Dead nvme Intel 600p -> Data Recovery -> now part is missing

[fzabkar]

You could try transferring the 25.000 MHz crystal, but I suspect that the controller may be dead.

[P.F.]

One other thing you could test is the MP5087GG load switch IC (marking "BDG" ?). Measure the voltages on the adjacent capacitors.

Just compared values there, it looks like there are differences as well. It will take me a bit to confirm and prepare image.
But it apears on the working ssd, at the all the capacitors ~3.5V, whereas at the patient drive its only at one capacitor ~3.5V, others 0V.

[fzabkar]

It could be that the load switch is controlled by the flash controller via the Enable pin. Check whether EN is on.

[P.F.]

It could be that the load switch is controlled by the flash controller via the Enable pin. Check whether EN is on.

EN is 3.06V on both ssds, vout is 3.3V on working ssd, and 0V on patient ssd. I should probably check VIN and VCC

[fzabkar]

It could be that the load switch is controlled by the flash controller via the Enable pin. Check whether EN is on.

EN is 3.06V on both ssds, vout is 3.3V on working ssd, and 0V on patient ssd. I should probably check VIN and VCC

Yes. It's starting to look like the load switch may be bad, or the input voltage may be missing, or the output may be shorted to ground.

[P.F.]

It could be that the load switch is controlled by the flash controller via the Enable pin. Check whether EN is on.

EN is 3.06V on both ssds, vout is 3.3V on working ssd, and 0V on patient ssd. I should probably check VIN and VCC

Yes. It's starting to look like the load switch may be bad, or the input voltage may be missing, or the output may be shorted to ground.

Vin is 3.3V on both, not sure how to measure Vcc safely, without shortening pins...

[fzabkar]

The supply pins are nearly always bypassed by a filter capacitor. You will probably find it easier to measure the capacitor rather than the IC. I always look for less "busy" access points.

Attachments

[P.F.]

The supply pins are nearly always bypassed by a filter capacitor. You will probably find it easier to measure the capacitor rather than the IC. I always look for less "busy" access points.

Hmmm,.....

[fzabkar]

Maybe you could test for continuity between Vcc and the nearby capacitors? Or could it be that Vcc and Vin are tied together by the circuit?

[P.F.]

Maybe you could test for continuity between Vcc and the nearby capacitors? Or could it be that Vcc and Vin are tied together by the circuit?

Thank you, its obvious, that I don't do this kind of work often

So, resistence between Vin and Vcc is 0 Ohm, is it valid to assume, Vin=Vcc in this circut?
Resistance Vout to GND is 33KOhm in working SSD, which has 2 NAND? chips = 128GB
Resistance Vout to GND is 27KOhm in defunct SSD, which has 3 NAND? chips = 512GB

Can we assume, the load switch is broken? If so, who can I ask to do the swap without frying the rest of the pcb?

[fzabkar]

Interesting! It does appear that the load switch is bad.

In some cases you can run a wire link between Vin and Vout, but I don't know what effect it would have in this circuit. Sometimes power sequencing is important, and things need to happen in the right order.

Anyway, you can always replace the IC if you don't want to risk linking Vin and Vout. You could ask at eevblog.com or badcaps.net. Some good component level techs hang out there.

The only other idea I have is that the IC could be shutting off due to an overcurrent. However, your resistance measurements don't seem to support this idea.

You could also check the SS pin (soft start). If the soft start capacitor is shorted, then this could prevent the load switch from working.

[P.F.]

Interesting! It does appear that the load switch is bad.
In some cases you can run a wire link between Vin and Vout, but I don't know what effect it would have in this circuit. Sometimes power sequencing is important, and things need to happen in the right order.

What is the function of the load switch? What would be the risk, except that it might not work? Could the chips draw too much power or something?
I'd try it, if the chances of burning anything are next to nothing.
Would be a too easy solution though, to be true

[fzabkar]

Did you read my edits? Check the SS pin.

[P.F.]

No, I meant the wire between Vin and Vout. But of course, first I'll check SS.

[fzabkar]

I don't feel 100% confident about shorting Vin and Vout. I have had two successes involving shorting load switches in another SSD during the past 2 weeks, but those switches had a different function. Yours is an unknown quantity to me. If it were my SSD, I would try it, but then all my data are backed up, so I would have nothing to lose.

[P.F.]

I don't feel 100% confident about shorting Vin and Vout. I have had two successes involving shorting load switches in another SSD during the past 2 weeks, but those switches had a different function. Yours is an unknown quantity to me. If it were my SSD, I would try it, but then all my data are backed up, so I would have nothing to lose.

Ha,Ha, if the data was backed up, we would not be sittin here
But for my understanding, does the load switch somehow provide higher current than the normal powersupply could deliver?
Or is it mainly for turning power on and off at specific times?

Would shortening Vin and Vout in the donor ssd give us some hint?

The SS pin to ground 68.2KOhm on the donor ssd.
The SS pin to ground 57.4KOhm on the patient ssd.

[fzabkar]

The load switch provides protection from shorted loads. It also has an undervoltage lockout, which means that it doesn't turn on until the supply voltage has risen above the UVLO threshold. It can also be used to put parts of the SSD to sleep in standby mode.

If the ENable pin is strapped to Vcc via a resistor, and is not controlled by the flash controller, then this would suggest that the load switch is not being used for power saving purposes. That is, it would suggest that it is always on. If so, then it would be OK to bypass it. I'm now wondering whether the function of the load switch is to protect the NANDs from an external 3.3V supply whose quality is unknown. All the other supplies are derived from the 3.3V supply via DC-DC converters, so they are largely immune to external quality problems.

So, in short, if you can find where the ENable signal is generated (either locally or from the flash controller), then we can have some confidence in what we are doing.

[fzabkar]

I'm wondering whether the load switch provides 3.3V power to both the NAND array and the controller. If so, then this means that the controller cannot control the Enable pin, which in turn means that the switch would be always on. Therefore it would be OK to bypass the switch. One could confirm this by measuring the voltages across the small bypass capacitors around the controller.

If after bypassing the switch the SSD doesn't come to life, then we would need to examine the PG (Power Good) pin. This signal is derived from the Vin and Vout voltages.

[P.F.]

Progress!!!

If you wanna call it that, I rather have the feeling I'just put the drive back into the state it was
when I sent it off for data recovery.

Thank you fzabkar for leading me thru the dungle this far.

I had a closer look at the connections of the load switch. I've tried a photo with my son's microscope.

I thought i quickly fix this with my soldering iron, well it took me the whole night until I got it right.
It turns out, the load switch wasn't broken, just bad connected.

Now the drive boots into ROM mode just fine.

Only in normal mode, the BIOS hangs for a while, then reports no drive.
For now I just let it sit, powered on in BIOS, to see if the firmware is gonna do some magic.
Reports suggest that powercycling sometimes has helped. We'll see.

Any suggestions on how to continue from here. Send it to recovery again?
Oh good, now I have to go with my kids to the swimming pool, without sleep, that kinda s...s.

[fzabkar]

I wonder if the "data recovery" shops are responsible for the missing resistor AND the load switch problem? If so, then I would have a closer look at your SSD.

BTW, if you have a CCD scanner (not CIS), you can make high resolution images of both drives. Then we can examine them for differences.