2TB Hitachi 3K7000 Power Conected in Reverse

[Monopole]

I have a 2TB Hitachi 3K7000 drive that I damaged by connecting it through an adapter to an external Molex power connector on my computer. Despite the Molex connector being keyed, the pins were able to touch when I accidentally tried to plug the connector in backwards. The result is 12V was applied to the drive's 5V rail, and the drive would no longer spin up to be recognized by the computer.

There's no data on the drive that I need to recover. I just want to see if I can get the drive working again for fun and to learn a bit about hard drives.

The PCB shows no visible damage. I discovered the 5V TVS diode was permanently shorted and the 5V fuse was blown. I replaced both of them, but the drive still doesn't spin up or get recognized.

I've attached a picture of my PCB with some voltages and resistances I measured.

I tried placing a short across the 0.1Ω (R100) resistor that measures 857Ω, but it didn't seem to make any difference.

Any ideas on what might be wrong or what else I can try?

I can make more measurements at your request. I also have an old 20MHz analog oscilloscope at my disposal.

Thanks!

Attachments

[fzabkar]

Could you trace the circuit around XY, XZ, R100, 1R0? That's the Vcore regulator.

There is an additional Vio regulator (Vio2 in attachment).

The two "100" coils form a Cuk regulator. This provides the negative supply for the preamp. This supply needs to be switched on by the MCU (which needs the Vcore supply).

Attachments

Vio2.jpg (63.68 KiB) Viewed 10380 times

[fzabkar]

What is the resistance between the 1R0 coil and ground? If it is close to 0, then the MCU would be suspect.

[fzabkar]

ISTM that R100 may be a current sense resistor in the ground circuit of the XZ transistor (2SD2679), in which case you shouldn't be linking it out.

http://www.s-manuals.com/pdf/datasheet/ ... 4_rohm.pdf (XY marking)
http://media.digikey.com/pdf/Data%20Sheets/Rohm%20PDFs/2SD2679.pdf (XZ marking)

[fzabkar]

Here is how I see the circuit (the broken output is caused by the broken forum software).

Edit: I worked around the bug by deleting leading spaces.

Code:

                 +5V
                  +
                  | 2SB1714
.--------.    XY  |
|        |      |<
| hi-drv |------|
|        |      |\    inductor
| motor  |        |   ___
|control |        +---UUU---+-------+---o Vcore
|        |        |         |       |
|        |      |/   1R0   ---     .-.
| lo-drv |------|          ---     | | Rfb1
|        |   XZ |>          | C    | |
'--------'        |2SD2679  |      '-'
    |             |         |       |_____.
    |            .-.        |       |     |
    |            | | R100   |      .-.    |
    |            | |        |      | |    |
    |            '-'        |      | |    |
    |             |         |      '-'    |
    |             |         |       |Rfb2 |
    |             '---------+-------'     |
    |                       |             |
    |                      ===            |
    |                      GND            |
    |                                     |
    |                                     |
    '-------------------------------------'

[Monopole]

Could you trace the circuit around XY, XZ, R100, 1R0? That's the Vcore regulator.

There is an additional Vio regulator (Vio2 in attachment).

The two "100" coils form a Cuk regulator. This provides the negative supply for the preamp. This supply needs to be switched on by the MCU (which needs the Vcore supply).

I've attached a schematic I came up with. The transistor arrangement looks wrong to me, but that's the way it appears to be according to their pinouts in the datasheets you linked to. I also attached a picture of the PCB with components labeled as they are in the schematic.

I measured the voltage at Vio2 and got 12.14 V. That point appears to be connected through vias directly to the 12-V rail on the back of the board.

What is the resistance between the 1R0 coil and ground? If it is close to 0, then the MCU would be suspect.

I measured 43 Ω.

Attachments

[fzabkar]

It appears that I got quite a few things wrong.

The regulator which I have incorrectly identified as Vio2 appears to be a boost converter (I should have known this from the polarity of the adjacent diode -- slaps forehead). I suspect that it produces approximately 18V for the VCM circuit (see the L7250 SMOOTH datasheet for an explanation).

L4 is the inductor for the Vio supply. It outputs 1.8V for the SDRAM and serial SPI flash memory.

L1 is the inductor for the Vcore supply. I would expect a Vcore voltage of around 1V.

L2 and L3 constitute the Cuk converter. There is an example in the L7250 SMOOTH datasheet:

http://pdf.datasheetcatalog.com/datashe ... Xyuswx.pdf

If this converter is enabled, it should produce -5V.

Your circuit is inconsistent with the two BJTs. Therefore it appears that I have incorrectly identified the XZ and XY markings. Can you perform diode tests on the BCE/GDS junctions?

Attachments

[Monopole]

It appears that I got quite a few things wrong.

The regulator which I have incorrectly identified as Vio2 appears to be a boost converter (I should have known this from the polarity of the adjacent diode -- slaps forehead). I suspect that it produces approximately 18V for the VCM circuit (see the L7250 SMOOTH datasheet for an explanation).

L4 is the inductor for the Vio supply. It outputs 1.8V for the SDRAM and serial SPI flash memory.

L1 is the inductor for the Vcore supply. I would expect a Vcore voltage of around 1V.

L2 and L3 constitute the Cuk converter. There is an example in the L7250 SMOOTH datasheet:

http://pdf.datasheetcatalog.com/datashe ... Xyuswx.pdf

If this converter is enabled, it should produce -5V.

Your circuit is inconsistent with the two BJTs. Therefore it appears that I have incorrectly identified the XZ and XY markings. Can you perform diode tests on the BCE/GDS junctions?

I found a list of semiconductors with "XY" and "XZ" markings:

http://chip.tomsk.ru/chip/chipdoc.nsf/vc1!readform&view=smd&cat=X&start=2001&count=500

The physical dimensions of the transistor packages on my PCB are 2.9 mm by 1.5 mm.

The INC5001AC1 and INA5001AC1, for example, have consistent markings, dimensions, and pinouts. Still, I don't even know if the transistors on my board are actually BJTs or MOSFETs.

http://chip.tomsk.ru/chip/chipfile0-x.nsf/all/2AEAD472AE55237BC62577E80060D226/$File/INC5001AC1_E.pdf
http://chip.tomsk.ru/chip/chipfile0-x.nsf/all/35E7302F3122E0A4C62577E80060C975/$File/INA5001AC1_E.pdf

For the purpose of checking the transistors with my multimeter's diode check function and reporting the results, I will assume the transistors are BJTs with terminals as marked in the attachment.

XY
B-C: 1.586 V
C-B: 0.661 V
B-E: 1.574 V
E-B: 0.641 V
E-C: 0.032 V
C-E: 0.025 V

XZ
B-C: 1.198 V
C-B: 0.612 V
B-E: 0.978 V
E-B: 1.579 V
E-C: 1.435 V
C-E: 0.537 V

I'm not familiar enough with the characteristics of BJT or MOSFET junction voltages to make much sense of these voltages, except that the XY E-C and C-E voltages look low.

Also, I checked the possible fuse you identified and got continuity across it.

Attachments

transistors.jpg (26.5 KiB) Viewed 10250 times

[fzabkar]

XY is testing like a PNP transistor with E & C reversed. The low resistance between E & C is due to the external circuit. That is, the Vcore output powers the MCU which presents a resistance of 43 ohms. This resistance appears low, but is actually in the normal range.

XZ doesn't test like a BJT.

IME these Vcore supplies, such as those on computer motherboards, typically use MOSFETs (with body diodes), so perhaps I have some fundamental misunderstanding in this area. Clearly neither of our circuits are correct, and none of our transistor choices are suitable for high frequency (1MHz) operation. However, it is clear that there is a problem in this area of the circuit. Perhaps you could probe the base (or gate) drive to XY and XZ. When everything is working correctly, you should see PWM pulses.

[fzabkar]

These complementary parts look like they might fit the circuit:

N0300P, Renesas, P-Channel MOSFET, marking XZ, -30V, -4.5A, SC-96:
http://pdf.tixer.ru/540531.pdf

N0300N, Renesas, N-Channel MOSFET, marking XY, 30V, 4.5A, SC-96:
http://datasheet.datasheetarchive.com/o ... 047462.pdf

Renesas was formed as a merger between Hitachi's semiconductor division and those of Mitsubishi and NEC, so Renesas parts would be likely candidates.

Edit: The dimensions in the datasheets match your measurements.

[Monopole]

XY is testing like a PNP transistor with E & C reversed. The low resistance between E & C is due to the external circuit. That is, the Vcore output powers the MCU which presents a resistance of 43 ohms. This resistance appears low, but is actually in the normal range.

XZ doesn't test like a BJT.

IME these Vcore supplies, such as those on computer motherboards, typically use MOSFETs (with body diodes), so perhaps I have some fundamental misunderstanding in this area. Clearly neither of our circuits are correct, and none of our transistor choices are suitable for high frequency (1MHz) operation. However, it is clear that there is a problem in this area of the circuit. Perhaps you could probe the base (or gate) drive to XY and XZ. When everything is working correctly, you should see PWM pulses.

I don't see anything that looks like a PWM signal at the gate/base of XY or XZ. I guess that means the MCU is probably damaged. However, it must be partially functioning if it is generating the 1.8-V supply.

Given how fast those TVS diodes are, I'm a little surprised the MCU apparently got damaged. It's also odd that I measured 857Ω across the R100 current sense resistor. It seems like it would have had to dissipate a relatively large amount of power to fail during the short time it took for the TVS diode to react. Perhaps I'll desolder that resistor from the board and see if it's actually completely open.

Thanks for all your help, fzabkar. You've been really helpful. It was also your site where I found information about replacing TVS diodes. Even though it looks like I won't get the drive working again, at least I learned a few things, like what a Ćuk converter is and the basic topology of a modern HDD PCB. That almost makes frying the drive worth it.

These complementary parts look like they might fit the circuit:

N0300P, Renesas, P-Channel MOSFET, marking XZ, -30V, -4.5A, SC-96:
http://pdf.tixer.ru/540531.pdf

N0300N, Renesas, N-Channel MOSFET, marking XY, 30V, 4.5A, SC-96:
http://datasheet.datasheetarchive.com/o ... 047462.pdf

Renesas was formed as a merger between Hitachi's semiconductor division and those of Mitsubishi and NEC, so Renesas parts would be likely candidates.

Edit: The dimensions in the datasheets match your measurements.

I bet those are the correct transistors.

[fzabkar]

Thanks for the feedback. I've seen this Vcore circuit many times, but I've never been able to get as far with it as we have in this thread.

BTW, I apologise for the initial distraction with BJTs. I never felt comfortable with my circuit diagram -- BJTs were always going to be inappropriate for this circuit. The current and frequency were too low, Vce(sat) was relatively high, and there was no internal diode for the collapsing coil current. Just wrong in every way.

FYI, I have written some articles that you may find helpful. I'll add the Hitachi Vcore and Vboost topologies in due course.

Tutorial - Linear and Switchmode Regulators used in HDDs:
http://www.hddoracle.com/viewtopic.php?f=59&t=231

HDD/SSD PCB protection devices:
http://www.hddoracle.com/viewtopic.php?f=100&t=1615

TVS Diode FAQ (with photos):
http://www.hddoracle.com/viewtopic.php?f=100&t=86