Was Optocoupler now PWM--Plan B Skip to plan B

[status1]

This is just a shot in the dark
I bought this power supply on ebay It is 115v input and 24v output at 5Amps
After I plugged it in it turned on for a second and that it went out than I tapped with my hand and I heard some arcing inside and started smoking
After arguing with the seller because there was no visible damage on the outside he agreed to let me open it and showed the part that smoked He identified the part as an optocoupler so he refunded my money
I am just curious if I can find this part and make the power supply work again

The part is an 8 pin ic with pcb mount pins Unfortunately there was no writing on top of the ic so not sure of the part number The only readable part was on the bottom "7a85" which is probably meaningless

I was wondering if it would be possible to somehow reverse engineer and find a part that would be appropriate for the design

From what I have seen most of them contain just one optocoupler and some of the pins are not used but from the traces I looked on the bottom it seems that all the pins go to different places making me think that maybe there are 2 optocouplers in this one 8 pin IC

I was wondering if that maybe narrows down the field of optocoupler types

[WattsThat]

Nope, doesn’t narrow the field at all. There is something that caused the coupler to fail, it’s a symptom, not a cause. Replace it and it and it will go poof a second time. Not worth the time or effort.

Recycle time.

[bob91343]

Most optocouplers have only four pins.  If this has 8 it may be something else.

Post a pic of the circuit.

[status1]

You are correct that this is a symptom

After removing it from the bottom part of the plastic case and looking at the solder joints I found a few pins of a TO-220 part where the solder was cracked I assume this caused the intermittent connection which I am guessing caused a reaction in the so called optocoupler One of the leads where the solder was cracked goes through a resistor to pin 8 of the optocoupler

[CaptDon]

Optocouplers commonly have 4, 6 or 8 pins. 4 and 6 is common for
single devices and 8 is common for more complex single devices
and very common for dual devices. So someone saying it probably
isn't an opto because it has 8 pins was not providing accurate
information. A vast majority of SMPS units use 6 pin devices which
very often are in a white plastic/epoxy encapsulation making them
very easy to spot on the circuit board.

[status1]

Thanks for the information
In a dual opto configuration is that arranged in a certain standard or not necessarily ?
I have seen some where there are some diodes from pins 1 to 4 and 5 to 7 would be the other side

I attached a few pictures in case it helps to identify the device that should be there
In the Ic blown picture you can see the device that was blown

in opto 004 you see the pcb after I removed the part and you can also see Q3 which is the TO-220 part I mentioned before This one I was able to read it and it says Magnachip MDF12N50 and it is an N channel mosfet

in opto 003 you see the solder side of the pcb At the top you see the solder that is cracked on the TO-220 pins of Q3
The one on the right is the Gate and that goes to the Diode D4148 which was also blown and I removed it
It also goes to that resistor marked 220 that looks ok but I measured it and it is open
The other side of the diode and resistor goes to pin 8 of the opto
It also goes through resistor R303 to the Source of Q3
The source of Q3 goes directly to pin 6 of the opto
This pin 6 was the most burned a melted away from the body of the ic

Not sure if all that helps to narrow down the device
I can provide more pictures if that helps

[CJay]

That's not an optocoupler.

It looks like it's the PWM controller and there's a mad selection of terrible soldering in the pictures you've attached.

[Alti]

That is not an optocoupler but a MOS gate driver, possibly with controller built-in.

[CaptDon]

Yes, in this case you can see it drives the mosfet
and would be some sort of PWM controller. Also,
It is very seldom to see a failed optocoupler with
a ruptured case!!!

[status1]

Thanks for the information
I only said optocoupler because when I showed the burned part to the seller he said
 "ok, the optical coupler was been burned"

The soldering is not too bad but the picture is bad because of the low resolution I had to use to attach it here
It's probably typical as it comes out of the wave solder the leads are somewhat on the long side but it is mounted in a plastic box so it's not going to short out
The biggest problem is the poor design or attachment of the  TO-220 transistor It's just screwed to a 1/8" aluminum plate with just one pin on the opposite end There is nothing else holding the plate in place so of course the plate will move around during handling and shipping and after a lot of vibrating and flexing I am not surprised the the solder cracked

So the consensus is that this is  some kind of mosfet driver ?
Well at least that is more than I knew before
Is there a way to narrow down the list of these mosfet driver types ?

[CJay]

Thanks for the information
I only said optocoupler because when I showed the burned part to the seller he said
 "ok, the optical coupler was been burned"

The soldering is not too bad but the picture is bad because of the low resolution I had to use to attach it here
It's probably typical as it comes out of the wave solder the leads are somewhat on the long side but it is mounted in a plastic box so it's not going to short out
The biggest problem is the poor design or attachment of the  TO-220 transistor It's just screwed to a 1/8" aluminum plate with just one pin on the opposite end There is nothing else holding the plate in place so of course the plate will move around during handling and shipping and after a lot of vibrating and flexing I am not surprised the the solder cracked

So the consensus is that this is  some kind of mosfet driver ?
Well at least that is more than I knew before
Is there a way to narrow down the list of these mosfet driver types ?

D4148 upper right corner appears to have been snapped off the board, the resistor marked 104 on the mid left appears to have little to no solder on the right hand side, it's awful.

Yes, you can make an educated guess as to the type of chip (it's not a MOSFET driver, it's a PWM SMPS controller) but you need to draw out the circuit or at least show better pics of a larger section if not the whole of of the board.

Are you absolutely sure there's not even partial markings on the chip?

[status1]

D4148 was blown as I mentioned it in the explanation of the picture so I removed it It was not snapped off
Resistor 104 I agree it has a slight angle and it may appear that it does not have enough solder but that is because of low resolution I looked at it with a loupe and it is well soldered

I am sure there are no markings at all on the top side just on the bottom as I mentioned earlier but I don't think that is helpful

I started looking around for this PWM SMPS controler I am just not sure what to look for
Does it have to have certain frequency or other characteristics to match the mosfet or I could use any PWM that would match the pin out  ?

[status1]

PLAN B
I did not have any luck finding a replacement pwm so I figured maybe I try to make something out of it with some known parts This is more like an experiment now

I attached a schematic that uses the UC3842N pwm and I am trying to see if it's possible to install it without changing too many parts I understand that I will have to change the pinout of the pwm I am just trying to identify the pins at the moment

The marks in red are the existing parts and pins of the ic as best as I can identify them
The marks in green are the substitute parts or proposed pins
The marks in orange are some of the unknown and not yet identified pins

For example on my board I had a diode in parallel with the 22 ohm resistor that was going from the gate of the mosfet to pin 8 of the old pwm now pin 6 of the new pwm

Is there any purpose for that diode ? Any problems if I leave it out ?

I plugged in the ac power and measured 168v at the cap and at the drain of Q1 so I figured the input section is working so if I get the switching circuit to work I should be able to get something out of it

There is still pins 5 and 7 that I am not sure where they are supposed to be connected

I am open to all comments suggestions or feedback
I am interested if someone can look at the schematic and perhaps comment on the unknown parts in orange

[status1]

Before going to Plan B I think I may have stumbled on the PWM ic that I need for this supply
I was watching a youtube video from a guy with a Russian accent reviewing a 24v 4A supply and as he was turning it around I was able to read the part number on the ic It was CR6841 The data sheet for that was in Chinese so I was not able to read it but I think I found an equivalent SG6841 which was in english and both of them matches my pin out so I ordered an SG6841DZ and should get it in a couple of weeks

I attached the pin out sheet
It has the gate going to pin 8 which is exactly what my supply has so hopefully I could install it and not have to change anything around it

[Alti]

This is most likely referenced to mains (it is on primary side of a converter). With differential isolated probe, experience and knowlege finding susbstitute of the noname chip would cost you 10x the price of off-the-shelf converter of similar parameters.

[Ian.M]

After a blow-up like that, on the primary side, every semiconductor, low value resistor and small electrolytic cap is suspect.  You also need to check the secondary side rectifier diodes.  If you miss a faulty part on the primary side there's >50% odds the new parts will blow immediately on power-on.   Also if the transformer has a shorted turn similar blow-ups are likely, so its very hard to determine if you've missed one, or if the transformer's bad and its B.E.R. so attempting to fix it can rapidly become a money pit.

[status1]

I appreciate the warning and I will try to be careful when turning it on
There are not that many parts to be damaged on this supply and I think because of the intermittent connection to the fet it blew up quickly so I don't think it had a chance to blow up anything on the secondary side but I will keep that in mind in case it doesn't work

I checked the regulator on the output with my meter on the diode scale and it seems to be ok
The transformer I can't check it because my inductance meter is not working but it's not open
The list of parts that are known to be damaged are the pwn , mosfet , the 22 ohm resistor and diode in parallel that was going from the gate to pin 8 of the pwm and the 0.24 ohm resistor from source to gnd pretty much the entire switching circuit section

The schematic I showed is not for my power supply so a lot of those resistor caps and diodes are not on my board I only showed that in case I need to replace the pwm with another part

Because this sg6841 is called highly integrated I am guessing that's why it does not need all those other additional parts

I am guessing the 168V value on the bulk cap seems to be about right for a 120v input so I assume the input section is working normally

I understand the repair cost value is an issue but for me this was more like a learning experience I learned a lot about how switching power supplies work This is just a one shot deal If replacing the blown parts does not fix it I will not go any further I will not be looking for a transformer if that turns out to be shorted
Actually the shipping cost more than the parts but you have to pay the shipping if you want to fix it
 

[Ian.M]

https://www.repairfaq.org/sam/smpsfaq.htm
is worth a *THOROUGH* read through, re-reading anything you don't understand and asking questions here if necessary (as the sci.electronics.* newsgroups have degenerated to pits full of trolls and spam), before starting your SMPS repair career with one with a fault that may be a tough challenge.

The supply decoupling cap for the control IC is particularly suspect as the current sense resistor in series with the MOSFET source blew, which would have caused a large positive going transient, blowing the gate oxide and putting 168V straight into the control chip output, which would have dumped it onto its supply pin till the chip exploded.  Small low voltage electrolytics don't like 168V, and if it had been sustained  the cap would almost certainly have ruptured.  As is, its seal and electrode separator may well be compromised, and as small electrolytics are too cheap to be worth removing to test, and as you probably don't have an ESR meter suitable for in-circuit testing or the experience to use it effectively and in-circuit capacitance value testing can be misleading, it should be replaced, with a low ESR 105 deg C rated part of the same capacitance and same or up to 60% higher voltage rating.

[status1]

Thanks for the link I will read through that while I am waiting for the parts

I am not familiar with the decoupling cap term Are you referring to the 22Uf cap coming from the smaller winding ?
I only have 4 electrolytic caps on this supply 1  big cap 100Uf 400V 1 small cap 22uF that I mentioned and 2 medium sized ones on the output 680 and a 1000uf

I attached a snapshot from the video which has a much closer resemblance although not an exact match to what I have The pin out of the ic is an exact match

For example the 82Uf that the pen is near is the one I am referring to as my 100Uf 400v cap

On the top where it shows 2 820k resistors in series I have 2 680k resistors in series going to pin 3

The 27 ohm res that goes from the gate to pin 8 I have 22 ohm and D4148 diode in parallel with it

On pin 6 I have a direct connection to the drain I don't have that resistor in between however I have a 30k in between pin6 and 8 of the pwm

On pin 4 I have the same 24k resistor to gnd and pins 1 and 2 are also a match except I don't have that cap going from pin 2 to gnd

Pin5 and 7 are a little different on my supply I have pin 5 going through a 100k resistor to the smaller winding on the primary winding where it also connects to one side the C22uF cap The other side of the cap goes to pin 7

Would you able to point out on this schematic which is the decoupling cap that you mentioned ?

[Ian.M]

The 22uF presumably electrolytic capacitor connected from pin 7 (Vdd) of the control IC to pin 1 (Gnd) is the one to be concerned about.  However you've described different connections for the negative side of the capacitor in your circuit, so if we need to investigate further rather than simply replacing it as I would recommend, you'd need to draw up a more accurate schematic of that section of the primary side circuit.  You can skip drawing everything before (and including) the bridge rectifier, and everything on the secondary side for now.

[status1]

Thanks for identifying the cap

As I mentioned that part of the schematic is a little different from what I have
One side of the cap goes to pin 7 but the other side does not go directly to gnd it goes to a 100 k resistor and than to pin 5 and perhaps internally is connected to gnd The side that connects the cap and resistor also goes to the smaller winding of the transformer

I attached a screenshot of the SG6841 block diagram to show what is inside this pwm in case that helps to clarify things

Would the power supply work without that cap ?

[Ian.M]

No.  Its responsible for smoothing the controller's Vcc, and if its gone high ESR, low value or has blown up or rotted off due to leakage, the controller would loose power and reset while attempting to output the MOSFET drive pulse, resulting in weak decaying gate drive rapidly blowing the MOSFET.

[status1]

Thanks I will try to get one

Is there a part or brand that you recommend ?

The part that is in there is a Hyncdz brand 22Uf 50v 105Degree cap
You said something about the ESR How important is that ? Could I get any cap with the same value or do I have to look more specifically about the low ESR ?

[Ian.M]

It only has to handle the ripple on the controller operating current so a super-low ESR isn't needed.   I'd prefer a 105 deg C cap because I don't want to have to re-do a repair in a couple of years, but an 85 deg C cap will do for testing.  I'd use whatever I had handy that's nominally low ESR, either 50V or 63V, of the same value, and if I didn't have any, hit the parametric search and add a pack of 10 to my next order from any of the major distributors.

If you cant source a cap reasonably quickly, *IF* you've got a full LCR meter or *both* a capacitance value meter (or capacitance range on a DMM) and a calibrated ESR meter, I might be tempted to leave it *IF* it shows no signs of distress or electrolyte leakage, its leakage current is insignificant, its value is within the range -10% to +20% (nom. +/-20% but I'm always suspicious that caps at the bottom end of the tolerance range may be on the way out) and its ESR is 'in the ballpark' for budget low ESR caps of its size, value and voltage rating (hit a distributors parametric search to find typical ESR values or trawl through https://www.eevblog.com/forum/beginners/esr-values-for-electrolytic-caps/ ).

[status1]

Thanks for clarifying about the ESR

I actually remembered that I got an old lc meter a while back from work that was getting thrown out so I took it home to see if I can fix it I found the on/off switch was somewhat loose so it was intermittently turning on and off but I fixed that and I put it in the basement since I did not have any caps to measure at the time
So I dug it out and measured the cap on the board and I got 22.2uf  so I figure that's close enough just to see if it works

Later on I found the exact one on ebay https://www.ebay.com/itm/2-5-10-20-50-100-Piece-22uF-50V-105C-Radial-Electrolytic-Capacitor-5x11mm/123976893765?hash=item1cdd994545:g:sNAAAOSwl8VdzwfR

So I ordered 5 pcs just to have some on hand in case I need it