xbox 360 blowing mosfet

[waldoalvarez00]

Originally Posted by keeney123

So why is is listed as phase 1, phase 2, phase 3? You will also notice the output looks back to the input through the V_CPUCORE as a reference. I have not look up the IC, but it would seem to me that the different phases are being sent up by what is being feed back form the V_CPUCORE and this is what is causing when the phases are sinking.

There are several phases as it lowers the frequency required for switching elements and spreads heat dissipated across them. Some VRMs have up to 10 phases! High end expensive ones. The higher the frequency the lower the output ripple. So many phases is something desirable to get that frequency high. The controller monitors many things. Current, Voltage, switching, etc. I ignore some of the details. But basically it does that, take high voltage and lower it in the most efficient possible way keeping it constant no matter power consumed by CPU.

The other circuit I am dealing with does the same. I removed some of the regulators and chips in their output path and nothing changed. Someone suggest me in another forum to change the controller chip same as stj. Let's see how I can get one of those new chips. I have no oscilloscope to see what it is spitting so that is my best choice I think.

[keeney123]

So Stj works on these on a continual bases and I do not. It would be worth following his advice to see what happens.

[keeney123]

I am thinking that the different phases are for different operations and different duty cycles for those operations. So I think they keep track of the phase shift of the power to determine which operation that is presently working. This may be done so that certain portions of an operation are not happening at the same time. So with three phase shifts that would be 120 degrees difference from one to the other. So the first phase for the first 120 of the PWM supply nothing but that phase one will turn on, then the second phase starts and at 180 degrees the first phase turns off. This is a bit of a guess. It just seems it would work like this or something similar.

[waldoalvarez00]

Originally Posted by keeney123

I am thinking that the different phases are for different operations and different duty cycles for those operations. So I think they keep track of the phase shift of the power to determine which operation that is presently working. This may be done so that certain portions of an operation are not happening at the same time. So with three phase shifts that would be 120 degrees difference from one to the other. So the first phase for the first 120 of the PWM supply nothing but that phase one will turn on, then the second phase starts and at 180 degrees the first phase turns off. This is a bit of a guess. It just seems it would work like this or something similar.

is simpler, the PWM is spread across all. One pulse in one line, next in next line and so on. Then the cycle starts again after the last one. The duty is given by the amplitude of the peak as always.

[keeney123]

My understanding is the duty cycle is determined on how long the pulse is on. The longer time on the higher the duty cycle. So say if the width of the pulse is 50 milli second and the on time is 25 milli second that would be 50% duty cycle. Of course this would all depend on if they are using the positive going and the negative going pulses in the positive direction or if the are only using the positive pulses. Only using the positive pulses one can only obtain 50% duty cycle max. It could also be the 120 degree phase shift and still start the first at 0 degrees the second at 120 degrees and the third at 240 degrees. The second pass the first would be at 0 degrees again. You could very the on time for each pulse for that operation. The overlap could be used to increase voltage to the CPU for when 0 and 120 degree operations were being performed at the same time or when 120 degree and the 240 degree operation was performed together or when 240 degree and the 0 degree operations were performed together. Both operations would have a window of 60 degrees where the voltage to the processor would increase by twice as much maximum. But you say it is not performed this way. You say that two pulses are never happening at the same time. I really do not understand the logic behind that as you could just have one pulse train.

[stj]

multi-phase psu's exist for the same reason as multi-barrel guns,
they can run faster than with a single fet / barrel without overheating.

[waldoalvarez00]

Allright check this out:

http://www.powerguru.org/multiphase-buck-converters/

and the waveforms here too:

http://www.tij.co.jp/product/jp/LP87...ed_description

some chips turn on/off phases too.

[waldoalvarez00]

They tend to be more expensive, but capacitor size reduction reduces price and volume on the other side. They are more efficient too.

[keeney123]

Why not just run a linear power supply? It will deliver all the power one may need at a extremely well regulated output and it will not over heat.

[stj]

an athlon / p4 vrm can generate 1.8v at upto about 70A
HOW BIG WOULD THAT BE WITH A LINEAR!!!

[waldoalvarez00]

Originally Posted by keeney123

Why not just run a linear power supply? It will deliver all the power one may need at a extremely well regulated output and it will not over heat.

Well that would likely be even more efficient and cheaper, but not too easy to upgrade. Since different CPUs use different voltages you would end up changing your power source anytime you want to change your CPU. The VRM just takes the 12v your standard PSU delivers and lowers it to the voltage value specified by the CPU. And that's it you don't need to change the PSU.

BTW I was thinking in this case since I don't have the chip around now. Just inject those 5V from an ATX PSU in the electrolitic capacitor pins, removing the FETs to isolate and see what happens. Problem I think is the common ground. Or another option just use another DC/DC converter and totally bypass the one in the XBox. To be certain and buy the chip later with confidence (or simply leave the patch ). No idea if timing would have something to do.

[keeney123]

Originally Posted by stj

an athlon / p4 vrm can generate 1.8v at upto about 70A
HOW BIG WOULD THAT BE WITH A LINEAR!!!

Linears are capable of delivering 70 A for a temporary time period. It is like if one needs 70A at 1.8 volts no matter if it is linear or not it will still require components that will have to handle that power. 1.8 X 70 is 126 watts a little more than a 100 watt light bulb. With a primary that can handle say 150 watts the secondary would have to handle 70Amp current using the same components, only less because it would not be so complicated. Because the demand for that type of current would be only for a very temporary basis the linear could be designed to do this while maintaining the 1.8 volts like a rock. As waldoalvarez00 mentioned it would have to be redesign which would not make it worth it.

[keeney123]

If the supply does not have anything to do this the phases of the pulses then that will not effect the processor. However if the CPU or the Graphic processor relies on a boost voltage through that chip then the power supply you use would have to be compatible with that. You would have to study these ICs and the power supply you are going to use to see if it is compatible.

[waldoalvarez00]

No no, CPU and/or GPU doesn't feeds from that. One side is 5v for USB + Drives (i think southbridge too). I saw what seemed to be like 3 regulators feeding from there too. The other side of the chip is for RAM. I think is 1.8v, gotta check this again. That is a problem if the chip is totally faulty. In that case would have to use two converters. Or at least a regulator if load ain't too high (otherwise I would burn it). I think that side was ok and the testpoint was giving proper voltage.

Power supply got the possible problem of ground dis-balance. A small swing at ground means a short circuit. So likely will not use PSU. I was reading in another forum that actually grounding the ground would be safe and would do away with that possibility.

[momaka]

Sorry for the late reply. Finally have enough time to read through this thread.

Originally Posted by waldoalvarez00

You see the one closer to the CPU next to the small black capacitor blew two times. No smoke no holes no nothing. The fan tries to start for a second then it stops entirely. It gets damaged and shows as to resistors on the tester. Maybe the controller chip? Just wondering. I don't have oscilloscope to test.

Yes, that is one possibility.
Though it could also be a short-circuit on that rail (shorted ceramic cap, or some other connector you are missing out). To test that, remove the shorted MOSFET and measure resistance between one of the legs of that toroid inductor and ground. Use the lowest resistance scale on your multimeter (that's usually 200 Ohms on manual meters). Report back what you get.

If there is a short-circuit, that could be the reason why your MOSFET keeps blowing. But if not, then maybe it is the buck controller for that rail that is bad.

Originally Posted by waldoalvarez00

Are you sure this gotta do with GPU reflow thing? That part of the circuit I think feeds USB and drives with 5V.

No. You are correct here.

Originally Posted by waldoalvarez00

Testpoints on all other parts give me 0v readings for RAM, CPU VRM and GPU VRM. Standby testpoints are the only ones good, 3.3v and 1.8v on the upper side of the board. No more voltage readings anywhere else.

That's because the Xbox PSU is shutting down due to an overload or hard short-circuit on the 12V rail from that MOSFET you mentioned above failing.

Originally Posted by waldoalvarez00

I know it gets +12 v on the upper Mosfet base for a short while if I remove it and start the console. Then goes back to zero and give the three rings.

That can happen if the motherboard is checking for all of the voltages, and when one is missing, it shuts down the PSU.

Originally Posted by waldoalvarez00

I think there are two things here. GPU problem and something else. I am looking to find that other problem and then jump into GPU reflow.

I concur.
That's how I would proceed with the troubleshooting as well.

[momaka]

Originally Posted by waldoalvarez00

So putted another mosfet and took the risk to power it on. Well this time it holded and didn't got damaged but the power brick is sensing something as the console spins the fans for a second and then the power brick turns the led red activating sort of an overcurrent protection mechanism.

That means there is still a short-circuit somewhere. I would bet that 5V rail again. Just because the MOSFET there is no longer blowing doesn't mean the controller is working right. It could be that the controller is trying to turn on both the lower and upper MOSFETs for that rail, causing a temporary short-circuit and thus shutting the PSU down.

Originally Posted by waldoalvarez00

Someone suggest me in another forum to change the controller chip same as stj. Let's see how I can get one of those new chips.

Which controller is that? I might have it on a spare junk 360 motherboard.

Originally Posted by waldoalvarez00

BTW I was thinking in this case since I don't have the chip around now. Just inject those 5V from an ATX PSU in the electrolitic capacitor pins, removing the FETs to isolate and see what happens. Problem I think is the common ground.

Excellent!
I've done that before on TV board power supplies, and I can tell you that this method works. But definitely do remove the FETs and even the toroid inductor before doing that. Better yet, first check the resistance on that rail with respect to ground as I suggested in my above post. If you have a dead short-circuit, even your ATX PSU will not like it.

As far as common ground: some Xbox 360 PSUs are grounded and others aren't. But either way, just connect the ground of your ATX PSU to the ground of the Xbox 360 motherboard and it will not be fine.

Originally Posted by waldoalvarez00

No no, CPU and/or GPU doesn't feeds from that. One side is 5v for USB + Drives (i think southbridge too). I saw what seemed to be like 3 regulators feeding from there too. The other side of the chip is for RAM. I think is 1.8v, gotta check this again.

Here's a diagram I made a long while back of the voltage rails on the Falcon board:
http://www.badcaps.net/forum/attachm...1&d=1461040574

These voltage rails will be quite similar on the Xenon, Zephyr, and Jasper as well, so you can still use the above diagram as a general guide.

[waldoalvarez00]

Ahh ok thanks pal, I am gonna check the resistance and report. I was just measuring with the meter in diode mode. Interesting, didn't thought about both FETs on at same time. The controller is ADP1823.

[waldoalvarez00]

Resistance is infinite in both pins of the output coil

[waldoalvarez00]

thanks for the testpoints, I had one like it but is a little bit lower resolution, yours looks better.

[momaka]

Originally Posted by waldoalvarez00

The controller is ADP1823.

I have it

Did you try inserting 5V at the output of that rail, though? If you do and the PSU trips, remove the MOSFETs on that rail (I believe that would be Q6F1 and Q6F2) and try again. Leave the coil in place.