Power Supply for Window Fan (BW2300-N)

[LiftedTrace]

So I plugged this fan in the other day and seen the display illuminate for a split second and then go dark. First thing I did was check the fuse in the plug, but the fuse is fine.
I found a website where someone tried to modify this exact fans behavior with an Arduino. They had the fortune of their fan still working so they were able to locate the displays power pins and check voltages on it.
I measured these two pins and have nothing at all on them. I checked for a short and there is no short between these pins. I have uploaded what I think is an accurate schematic of the power supply.
There is a few things I left out which I suspect is just the circuity that controls the relays and powers the motors. I might have left something important out, but with trying to not waste to much time drawing the entire board out, I stopped with what I think was necessary. (I am most likely missing something necessary.)
Anyways, In hopes I can get some help from people much more knowledgeable than I, I have uploaded my schematic.

Does this make any sense whats going on, how is there supposed to be 5V at the output when connected directly on mains, etc. What might be bad in this? I checked the transistor with a meter and it tested good with that. Diodes are good too as far as my meter says.

[Kim Christensen]

how is there supposed to be 5V at the output when connected directly on mains

It looks like a capacitive dropper type supply. They use the reactance of the capacitor pictured in the lower left of your schematic to drop the line voltage down without wasting a bunch of power in a resistor as heat. Hard to comment more without proper part designations on your schematic, but it looks like a bipolar supply giving both + and - voltages with a -5V regulator as well.
These power supply types are non-isolated and not safe to touch the output lines (5V, etc) because they are connected to the mains supply.
 

[eurgenca]

I agreed.
Check this capacitor they often loose capacity and PS don't work.

[LiftedTrace]

I uploaded a new schematic with values.
While removing components to see what they were, I happened to find diode D5 doesn't seem normal. Diode check both directions shows .002v on the meter which I gather is not correct. A resistance measurement to see if its shorted shows 3.2 ohms.
The last number is almost gone so I dont know if its a 5,6, 8, or 9, but I think replacing it with a 5 looks to be a safer option looking at the datasheet. ( https://www.mouser.com/datasheet/2/268/SA5_35-1592030.pdf ) It has a lower zener voltage and a higher surge current.

Do you see anything wrong with me going this route?

[Ian.M]

That's not a good idea.  R4, D6 and the transistor act as a regulator for the 5v output, and we've got no idea how much headroom it requires.  It has to also have additional headroom to allow for the half-wave ripple on  C4 which is load current dependent.  The original D5 Zener was probably closer to the top end of the 1N473x voltage range than the bottom.

[Kim Christensen]

The last number is almost gone so I dont know if its a 5,6, 8, or 9

I agree with Ian... It needs to be more than 5V because of the 5V regulator (R4, Q?, & D6) that comes after it. A 1N4739 makes more sense if C4's voltage rating is something like 16V or so. But if C4 has a 10V rating, then it may be a 1N4738 to give a bit more headroom for C4.

[LiftedTrace]

Im not sure what to do then without knowing what the original one was. I found a C5Vx on a board that looked pretty beefy and installed it before I seen this post, just to see if I got some output. I ended up getting just a tad under 5V on the output and the fan seemed to have been working but I didnt test it for long, I wanted to see if that was the only thing wrong.
C4 is a 25v cap and C5 is a 16v cap.

If I leave in what I put, and its working, what do you suppose might happen?
The board I got the C5V off also had a C9V on it. Perhaps I should switch it out with the 9V one? Otherwise, looks like Ill have to put an order in for the N4739

[Kim Christensen]

If I leave in what I put, and its working, what do you suppose might happen?

If D5 is a 5V zener, when you load it, the 5V output will drop significantly below 5V and have some AC ripple on it as well.
Lets assume Q? (S8550) beta is 45 as the worst case and D5 has exactly -5Vdc across it. We'll also assume you want to draw -5mA off the -5V output. This means that Q? base current will be -111uA which will drop 0.37V across R4 so there will be -4.63V on the base of the S8550. But then there will also be 0.7V across the base-emitter junction of S8550 so the voltage at it's emitter, and thus the output, will be -3.93V. (With only 4.63V across it, D6 does virtually nothing)
Now if you let D5 be a 9V zener, then D6 will come into play and will have apx -5.6V across it (and 1mA though it) which will give enhanced regulation and ripple rejection. Now the output should be around -4.9V or so under load.

[Algoma]

Is there physical room to simply replace the 5V DC source / whole control board with something more... repairable.  The AC fans are likely fine and can be controlled with a few relays.

I once looked into fixing one of these fans and scrapped the whole thing.. The whole electrical design was so cost oriented, that failure was inevitable. The microcontroller's IO was baked to a crisp as the DC regulation failed. It was so clearly designed to last only a few years at best.

[LiftedTrace]

If D5 is a 5V zener, when you load it, the 5V output will drop significantly below 5V and have some AC ripple on it as well.
Lets assume Q? (S8550) beta is 45 as the worst case and D5 has exactly -5Vdc across it. We'll also assume you want to draw -5mA off the -5V output. This means that Q? base current will be -111uA which will drop 0.37V across R4 so there will be -4.63V on the base of the S8550. But then there will also be 0.7V across the base-emitter junction of S8550 so the voltage at it's emitter, and thus the output, will be -3.93V. (With only 4.63V across it, D6 does virtually nothing)
Now if you let D5 be a 9V zener, then D6 will come into play and will have apx -5.6V across it (and 1mA though it) which will give enhanced regulation and ripple rejection. Now the output should be around -4.9V or so under load.

Thanks for explaining that a bit for me. It is hard for me to look at that power supply and grasp whats going on. Given that, i think ill swap out the one i just put in for the 9v one.

Is there physical room to simply replace the 5V DC source / whole control board with something more... repairable.  The AC fans are likely fine and can be controlled with a few relays.

I once looked into fixing one of these fans and scrapped the whole thing.. The whole electrical design was so cost oriented, that failure was inevitable. The microcontroller's IO was baked to a crisp as the DC regulation failed. It was so clearly designed to last only a few years at best.

There is actually a good amount of room in there and someone has already mapped the logic for controlling the relays. Other than what three triacs are doing inside, it looks like the motors are connected directly to mains thru relays. Maybe the triacs juice up multiple windings depending on the speeds?
But given some wiring and a maybe a pi nano for relay logic, i think the 5v source could be replaced easy enough.