Voltage drop after few minutes turned on

[jtb1982]

Hello every one ?

I have this case is a nvme and after few minutes turned on has a voltage drop, from 3,3v to 2,2V. Even the led from the usb adapter dims.
My first measurement point was the protection diode, I even remove the diode and measure directly in the pad for more cetainty.
Close to anothers voltage regulators i measure 2.8v but when the voltage drop in the power input i only measure 0v.
I dont have a termal camera but i try to measure with a thermocouple the contrller and other components. The controller temperature rise to 43°C
in this moment the voltage drop and the temperature too. The integrated circuits that i imagine are the voltage regulators are to small to measure
the temperature with the thermocouple.
The nand flash and the ram i dont sense any temperature change.
With i microscope in the possibles voltage regulators i read FT 71.

The client bring me the nvme without the label but the brand is intel.

If any one can giveme a hint, I would be very grateful.

Thank you very much to all who read my post

Attachments

[Lardman]

after few minutes turned on has a voltage drop, from 3,3v to 2,2V.

Always the same time does leaving it idle or reading it impact on the time it stays alive? It has a thermal feel to it - can you actively cool the controller and nand to test ?

[fzabkar]

It's an Intel 660p:
https://www.anandtech.com/show/13078/the-intel-ssd-660p-ssd-review-qlc-nand-arrives

https://images.anandtech.com/doci/13078/IMGP1855.jpg

[fzabkar]

MP2152GQFU, Monolithic Power, adjustable 5.5V, 2A, Sync Step-Down Converter, marking FT, SOT563:
https://au.mouser.com/datasheet/2/277/MP2152-1624343.pdf

I would use a can of spray freeze to restore the 3.3V rail after it drops to 2.2V.

I suspect that 2.8V is the NAND supply.

[fzabkar]

The datasheet for the converter IC specifies a rising undervoltage lockout (UVLO) of ~2.25V and a falling UVLO of ~2.1V. This would suggest that all converters would be switched off when the supply voltage drops to 2.2V.

If this is actually the case, then the cause of the overload must be before the converters. Alternatively, it could be that an overloaded converter is hunting about its UVLO threshold. That is, the converter experiences an overload and this overload then drags down the 3.3V rail. The supply drops below the UVLO threshold, switching off the overloaded converter, and the 3.3V supply then begins to rise again. The UVLO threshold is now exceeded and the converter switches back on again, only to see the same overload and repeat the cycle.

[pepe]

I saw caps to fail several times on SSDs. They can produce a nice amount of heat if the supply is strong enough, they can go beyond 100c easily. But it may not happen if your USB bridge limits current on Vcc.
I would try injecting the appropriate voltage past the regulators from a lab supply with current control and see if i can feel heating up somewhere...
pepe

[jtb1982]

MP2152GQFU, Monolithic Power, adjustable 5.5V, 2A, Sync Step-Down Converter, marking FT, SOT563:
https://au.mouser.com/datasheet/2/277/MP2152-1624343.pdf

I would use a can of spray freeze to restore the 3.3V rail after it drops to 2.2V.

I suspect that 2.8V is the NAND supply.

fzabkar, really, thank very much for all the help all the data.

I will try everything you told me.

[jtb1982]

I saw caps to fail several times on SSDs. They can produce a nice amount of heat if the supply is strong enough, they can go beyond 100c easily. But it may not happen if your USB bridge limits current on Vcc.
I would try injecting the appropriate voltage past the regulators from a lab supply with current control and see if i can feel heating up somewhere...
pepe

Thank very much pepe, in this moment i dont have a lab supply but i will try to measure the current with a multimeter

[pepe]

The point of the Lab supply is to have control over the current. A normal pc psu will keep 3.3V stable feeding lots of amperes into the target device. Which means things can take a bad turn and something may burn. So measuring the current is not enough...

pepe

[fzabkar]

An alternative approach might be to heat up the SSD with a hair dryer in the absence of power. Then measure the resistance between ground and each of the 4 inductors. Hopefully the heat will cause a capacitor to go short circuit, and the affected converter can be found with a multimeter. Once you have found the converter, then use the multimeter to detect all the capacitors that are connected to this supply. Then apply a soldering iron tip to each capacitor while monitoring for shorts.

[Lardman]

I would have thought something like a shorted cap would have manifested itself without a few minutes delay. What happens if the drive is given an immediate power cycle, continue at 2.2 or back to 3.3 for a few minutes and then back to 2.2?