I’m in the middle of very slowly recovering data from Midori’s 250GB Fujitsu hard drive. Over the last 6 months the problems started out innocently, as if there was a bad sector. The drive would freeze with the HDD indicator solid, then after 2 mins it would recover. Running a full HDD Regenerator scan revealed no bad sectors, but still it happened, albeit not very often.

Then suddenly it has got worse, doing it every 2 minutes, blue screening the laptop. Putting the drive in my USB caddy the drive actually freezes and then completely disconnects itself from the USB bus if I attempt to write to it or cut and paste from it, as if I’ve used Safely Remove Hardware and turned it off. It will read data from itself, but very very weakly and slowly, starting at 800KB/sec, creeping up to a max of 8.56MB/sec after 5 minutes, when it should be at over 15MB/sec, normally starting out at 24MB easing to between 11 and 15MB.

It seems the MCU (Micro Control Unit, Main Control Unit, or Micro Code Unit) is failing, or there’s a voltage regulation issue between that and the heads. Modern hard drives only have three main chips as well as resistors, diodes and capacitors on their PCB”s. You have the MCU (main processor), the Cache memory chip, and the motor driver chip, often a SMOOTH chip, that spins the drive up using high current, then tapers off keeping it steady once it’s spun up to full revs. Remember when HDD’s had massive boards with lots of chips and electrical gubbins? That single MCU does the work of most of those, an awesome example of modern integrated electronics!

While we’re on the subject of Fujitsu, let me tell you a little bit of my data recovery background involving them…

Hard disk drive faults can occur for any number of reasons, sometimes wear and tear on the mechanical parts of the drive’s internals can lead to a drive failure, in other cases electronic faults on the drive’s PCB can lead to the failure of the drive, or even a mixture of both. Even a drive that is mechanically and electronically sound can fail, often leading to confusion in determining exactly what the cause of the failure is. The answer lies with the software that controls the hardware, that is stored on the platters, in the MCU, or both.

Quite a few years ago, when the data recovery industry was really taking off, new failures started cropping up, drives would spin up, make sounds as if initialising, and then…? …Nothing.

But what could be the cause? There was a very well known failure that appeared around the same era that data recovery companies started to appear like they were being mass produced from a factory! This failure was found in a popular brand of consumer desktop hard disk drives manufactured by (of all guys, Fujitsu!). These series all had model numbers beginning with either MPF or MPG. Before long the following drive models started failing, going into failure territory like no drive had been known to before:

  • MPF3102AT
  • MPF3102AH
  • MPF3153AT
  • MPF3153AH
  • MPF3204AT
  • MPF3204AH
  • MPG3102AT
  • MPG3102AH
  • MPG3204AT
  • MPG3204AH
  • MPG3307AT
  • MPG3307AH
  • MPG3409AT
  • MPG3409AH

These drives weren’t of the modern simple three chip design, they had big PCB’s with lots of circuitry. Once failed the Fujitsu hard disks behaved normally, spinning, apparently initializing, but not becoming ready. Whilst common in all drives of the above series the problem was particularly common in the MPG family, especially the 40GB and 20GB models, MPG3409AH, MPG3409AT, MPG3204AT, MPG3204AH.

To repair these drives access to the micro-program that starts and controls the drive (the firmware) was required. Once access to the drive had been gained via the manufacturer’s own unpublished ATA command-set the job of checking each of the firmware modules began. In most cases a temporary repair of the drive in order to extract a full clone onto a working device could be performed, involving repairing certain logs in the drive’s own firmware by replacing the contents with those from a known working drive of the same firmware revision. Results were often instant and long-lasting, but, once a drive had failed once there was only a finite length of time before it would fail again.

A good few years after the first problems Fujitsu finally admitted there were issue with the hard drives, cowardly blaming component manufacturers for the fault. The MPF and MPG series of drives showed excellent promise, with good performance, low price point and good build quality to boot, they should have really cemented Fujitsu’s foundations in the consumer desktop hard disk drive business, though it lead to Fujitsu calling it a day on further desktop hard drives instead concentrating on notebook and Enterprise class devices.

Even today they are still utter crap. They use the same Marvell processors that a lot of Samsung and Western Digital drives do, but WD and Sammy drives seem much more reliable. Samsung and WD boards that match failed ones are also easier to source as you don’t need to match serial numbers (embedded in ROM and on the platters on Fujitsu’s) because WD’s firmware and serial are just stored on the platters, as I think Samsung’s still are.

Be Sociable, Share!
Leave a Reply