CE46, CE47, CE48, CE49: UCC KZG 1000uf 16v all in parallel WV=12v ATX
CE51, CE52: Ltec 820uf 6.3v
CE29, CE30, CE31, CE32, CE33, CE34, CE35, CE36, CE37, CE38, CE39: Fujitsu Poly 560uf 2.5v all in parallel WV=Vcore
Q19, Q20, Q21, Q23, Q27, Q28, Q31, Q32, Q33, Q36, Q37, Q38: Infineon IPD06N03LA TO-252 DPAK, MOSFET pairs on back are in parallel
U30, U31, U33, U34: Analog Devices ADP3418J dual buck MOSFET driver
U32: Analog Devices ADP3180 2/3/4 Phase Synchronous Buck Controller Sequencer
D26, D??, D??, D29: JC possibly a BAW56WS
PWM wiring from U32 ADP3180 through each ADP3418 to the MOSFETs:
PWM1: U32-27 PWM1 -> U30-2 IN, U30-8 DRVH -> Q19-gate, U30-5 DRVL -> Q20-gate+Q21-gate
PWM4: U32-24 PWM4 -> U31-2 IN, U31-8 DRVH -> Q23-gate, U31-5 DRVL -> Q27-gate+Q28-gate
PWM3: U32-25 PWM3 -> U33-2 IN, U33-8 DRVH -> Q31-gate, U33-5 DRVL -> Q32-gate+Q33-gate
PWM2: U32-26 PWM2 -> U34-2 IN, U34-8 DRVH -> Q36-gate, U34-5 DRVL -> Q37-gate+Q38-gate
The AD3180 datasheet has a sample 3 phase Vcore power supply.
The AD3188 datasheet has a sample 4 phase Vcore power supply.
Richtek RT9248 RT9603 duo is similar and the Richtek RT9248 datasheet has a sample 3 phase Vcore power supply.
The Fairchild FAN5009 crosses to the ADP3418K which should be similar to the ADP3418J.
Phantom MOSFET shorts: none
CE46, CE47, CE48, CE49, CE51, CE52 were bulged. I replace those and the board shuts down immediately on power up as if shorted. Several MOSFETs had some shorts but Q31 tests shorted on all terminals so it goes first. I replace Q31 and the phantom shorts on the parallel MOSFETs disappear. The board shuts down immediately as before and the new MOSFET Q31 is shorted on all terminals. SMPS MOSFETs drain and source typically hook to 12v power, ground, or Vcore. Many MOSFETs are in parallel so a problem on those terminals which could only come from a faulty ATX 12v power supply should burn out all MOSFETs at once. The only practical way to quickly destroy an individual MOSFET before it can overheat is through the gate. Q31-gate traces to U33-8 DRVH so I replace U33. Now Q31 does not burn out. The board runs a bit longer and something in the Vcore area emits a half second audible buzz before shutting down. With no more shorted parts making themselves obvious I replace U32 hoping for a lucky fix. The shut down behavior does not change.
As I tested MOSFETs I would get shorts then not shorts. It wasn't until I read a tutorial on how to test MOSFETs properly that I realized why it was happening. A meter can charge the gate enough to turn a MOSFET on and the gate capacitance can hold the charge to keep it on after the probe is removed. A conducting low Rds(on) MOSFETs tests as almost shorted. Until I knew this, frustrated by the inconsistent presence of shorts, I pulled an apparently shorted Q23 only to have it test good on the DCA55.
Thinking that the faulty U33 might have also destroyed Q32 and Q33 I replace them but the pulls test good and the replacements behave no differently. A voltage test on the 4 front side gates of Q19, Q23, Q31, Q36 shows that the gate drive signal is reaching each one. A voltage test on the back side gates via each ADP3418J pin 5 shows that the gate drive signals are reaching there too. No scope is available to see if any gate drives have an out of place waveform.
The buzz comes from the vicinity of the processor and sounds like excessive current from over voltage which the ADP3180 detects and shuts down. The MOSFETs test good so I figure that one of the ADP3418J is over driving a MOSFET. I can easily replace them all but this is a learning experience so I want to figure out how to tell which one has failed. With my ear close to the board to narrow down where the buzz is coming from I accidentally put my lip on Q36. As I press the power button to make yet another buzz I notice that the temperature of Q36 rises surprisingly fast. My fingers are not sensitive enough to detect the small temperature rise but my lip did easily. No room for lips on the rest. According to the infrared temperature gun Q36 jumped 5*F in the half second of buzz. The remaining top side MOSFETs jumped 1*F or less. Looks like Q36 is carrying too much current but it is not shorted so it can only carry too much current if told to by a faulty U34.
I replace U34 and the board works. Success with 6 caps, 1 MOSFET, a destructive dual buck MOSFET driver, a non destructive dual buck MOSFET driver, some other parts that were not necessary to replace, and a little lip.
Though replacing U32 did nothing useful it turns out I selected an ADP3188 from a Dell GX280 donor board to replace the ADP3180. Fortunately the ADP3180 and ADP3188 look pin compatible so the accidentally substituted part should not be damaged.
The board ran MEMTEST for several hours on the ADP3188 so I don't see the need to change it back.
This board was serviced 7 months ago and has not returned so the fix and unwanted subs must work.
CE51, CE52: Ltec 820uf 6.3v
CE29, CE30, CE31, CE32, CE33, CE34, CE35, CE36, CE37, CE38, CE39: Fujitsu Poly 560uf 2.5v all in parallel WV=Vcore
Q19, Q20, Q21, Q23, Q27, Q28, Q31, Q32, Q33, Q36, Q37, Q38: Infineon IPD06N03LA TO-252 DPAK, MOSFET pairs on back are in parallel
U30, U31, U33, U34: Analog Devices ADP3418J dual buck MOSFET driver
U32: Analog Devices ADP3180 2/3/4 Phase Synchronous Buck Controller Sequencer
D26, D??, D??, D29: JC possibly a BAW56WS
PWM wiring from U32 ADP3180 through each ADP3418 to the MOSFETs:
PWM1: U32-27 PWM1 -> U30-2 IN, U30-8 DRVH -> Q19-gate, U30-5 DRVL -> Q20-gate+Q21-gate
PWM4: U32-24 PWM4 -> U31-2 IN, U31-8 DRVH -> Q23-gate, U31-5 DRVL -> Q27-gate+Q28-gate
PWM3: U32-25 PWM3 -> U33-2 IN, U33-8 DRVH -> Q31-gate, U33-5 DRVL -> Q32-gate+Q33-gate
PWM2: U32-26 PWM2 -> U34-2 IN, U34-8 DRVH -> Q36-gate, U34-5 DRVL -> Q37-gate+Q38-gate
The AD3180 datasheet has a sample 3 phase Vcore power supply.
The AD3188 datasheet has a sample 4 phase Vcore power supply.
Richtek RT9248 RT9603 duo is similar and the Richtek RT9248 datasheet has a sample 3 phase Vcore power supply.
The Fairchild FAN5009 crosses to the ADP3418K which should be similar to the ADP3418J.
Phantom MOSFET shorts: none
CE46, CE47, CE48, CE49, CE51, CE52 were bulged. I replace those and the board shuts down immediately on power up as if shorted. Several MOSFETs had some shorts but Q31 tests shorted on all terminals so it goes first. I replace Q31 and the phantom shorts on the parallel MOSFETs disappear. The board shuts down immediately as before and the new MOSFET Q31 is shorted on all terminals. SMPS MOSFETs drain and source typically hook to 12v power, ground, or Vcore. Many MOSFETs are in parallel so a problem on those terminals which could only come from a faulty ATX 12v power supply should burn out all MOSFETs at once. The only practical way to quickly destroy an individual MOSFET before it can overheat is through the gate. Q31-gate traces to U33-8 DRVH so I replace U33. Now Q31 does not burn out. The board runs a bit longer and something in the Vcore area emits a half second audible buzz before shutting down. With no more shorted parts making themselves obvious I replace U32 hoping for a lucky fix. The shut down behavior does not change.
As I tested MOSFETs I would get shorts then not shorts. It wasn't until I read a tutorial on how to test MOSFETs properly that I realized why it was happening. A meter can charge the gate enough to turn a MOSFET on and the gate capacitance can hold the charge to keep it on after the probe is removed. A conducting low Rds(on) MOSFETs tests as almost shorted. Until I knew this, frustrated by the inconsistent presence of shorts, I pulled an apparently shorted Q23 only to have it test good on the DCA55.
Thinking that the faulty U33 might have also destroyed Q32 and Q33 I replace them but the pulls test good and the replacements behave no differently. A voltage test on the 4 front side gates of Q19, Q23, Q31, Q36 shows that the gate drive signal is reaching each one. A voltage test on the back side gates via each ADP3418J pin 5 shows that the gate drive signals are reaching there too. No scope is available to see if any gate drives have an out of place waveform.
The buzz comes from the vicinity of the processor and sounds like excessive current from over voltage which the ADP3180 detects and shuts down. The MOSFETs test good so I figure that one of the ADP3418J is over driving a MOSFET. I can easily replace them all but this is a learning experience so I want to figure out how to tell which one has failed. With my ear close to the board to narrow down where the buzz is coming from I accidentally put my lip on Q36. As I press the power button to make yet another buzz I notice that the temperature of Q36 rises surprisingly fast. My fingers are not sensitive enough to detect the small temperature rise but my lip did easily. No room for lips on the rest. According to the infrared temperature gun Q36 jumped 5*F in the half second of buzz. The remaining top side MOSFETs jumped 1*F or less. Looks like Q36 is carrying too much current but it is not shorted so it can only carry too much current if told to by a faulty U34.
I replace U34 and the board works. Success with 6 caps, 1 MOSFET, a destructive dual buck MOSFET driver, a non destructive dual buck MOSFET driver, some other parts that were not necessary to replace, and a little lip.
Though replacing U32 did nothing useful it turns out I selected an ADP3188 from a Dell GX280 donor board to replace the ADP3180. Fortunately the ADP3180 and ADP3188 look pin compatible so the accidentally substituted part should not be damaged.
The board ran MEMTEST for several hours on the ADP3188 so I don't see the need to change it back.
This board was serviced 7 months ago and has not returned so the fix and unwanted subs must work.
