Trying to determine NPN or PNP in a circuit

I have an inverter which I need to replace a SMD transistor. Since the burn hole slightly obliterated the letter I can not tell if it is a 3D PNP or a 3B NPN

tracing the circuit
collector has a electrolytic cap 100 uf with negative side direct to collector
emitter has electrolytic cap 220 uf with negative lead direct to emitter

so would it be NPN or PNP?

my drawing

pic of the problem

 

Ok, I diagrammed out this part of the circuit.
Polarity as shown on board and components
Item in question is an SMD transistor either 3D or 3B, a hole has partially burned out the ID!
3B is NPN
3D is PNP
So is the trans a NPN or PNP?


This trans appears to drive the piezo buzzer.
The 2SC3205 is an NPN transistor

 

You expect people on a boat building site to know transister theory? LOL. I went to electronics school in 1965, and they cover transisters but I haven't used it since. I suggest you call the manufacturer of the inverter and ask them or find out if they have a schematic of the circuit.

 

It looks like it was a NPN , the lm324 is a comparator and im guessing when the inverter got to a preset level , (maybe low volts or some other alarm) the buzzer was turned on. The emiiter and -ve side of the 220uf i suspect is also ground. Since it is only a buzzer probably any general purpose transistor would do. eg BC847. This has a max Ic of 100ma. I would check that buzzer before i turned it back on. It was a lot more than 100ms that burn the track and let the smoke out. I would suspect the buzzer is shorted and you had the full supply volts applied to that track.
Thanks Don

 

Ive burned a few invertors and never ever been able to get them repaired.

 

yes, I lean towards NPN.

I have a NTE199 NPN trans I could try in the circuit, think that would just work?

Buzzer measures 4k ohms.

I was hooking up a remote switch and accidently sent 120vac into the 12 vdc switch circuit. Heard a big 'boom', then it let out the magic smoke. 2 exploded capacitors made the boom 220uf 25vdc electrolytics got 120vac.

Only one of the power mosfets also burned, they are in a long line of about 15 on each side.

Inverter is a 3000 watt MSW. It is made by combining two 1500 watt boards together. One board is slaved to the control board.





showing an exploded cap.
This slave board shows a buzzer symbol on the board with polarity marking.
The empty plug is what took in the 120vac grid power.

 

Reviving the dead never works.
The other mosfets all have their gates blown, they are reduced to ordinary fets now. Their was a cascade of events killing most (or all) semiconductors on the board.
So make funeral preparations!

 

Thanks for that, yes it is a ground. I just checked from the negative terminal for power hookup to that point on the circuit is .2 ohms.

So must be NPN then.

 

After seeing the new pics and hearing about the 120vac i am sorry to say that CDK is correct. It's beyond repair. All of the semiconductors on the 12vdc supply will be dead. They only have a threshold of 30vdc or so. They may not look damaged but they surely will be. That pic with the burnt fet shows that the fiberglass board has burnt. You cannot replace parts as the pcb is now impregnated with carbon from the burn and is a short. The only way to get rid of that is grind it out and that leaves you with no pcb.Its not worth the time or trouble. I recently bought a 3000w sine wave inverter off ebay for $250 and it works fine. My suggestion is get a new inverter and give the old one a burial at sea.

Thanks Don
Thanks Don

 

Ok, after those posts, I checked some of the mosfets.
So far only one other is bad. I pulled out 2 of the mosfets next to the burnt one.
Following this test,
http://www.4qdtec.com/mostest.html
one still functions perfectly that was not sharing the heat sink clamp. The one that did share the heat sink clamp is bad (to right of burnt mosfet, it's surface had an overheat mark). So I think that perhaps just two mosfets are bad the ones sharing the heat sink clamp. They are about $2 - 3 each. Allied has them for $2.29 in Va Beach which is an hour drive away for me.

Any how I ordered some ebay parts which will take a couple weeks to arrive and will post back what happens. So far have spent $7 on it.
So far it needs 2 -220uf caps, 2 - F1010e mosfets, 1 generic type NPN transistor to turn on the buzzer.
The carbon burn on the board I could scrape a little but it is still very hard not really carboned very deep. For some reason that did not concern me at all. Perhaps put some shrink wrap on the legs and overdrill the mounting holes.

The tiny burned looking resistor next to mosfets is a 4.7 ohm 5% tolerance. It still reads 4.4 ohms.

 

I found 2 fuses blown. On this inverter the drains from the mosfets are all fused with 30 amp fuses, 8 on each board, each mosfet must run in unit pairs since each board has 16 mosfet, then sharing the 30 amp fuses.

Perhaps that helped all the other Mosfet from failing?
showing the fuses


showing an overall view

 

You may have been extremely lucky and some parts may still be ok. From the pics there doesnt appear to be too many ic's They also appear to be through hole components and are in sockets so thats a bonus. Measure all transistors, fets and diodes and i would replace those ic's. When you come to turning it on i would remove the fuses and substitute a low value resistor (rated to several watts) for each fuse. Dont have any load connected and check the waveform for frequency and voltage. With no load the pulse will be quite thin but this way will allow you to test without too much risk.(i dont know if its a true sine wave or square wave/modified sine wave inverter. I am not convinced there are not many more fried parts but if you are keen on investigating further then good luck to you.

Thanks Don

 

So far finding out what is bad.
I pulled and checked all 1010 mosfets.
8 are bad.

There are 3 optical couplers EL817's, which are likely bad. These join the DC side to AC side on the control board using light instead of electricity, in effect isolating the AC from the DC to help fools like me.

6 2sc3205 NPN transistors bad
4 A1273 PNP transistors bad, only cross I can get is NTE12.
These transistors control the GATES of the 1010 Mosfets.

one piezo buzzer which I yanked off a motherboard.

Would like to know about these 1010 Mosfets.
question on diode test reading following http://www.4qdtec.com/mostest.html

How about those that display a lower voltage reading when they are turned on with my meter?
some display .3v and others display up to 1.1v

All the 'good' ones function as in they turn on and they turn off. Following the test procedure from here called 'simple test'.
But I was wondering if those that display lower readings are worse compared to those showing higher readings?

 

I would be inclined to replace all Mosfets. I am not sure whether any of your readings indicate a good fet. The fet has a diode connected between drain and source and by connecting 120vac to the 12vdc supply i would not be surprised if all are damaged to some extent. If you have to drive for an hour to get parts it would be more ecconomical to swap the lot. The basic operation of an inverter is to switch the supply alternatively through the winding of a center tapped transformer. If the fets switching on side are good and the ones switching the other side are bad you may damage the good ones. Also the inverter will have multiple fets in parrallel switching each side. the reason for putting the fets in parrallel is to lower the on resistance that occurs in the fet junction. If you have a dud fet in parrallel with a good one the load sharing will not be equal and again the good fet may be overloaded and damaged. If you spend the extra few dollars you will at least know 1 section is good and you can go on from there. You also want to make sure that the drivers are alternating and not all on at once. This will damage your transformers. Disconnect the fet stage and connect a 10k or so resistor from the transistor output to the supply and check the waveforms.

Thanks Don

 

Also getting a ST SG3525A Pulse Width Modulator control circuit
and a LM324 (four independent high-gain frequency-compensated operational amplifiers).
Those are on the 12v side.
I pulled some components from the AC side and they work.

My final total parts cost maybe $45. I can see how it may not be worth taking in something like this for a repair. Of course you balance that cost against the cost of a new unit will still be somewhat cheaper.

The threshold vgs of the 1010 mosfet is 2 to 4v, my meter puts out 2.9v in diode check mode, so the mosfets are likely not being fully turned on. Was thinking of running 3 AAA batteries into gate for 4.5v to see if they turn on more fully.

I also found a good sub for A1273 PNP. The new ones will be 2SB892. They look almost identical in performance.