The immediate problem is that the 3.3V supply is faulty. We don't need to look elsewhere yet. Once we take some measurements, then we can determine what happened and where to go from there.

_________________
A backup a day keeps DR away.

Thanks fzabkar




About Vol, sorry you was on true
L1=0,97V
L2=2,79V

Info about IC
IC at U7
POWR607
01SN321
A146KK10


Info about U22 IC
F
PB98F
FDS
6898A


markings on U3 (I think now you can see it in atached new picture)
9A27
markings on U4 (I think now you can see it in atached new picture)
43t

measure the resistances of R4, R5, R22, and R23 (Measured inside PCB not out of PCB)
R4
71.3 x 200k Ohm
14.260 OHM


R5
71.8 x 200k Ohm
14360 OHM

R22
107 x 200k Ohm
21.400 OHM

R23
400 x2000K Ohm (but increasing measure to infinit)
800.000 OHM (but increasing measure to infinit)






About stuff , i was cleanig but not sure about it. I think metalic ink mark from factory

About POWR607 Pinout i am not sure if you can see new atached PIC, can you repeat me about testing pinout.

witch pins you want to test voltage?




in this case we can add more power to recover data?

Thanks BlackST but i think that fzabkar advices are very interesting.
Maybe in this way we can fix SSD problems, and take out data.

But thanks for your comments

About temperature. i atach pics

Just my opinion, I see it everyday. Here the rule is always no money = no data, but when it's no data = little money - just the diagnose fee. That's why I exclude tinkering and guessing.

HI BlackST

You really are on true, but i like to study each case until found the problem.

Fig 26 on page 17 of the TPS62065 datasheet shows a typical configuration.

The output voltage is given by ...

Vout = Vref x (1 + R1/R2)

... where Vref = 0.60V

In the L1 case ...

Vout = Vref x (1 + R5 / R4) = 0.6 x (1 + 71.8 / 71.3) = 1.20V

For L2 ...

Vout = 0.6 x (1 + R23 / R22) = 0.6 x (1 + 400 / 107) = 2.84V

Neither of these numbers feels right. The resistances are difficult to measure in-circuit because they are high enough to be influenced by the resistances of the respective loads. Therefore I wouldn't place too much faith in these calculations except to say that they are in the ballpark.

Your temperature measurements seem OK to me. I would have thought that, if a bad controller or NAND flash IC were responsible for any droop in the supply, then it should be very hot to the touch.

The pinout you have marked out for the POWR607 is correct. That said, I would refrain from measuring the voltages right at the pins of the IC since a slip of the probes will probably destroy the board. Instead I would probe the less "busy" areas.

The JTAG pins are probably associated with the POWR607. They would be used at manufacturing time to program the IC. I suspect that the "stuff" on the chip may be a marker to indicate that the chip has been programmed.

U22 is a FDS6898A dual MOSFET.

FDS6898A, Fairchild, Dual N-Channel Logic Level PWM Optimized PowerTrench MOSFET:
http://datasheet.octopart.com/FDS6898A- ... -46434.pdf

U22 appears to be operating as a power switch. Each of its drain pins are connected to Vcore and Vio. The POWR607 has two high voltage outputs. I suspect that these are used to switch the gates of the two MOSFETs. If you could measure the voltages on pins #1 and 3, then this should tell us whether the MOSFETs are switched on. Pins #5 - 8 will be the inputs.

It appears that the 3.3V JTAG supply, and the Vcc supply on pins #4 and #21 of the POWR607, may be provided by U3 (9A27). I haven't been able to identify this chip, but I suspect that it may be an LDO regulator. If you could measure the voltages on each of its pins, then that might help to identify it.

The bottom line is that the Vio voltage (2.8V) still doesn't look right, but I'm wondering if that is indeed the correct voltage. One way to determine whether the "OFA" chips are regulating correctly would be to measure the feedback voltages at the junctions of R4 and R5, and R22 and R23. This voltage should be 0.60V.

_________________
A backup a day keeps DR away.

thanks fzabkar,




U22
Betwein pins #1 and 3 = 1.82V

Betwein pins #5 - 8 = 1.82V too



About U3 Regulator measures to Ground i atach pic




voltages at the junctions of R4 and R5 (betwen)
None V

voltages at the junctions of R22 and R23 (betwen)
None V

strange

You've measured the wrong voltages at U22, but in any case they are OK. I was wanting to see the voltages with respect to SATA ground, not with respect to each other. As you can see, U22 consists of two MOSFETs. Each MOSFET switches Vin to Vout. Vin and Vout are the Vcore and Vio inputs and outputs. If each switch is working correctly, then Vin = Vout.

U3 is indeed an LDO 3.3V regulator. It provides the supply for the POWR607.

The R4/R5 and R23/R22 junction readings are very strange. It's as if the FB (FeedBack) pin were sitting at ground. Are you using SATA ground as your ground reference for measurements? If not, have you been using pin #4 of the "OFA" chips as your ground. If so, then this would explain why Vio is reading approximately 0.6V lower than it should be.

_________________
A backup a day keeps DR away.

sorry for inconvenience i have done bad measures i two cases, bada measures in U22 and bad measures in R(nº).
i'll do other measures.


about U22

1 t ground 0,98v
3 t ground 2,81v
5 t ground 2,81v
8 t ground 0,98v

about R(nº) see pic

hhddrec, do you have another (similar to it) SSD OCZ drive?
If you do, then what voltage for NAND chips it is provide? Can you measure it?

And, what exactly written on U5 chip? Is it SPI flash?
If it is, then, do you have a SPI programmer to read it?
If you do, then it is possible to try to analize a dump from it.

But again, chances to get data are low, it is difficult case.

Hi Randy,

Thanks for your news

Info about U5
ATH130
2ECL 8
186931P

It's not a SPI flash IC. Seems it's ATMEL 256 Kbit EEPROM.
Datasheet: http://www.atmel.com/Images/doc8568.pdf

I doubt, that it contains a firmware code. Most likely, it contains some firmware constants, working data, smart data, counters etc. But i may be wrong.
If you'll be able to make a dump, then we can see, what inside. And make suggestions, what it is.

fzabkar has done a good analisys, and i suggest to you first follow his instructions. At least, to make sure, what NAND array is powered with correct voltage.

Even if U5 is just a some kind of configuration storage, because of power failure (if), it may contain now a "garbage".
Make sure also, that U5 is also powered with correct voltage.

Randy, thanks for chiming in. I was getting a bit lonely. :-)

Do you think that the controller's Vio runs off the same supply as the NAND flash, or do you think that the 3.3V LDO provides the Vio? I would think that the Vio must come from the OFA chip, otherwise the POWR607 wouldn't be able to switch it on and off.

ISTM that both buck regulators are regulating correctly. If 2.8V really is too low, then the only culprits I can see would be R22, R23, C18, or C20. Obviously MLCC capacitors don't leak in the same way as liquid electrolytics, but they do exhibit failures related to manufacturing stress. These can eventually develop into short circuits.

I can't offer any more observations or suggestions except for hhddrec to confirm the NAND array voltage on another board, as you have already suggested. If anyone has a datasheet for the SandForce controller, then that might also tell us whether 2.8V is OK.

Another way to find out whether 2.8V is the design voltage would be to determine which of the POWR607's I/O pins is the Power On Reset pin for the SF-1222TA3. This will at least tell us if the POWR607's PSU supervisor logic thinks that everything is OK. Unfortunately it appears that the POR pin is programmable, so I can't see which of the I/O pins to check. In any case, I don't know whether POR would be active high or low. Perhaps you have a suggestion?

_________________
A backup a day keeps DR away.

Some nands need 2.8V to read properly. I think hhddrec can desolder one chip and check ID in reader - maybe we can find in database something more about this nands.

_________________
Flash Killer - everyday new resources (pinout, XOR, ECC,config) for flash devices

I'm wondering whether the POWR607 chip would still switch on U22 if it believed that Vio was too low. FWIW I examined some Samsung NAND flash datasheets and found that the voltage range for their 3.3V parts is 2.7V - 3.6V.

_________________
A backup a day keeps DR away.

In a nutshell all these stories end with this conclusion : when it is not an evident problem that cannot be classified as failure i.e. Fuses or alike, no low cost solution... (probably they invented the week because it's never always Sunday ).
@OP , SSDs are extremely difficult to deal with, if you really want to give services in this field to customers, seriously consider investing in equipment and know how .
P.s. Up now, on more than 1000 case, only few were power related and the outcome was bad (damage went beyond to the flash chips), on all the other the SSD were more or less working and the problem was ELSEWHERE. Having other identical devices at hand also helps.

I found numerous studies of multilayer ceramic capacitor (MLCC) failures at low voltages, eg ...
http://downloads.hindawi.com/journals/a ... 147949.pdf

They suggest that typical failure modes are short circuits or low resistance. Leakage currents of several uA are typical in the period prior to total failure.

ISTM that the OCZ design is vulnerable to problems arising from such failures. The high resistances in the feedback paths of the two OFA regulators mean that it doesn't take much leakage in either of the Cff capacitors (C18 and C8) before the output voltage drops significantly. HDDs, OTOH, generally have feedback resistances of the order of 100s of ohms.

In the Vio case, the current through R23 and R22 appears to be around 5uA. AFAICS it would only need around 1uA of leakage in C18 for Vio to fall from 3.3V to 2.8V. If this is indeed what has happened, then I wonder if this is a common problem. In fact I'm wondering whether the Vcore supply has been affected as well.

_________________
A backup a day keeps DR away.

Thanks to:
Randy, fzabkar, arvika, BlackST,

I am surprised with your responses, all of them are very smart and interesting, including BlackST responses because I understand you:
Profesionally quick responses are needed, if not, customer get you in troubles, and customers do not accept problems, investigation, tests, and so on. So, with this kind of cases, you don't get positive feedback from customers

Thanks Randy


Good suggestions



yes the best analisys, thanks Frank



I think measure here. What you think?

see picture