Experiences from fixing many WD800JD-75MSA3 drives

[Rainbow]

During past years, I've fixed many WD800JD-75MSA3 drives in Dell machines we're running. These drives are very reliable mechanically but the PCB is prone to failures.

The most common problem is faulty R050 resistor (current sense for Marvell 1.37V core voltage). Its resistance increases over time, causing voltage drop and various problems - e.g. timeouts, drive stopping and starting again or other weird behavior (like drive working fine at SATA I speed but timing out at SATA II speed). This failure is so common that I'm now checking it on all drives I get in my hands: voltage drop on R050 must be around 16.6mV (bare PCB). If it's too high, replace R050.

U1 (SMOOTH L6283 1.3) burns sometimes. The cause is uknown. If the PCB is not burned, U1 can be replaced. Sometimes it just dies without burning anything (a reliable check is to measure voltage drops from GND to motor output pins).

Other resistors can also fail, causing voltages to be too low or too high.
1.37V too low: bad R74 (8660 ohms)
1.37V too high: bad resistor right of D3 (3160 ohms)
-5V too low: bad R73 (10.2K)
3.3V too high: bad R41 (2.32K), often results in burnt U4 (RAM) or U2 (Marvell CPU)
spins up, clicks 3 times, spins down, repeats: bad R33 (1K)
When R120 (-5V current sense) fails, it can lead to slow write performance (like 10MB/s) and many unreadable sectors.

Due to sudden PCB failure, the drive can stop with heads not parked (when U1 burns or CPU stops suddenly because of faulty R050). Once the PCB is fixed (or replaced), the drive can recover from this state (many drives are unable to spin up in this state) but a surface check is needed. Depending on drive handling, there can be no bad sectors or many of them, requiring a selfscan. Sometimes SMART gets corrupted and must be cleared.

[fzabkar]

Very interesting! Thanks!

I guess these resistors are probably carbon types?

[Rainbow]

My theory behind the failing resistors is that the foam between the PCB and drive body contains sulfur which is known to damage SMD resistors.

Most of the drives (a couple of hundreds) are running 24/7 since 2008 so they're over 120000 hours. Most of the machines have these WD drives, but there are also Seagate ST380815AS (7200.10) and Hitachi HDS721680PLA380 (and a small amount of Samsung drives). There were no PCB failures on them but they fail in different ways (except Samsung - no failures).

Seagate drives suffer from bad sectors - sometimes a self-scan can fix that, but not always (probably weak heads). Hitachi drives have bad sectors too sometimes. If they can be reallocated by rewriting, they can be then moved to P-List (using format unit script for MHDD). But there are often slow sectors which don't reallocate and so I'm unable to fix.

[fzabkar]

Do you see this kind of problem?

Oxidisation on Western Digital PCBs:
http://www.hddoracle.com/viewtopic.php?f=86&t=649

[Rainbow]

Yes, the pads are oxidised on almost all drives. Some are better, some worse. But it never caused any problems - until the PCB is removed. The pads must be cleaned before putting the PCB back.

When I remove the PCB, I gently scrape the oxidised layer off the pads. I've tried pencil eraser and alcohol before but it was not a reliable fix.

[pepe]

wondering what the purpose of repairing these is
the most problematic was the feedback path of -5V supply, as it resulted in preamp damage and data/servo erasure in many cases. It didn't only affect the resistors but the vias as well.

[Rainbow]

I've had only one case of preamp damage. Burnt U1 (SMOOTH), drive didn't work with known good PCB. Checked pins using a multimeter diode test and found them to be very different comparing to a working drive.

[fzabkar]

@Rainbow, just in case you were not aware, there is a publicly available datasheet for the L7250 SMOOTH controller which provides a useful insight into the workings of these ICs.

[Rainbow]

Thanks. Another good document is: ST AN1139 APPLICATION NOTE (L6254 - L6268 - L6269 12V DISK DRIVE POWER COMBO IC)

[fzabkar]

Thanks. Another good document is: ST AN1139 APPLICATION NOTE (L6254 - L6268 - L6269 12V DISK DRIVE POWER COMBO IC)

Nice, thanks. I found it at datasheetarchive.com.

Attachments

DSA0027053.pdf

(622.92 KiB) Downloaded 205 times

[Amarbir[CDR-Labs]]

wondering what the purpose of repairing these is
the most problematic was the feedback path of -5V supply, as it resulted in preamp damage and data/servo erasure in many cases. It didn't only affect the resistors but the vias as well.

Well,
these drive never have a Data/Servo Erasure ,Infact i have never seen a drive do this in my 7 years of experience with dealing with mechanical drives , 2 to 5 cases a day .

[pepe]

PCB 701335 and its PATA version were the most prone to produce these issues, at least in my region. Perhaps climate affects it too or India got pcbs or foam from other source, who knows...

[Doomer]

\

Well,
these drive never have a Data/Servo Erasure ,Infact i have never seen a drive do this in my 7 years of experience with dealing with mechanical drives , 2 to 5 cases a day .

I've seen more than 20 (of this model), this is one of the few models that has it.