Post by VanguardLH Post by J. P. Gilliver (John)
Rather than move the platters, why not move the controller (from the
good drive to the dud), if you think that's what's faulty? Doing that
might also be possible without breaking the seal on the housings.
The problem with swapping PCBs (assuming you can find a replacement that
matches the old one) is the calibration and low-level bad-sector mapping
recorded by the factory during manufacture and testing won't match from
the replacement PCB to what is on the failed drive's PCB. Sectors
marked and masked out by the replacement drive's minicontroller will
prevent access to sectors for files you want to recover on the failed
drive, and you would end up trying to use the bad sectors no longer
mapped out to the minicontroller on the failed drive.
Timemark 5:40 - Swap doesn't work.
Timemark 7:12 - Gotta swap the ROM chip.
You'll end up having to move the ROM chip, if still usabled, from the
failed drive's PCB to the identical replacement PCB. Easier and more
likely to succeed by repairing the failed drive's PCB, like replacing a
burnt TVS diode, than to replace the PCB and somehow transplant the
calibration and bad-sector tables to the replacement PCB.
Since the OP is asking about using a recovery lab on his failed drive, I
doubt he has the skills and gear to swap the ROM chip assuming he finds
a donor drive with EXACTLY the same PCB (same minicontroller, same
firmware) and even knows how to identify which is the ROM chip to move.
There are lots of urban legends out there on swapping PCBs and magically
the replacement PCB on the failed drive suddenly works. The success
rate of a simple PCB swap is rare. Go to your nearest casino and you'll
have better odds of winning enough money to pay the recovery lab.
According to this, the main chip has firmware.
As well as the external 8-pin 25P05AV flash chip.
The 25P05AV is a 64KB chip, too big to be a config ROM, too small to
hold a decent amount of code. I looked at an 80GB Seagate here,
as the first drive I could find that has a chip like that.
("Modern" drives flip the PCB upside-down, so easy visual
inspection is not possible on those.)
Some of the eight pin chips, are actually power MOSFETs
and not ROMS at all. So if you spot an eight pin one, get
out your magnifying glass and look up the part number.
That's how I got to this thread, using 25P05AV as a search term.
As near as I could tell, the main chip (the "MCU") is a mask ROM
processor of some sort. And I don't think flashing ever changes
that chip. I got this idea, from seeing different MCUs with
different 9 digit part numbers on them. Normally, silicon companies
don't like to make a ton of SKUs like that. Bad for business.
Which implies a standard part, with a custom layer 1 metal
for the ROM. Using ROMmed parts like that, makes sense if you
plan on buying 50,000 of them (you have to place a minimum order).
For some reason, there's a price penalty for having parts with
EEPROM inside instead. A masked ROM is about the ugliest way
of doing things, you can think of (i.e. not very practical,
Adding an external EEPROM, implies the product has variants,
and you weren't clever enough with the ROM code to cover
all of them.
My failed 40GB Maxtor didn't have a problem with this.
If the MCU cannot access the Service Area, the MCU declares
"this drive is 10GB in capacity and the product name is Falcon".
Which of course, it isn't. That's the internal critical data,
before the MCU loads the Service Area. Once the Service Area
is loaded, then it changes to a Maxtor part number and the
proper 40GB capacity. The implication was, that controller
could have worked with a single platter 10GB setup, or
a four platter 40Gb setup.
I wouldn't know this, and the part wouldn't normally do that,
but, while that drive was powered, it was dying while running.
And it revealed its new name, because it was already spinning
and didn't have the common sense to stop responding. That's
how I got the name "Falcon" out of it. It shouldn't really
have responded and given away its "secret identity".