ThinkPad T490 20N2000CSC NM-B901 no signs of life - short on 5V SMPS, ThinkEngine-3 hot - *speculations*
Attached schematic 🙂.
Laptop gives no signs of life. No fan and no CPU heat.
Think Engine BD4179MWV is hot on plugging in *AND* pressing the power button.
Discovered short to ground on VCC5M power delivery IC NB690GRP (5 volt line, not the high current VCC5M_PD) which is one of the ways that chip is powered. I'm puzzled.
Nothing else is warm on the board. CPU resistances are OK, 9V VSYS line is present from the BMIC. All misc and cables are disconnected.
What do you guys think happened? I've got two theories:
If the think engine shorted to ground first, how could have it caused the 5v low side FET to fault by shorting to ground? Those things rarely if ever happen on current controlled PMICs. One speculation I have, is the sudden voltage increase on the node pin due to current increase EMI. If the normal load is ~800mohms, and the short is ~200mohms, then the combination is ~160mohms which at 5v is ~30A of current. This is a change of 24 amps in what, like 2 micro seconds on that line. U = -L di / dt that's high. If this happened during the low-high dead time, the voltage on the node would have skyrocketed and possibly shorted the FET. Note how this doesn't make sense, since all di current would have been supplied through output capacitors at first, increasing the dt. Also, the timing would have to have been PERFECT, with the short happening at exactly the half H-bridge dead time. At 60kHz that's ~1.6us.
Mind you, the 3v3 LDO of that same 5V PMIC was also shorted to ground, even though there is only one resistor to one mosfet on it (first time seeing that happen). That means the chip either uses the common leads for the voltages internally, or multiple 'compartments' got damaged at once. I'd go with the second one.
The second possibility is the low side FET dies first, and the ThinkEngine IC keeps getting digital signals from other ICs. Every IO pin on a chip has 2 diodes in reverse polarity to GND and VCC. If VCC = GND then we have a diode between any pin straight to GND (this is a probable cause if the IC stayed half-functional). The combined current from communication and other logical nonsense slowly cooked the chip over time until something went off.
The funny thing is how that same IC turns the input power FETs so it had to be functional at first, although very hot to the touch. Well, after leaving it on for a minute or two I heard a high pitch inductor resonance sound which lasted for good 30 seconds then came down. After that no power has been distributed on the board, and no signal has been sent to power the FETs on (main PMIC did, but ThinkEngine didn't) so ThinkEngine is dead too.
Ordered both the BD4179MWV and NB690GRP from AliExpress. Now we're waiting. Do You think replacing them should work? EC seems fine.
NOTE how if the above is correct, I can't test either components separately. I MUST soldier them both on and test at once. Otherwise it either won't work, or it will cook the chip again.
What are Your thoughts people? I might edit this to add screenshots from the schematic tomorrow.
Any experiences? This is my most complex issue so far.
Thank you,
Feliks
Laptop gives no signs of life. No fan and no CPU heat.
Think Engine BD4179MWV is hot on plugging in *AND* pressing the power button.
Discovered short to ground on VCC5M power delivery IC NB690GRP (5 volt line, not the high current VCC5M_PD) which is one of the ways that chip is powered. I'm puzzled.
Nothing else is warm on the board. CPU resistances are OK, 9V VSYS line is present from the BMIC. All misc and cables are disconnected.
What do you guys think happened? I've got two theories:
If the think engine shorted to ground first, how could have it caused the 5v low side FET to fault by shorting to ground? Those things rarely if ever happen on current controlled PMICs. One speculation I have, is the sudden voltage increase on the node pin due to current increase EMI. If the normal load is ~800mohms, and the short is ~200mohms, then the combination is ~160mohms which at 5v is ~30A of current. This is a change of 24 amps in what, like 2 micro seconds on that line. U = -L di / dt that's high. If this happened during the low-high dead time, the voltage on the node would have skyrocketed and possibly shorted the FET. Note how this doesn't make sense, since all di current would have been supplied through output capacitors at first, increasing the dt. Also, the timing would have to have been PERFECT, with the short happening at exactly the half H-bridge dead time. At 60kHz that's ~1.6us.
Mind you, the 3v3 LDO of that same 5V PMIC was also shorted to ground, even though there is only one resistor to one mosfet on it (first time seeing that happen). That means the chip either uses the common leads for the voltages internally, or multiple 'compartments' got damaged at once. I'd go with the second one.
The second possibility is the low side FET dies first, and the ThinkEngine IC keeps getting digital signals from other ICs. Every IO pin on a chip has 2 diodes in reverse polarity to GND and VCC. If VCC = GND then we have a diode between any pin straight to GND (this is a probable cause if the IC stayed half-functional). The combined current from communication and other logical nonsense slowly cooked the chip over time until something went off.
The funny thing is how that same IC turns the input power FETs so it had to be functional at first, although very hot to the touch. Well, after leaving it on for a minute or two I heard a high pitch inductor resonance sound which lasted for good 30 seconds then came down. After that no power has been distributed on the board, and no signal has been sent to power the FETs on (main PMIC did, but ThinkEngine didn't) so ThinkEngine is dead too.
Ordered both the BD4179MWV and NB690GRP from AliExpress. Now we're waiting. Do You think replacing them should work? EC seems fine.
NOTE how if the above is correct, I can't test either components separately. I MUST soldier them both on and test at once. Otherwise it either won't work, or it will cook the chip again.
What are Your thoughts people? I might edit this to add screenshots from the schematic tomorrow.
Any experiences? This is my most complex issue so far.
Thank you,
Feliks
Comment