Experiment to recover data from the FUJITSU HDD (Model MPG3409AT-EF)           - Continued -

What happened?
The other day my wifefs PC broke down. She could not save anything into the C drive and her PC hung up. The BIOS didn't recognize the HDD at all upon power up. As a last resort I reloaded all files that I backed up two months ago. Now she can use her PC. After I removed the HDD from her PC I found that it was Fujitsufs Model MPG3409AT-EF. I searched the web and found that this HDD was a notorious HDD (This site is closed as of 4/13/2006), which was recalled by Fujitsu in July 2002 in Japan. 

Also see Fujitsu's Public Statement on faulty drives (This site is closed as of 5/5/2005)


What went wrong?
@I searched the web further and found that the root cause of this malfunction. The Maxim Integrated Productsfreport says that the Sumitomo EMExxxxU Series molding compound was the direct cause of this failure mechanism. It is said that the Cirrus logicfs HDD controller in this HDD uses Sumitomo EMExxxxU Series molding compound and if you keep using this HDD for a long time adjacent pins in this controller will short and malfunction will occur. Since the HDD controller chip in this HDD causes this failure. However, HDD disk itself is not broken. So there is a possibility that it can be repaired.  

Cause of the failure:
   I'll explain why adjacent pins in the HDD controller get shorted in time. This HDD controller is housed in Sumitomo
EMExxxxU Series molding compound, which contains phosphorus as flame retardant. Phosphorous particles are coated with dielectric thin film. However if this dielectric film is broken from the very beginning or with any cause if it ruptures in time phosphorus becomes phosphoric acid by being combined with moisture in the molding compound. Phosphoric acid attacks copper in the lead frame material or silver,plating material at the second bonding area and ionizes copper and silver. If there is electric field between adjacent pins copper ions or silver ions move in the electric field along with the boundary of molding compound and form dendrite (X ray picture,
Silver dendrite which I made accidentally during analysis) between adjacent pins of the HDD controller. When dendrite bridges two adjacent pins electrical short circuit with some resistance is formed and HDD controller malfunctions. Most papers I read claims that silver dendrite caused the short but copper dendrite can be formed also and can short adjacent pins. Since dendrite formation is accelerated in the highest electric field region there is the highest possibility that I can find electrical short between power pins and ground pins and I might be able to fuse open such dendrite easily because dendrite is very fine element. If I know the shorted pins there is a possibility that I can save my HDD. Please read following explanation with pictures in the report.

Experiment and results:
 You need following items to do experiment.

1)   
Probing jig with 0.5 mm gap at the tip of the needles. I made this jig using two sewing
     needles and attached them to a ruler with instant glue and soldered two wires.

2)   Multi-meter
3)   Magnifying glass
4)  Package pin assignment diagram

Before this experiment I looked for the datasheet of cirrus logicfs SH8671 but I couldn't find it. So I just draw package pin assignment diagram by myself. Then I measured resistivity between adjacent pins one by one using the probe jig and the multi-meter. If I use x1~ x100 resistivity range I read 500K ohms for most of pins but some pins showed lower resistivity like 20~30 ohms. Also some pins showed 0 ohm. I checked with these low resistive pins with a curve tracer later and found that the pins that showed lower resistivity had diode characteristics. When you measure the resistivity between adjacent pins with a multi-meter you are measuring the resistivity beyond the forward voltage of diode and you will see lower resistivity because multi-meter applies 1.5V (x1 ~ x100 receptivity range) or 9V (x10K range) between pins. So I had to probe all 200 some pins. I couldn't confirm which adjacent pins were actually shorted by dendrite because this HDD controller was still on the PCB and many other chips were on the same PCB. However, BIOS recognized the HDD after I measured resistivity between all adjacent pins with the hand made probe jig. Even though I couldn't recover my wifefs mail folder because it was corrupted I could format this disk successfully. As of writing this report it is still running without problem. As long as I donft power up this HDD for long time I think I can use this HDD as a back-up HDD.

Good news and bad news (10/31/2004 Updated)
I got a good news from "Nono-chan" who read my report and tried my data recovery method. His experiment was a success and he could recover complete data from two HDDs. One of them is Fujitsu's MPG3409AT(*) and another one is
Seagate's ST34342A.(*) Another person from Texas USA could recover data from Fujitsu's MPG3409AT. However, one guy from also USA zapped IC chip. It looks like you need to limit electric current to do this experiment. A multi-meter is a good choice.
Note(*): These web sites are written in Japanese but you can see what kind of jig he used to recover data from faulty HDDs.He recommends to use "dq" and "ec" software when you encounter corrupted folders or files.


Type of HDD which was recovered by my probing technique. (4/13/2006 Updated)

Manufacturer Type of HDD QTY
Fujitsu MPG3409AT-EF 9
MPG3409AT-SY 1
MPB3032AT 1
MPG3409AT-EU 1
Seagate ST34342A 1
Total 13

How many person tried this technique (probing) and could recover data from HDD?

Success Failure
Country person Country person
Japan 8 Japan 0
Other 3 Other 1
Total 11 Total 1

Other technique used:20 HDDs
1) Controller board swapping - MPG3409AT-EF - 3 HDDs (India/Japan/Germany)
2) Fujitsu software - 13 HDDs
3) Mechanical stress repairing method (Hot air ) - MPG3409AT-EF - 2 HDDs (Japan/Netherlands)
 "Tried the cigar lighter from my car to heat up Fujitsu hard drive. It worked!"
What a brave man! He also tried other HDD, Maxtor (diamondmax plus 5T040H4) but it didn't work. I don't think Maxter HDD uses phosphorous doped mold compound.
4) Mechanical stress repairing method (dry/cooling ) - MPG3204AH-E - 1 HDD (Russia)
"I didn't believe it - it works!!! I've managed to copy ALL my data from Fujitsu!!!!"
- Mechanical stress repairing (dry/cooling ) : MPG3409AT-EF: Controller board swapping : Fujitsu software-> No recovery -> cigar lighter -> No recovery -> Cooled/dried in the fridge -> BIOS detected HDD -> HDD stopped during data backup -> Put CPU cooler on the Controller chip -> Data was fully recovered.
" I almost gave up to recover data from faulty HDD because repair cost by repair shop was too much until I found this web site."
5) MPG3409AT-EF: Controller board swapping ->No data recovery -> Fujitsu repair software -> recovered some folders -> HDD died again -> Controller swapping -> Repaired HDD.
It looks like this is a real phosphorous related failure because HDD died again in short time. He thinks that since Fujitsu repair software modifies parameters in the HDD there is a risk to use the Fujitsu repair software first. That's why he tired Controller board swapping method first. He paid 40 Euro for this Controller board.

Total 33 HDDs were repaired as of 4/13/2006

As of 10/30/2005 12 people successfully repaired their HDDs using Fujitsu' data recovery software. If you are interested in the data recovery software for Fujitsu HDDiMPG3409AT-FFjplease follow this link (This page is written in Japanese). iTitle: Fujitsu MPG series HDD Recovery Tool Version 2.02Aj@I got the same softwareiZIP,1.36Mb Virus scanned) from foreign country the other day. According to the person who sent this software to me he was very happy because he could recover all family pictures he took during his trip to foreign country by just running this software. I think it's worth trying.

HDD controller board swapping is the most reliable way of data recovery. One guy from India and one Japanese fixed his HDD by controller board swapping technique.

Please see this page for more repair methods.

If you have any question about my experiment please send me a mail.

@
            
Original : 11/25/2003
Updated: 4/13/2006