LM79 datasheet, and it looks very similar to LM78.
The only possible explanation I have left is that this transistor is necessary for isolation - if LM78 pulls the emitter low, it opens the transistor, and it works as a sink to pull the line down. But when the emitter is high, the transistor is closed, and the line can be pulled up or down independently.

hey, that address looks familiar 😀
/SMI is Open Drain = chip only pulls it low, no level conversion or buffering is needed at all. It should go directly to PIIX4 southbridge. Makes me even more dubious about purpose of this transistor.
P2B mobos for example connect SMI directly to chipset. There is also 3.3V pullup, small cap to ground and Header pin.
edit: looked into other schematics I have,802, 815 and VIA boards gigabytes. Not a single one has any thermal alarm functionality meaning they all either omit it completely or route extsmi to a header.

Pics

So from that picture my assumption would be "04" as a marking, which leads to the DTC114 NPN transistor (with integrated resistor, I've seen this last years ago in consumer AV electronics) - with it's pinout I'd assume something like the following:

Could be a way to disable the SMI coming from the LM78 to the chipset on demand. However it somewhat surprises me, since the SMI could also just be disabled in the LM78 via a register setting.
There is for sure another player in the game, because there is a via at the resistor divider (marked as SOMEWHERE) that has the power to disable the transistor (normal on).

pertinent question is does the board post now? because this transistor has zero influence on the rest of the system, and its lack should not stop it from 99% normal operation, that remaining 1% being not working bios voltage/temp monitoring alarms and case Intrusion switch.
if eh board works I would just leave it be missing
if it doesnt this component is not the source of your problems

PcBytes wrote on 2026-03-22, 18:42:

Pics

Thank you! 🙇🏻‍♂️

tehsiggi wrote on 2026-03-22, 19:17:

So from that picture my assumption would be "04" as a marking, which leads to the DTC114 NPN transistor (with integrated resisto […]
Show full quote

So from that picture my assumption would be "04" as a marking, which leads to the DTC114 NPN transistor (with integrated resistor, I've seen this last years ago in consumer AV electronics) - with it's pinout I'd assume something like the following:

The attachment Screenshot from 2026-03-22 20-14-56.png is no longer available

Could be a way to disable the SMI coming from the LM78 to the chipset on demand. However it somewhat surprises me, since the SMI could also just be disabled in the LM78 via a register setting.
There is for sure another player in the game, because there is a via at the resistor divider (marked as SOMEWHERE) that has the power to disable the transistor (normal on).

Sorry, I forgot to check this trace 🙁. I'll check it and post results ASAP.

rasz_pl wrote on 2026-03-23, 01:56:

pertinent question is does the board post now? because this transistor has zero influence on the rest of the system, and its lack should not stop it from 99% normal operation, that remaining 1% being not working bios voltage/temp monitoring alarms and case Intrusion switch.
if eh board works I would just leave it be missing
if it doesnt this component is not the source of your problems

I didn't try to run it yet 'cos it needs recap and I'm waiting for delivery.

tehsiggi wrote on 2026-03-22, 19:17:

So from that picture my assumption would be "04" as a marking, which leads to the DTC114 NPN transistor (with integrated resisto […]
Show full quote

So from that picture my assumption would be "04" as a marking, which leads to the DTC114 NPN transistor (with integrated resistor, I've seen this last years ago in consumer AV electronics) - with it's pinout I'd assume something like the following:

The attachment Screenshot from 2026-03-22 20-14-56.png is no longer available

Could be a way to disable the SMI coming from the LM78 to the chipset on demand. However it somewhat surprises me, since the SMI could also just be disabled in the LM78 via a register setting.
There is for sure another player in the game, because there is a via at the resistor divider (marked as SOMEWHERE) that has the power to disable the transistor (normal on).

It doesn't seem to go anywhere. At least there are no traces from that via on the back side...
Back side (mirrored)

Front and back together, false colors

It's a multi-layer board (at least 6-8 layers) - so it's surely somewhere inside.