Fujitsu MHV2100AH: head fails

[repfanatic]

Hi, this is my first post here. I love to repair all sorts of things (sometimes it fails), from cars to watches.
It is for me always a big satisfaction to bring something broken to work again.
After reading in this forum, I learned that repairing a hdd is a BIG challenge, which only gives me more incentive to try it.

I have a 2.5" 100Gb IDE hdd, model Fujitsu MHV2100AH , which has been wrongly powered with 12V instead of 5V, and was'nt responsive anymore.
Data on the drive is not too important, even if I'd like to recover it.

Hoping just the electronic PCB was gone, I found an identical drive, same country of origin , with most numbers matching.
With swapped PCBs, the motor is now speeding up but absolutely no noise comes from the heads: no clicking, nothing!.
Donor PCB is ok: put back on the donor drive it is still working.

Here are fotos of the broken drive and its pcb



And of donor drive and pcb.



As electronics tech and engineer, I think in any case the servo arm with heads should be replaced, but want also a wise advice from specialists, the big question beeing: would this be sufficient ? I've been searching in this very interesting forum for hours, but without finding an explicit answer.

Thank you for your help.

PS: I have access to a clean room, where I've already repaired the cam (lens mecanism) which took the hdd fotos.

[pcimage]

No specific “adaptive” info in these. The PCB’s appear compatible to me and should have worked if preamp survived.

Preamp failed IMHO, so requires head swap.

To be sure, it may be an idea to fix the original PCB.

There is a fuse on these boards, so please show the component side of the bad PCB.

[fzabkar]

Could we see the other side of the PCB, ie the side with the electronic components on it?

[repfanatic]

Many thanks for your answers

About the fuse: I found the info in this forum (did'nt imagine there is a fuse there): SMD marked with N and connected to TP1.
Measured it on the broken drive: blown .
So I just applied 5V power with an USB-IDE adapter, and shorted the fuse. Motor did'nt spin up, but MCU (chip 88i6632-BAN1) became hot, so it is probably dead, although there is no visual damage.

PS: Will post foto of the components side (don't have the drive here at the moment)

[repfanatic]

Hi, here you have a foto of PCB components side, if it can be of some help.



Both PCBs are identical in components.
Fuse is component with N mark in red circle, and the yellow square seems to show there is no ROM here.

As my understanding, only thing I can try now is to swap heads ?

[fzabkar]

I would try to confirm whether the preamp may have been damaged. To this end I would start by measuring the Vcore and Vio voltages on both PCBs.

I would then determine if there is any continuity between the pins of the HDA connector and Vio and +5Vin. This will tell us the power supply requirements of the preamp on the headstack (either +5V or +3.3V, or both). If the preamp gets a positive supply directly from the external +5V input (via the fuse), then it has probably been damaged.

I would compare the resistance between the relevant supply and ground of each of the donor and patient HDAs (by mounting the donor's PCB on each HDD). This will enable us to compare the resistances of the donor and patient preamps.

BD9109FVM, Rohm, 4.5V - 5.5V, 0.8A, 1ch Buck converter:
http://rohmfs.rohm.com/en/products/data ... 6fvm-e.pdf

Attachments

HDA_conn.jpg (90.42 KiB) Viewed 12076 times

[repfanatic]

Thank you so much,fzabkar.
This is a very valuable information . I will check all this this week-end.

There is one thing I do'nt understand:
Various posts here say that without (working) preamp, the disk woul'nt spin up. In my case, it IS spinning up with the good PCB: so in this case the preamp should be OK.
BUT as the heads do'nt make any noise at all, something would be failing in the servo actuator.
Thinking earlier that the servo winding is driven directly from the pcb, I tried unsuccessfully to find its resistance directly on the drive side HDA pins. I am not sure I tried it also on the good drive, fearing the multimeter could induce a failure inside.
But the winding could also be fed by the preamp...

[helpdisc]

i would try with same FW version of pcb, or replace the mcu on good pcb

[fzabkar]

i would ... replace the mcu on good pcb

MCU (chip 88i6632-BAN1) became hot, so it is probably dead, although there is no visual damage.

[helpdisc]

yes you are right.
Did not read it good.

[fzabkar]

I would measure the resistance between ground and each of Vcore and Vio, and then compare the readings against the donor's. An excessively low reading for Vcore would confirm that the MCU is dead.

[fzabkar]

Various posts here say that without (working) preamp, the disk woul'nt spin up. In my case, it IS spinning up with the good PCB: so in this case the preamp should be OK.

A typical failure mode of some preamps is a short between supply and ground. This will prevent the PCB from powering up.

Some drives need to detect a preamp before they will spin up, others don't. I'm not a data recovery professional, so I don't know how your model should behave. However, you could try to isolate the HDA contacts on you donor drive with a business card, assuming that the motor contacts are still able to make solid contact.

BUT as the heads do'nt make any noise at all, something would be failing in the servo actuator.

If you wish to rule out this possibility, measure the resistance between the VCM pins with the PCB on and off the HDA.

Attachments

[fzabkar]

To see if this is true for your model pick the good one and insulate the head contacts. If the drive spins and doesn't move heads then it's necessary for pre-amp to ID itself.

If you isolate the entire HDA connector, then you would be isolating the VCM contacts as well.

[fzabkar]

Here are hi-res photos of the PCB:

Fujitsu CA26332-B43104BA hires PCB scans:
http://www.hddoracle.com/viewtopic.php?t=2237

[fzabkar]

VCM = Voice Coil Motor, not spindle motor.

[jono-ats]

VCM is part of head stack assembly.

Heads need replacement if FW matches (as it does).

This is where things get a little trickier . . .

[fzabkar]

From the manual ...
http://www.fujitsu.com/downloads/COMP/fcpa/hdd/mhv2xxxah_maint-manual.pdf

Attachments

[repfanatic]

Thanks for all your input, I've learned a lot.

So, this is what I measured:

Vio on both PCBs: 3.3V (connected or not to drive)
Resistance from Vio to gnd:
- donor PCB: 5.33k not connected, 4.9k connected to drive
- failing PCB:5.25K
No continuity from Vio to HDA pins.

Vcore on donor PCB: 1.24V, failing PCB: 3.44V (connected or not to bad drive) .
Not a BD9109FVM regulator I think, but has matching pinout.
Resistance from Vcore to gnd:
- donor PCB: 166ohm does'nt change connected or not to drive
- failing PCB: was first 98ohm, 48.5ohm after powering a few seconds to read voltages, then 15.3 ohm after powering another time ( sign of melting MCU?)
No continuity neither from Vcore to HDA pins.

There is 6.6ohm between VCM pins on both drives. On both disconnected PCBs, its near 160k, so servo winding should be OK.

Interesting thing: good drive spins up even when HDA disconnected -> it does'nt need to detect preamp.

My deducing is: failing core voltage regulator -> dead MCU + dead preamp, servo winding OK (does'nt help much...) Head stack replacement needed

[fzabkar]

Sorry, the hires scan identifies the Vcore regulator as a BD9102FVM.

BD9102FVM, Rohm, Synchronous Buck Converter with Integrated FET, 4.0V - 5.5V, MSOP8, Vout = 1.24V:
https://www.marutsu.co.jp/contents/shop/marutsu/ds/bd9102fvm-e.pdf

You might also like to measure the Read/Write Channel test point voltages (T3/T4 & P15/P16) on the donor PCB when mounted on each of the donor and patient HDAs.

Preamp Block Diagrams:
http://www.hddoracle.com/viewtopic.php?f=59&t=229

Attachments

[repfanatic]

Thanks again for all your help .

Anyway, even if we know now it is not a DIY situation, I am very willing to go on for a head swap...

The most important files on this drive are some work I could do again, so losing them is'nt dramatic. It's already some years this drive broke and I never needed it really since.

Also:
- I have access to a clean room,
- I am conscious some training would be needed: planning to do this first with old 3.5' dead drives without needing the clean room,
- this would show me what are the issues, and let me search how to solve them (if possible).
- probably I would need appropriate tools, some are already identified:
+ a stable workbench (could be DIY one),
+ a system to unmount cleanly the VCM magnet (I know how strong they are, surely my fingers remember it )
+ the corresponding headcomb I can't build myself, but some are not really expensive (even for a one-time use)

Last but not least, supposing a perfect head swap can be done, is there another necessary step then, like adjusting some parameters in RAM,
or would the drive straightforward work again .