Unknown 3-pin part in a PSU trace, datasheet hunt or sub an equivalent?

I sub in equivalents almost every time, and it's served me well. Last year I had a TV power board with a little three-pin part near the standby section I couldn't match to anything either. The markings looked like house codes from the original manufacturer, and after two evenings of searching I gave up on identification and just measured what the circuit was doing instead.

With mains-derived circuits like this, the LED side of an optocoupler is ussually fed thru a resistor chain sized so a few milliamps flow when line voltage is present. So I measured the voltage across the part in circuit, worked out roughly what current the resistors were limiting to, and picked a common optocoupler whose LED forward voltage sat comfortably in that range. Board powered up, standby detect worked, done in an afternoon instead of a weekend.

The one thing I'd be careful about is measuring with the board live, since that section sits at mains potential. I use a cheap isolation transformer before I poke around anything fed directly from the wall, and I keep one hand in my pocket out of old habit. For a schematic you're sharing publicly, I'd just label it as 'functionally equivalent, original part unidentified'. That's honest and it saves the next person the same hunt. If the replacement ever ran hot or drew noticeably more current than the measurements suggested, that's when I'd go back and keep digging.

I'd push back on just subbing something in, at least for this part. I did exactly that on an Xbox One S brick a couple years back. The AC detect optocoupler had a house code I never cracked, so I dropped in a PC817 becasue the pinout matched. Worked fine on the bench for an hour, then the console started randomly waking from standby. Turns out the LED current through that big resistor chain was way lower than what my substitute wanted, so the output was half on, half off. The original was clearly a low-current type and I'd never have guessed that from the pinout alone. Since you've allready redrawn the section, you're most of the way there. Measure the voltage across the LED side while it's running and work out roughly what current the resistors are feeding it, then pick a part that's happy at that level. That's not sloppy, that's just reading the circuit.

Where I'd draw the line is documentation. If other people are going to follow your schematic, label it honestly as 'equivalent, original unidentified' rather than presenting the sub as fact. A wrong part drawn as correct will outlive all of us on some forum somewhere.

So I'd give the marking one focused evening, maybe try a few code lookup sites, and if it stays silent, sub in with measurements in hand and a clear note on the drawing.

Same approach worked for me on a lamp-dimmer style board years back, so I'd side with just subbing it in. Mine was a little opto feeding a standby indicator, markings were some custom code I never cracked. I traced the resistor chain on paper, wrote the measured voltages right on my redrawn schematic with a pencil, and dropped in a generic part that matched the pinout. That annotated sketch ended up more useful than the original would have been anyway. One trick I'd add: check the replacement's current draw at low line first, not just full mains. If it only behaves right at 240V you'll chase ghost faults for days.

One thing I'd check before picking a sub: is it actually a single LED inside, or one of those dual ones wired back to back? Mains detect circuits sometimes use the type that lights on both half cycles, and a regular single-LED opto will only catch every other one. I got burned by that on a boiler control board years back, the detect signal looked fine on my meter but the controller saw it as intermittent. Also, if you snap a clear photo of the markings and post it, odds are decent someone here recognizes the code. Half these house numbers turn out to be house chips some guy on a forum has seen before.

Update: checked the part again with a meter and it looks like it might be one of those dual LED types, back to back inside. That lines up with what someone said about mains detect circuits, and would explain why a plain single LED opto could miss every other half cycle. Going to test both directions tonight before picking a substitute.