Dead Hitachi HUA723020ALA641 2TB 3.5" 7200rpm (Jun-2012) HDD

[nc513]

If I decide to use the services of a professional data recovery company ... Does anybody happen to know if it would be useful to send a HUA723020ALA641 drive manufactured in September of 2011 along with the bad JUN-2012 drive? That is, could it most probably be used as a donor in a bunch of different failure scenarios? Impossible to tell?

[nc513]

Done some more tinkering ...

My good FEB-2012 drive that I stole the PCB from earlier is now back up and running, using a SEP-2011 PCB that I transferred the FEB-2012 ROM chip to.

I suppose this might indicate that the PCBs they used during the period September 2011 to June 2012 are all compatible with each other (and there is e.g. no further encryption to worry about).

However, I guess it's still possible that the drive mechanics may differ somewhat, if you compare e.g. a SEP-2011 drive to a JUN-2012 drive ... ?

Btw, I dumped also the SEP-2011 ROM, if anybody would happen to be interested (see the attachment). It actually seems to contain a lot more code/data than my earlier two .bin files.

Attachments

ROM_dump_sep2011ebaydrive_nc513.zip

(223.25 KiB) Downloaded 544 times

[fzabkar]

The code segment between 0x8000 - 0x2FFFF is replicated at 0x31000. It looks like a backup copy. There are also additional copies of those smaller segments which I extracted in my earlier post.

This would be very useful if one needed to recover a ROM with a bit error.

[nc513]

Ok, thanks!

Do hard drives in general try to read anything at all until the motor has reached full speed?

[fzabkar]

I would think that the headstack would remain on the ramp until the spindle servo has locked onto the target RPM.

If you read the SMOOTH L7250 datasheet, you can see that the state of the spindle is reported to the MCU via 4 Spstate bits. There is also a flowchart which shows how the spindle servo transitions between each of the states during power-on.

https://pdf.datasheetcatalog.com/datasheet/SGSThomsonMicroelectronics/mXyuswx.pdf

[nc513]

Interesting.

But right now I'm feeling rather confused. I bought another one of these 2TB UltraStars from eBay. It was sold as fully working, with screenshots from CrystalDiskInfo. Well, guess what. It's misbehaving just like my own dead drive; the BIOS doesn't see it, and the only signs of life it shows are 6 x BZZZ and then a faint bump+click sound. I tested by connecting the drive directly to the motherboard as well as through a SATA-USB bridge.

Unbelievable. What could be causing this? Should I remove the PCB and clean all contacts, as well as the SATA connector (long shot)?

[fzabkar]

This is another weird thread without a proper resolution:

https://forum.hddguru.com/viewtopic.php?f=20&t=22546

I incorrectly thought that there may have been a spindle current sense resistor that we could "fudge", but it appears that the current sensing is internal to the IC, perhaps implemented with senseFETs.

[fzabkar]

What resistance do you measure between the common motor terminal and ground on your PCBs (with motor disconnected)?

[nc513]

What a shame that that older thread about those Ultrastar A7K2000 drives never got resolved. The symptoms sounded very similar to what I'm experiencing with these two A7K3000 drives now (one of which should've still been in excellent condition and had very low mileage).

What resistance do you measure between the common motor terminal and ground on your PCBs (with motor disconnected)?

In other words, check the PCBs of the two non-working drives and the PCB of one working drive, by removing the PCBs from the drives and measuring between pin 1 and GND (see the pic), while not applying power?

[fzabkar]

In other words, check the PCBs of the two non-working drives and the PCB of one working drive, by removing the PCBs from the drives and measuring between pin 1 and GND (see the pic), while not applying power?

Yes, do that. Also check the boost voltage when the drives are powered up and trying to spin. I expect it will be 12V when idle and ~18V when running.

Measuring VCM and Spindle Motor Current:
https://www.hddoracle.com/viewtopic.php?p=484#p484

I think this is your scenario:
https://www.hddoracle.com/viewtopic.php?p=11315#p11315

[nc513]

Yes, do that. Also check the boost voltage when the drives are powered up and trying to spin. I expect it will be 12V when idle and ~18V when running.

What would be the best place to measure V_boost? There doesn't seem to be any vias right where you wrote V_boost with yellow letters in this picture:

Btw, I got the pin numbering wrong in the pic in my previous post. This is the correct numbering:

[fzabkar]

The boost voltage must connect to a filter capacitor, so I expect that the small component to the left may be an inductor (for filtering), in which case there is a filter capacitor at the top left of the motor controller.

As for the motor terminal, just measure them all.

[nc513]

The boost voltage must connect to a filter capacitor, so I expect that the small component to the left may be an inductor (for filtering), in which case there is a filter capacitor at the top left of the motor controller.

So e.g. here should be good?

[fzabkar]

Yes, I expect so. You can confirm that 12V is present when the PCB is not connected.

[nc513]

** Motor terminals on PCB, PCB loose, no power;

* JUL-2012 eBay drive that should've been good, but isn't:

pin 1 to GND: 38.6 kOhm
pin 2 to GND: 0.3 Ohm
pin 3 to GND: 0.3 Ohm
pin 4 to GND: 0.3 Ohm

* SEP-2011 eBay drive that is working perfectly:

pin 1 to GND: 37.9 kOhm
pin 2 to GND: 0.3 Ohm
pin 3 to GND: 0.3 Ohm
pin 4 to GND: 0.3 Ohm

** V_boost behaviour when switching power on (SATA-USB bridge with USB cable disconnected);

* JUL-2012 eBay drive that should've been good, but isn't:

17.60V for a split second
11.93V during 6 x BZZZ
12.00V when it's given up

* SEP-2011 eBay drive that is working perfectly:

17.60V for a split second
11.70V (fluctuates slightly) during proper acceleration
23.88V for maybe about 12 secs after having reached full speed
11.90V when still spinning at full speed, but supposedly not doing any other work

[fzabkar]

It looks like the boost voltage is only used when the drive is seeking. That is, it is required by the VCM circuit, not the spindle. This is consistent with the L7250 datasheet.

The low resistances for the motor phases are the result of residual charge on the gates of each of the low-side MOSFET drivers, so that's normal.

The only thing I can suggest is that the spindle bearing may be sticky. You could try heating the spindle with a hair dryer to free it up. I'm not holding out much hope, though.

[nc513]

Do the electronics engage the VCM circuit during that first split second (some sort of self-test, maybe?), or does the VCM stay completely idle until the motor has reached proper speed? Does e.g. unparking the heads cause a considerable current spike?

I guess I could try the hair dryer trick on the JUL-2012 eBay drive, but maybe not on my own bad drive that actually contains some semi-interesting data.

[fzabkar]

Maybe the boost voltage is tested as part of the drive's POST? Otherwise, I'm wondering whether the VCM nudges the headstack towards the rear of the ramp for some reason???

In short, I don't know what is happening.

FWIW, some old Toshiba (?) models had a sticky spindle bearing that responded to being warmed up.

Edit:

The R470 resistor (0.47 ohms) is probably the VCM current sense resistor. You can measure the voltage across this resistor to see what the VCM is doing. However, a multimeter wouldn't probably tell you much, but I expect that you may not have a DSO. :-)

[nc513]

I do own a Hantek DSO5102P scope, but I have zero experience using it.

[fzabkar]

You would need to measure the voltage across the resistor in differential mode (ie subtract channel 1 from channel 2).

BTW, did you disturb the eBay drive in any way before it went bad?