Corsair HX620W - dead APFC MOSFET

[amartolos]

Recently I acquired this non-working Corsair unit. It is described very well here:

Hardware Secrets review

and some additional info was posted here. Mine has a Rubycon as mains cap.

The fuse (10A, fast) is blown, and also the inrush current limiter looks damaged. A schematic of the primary side was posted also at hardware secrets. The PFC switching is done by two parallel 24N60C3 FETs, and one of them seems to be shorted (measured in circuit, but shows short even between gate and other terminals, unlike his "brother").

Now the question: if I find the same FET as replacement, should I buy them in pairs?

I could also replace both of them with a different type. Any idea what the naming of this Infineon FETs means, i.e., can I put two 20N60S5 (available at the local stores)? They seem to have similar specifications (aside from the lower max current), but they do differ slightly.

From what I've read in the forums, a bad PFC FET might have as consequence a bad control IC. In this case, should I check it before replacing the MOSFET? Other suggestions as what to check are welcome.

What's nasty for me about this PSU is that the PCB is double sided, meaning that the components' legs have solder (lead-free) on both sides. This makes them very difficult to remove, because I am not able to melt all the solder. I thought, no problem, just use a bigger gun, but with 60 W I managed to burn the PCB (it gave an awful smell) and to melt the glue that holds the copper layer so I would be grateful for tips here. I already dread the thought of removing the primary heatsink, and I am wondering whether I could remove just the FETs in question.

EDIT: I also noticed a SMD component on the backside that was a bit unusual. It was covered with some black residue, and this residue was also visible on the plastic sheet that isolated the PCB from the bottom panel. It is marked as D309, and with the DMM tests OK in diode mode. It also has one of the terminals marked with a line, for the cathode. I just could not find any data about it. There seems to be written on it LT7216 US1D (barely visible), and the closest connection I could find was with a transient voltage suppressor, see this page. It seems to be connected to one of the pins of the main transformer, but it is hard to see where it goes due to the double-layer PCB. Any thoughts?

Some background info: according to the owner, the unit failed during an installation of Windows, upon a restart. It just made a pop and gone it was. This was a machine with about 7 HDDs connected. It looks pretty nice on the inside, no dust bunnies, just some fine dust on the components; might have been just cleaned.

This will be my new primary unit, if I can make it work

[Francesc V.]

Re: Corsair HX620W - dead APFC MOSFET

Hi:

I am recovering this thread as it's happening the same to me.

Why might be the cause that one of the APFC mosfet blows "again and again"? Although I am using a light bulb current limiter, I am getting rid of 24N60C3

First thing I've seen is that Q302 and Q308 were shorted.

Q302 was marked as W2F (PMBT2907A??)-> I've replaced with M7A
Q308 was marked as W1P (PMBT2222A??) -> I've replaced with M2A


Changing the transistors hasn't made any good difference. I switch on the power supply and the light bulb lights for an instant (good). The 5Vstby is there (good). As soon as I connect the ATX tester..........the bulb shines like the sun (no good)

Any idea what to look at? Any suggestion about how to trace this error?


I've placed two pictures with some measurements. Zoom is needed

[davidebaldini]

Re: Corsair HX620W - dead APFC MOSFET

@Francesc V.

I think your description of symptoms so far matches the conditions of my failed ATX power supply with shorted APFC mosfets that I repaired here. Maybe you can repeat the same tests I did.

[Francesc V.]

Re: Corsair HX620W - dead APFC MOSFET

Grazie Davide.

I will have a look. So far the main capacitor looks ok (at least from capacitance). I will measure the ESR on it (pending).

In the meantime I've realised that also two resistors were bad (R302 was broken -10K, R304 was blown - 4R7). Last one replaced with 5R1

I will keep on reviewing it until new mosfets arrive......

Regards,
Francesc.

P.S: Pendind to take a more detailed high resolution picture......and also cleaning the PCB with some isopropanol.

[momaka]

Re: Corsair HX620W - dead APFC MOSFET

Another thing is, APFC circuits do NOT like series incandescent bulbs - not low-wattage anyways. Try using something like a 200-300 Watt Halogen bulb instead. Or heating element from appliance, rated for 200-500 Watts at your line voltage (120V or 230V, depending on where you live.)

Also, if you had open resistors in the Gate drive circuit to the APFC MOSFET, chances are the IC driving it is dead too. And with that dead, it may be getting too high of a voltage on its Gate and killing the new MOSFET too.

[Francesc V.]

Re: Corsair HX620W - dead APFC MOSFET

Some follow up on this power supply.

- I've finally put a 4R7 resistor in R304 to have both mosfets with the same gate resistor.
- I have check that both gates receive the same voltage (10.6V) (Gate-Source voltage) - Source is conected to "big cap" negative terminal.
- The SMPS works perfectly without APFC circuit (mosfets are not there).
- I have connected the oscilloscope in both gates but not sure if the picture is the one I was looking for.......

What do you think........is the Power supply ready for installing again 2 24N60C3 and see them blowing again?


Regards,
Francesc.

[Francesc V.]

Re: Corsair HX620W - dead APFC MOSFET

Any idea? I mean.........I thought that the pair of mosfets will work alternally, (while one is on the other is off)..........looking at the signals they look like they are working in parallel. Is it a different design?

I guess I will need to sacrifice a pair to see the answer :-)

[Francesc V.]

Re: Corsair HX620W - dead APFC MOSFET

I double checked the PCB and I realised that Mosfets are connected in PARALLEL. Between gates there's only 40ohm and both drain and sources are directly connected. So Drive is the same. That is what I was seeing in the oscilloscope.

I put 2 new mosfets and this time the SMPS worked fine. (only 6 mosfets lost in the battle )

Regards,
Francesc.

[Mikenhauzen]

Re: Corsair HX620W - dead APFC MOSFET

Hello Francesc, what is the conclusion, chaning those two mosfets in the same time solved the issue? I happend to change mosfets one by one and everytime one of them blew up, I had privously some other diodes shorted, replaced those and currently I am having 1 out of 2 mosfets shorted, wondering if I should spend another money for 2 more mosfets…

[momaka]

Re: Corsair HX620W - dead APFC MOSFET

Hello Francesc, what is the conclusion, chaning those two mosfets in the same time solved the issue? I happend to change mosfets one by one and everytime one of them blew up, I had privously some other diodes shorted, replaced those and currently I am having 1 out of 2 mosfets shorted, wondering if I should spend another money for 2 more mosfets…

Always replace both MOSFETs at the same time if one of them had shorted.
Also check the main capacitor. If it's bad, you'll keep blowing MOSFETs and/or possibly damange something else on the primary side too. In fact, the PWM or APFC controller could already have been damaged after those MOSFETs failing. Either way, DO NOT plug in the PSU directly to the wall even after replacing the MOSFETs and checking the primary cap. Use a 200-300 Halogen lightbulb instead in series with the AC Live. The Halongen bulb will limit the current and power in the even there is still something else wrong, thus would spare you a blown fuse and likely the APFC MOSFETs too. If you don't have a 200-300 Halogen bulb, you can also use a heating element with a power rating of up to 450 Watts (but no less than 100W.)

[Clunche]

Hello,

I have spent time on a Corsair HX620 in which the fuse was blown as well as one of the two SPP24N60C3 like reported in one of the posts. I've successfully removed the heatsink which hosts the two MOSFETs, the diode (a STTH8S06D) and the two MOSFETs for the 2-switch forward converter. None of the small bipolars in the driving stage were dead and waveforms from the UCC28515 daughter board were good, despite quite a large duty ratio in lack of feedback (transistors removed, 230 V applied). The standby power supply needs 230 V ac to operate, it did not start up with dc applied to the bulk cap. I thus replaced the defective SPP MOSFETs (I soldered two new parts) and reapplied power. Same effect, one of the two transistors failed in short circuit immediately.

Rather than retrying again with all transistors replaced, I wanted to check the PFC stage alone (and unloaded) by wiring a fast diode and an external MOSFET in a quick 3D wiring. When probing PCB tracks with the ohm-meter to identify the connections, I realized that the track from the two cathodes (catch diode and bypass) were going to the bulk cap. via a NTC. I unsoldered the NTC and measured it at 15 kOhms so no wonder why the drain-source voltage goes to the sky at turn off, making one of the two paralleled MOSFETs who has the lowest BVdss break first, protecting the second. I replaced the NTC by an equivalent of the SCK 054 (4 ohms at 25 °C) as described here. When doing this operation, make sure the new NTC is well protected with a heat-retractable sleeve otherwise it touches on both sides the switching transformer ferrite and the bulk capacitor. Truly packed!

This NTC is supposed to limit the in-rush current and is usually on the ac side of the power supply. To improve efficiency, designers look for places where the ac ripple - which partially contributes to the rms content - is smaller and one option is to put it in series with the bulk capacitor (then shorted later on) or in series with the PFC diode as in here. If in series with the ac, when it fails, the whole power supply safely stops and replacing the NTC is benign. But here, the NTC being open, the destruction of the MOSFET by over-voltage is immediate which is not cool. Now, I read in one comment that you need to replace both MOSFETs when one of them fails. Ideally, yes, they should come from the same lot for the best sharing option but MOSFETs have positive Tc which helps for current sharing anyway. Because I only had two MOSFETs in spare that I soldered in the first attempt, I had to reuse the one MOSFET that did not originally blow, no choice.

I then powered the board and started the standby power: the fan did operate and all dc supplies were present. The only thing I noticed is the rectified bus which is at 375 V dc or so which unusually low, as the bus is usually 380-400 V. The UCC may embed a kind of no-load mode which reduces the dc rail, I will double check this. I now need to test fully loaded but so far so good!

- Cheers

[temyth]

Recently I acquired this non-working Corsair unit. It is described very well here:

Hardware Secrets review

and some additional info was posted here. Mine has a Rubycon as mains cap.

The fuse (10A, fast) is blown, and also the inrush current limiter looks damaged. A schematic of the primary side was posted also at hardware secrets. The PFC switching is done by two parallel 24N60C3 FETs, and one of them seems to be shorted (measured in circuit, but shows short even between gate and other terminals, unlike his "brother").

Now the question: if I find the same FET as replacement, should I buy them in pairs?

I could also replace both of them with a different type. Any idea what the naming of this Infineon FETs means, i.e., can I put two 20N60S5 (available at the local stores)? They seem to have similar specifications (aside from the lower max current), but they do differ slightly.

From what I've read in the forums, a bad PFC FET might have as consequence a bad control IC. In this case, should I check it before replacing the MOSFET? Other suggestions as what to check are welcome.

What's nasty for me about this PSU is that the PCB is double sided, meaning that the components' legs have solder (lead-free) on both sides. This makes them very difficult to remove, because I am not able to melt all the solder. I thought, no problem, just use a bigger gun, but with 60 W I managed to burn the PCB (it gave an awful smell) and to melt the glue that holds the copper layer so I would be grateful for tips here. I already dread the thought of removing the primary heatsink, and I am wondering whether I could remove just the FETs in question.

EDIT: I also noticed a SMD component on the backside that was a bit unusual. It was covered with some black residue, and this residue was also visible on the plastic sheet that isolated the PCB from the bottom panel. It is marked as D309, and with the DMM tests OK in diode mode. It also has one of the terminals marked with a line, for the cathode. I just could not find any data about it. There seems to be written on it LT7216 US1D (barely visible), and the closest connection I could find was with a transient voltage suppressor, see this page. It seems to be connected to one of the pins of the main transformer, but it is hard to see where it goes due to the double-layer PCB. Any thoughts?

Some background info: according to the owner, the unit failed during an installation of Windows, upon a restart. It just made a pop and gone it was. This was a machine with about 7 HDDs connected. It looks pretty nice on the inside, no dust bunnies, just some fine dust on the components; might have been just cleaned.

This will be my new primary unit, if I can make it work

i've had the same problem with a 750w corsair power supply (i dont remember the exact model number).
the fuse was blown and one of the PFC mosfets was totally shorted, all of this happened after a surge in the electrical network
i just bought a new pair of mosfets and a new fuse and replaced them (i replaced with the same exact mosfet, wich were SIHG33N60EF-GE3, quite hard to find here in Brazil)
after the replacement the power supply worked just fine

[ndarjo21]

Re: Corsair HX620W - dead APFC MOSFET

Some follow up on this power supply.

- I've finally put a 4R7 resistor in R304 to have both mosfets with the same gate resistor.
- I have check that both gates receive the same voltage (10.6V) (Gate-Source voltage) - Source is conected to "big cap" negative terminal.
- The SMPS works perfectly without APFC circuit (mosfets are not there).
- I have connected the oscilloscope in both gates but not sure if the picture is the one I was looking for.......

What do you think........is the Power supply ready for installing again 2 24N60C3 and see them blowing again?


Regards,
Francesc.

how do you disable the PFC circuit if the PFC controller integrate with PWM controller. will it work in 80% load?

[momaka]

how do you disable the PFC circuit if the PFC controller integrate with PWM controller. will it work in 80% load?

If you live in a country with 230V or 240V AC, you can do that by either removing the APFC MOSFETs or just by disabling their Gate drive circuit.
However, some APFC controllers may not be happy just with this. For these, you might also have to "trick them" by changing a few resistor on their voltage feedback pin so that the APFC IC "thinks" the rectified 230/240V (~340V DC) is actually the proper level the APFC is trying to boost to.

As for weather the PSU can be loaded up to 80%... or even that much.. that will depend on the PSU itself.
I tried it with an Enermax 500W PSU, and I found the max power the PSU could do with the APFC disabled is a lot lower - barely over half the rated power.

[ndarjo21]

If you live in a country with 230V or 240V AC, you can do that by either removing the APFC MOSFETs or just by disabling their Gate drive circuit.
However, some APFC controllers may not be happy just with this. For these, you might also have to "trick them" by changing a few resistor on their voltage feedback pin so that the APFC IC "thinks" the rectified 230/240V (~340V DC) is actually the proper level the APFC is trying to boost to.

As for weather the PSU can be loaded up to 80%... or even that much.. that will depend on the PSU itself.
I tried it with an Enermax 500W PSU, and I found the max power the PSU could do with the APFC disabled is a lot lower - barely over half the rated power.

thanks. perhaps for this PSU, the feedback for the APFC ic is the VSENSE. maybe i will try do it in the future and test with heavy load

[Loulou31]

I have a Corsair HX450W with fuse blown ( exploded)
WIth first vérification both MOSFET seems to be shorted; would like to have a view of the schematics, but I cant see it on Hardware secrets review. Can you post me a diagram to have a view of the principle of operation of APCF. What is the IC based circuit ?
Thanks and best regards

Jean-Louis

[Clunche]

Hello Jean-Louis, the board I fixed was driven by a UCC28515 which is a combo PFC + 2SW forward from TI. Once the MOSFETs are removed - you need to extract the whole heatsink at once, with 5 TO-220 to unsolder - you can supply the daughter board with a 12-V dc supply and make sure the driver signals are observed at the MOSFETs gates:



The standby power supply uses an ICE from Infineon but needs a sense of the input line to start hence the external 12-V dc that I applied at the auxiliary Vcc cos' I did not want to apply ac:



The issue on my side was the in-rush NTC, stuck between the bulk capacitor and the transformer. You can check its value immediately by measuring the resistance between the PFC diode cathode and the bulk capacitor. It should be a few ohms but was kOhms in my case. I replaced the MOSFETs and the NTC, back in business : )



Good luck and please use an isolation transformer when powering this beast while discharging the bulk cap. through a resistor at power off.

Christophe

[Loulou31]

Hi Christophe,

Thanks you very much for your response and all details and photos.

Best regards

Jean-Louis

[stj]

2 scope workbench - nice!
everybody needs multiple scopes.
i have 6 now
what can i say - they are like pokemon, gotta get them all!

[momaka]

The issue on my side was the in-rush NTC, stuck between the bulk capacitor and the transformer. You can check its value immediately by measuring the resistance between the PFC diode cathode and the bulk capacitor. It should be a few ohms but was kOhms in my case. I replaced the MOSFETs and the NTC, back in business : )

I really dislike those designs with a passion.
There should be absolutely NO other component between the APFC MOSFETs, APFC forward diode/rectifier, and the bulk cap. Putting anything there that can go open-circuit is asking for trouble... and usually that's what happens - APFC MOSFETs blow, because the APFC inductor has nowhere to dump its energy after being charged up from the FETs. As a result, the voltage across the APFC MOSFETs sky-rockets and shorts them out. Really a shit design.

The NTC should be up front, right after the fuse... and by-passed with a relay when the PSU switches On, to prevent the APFC from unnecessarily over-loading the NTC. Also, beyond ~100 Watts of power usage, a relay becomes more economical (energy wise) than an NTC. But I guess a couple extra parts is too much for many PSU manufacturers. :\

2 scope workbench - nice!
everybody needs multiple scopes.
i have 6 now
what can i say - they are like pokemon, gotta get them all!

After all these years fixing electronics, I still have none.
(Well, technically, I do have one now, but it's not working and that's how I got it. Gonna fix it some day though... )
IDK, I find that scope isn't that essential for the stuff I troubleshoot. Most stuff can be done with a multimeter and a bit of creativity.
Strangely, I find wall power meters to be more useful. Need to get one for 230V. The P3 Kill-A-Watt I have is 120V only, so have to go 230=>120 stepdown traffo, then to the Kill-A-Watt, then back up to 230 with a step-up if the device wants 230V AC... and account for the core losses of the 2nd (step-up transformer), not to mention I can't get a PF reading for the DUT due to the step-up traffo. Can only count the Whats... errr, Watts.
Goofy setup for sure, but better than none.