This is the story of a job that should never have been needed,
were it not for a couple of bad ideas that combined to snatch
defeat from the jaws of victory. It is also related to the story
of Repairing a Thumbdrive
in that they involved surprisingly similar circumstances, and
they both happened over the course of a week.
I was at The Hackery again
when a call came in from a teacher at a local school. Another
teacher there had suggested she call us about a big problem that
had developed. Her external hard drive was not working, and surprise
surprise, it contained the only copies of a bunch of important
data. Bad idea number one. The data recovery place she'd called
had suggested it might cost a thousand dollars to repair. So she
was asking around for a second opinion. We told her to come down.
She soon brought in a 500 gigabyte Verbatim 2.5" USB hard
drive. Sure enough it failed to detect on our machines. So with
the silicone rubber shell cut open, we slid the hard drive out
with the intent of plugging it straight into a SATA port to see
if that would bring it back to life. Unfortunately what we found
inside was bad idea number two.
Traditionally, external hard drives have been nothing but standard
laptop or desktop hard drives mounted in a case with a small adapter
plugged into them to convert between SATA and USB. This meant
you could pull off the adapter to get a standard drive, which
is very handy if your USB to SATA adapter stops working. But some
"genius" decided that this was too costly, and so an
increasing number of external drives are being manufactured that
have the entire USB converter integrated into the drive PCB. This
means there's no SATA plug anywhere on them, and no way to separate
the USB adapter from the drive itself. It was one of these drives
which our customer had brought us. Joy.
So we had a drive that couldn't be isolated from its USB controller.
And when we examined it beneath our magnifying lamp, we could
see that it was indeed the USB controller which was at fault.
A blow to the USB port had not only broken its connection to the
PCB, but also bent the PCB to such a degree that many traces on
the opposite side (Where the SATA-USB bridge chip was located)
had torn right off. A traditional repair would be very difficult.
Instead I proposed a more radical solution. I would attempt to
disconnect the USB-SATA bridge chip from the drive controller
IC and tap into the SATA signal lines, so that the drive could
be accessed by way of a PC's SATA connector. In effect I was going
to transform it back into a normal laptop drive. Power would still
be supplied over the USB port though as the power lines had remained
connected. Were it not I would have also wired a standard molex
socket to the drive.
The first step was to look up the bridge chip. The markings on
the top indicated that it was a JM20329 by JMicron. And a quick
trip to Google unearthed the datasheet.
Then I went looking through the pinout of the chip until I located
what I needed. Page 6 showed that pins 20, 21, 25 and 26 were,
respectively, the positive and negative RX lines, and the negative
and positive TX lines for the chip. These are the 4 pins that
make up a SATA data cable. And that meant the traces which lead
to these pins were carrying the signals which would normally be
found at the drive's SATA connector, if it were so equipped. Jackpot.
The datasheet also helpfully told me that the controller expected
to be connected with a capacitor in series on each of those lines,
so I knew that the 4 capacitors to the bottom/right in the picture
above would need to be removed if I wanted to connect the drive
directly to a PC. A good thing, as it presented a handy location
to solder to.
I cut the end off of a standard SATA data cable and stripped
it down. Beneath a foil shield was 2 wire pairs, one of which
carried data from the drive to the controller, and the other which
carried data in the opposite direction. I neatly stripped a tiny
bit of insulation off the ends of each.
Then it was time for some extremely delicate soldering. I had
brought my stereo microscope from home for this job, as I couldn't
see clearly enough with my naked eyes.
Here I am first removing the tiny surface mount capacitors, then
soldering the 4 wires to the pads the capacitors had occupied.
And here's the end result, with the electrical tape holding the
SATA cable roughly in place to keep the tiny solder joints from
breaking. Were I to do this again I would use thin magnet wire
as a bridge between the extremely stiff SATA wire and the tiny
solder pads. But it was attached, and that meant it was time to
We carefully plugged the drive into the drive test station...
And it worked! The drive spun up and mounted! We began copying
the data onto another drive the customer provided as fast as we
could. And a good thing, too, for the tiny traces on the PCB tore
off when we disconnected the drive at the end to show the customer.
It held out just long enough and not a minute longer.
So what can we learn from this? Well, I know it's been said many
times before, but please, BACK UP YOUR DATA! The world's most
reliable hard drive can't replace a backup. But I think there's
another good lesson here too. If you have the option, it shouldn't
cost much more to buy a drive that uses a normal SATA mechanism
inside. And an extra few dollars or an extra 1 cm in case length
are a small price to pay for having the flexibility to directly
communicate with your drive if something goes wrong. Unfortunately
it can be a little hard to tell from the outside of the box what's
lurking on the inside. It's for this reason that I usually assemble
my own external drives using separately purchased hard drives
and drive boxes. And that's a great little project for someone
who wants to gain some technical experience.
Page created July 24th 2013
Seemed like the real thing
only to find... Mucho mistrust, love's gone behind.