SSD not recognized - Hardware verification

[bod]

Hello liebe EEVbloggers,

I received at my work a SSD (Intenso SATA iii TOP) with NAND flash from Sky hynix (h27q1t based) with a MCU (SM2246EN G AA).
Impossible to detect the device through Windows, and the BIOS / UEFI. I tried to connect via an adaptor SATA-USB and directly SATA but still nothing.
I decided to open the case and look at the components :  The only information I can get is this red LED that lights up permanently and that all components heat up normally.

https://zupimages.net/viewer.php?id=23/25/hdm8.jpg |||||| https://zupimages.net/viewer.php?id=23/25/ybja.jpg

I think there is a problem with the MCU because of the non communication process with the computer, but as I never did repairing before, I don't know how to look at it precisely.

Do you know in which steps should I go to deal with it ?

Thanks a lot !

[fzabkar]

What typically happens is that bad NAND causes the firmware to panic or remain busy. Sometimes you can recover the SSD by allowing it to repair itself while sitting in the BIOS setup screen.

You can check whether the MCU has basic sanity by powering up the SSD in "ROM mode". You can do this by shorting the ROM mode test points with tweezers during power-on and then releasing the short. These two test points are labelled "Rom" on your PCB, and they are located above the 2x4 JTAG (?) pads. In ROM mode the SSD identifies itself with a factory alias, usually the name of the flash controller.

[Haenk]

I had SSDs with firmware bugs (been a while, SanDisk AFAIR), which eventually fell off the bus, forever. Maybe google for the model, if you find a hint at a common failure.
Electrically, I would check the caps for shorts.

[bod]

Hello again,

I shortcuted the ROM test points while pluging in USB, but nothing seems to react...
I checked then the capacitors as Haenk suggested and i found this :

[fzabkar]

I have no idea what that component is.

MT3012 is the step-down converter marked "3012". You can measure the output voltages at the inductors.

MT3012, M3tekic, 5V 2A 2.25MHz Synchronous Step-Down DC/DC Converter, Vin = 2.5V - 6.0V, SOT23_5L, marking 3012:
https://web.archive.org/web/20160402175159/http://m3tekic.com/upload/2016/03/20160310101535.pdf

Here are several Silicon Motion SSD Reference Circuits and Datasheets:
http://www.hddoracle.com/viewtopic.php?f=46&t=3175

... and my SSD/HDD IC database:
http://users.on.net/~fzabkar/Datasheets/DATAURLS.HTM
http://users.on.net/~fzabkar/Datasheets/datasheets.txt

ROM mode screenshots:
http://vlo.name:3000/hw/ssd/rommode/
http://vlo.name:3000/hw/ssd/rommode/sm2246en_rom.png

[bod]

Well, these are good informations. Thank you

I checked the voltages on the MT3012 which are corresponding to the PWM_1 for VCCK/A1V2 1.2V etc. (U5, U15... based on the datasheet http://www.hddoracle.com/download/file.php?id=9657&sid=b04f46e99007e00b582bc5be84268f73) :



[ Specified attachment is not available ]

It looks healthy/fine for me. But still there are weird things happening, for example, when I was measuring the RA2 resistance :

[ Specified attachment is not available ]

I could hear a little noise similar to a square signal, showing the resistance on the Ohmeter from 10k to 30k. really weird...
I will try to figure out later if the quartz of the MCU is still working on my analog oscilloscope.

[bod]

attach 2

[bod]

attach 3

[fzabkar]

You only need to measure the output voltages at each inductor. It would appear that these are 3.3V, 2.5V, 1.8V, 1.5V and 1.2V. That's normal.

As I said at the outset, the problem is most likely due to bad NAND blocks.

I would try to invoke ROM mode with the SSD connected via SATA rather than USB.

[wraper]

This little grey component shows 0.00 on my multimeter. It looks like a capacitor for me, but how can I be sure of that ?

Looks more like ferrite bead to me. Those are normally short. However they are normally darker color, so may be a normal inductor, or its just lighting/photo makes it look like this. The only type of MLCC that you can find in grey color are oddball NP0 (usually they are almost white to light grey) and similar, which is super extremely unlikely to find in this size on SSD. Usual decoupling caps like X7R, X5R, Y5V and similar are light to dark brown. Generally on darker side for high capacitance for the size.

[bod]

I read that ferrite bead are classified as inductor. There are only 3 components like this on the board.
The quartz oscillator is looking healthy with 2 VP-P and approximately 50 MHz like it should be !


I didn't understand as well that the inductor were these brown SMDs. I checked again, and that is indeed normal.
I tried to invoke this ROM mode but nothing is happening.

Sorry, maybe I am missing something clearly fzabkar, but I don't really understand what you mean with this ROM mode and what should I expect from it. Like, that afer the power-on I can see it in the UEFI ?

[wraper]

I read that ferrite bead are classified as inductor.

Ferrite bead is sort of an inductor, however it's far from normal and is intentionally made very lossy. Generally it is used for noise filtering and will not work anywhere where actual inductance is important.

[fzabkar]

When the MCU powers up and checks the ROM mode GPIO pin(s), it switches to "firmware download" mode. This mode ignores the NAND array. If you cannot get the MCU to identify itself in this mode, then that would point to a problem with the MCU or its supplies or clock or reset, etc.

Are you shorting the pads before powering on the SSD? If you short the pads after powering on the SSD, then the MCU won't see them.

[bod]

OK, thank you, I understand more.
As I checked the power supply, the clock etc everything was fine so I tried again to do this ROM mode, aaand it worked this time!

I could see in the UEFI the SATA:SM2246AA with 1023 Mbytes of space wich is the data, instruction cache ?
Quite good to see a response )

You mentioned that sometimes one can recover the SSD by allowing it to repair itself while sitting in the BIOS setup screen. How could I do that ?

[fzabkar]

Your testing confirms that the MCU has basic sanity. Unfortunately this also confirms that the problem is in the NAND.

To see if the SSD will repair itself, just power up your computer and enter the BIOS setup screen. This will prevent anything from accessing the SSD. Let the machine sit for 1 hour and then restart it.

Alternatively, just keep the SSD powered on with the SATA data cable disconnected.

[amyk]

According to the data shared by fzabkar above,  SM2246 has a UART - Tx is on the same pin as the one that's grounded to go into ROM mode, so perhaps take a look and see if there's any output there in normal mode.

[bod]

Thank you,

Yes it is looking bad for the NAND flash. I let the machine sit 1 hour with and without the SATA but at the restart, still nothing is happening.
I could try to look at the UART Tx indeed, I hope that my analog oscilloscope can see it

I'll keep you update

[fzabkar]

This is a procedure for recovering the data from an SM2258XT based SSD that is stuck in the BUSY state due to bad NAND. Unfortunately your SSD requires a different set of tools.

DIY recovery from busy SM2258XT SSD:
https://www.hddoracle.com/viewtopic.php?f=106&t=3238