Data recovery attempt from laptop's Sandisk SDINADF4-64 eMMC chip

[ac1dg]

Hi everyone,

my girlfiend's laptop (Odys Winbook 13) just went black when I plugged in an external HDD and now won’t start. When plugging in the charger, the charging light goes on. It will get hot and drain the battery to zero when I unplug the charger. I opened it up, unplugged the battery and plugged it back in. Now it stays cool, but the charging light won't light up, when I plug the charger in. As soon, as I press the power button, it will get hot again and the charging light will light up again. I read this probably is a dead CPU, so the PCB is trash (CPU is soldered).

Apparently, my girlfriend didn't use Dropbox or any other backup strategy for some of her stuff (shame on me, didn't ask as the IT guy). So I am currently trying data recovery by reading the internal SanDisk eMMC (SDINADF4-64) flash directly. This is what I am looking at right now:


I tested all pins twice and they are connected properly. However, the drive does not show up, when I plug the "SD card" into my card reader. I am using Linux and no messages are logged in dmesg. It does try to communicate as I can see the CLK line active on the oscilloscope.
One thing I noticed: There are some spots on the chip which look like there are pads missing:


However, when looking on the datasheet, it seems like they would cause shorts, so I am confused if there actually should be pads or if it is fine like it is...

Has anyone of you ever done something like this and can tell me, if I am doing something completely wrong?

[evac]

I see VccQ, but aren't you missing VssQ, or are my eyes tricking me?

[ac1dg]

VssQ is internally connected to VSS. If you look closely, you can see that all VssQ pins are connected to the common ground plane.

[goaty]

Is it really that easy to adapt ?
I just found that:
https://ripitapart.com/2016/10/28/emmc-adventures-episode-1-building-my-own-64gb-memory-card-with-a-6-emmc-chip/

I´m not sure the interfaces go well with each other.

Also the words "200MHz" and "impedance controlled 50Ohm" don´t seem to make it any easier.

Better glue the chip directly into the adapter and then use thin enamelled wires.

[senso]

Not all adapters work equally, also, wire a couple more Vcc and Gnd pads, and add some smd caps between the extra power supply wires, then you might have a chance.

[evac]

VssQ is internally connected to VSS. If you look closely, you can see that all VssQ pins are connected to the common ground plane.

Indeed, you're right, but senso might have a point in connecting some additional Vcc pads, e.g. M4 and N4 should both be VccQ, but they don't seem to be directly connected, at least on the visible RDL layer.

The "missing pads" you mention are also strange. From the picture it's hard to tell, but are those dots actually tinned, or are they not conducting?
Otherwise it would appear that e.g. N4 and P3 VccQ are being shorted to N2 VssQ.

[testpoint1]

I checked the picture between you and orginal, seems the pad you already destroyed when you desoldering it.

[abyrvalg]

Are you sure that your card reader supports MMCs? Some readers (i.e. Kingston FCR-HS3) have MMC mode disabled in firmware and support SD only. Try some other reader to be sure.

There are no pins in your marked locations, those are some test jumpers cut by milling at the factory, ignore them.

[shakalnokturn]

I checked the picture between you and orginal, seems the pad you already destroyed when you desoldering it.

Surely only the ones on the grid array must be for final use, the ones that seem missing and out of the array must be taken out during production?

[ac1dg]

Are you sure that your card reader supports MMCs? Some readers (i.e. Kingston FCR-HS3) have MMC mode disabled in firmware and support SD only. Try some other reader to be sure.

There are no pins in your marked locations, those are some test jumpers cut by milling at the factory, ignore them.

I tried several other cardreaders and had no success. Not even an error in dmesg.

I added some more wires for ground and VCC with caps. I also shortened the wires, but still no success...
Seems like the chip is dead. Since the data is not that critical, I won't spend any more time in this.

Nevertheless, thanks a lot for you help!

[ginbot86]

Looks like you might have missed extra decoupling closer to the chip (I'm the author of the blog post further up the thread), but could also be termination issues. Or the chip might actually be dead.

If you're interested, I have an eMMC test socket so it might be feasible to send the chip in the hopes the chip is readable otherwise.