Fwd: article: Why Fujitsu hard drives break
ZXNet echo conference «zxnet.pc»
- --- begin of forward ---
Newsgroups: zxnet.pc
X-Comment-To: All
From: Boris Kreynin
Date: Sun, 28 Sep 03 23:15:52 +0400
Subject: article: Why Fujitsu hard drives break
Hello everyone!
Finally, I discovered the subject :-)
How do you recommend cleaning the board from flux?
*Why hard drives break*
How do we imagine an ideal hard drive? I'm sure the majority
will answer
something like this: to be fast, quiet, cool and reliable. In general,
a pipe dream. But there was a time when she was so close that she
could be held in your hands. The distant year 2000. After failures with Quantum CX
User confidence in modern screws was completely undermined. Here
here
The Fujitsu MPG series appeared on the scene.
Compared to analogues from Quantum and Seagate, these screws were slightly different
overpriced, just like another great hit of that time -
IBM DTLA. But, being fast, not very loud and relatively cold, these
the screws quickly won the sympathy of users. Shortly before this, the company had
several very successful developments in hard drive manufacturing, and rumors about the legendary
the reliability of their drives was already known to everyone.
And although relatively few Fujiks were brought to Russia, there were
lucky people who purchased this device and were sure that they would finallythe dream of an ideal storage device has come true. And when did the massive fall begin?
"woodpeckers", users' faith in Fujitsu has become doubly stronger - despite everything,
them
the screws continued to work normally.
The service workers were also pleased: there were no serious damage to these screws, and
repairmen for a long time were spared from hemorrhoids associated with studying them
secrets, since Fujitsu has always been distinguished by the originality of its solutions. However
less, even with the naked eye it was clear how perfect these
products: small number of chips, good connector location,
connecting the board to the HDA (unlike "woodpeckers"), and fantastic
smooth reading schedule. how many thousands!
But all good things come to an end sooner or later, and dreams tend to break
about harsh reality. It was the same this time - buying beautiful boxes
with the Fujiks, no one could have imagined that the clock mechanism had already been cocked
bombs, and these propellers are doomed to death, inevitable, like fate itself.
And the owners of these disks, who lost everything in one day - the screw, the information, and
trust in this company, more than once I had to regret my once beloved
purchase - in the spring of 2002, the number of defects in MPG series screws exceeded
all conceivable limits, leaving far behind its competitors. And despite
that Fujitsu officially admitted its mistakes, many detailsof what happened remained unsolved, including the most important of them -
the real reason for the failure.
Therefore, now we will travel back in time, exactly one year ago,
when most of these screws were still alive. And let's look at their history
death with a new look - both from the side of a simple user and from outside
hard drive repairman - after all, the beginning of this epic was so strange and
confusing, which contributed to the slow response of society to this problem.
Unlike the "woodpeckers", these screws continued to be praised all the time, buying them before
until... their stocks in warehouses ran out. And then there was a mass of them
and absurd death, there were discussions in the forums, and the information unearthed
professionals, suddenly became closed, which gave rise to fiction, many of
which had nothing to do with the real state of affairs.
And this article would never have appeared if its author, having woken up one day
beautiful morning and turning on my computer, I didn’t witness what I’ve already experienced
hundreds of users are still worried - the machine refused to boot,
emphasis on not seeing the hard drive. Yes, yes, it is the screw of this long-suffering series,
dead, but the only one and with important files, became the reason to study them
problems, which subsequently led to its successful repair and rescue
information.
UsersThe first glitches with these hard drives appeared about a year after purchase, in the fall
2001. At first, no one attached any importance to this - it was too unusual
were
their manifestations: at random times the screw “disappeared” from the system,
was vaguely detected in the BIOS, and then appeared again. This happened often
directly in Windows, which caused the system to freeze when trying to access
disk.
But after a while the screw behaved normally again, and this was forgotten.
It never occurred to anyone that the hard drive was to blame, and at first
users turned the arrows on bad contacts in the cables, mastday glitches and
own carelessness.
Soon, however, another unpleasant feature became clear - some screws
began to click loudly when copying files, which led to damage
a small black microcircuit with the inscription HA13627 - it exploded from overheating,
because, despite its temperature of 80-100 degrees, for some reason there is no
it was
radiator
In this case, the screw naturally died. And even after that no one believed that these
drives are bad. Many tears were shed for those innocently lost because of
improper nutrition for devices - this was not only the generally accepted opinion,
but
and conclusion of branded service centers.
After replacing the power supply, we often bought exactly the same screw: people
characteristicto believe in the best, but it was somewhere nearby, like a blue bird... And even leaving
Fujitsu's departure from the desktop market was no cause for concern.
But the joy of sellers who do not accept screws with thermal
damage was short-lived. Quite a bit of time has passed, and the warranty
departments were full of these screws, in perfect external condition, but, alas,
not working. However, unlike the previous wave of marriage, here it has already become
some pattern can be traced: most of the dead hard drives were
exactly
AT series, at 5400 rpm.
These drives had a common fault - they were not detected in the BIOS at all or
quacks were displayed instead of the model name during auto-detection. Soon
found out
and the reason for this behavior is that Fujitsu has officially admitted that the early
AT-series screws had a serious error in the firmware, leading to this
amazing effect - the screw forgot about the size of the error logs, and then, when they
overflow, overwriting other vital service modules.
And the included SMART, which periodically updated its tables, greatly influenced this
contributed. The faster brothers of the AH series, at 7200 rpm, had
completely different microcode and were free of these shortcomings. For screw treatment
the company released a patch updating the buggy firmware version 80B5 to completely
normal 80C2.So, if your propeller is the MPG3xxxAT series, that is, 5400 rpm, look
look carefully at the large sticker on the HDA, the one where the word is written
"Fujitsu". At the very bottom there is a line that looks something like
like this: CA9-80B5. The last four digits are the firmware version. And
if it is not 80C2, then it is highly advisable to update it - in this case there are problems with
auto-detect can disappear, and the screw can return to normal.
How wrong were the users who perceived the incident with the firmware as
another ridiculous accident! “With whom it does not happen,” they said, “after all, with those
"Dyatlov" had similar problems." But after a while "Fujiki" again
they began to fall out, all of them indiscriminately. And every such screw was unhappy
in their own way: some clicked during operation, others blocked the system, others
were covered with terrible bad things, and there were also specimens in which the
the recording function works.
And jokes about a 20-gigabyte CD “with virus protection,” although they had
place in the forums, but did not cause anything but tears from the owners of such hard drives.
But Russian craftsmen never give up, and soon various
traditional methods of retrieving information from dead devices - by putting them in the refrigerator
before warming it up on the electric stove - and it really helped! In general, in this
there were many mystical moments in history.How, for example, can we explain the fact that half of all hard drives started
Is it normal to work after “keeping it in check” for several days? And sometimes
Even dancing with a tambourine around the propeller and shamanic spells helped - after
this
The screw calmly allowed the information to be drained from itself and again fell into a stupor.
No one doubted that this series was defective; the question was different:
where and how to fix them in order to at least save your information? This is where it gets to work
repairmen, both professional and not very professional, from among the advanced
users.
Repairmen
“The electronics are dying on these screws” - this was the verdict made by many
specialists from all over the world, which was, in principle, correct. "That means you just need
change the board, and everything will be okay,” ordinary users decided: after all, for all
without exception, on Quantum screws, this method worked with a bang, and was quite logical
there was a desire to do this at Fujitsu.
Dozens of unhappy people rushed to the markets to buy junk to
buy
the second same screw and finally get your information. This is where the first one was waiting for them
blow - boards rearranged with live screws did not always work.
The reason was that Fujitsu MPG propellers, unlike Quantum, have
There are many types of ROM firmware. There were 25 types, and if the ROM version
on
the board did not coincide with what was in the service area on the pancakes - the screw onreadiness did not come out.
Therefore, we had to select boards from live screws with the same firmware version,
the same as the deceased. And still, the percentage of saved drives was very
small - even this selection of boards did not always help. For example, on many
copies of the AT series, after replacing the board with a non-original one, the screw started so loud
knocking heads that they immediately turned it off with horror, fearing damage
mechanics. There was even a rumor spread that the company had made the connection on purpose
boards to the HDA serial number so that they do not need to be changed. third category
dead screws that were not detected in the BIOS even after replacing the board.
The reason for this was the destruction of service information on the pancakes, and
this is due to an electronic problem. As you know, the service area of any HDD
consists of several modules, each of which is responsible for a specific
set
functions.
There are vital modules, for example, defect lists, translator, adaptives, and
There are also secondary ones - serial number, SMART tables, error logs. And
An unpleasant feature of the Fujitsu MPG screws was that for their normal
the work had to fulfill a mandatory condition - all without exception
modules
service records had to be read without errors.
This was checked by comparing the contents of the modules with their checksums,
and if they did not match, the screw did not unlock the interface andremained unavailable. Usually the modules were destroyed at the moment when the screw
updated them, for example, by rewriting SMART statistics. If at this moment
the electronics malfunctioned, recording to the pancakes was interrupted, and the module turned out to be
unfinished.
In this case, the screw might not produce any errors, and often for the user this
remained unnoticed - after all, the module remained in the controller’s memory, and
storage device
continued to work normally with an uncorrupted copy until
the power was turned off. And when turned on, the screw could not read them and did not go to
readiness. This explains the strange fact that these screws are mostly
died when turned on. Why did all these troubles happen?
For a long time, repairmen blamed the HA13627 chip for everything, which is very
It got very hot during operation and sometimes swelled. And her replacement returned
the screw is in working order. But it soon became clear that she had nothing to do with it -
the repaired Fujitsu screws again began to fail terribly, and in similar
Using Samsung screws, this chip worked flawlessly. And more was done
interesting observation - Fujitsu screws died even from simple storage on
warehouse!
Then careful research was carried out, which revealed the true reason
all problems - violation of the technology of soldering board elements. It turned out that
almost all factories for assembling these screws used some kind of veryaggressive flux, which was not removed from the installation after the board was assembled. Over time
it destroyed the solder connections of the parts' pins with the copper traces of the board, and
heating the chips contributed to this.
The Cirrus Logic CL-SH8671-450 multifunctional chip especially suffered from this:
evaporating, the flux penetrated through the pores of the plastic housing of this microcircuit,
with
over time, disabling its crystal, and glitches began. But the main reason
there was precisely a break in the leads - between them and the board tracks a
oxide
film, and the electrical contact was broken.
And this is precisely what explained all the mystical phenomena that occurred with these
hard drives: thermal and mechanical effects led to temporary
contact was restored, and the hard drives briefly came to life.
Such drives must be repaired by unsoldering the microcircuit and carefully
washing the board from factory flux residues, after which the chip is soldered
back to your place. The percentage of successful recovery is quite high -
theoretically this stops the destruction of the board, and the screw can then
work without any breakdowns at all.
Practical advice
If you are the “lucky” owner of a Fujitsu MPG propeller, then try
observe the following recommendations.
If your propeller is an AT series, check if it needs a firmware update and
update it if possible. "Re-threading a screw" is updating itmicroprograms In this case, a piece of executable code on pancakes is rewritten, and
sometimes a Flash-ROM chip on the controller board. The whole process takes
a few seconds and does not affect the state of the information stored on the screw. However
However, when performing it, you must observe the same precautions as
at
flashing any other device - remove all overclocks, check
reliability of connecting the loops; if there is a UPS, power the computer through it.
Fujitsu branded flashers create a boot floppy disk that can
become a weak link, so it is advisable to check its surface "Disk
Doctor." If your screw is detected incorrectly in the BIOS, with cracks or not
determined at all, there is a small probability of its recovery after
this procedure.
WARNING! Everything you do is at your own risk. The author and editors are not
carry
no responsibility for any consequences of these and the actions described below!
If the BIOS does not see the screw, but the pancakes are unscrewing, listen carefully to
its sounds immediately after switching on - this will allow you to assess whether it is destroyed
official
information or not: a hard drive with a working service should make a sound like
“click-trrrrr”, and with the destroyed one - “click-pause-tick-tick-tick-tick”.
If you have the second case, you can try to update the hard drive with a proprietary utility,and if this does not help, refer it to a qualified specialist. Hope for
the fact that the screw will come to life on its own is useless in this case - you need to contact
to a repairman who will restore service information, for example, by rewriting
destroyed modules from another screw.
If your drive is still alive, take a close look to see if the light is on
HDD indicator (on the system unit) continuously or somehow strangely. Long-term
its continuous glow in the absence of disk activity indicates
looming problems in the controller, and it is better not to use such a screw for
storing important data.
It is advisable to install a radiator on the HA13627 chip or cool the entire
fan board. True, this does not eliminate the root cause of screw failure, but
The likelihood of it breaking slightly reduces. If the screw starts to click, immediately
turn it off and take it for repair;
If the pancakes do not unwind or stop during operation, then try
not
turn on the hard drive once again, as it may mistakenly unpark
heads and scratch the surface;
If the service information of the screw is correct, as evidenced by the normal
recalibration sound, that is, there is a high probability of removing information from such a hard drive. For
To do this, you can try letting the screw sit for a couple of days, then turn it on, and if
he decided in the BIOS - run to merge the data.There is no need to put the screw in the refrigerator, because condensation will form from the cold,
which will ruin it. You can try connecting it to different IDE channels,
switch master/slave jumpers, disabling other IDE devices - no matter how
strange, but sometimes it helps.
Just be sure to turn off the power to the hard drive, without disconnecting the cable, before each
an attempt to detect it in the BIOS - this causes a restart of the service
screw firmware, which increases the likelihood of success. You can put the screw on
flat surface and, lightly pressing the Cirrus Logic chip, try
redefine the screw by turning off and on the computer's power again.
Just don’t touch the terminals of microcircuits and other parts with your hands,
because a short circuit may occur and the hard drive will malfunction.
If all these steps do not help, and there are no repair shops nearby,
you can
try more extreme methods: find a working board from the same
screw and carefully resolder the Flash-ROM chip from the original board to a working one.
This can be done with a soldering iron with a flattened tip (and before that it is advisable
experiment on other unnecessary boards).
Use an alcohol solution of rosin as a flux. Under no circumstances
you should try to solder the microcircuits with a gas soldering iron and other folk
ways. This will only lead to damage to the contents of the ROM due to overheating! In general
II say this more for information, because it’s better to refrain from
similar experiments, and, if possible, contact specialists who
they do this every day.
Unfortunately, recently there have been cases of irreversible damage to these
screws
insufficiently experienced repairmen - they rewrite service information
new one without saving the old firmware, which leads to the loss of adaptive devices.
But with other people's adapters, these screws cannot be restored, since they have
there is no selfscan, and they cannot create them themselves - screw the field of such
repairmen usually flies into the trash bin. Therefore, please contact only
to trusted repair organizations, especially if valuable items remain on the hard drive
information, and always require saving the old firmware. This is mandatory
condition for any competent repairman, and if he ignores it, then
You shouldn't trust him.
Currently, Fujitsu has officially admitted its mistakes and has begun
slowly recall these hard drives. Until they changed their minds there in their office,
it makes sense to think about an exchange, or at least try to do it.
And if your screw is still under warranty, then try contacting the seller - no
It is impossible that it will be replaced with something more reliable. Try it in the end
After all, you no longer risk anything, and the chance to get a new hard drive is still
there is anyway, and it’s not bad enough.Don't neglect the little things: the case of Fujitsu hard drives is clear
showed us how a trivial mistake can kill anyone, even the most perfect
development. But as you know, only those who do nothing make no mistakes.
does.
And Fujitsu does a lot of good things. Who knows, maybe refurbished
MPG series screws thanks to their cool mechanics even after many years
will they stay with us and continue to delight us with stable work?
Source: Computerra, "Internet Zone"
Boris
- --- end of forward ---