Hitachi ARM DT01ABA100 bad NV-RAM

[deftrue]

Hello everybody,

I have a Toshiba/Hitachi DT01ABA100 not spinning up. PC-3000 is not helping me that much with this drive for some reasons. However, I think it is just an NV-RAM corruption or bad checksum. I wrote a working DT01ACA100 NV-RAM and it spins and seems to callibrate, but with no sector access of course.

Has anybody got a tool for checking these NV-RAMs or at least calculating their checksums? They are E2PR - 4 kB long, and the Hitachi utilities does not seem to help on this newer drive.

Thanks in advance

[deftrue]

Maybe if someone can share NVRAM resources from similar drives... HDS5C1010DLE6 kind or DT01ABA100. That might help.

Thanks.

[shahij]

Oh my god. At least I found someone regardingg NVRAM issue of ARM series.

I faced several cases regarding NVRAM dead issue of ARM series, most of them were CLA series. All were unsolved. Only person DR-Kiev replied it's possible to recover data from NVRAM issue of CLA series, no one else had any positive response regarding this issue.

Ref: viewtopic.php?f=1&t=27412

[fzabkar]

Maybe if someone can share NVRAM resources ...

Could you at least start by uploading your good and bad NVRAM images?

[deftrue]

Of course, I will do it tonight. It is strange that editing headmap in utility (NVRAM editing) doesn't seem to update any checksums.

This drive suffers from pretty strange things : ( I first thought about preamp failure because it was not spinning with HSA attached, and opened the drive in the clean room and... well the top surface was partially black + rainbow coloured and I said WTF? Preamp must have not just failed but flashed and badly burnt! Mysteriously, there were no evident signs of damage on the bare-type chip, and I was able to restore platter mirror shining look. Now, with new HSMA and new PCB, but original NVRAM, it displays the same behavior. A foreign NVRAM makes it spin...

Well as I said I'll upload both tonight.

Thanks for your interest.

[fzabkar]

It is strange that editing headmap in utility (NVRAM editing) doesn't seem to update any checksums.

ISTR that, at least on earlier versions of PC3K, the NVRAM checksum was not automatically recomputed whenever the NVRAM contents were edited. This needed to be done by way of an add-on utility supplied by Ace Labs.

[deftrue]

Just like I promised, here are the two NV-RAMs. The xACA100 one is foreign, and makes the drive to spin. The xABA100 one is the one retrieved from the patient drive - with this one it will not even try to spin. I understand PUIS mode would give ID or some sort of ID, as oposed to the behavior this drive displays...

Thanks for any suggestion - fzabkar, I happen to know old PCI PC3K and older versions from UDMA, but when I started working on Hitachi drives, checksums were calculated by utility transparently. I'll check on monday with ACE to see if they can provide such tool for this NV-RAM! In the meantime, any help is greatly appreciated.

Cheers

Attachments

NVRAMs-ABA_ACA.zip

DT01ABA100 and DT01ACA100 NV-RAMs.

(3.83 KiB) Downloaded 461 times

[Doomer]

ACA NVRAM is from a HDS721010DLE630 model and as you may guess it is not the same as HDS5C1010DLE630
If drive spins with it then probably that PCB is not original

[deftrue]

That's weird. I have bought another PCB from a DT01ABA100 and a brand new DT01ABA100V (whole disk), transferred NV-RAM from original and same result on both cases. The label from the patient says DT01ABA100 which is 5C1010 and it had no signs of being changed or opened.

The only 721010 I might have was the DT01ACA100 I used just to give the other one a try.

What is going on here?

[Doomer]

You need to figure out correct diagnosis of why the drive doesn't spin.
Swapping PCBs randomly is not a correct way of doing that.
You probably don't have enough knowledge to work on this case.

[deftrue]

You need to figure out correct diagnosis of why the drive doesn't spin.
Swapping PCBs randomly is not a correct way of doing that.
You probably don't have enough knowledge to work on this case.

Doomer,

The patient PCB had short circuits on several components. I know how to do the diagnosis, and I'm very confident on my job and which PCB and NV-RAM corresponds to each drive - at least, the way they came to me. Serial number on NV-RAM matches, to the extent possible, serial number on sticker. Also, the customer talked about a power surge happening on their building. So first thing I did was to test PCB and try to fix it. And then, start thinking about other damages since I could not make the drive spin with HSA attached.

Ultimately, this is an extremely rare and expensive drive to find a donor for, so resources are limited and since knowledge comes from experience, I doubt about lots of pros working on this specific drive with this very same sympthoms.

[fzabkar]

The patient PCB had short circuits on several components.

Could you upload a detailed photo of the PCB and identify the shorted components? That will help us to understand the nature of the fault. For example, if the power surge hit the +5V side of the electronics, then one would have to be concerned about the preamp, especially if the zero-ohm resistor attached to the 5V TVS diode measured open circuit.

Clearly the preamp is not pulling down the onboard supply voltage, so if it is faulty, then it obviously isn't shorted. However, there are simple measurements that you could do which would tell us whether the preamp is at least partly functional. For example, I would measure the write channel pins with the board mounted on the HSA connector and compare them against a working drive. IIRC, these voltages should be around 1V or more. Since these pins are outputs from the preamp, then a missing voltage would confirm that the preamp is dead.

As for your earlier question regarding the NVRAM, I confess that I can't immediately see how to compute the checksum/ECC. However, I suspect that the last 6 bytes of the NVRAM are probably involved. If your model stores the PUIS flag in NVRAM, then one way to determine the location of the checksum would be to enable PUIS and compare the NVRAM dumps before and after. Any differences will be in the PUIS flag and the checksum/ECC bytes. I would then take a working drive and see how it behaves after invalidating the NVRAM checksum, but without disturbing the NVRAM data.

[Doomer]

I know how to do the diagnosis, and I'm very confident on my job

ok then
good luck

[deftrue]

I know how to do the diagnosis, and I'm very confident on my job

ok then
good luck

I can send it to you if you think you can solve this case.

Well, yes, the zero ohm resistor, it was burnt. I will attach some pictures, although they will not reveal anyhing. I repaired the patient PCB in the end.

[fzabkar]

If the resistor was "burnt" rather than merely OC, then this sounds serious. Normally these resistors fail like fuses, without too much distress. Furthermore, these resistors don't appear to protect the rest of the drive, at least in some Hitachi models. AFAICT from previous discussions, these resistors fuse when the TVS diode goes SC, but they do not interrupt power to the motor controller or the preamp. Your model may be different, though.

BTW, I can help you identify the HSA connector pinout if you need me to. I have been through this process for other Hitachi models.

Last edited by fzabkar on February 2nd, 2014, 17:30, edited 1 time in total.

[Doomer]

I can send it to you if you think you can solve this case.

I don't follow you
Are you confident in what you doing or you actually want to outsource it?
Anyway, if you want to outsource it, I'm sure you can find people who know how to fix it, including company I work for
Just ask for it

[deftrue]

Being confident with my work doesn't mean I can fix it, or even whether it is possible or not for reasonable amount of money or in a reasonable amount of time, or even if it's possible at all. I seriously would outsource this if I had the time and some guarantee it is actually possible in such terms. But my customer will have redone all the work they've lost in about one week from now if I don't tell them good news, so I'm out of luck outsourcing it to the US.

Well, about the 0 ohm resistor - sorry for the hyperbole, it was just OC. I will apreciate your help with the HSA pinout. With respect to the checksums, yeah I suspect from the final 6 bytes, but I don't know about any 48 bit checksum algorythm and there are several 6 byte fields resembling checksum digests every other chunk of data. I've tried several CRC polynomials outputting 6 byte digests, but there are lots of variables and I couldn't find a matching one. Bruteforcing it may be an option! I will test the PUIS option with a good drive - but it won't be an ABA100 drive; as I said, these are rare.

Thanks for your interest in the case, and sorry for the typos if there are, I'm on my smartphone.

[fzabkar]

Try FsumFrontEnd:
http://fsumfe.sourceforge.net/

Fsum Frontend is a free and easy-to-use tool that allows to compute message digests, checksums and HMACs for files and text strings. ...

It supports 96 algorithms: ...

[Doomer]

I would first confirm that writing NVRAM to a drive with PC3000 does not recalculate the checksum. I wonder how many threads about non-spinning drives we would have if that would be true.

[deftrue]

Thanks fzabkar. Interestingly, both NV-RAMs have a XOR-8bits sum equal to zero (1 byte). I don't think this is coincidence! I've programmed a XOR-Nbytes algorythm to check for similarities on the N byte space (N=6 for instance), having checksum = 0 with N=1 and N=2 for both NV-RAMs, but I haven't got a 6 byte zero by now... and it's time to go to bed.

This mini-test, however, points towards a consistent NV-RAM.

Thanks for your help.