Need advice on how to attack a HDD problem

[Old Tech]

Hi Old Tech,
Read my message carefully! I wrote " don't touch the PCBA", not inhard!
.

I understood your point and I want to repeat that I was not trying to be insulting or insensitive in my reply to you. I just feel a bit frustrated with regard to the mystery surrounding the inner chamber of these drives.

I had not intended to do anything to the PCB as far as shotgunning the board. That's a reference to a hacker method of repairing electronics by changing all the parts till you find the problem. In order to troubleshoot a circuit like this, you really need to understand the circuit, and that means having a schematic. Since none are available, I'm making my own.

Once I have the circuit, there are various means of attacking the problem, even though I can't see inside the chips to know what they are doing exactly. Once I'm satisfied that a chip/part is defective, then I'll try to change it.

So far, I've traced half of the 176 pin DSP. The 40 pin input cable occupies pins 43 to 88 and the memory chip occupies pins 1 to 42. The rest of the pins will be the most interesting because they will interface to the spindle and voice coil motors as well as the read/write mechanism. The PCB also has a generous supply of reference symbols that give an idea what is going on.

I would still appreciate help from anyone who can get me data sheets on the DSP (D741667APGF), the SH6770C, and the two Agere chips (MS453 and the preamp 541008D6). In return, I'd be happy to publish my results on the forum.

[Old Tech]

As far as I know Maxtor DSP first check the PCB itself and then starts the motor.

I may be wrong, but my understanding is that it's the housekeeping you describe that is the problem. We want the drive to just go read the tracks under the command of an app that can recover data. It's interesting that you say the DSP checks the PCB then starts the motor. Since my motor is starting OK, it would seem there is an issue with it's ability to read the disk. As you say, that could be a head, the preamp, damage to the service area of the disk or a corrupted data.

When the DSP checks the PCB, it needs feedback from the other chips to tell it to proceed. It would not be hard to fool it into thinking everything is fine.

[Old Tech]

That is correct. But there is no way to check all components on the pcb...

why do you say that? There are basic digital and analogue troubleshooting techniques to test any components. If you know what the circuit is trying to do, you can either trace it's own data or inject data of your own to trace.

By knowing the ATA standard, you can inject data and codes to make the DSP respond. By inference, you can trace through the chips doing that. Think of the circuit as trying to do three things: transfer data (data bus), establish addresses and codes (address bus), and control signal routing (various control bits). You can set up a jig to do that with dip switches and a power supply. If you need clock signals, a cheap 555 timer will do the trick.

Remember that the data entering the PCB is simple, parallel 16 bit data. Depending on the code you put on the 3 bit address bus, the DSP will do the rest for you. With those codes, you can tell the heads to seek, read or write. You should be able to trace the appropriate path through the electronics depending on what you enter for data and addressing.

I don't know about this DSP, but many PICs have a single-stepping feature. That means you can toggle a bit and get the DSP to run one step at a time instead of the 2 gig speed (crystal speed) it normally runs at.

[Dmitry Postrigan]

That is correct. But there is no way to check all components on the pcb...

why do you say that? There are basic digital and analogue troubleshooting techniques to test any components. If you know what the circuit is trying to do, you can either trace it's own data or inject data of your own to trace.

Oh, I mean by the pcb itself

[niddo]

Head changes on these drives are very difficult

Last edited by niddo on March 22nd, 2006, 16:23, edited 2 times in total.

[Old Tech]

I'm assuming with your level of experience that you have access to a storage scope, just use a differential probe on the RDN and RDP test points to determine whether there is a good signal coming from the preamp.

thanks for the informative response. I wondered what the RDN/RDP test point were and I just happened to see them referenced on a data sheet I found for a 6 head Sony preamp. I also have a detailed explanation of the pre-amp for another device but unfortunately I have so much to read and digest.

I don't have a storage scope, or access to one because, I'm working as an electrician right now. I have an older 10 Mhz dual trace scope and I'm wondering about the possibility of finding a device that could interface between my scope and the higher demands of the high speed data. I'll look into finding a used storage scope, or perhaps some newer technology using data capture and a PC.

With regard to the contaminant issue, drives you have looked at, opened and exposed to contamination in the past were probably not using GMR technology (possibly thin film or something similar). GMR is extremely sensitive to ESD events and proper precautions should be taken.

Although I have challenged the need to be paranoid about working on open drives, I have never claimed that one could be sloppy. I am always conscious of static while working on any electronic device, making sure I am well ground before touching anything with my hand. Fortunately, I live in a damp environment where large static charges seldom occur.

Also, static doesn't damage a board or a part randomly. You have to introduce it across a MOS gate. I realize that static can be as high as 40,000 volts in magnitude off a human body, and I have seen static discharge off my own body that was a blue spark. When you see blue in a discharge, it's getting up there in magnitude. In fact, on one site, in a very dry environement, I managed to shut a computer down by touching a vynil table top the computer was sitting on.

The biggest danger with static, IMHO, is the static on polyester or wool clothing. As you walk, especially on a carpet, you are generating static. But even standing still and moving your arms with certain clothing on, you can generate it. It's important to get into the habit of touching a good ground each time you come in contact with electronic devices. It's just as important to get into the habit of handling the parts so as not to come in contact with the leads or PCB traces.

The concern I have right now is the use of isopropyl alcohol to clean up the platters. The best way would be to pull the platters and heads apart, and I have heard of this being done, but I don't know anything about the drive's ability to realign itself.

Particulate contamination is only a secondary concern, this has always been known - however manufacturers and IDEMA put handling controls in place for a reason it is irresponsible to think that they can be ignored (not in your case where it is personal data OldTech) when dealing with client's drives in a Data Recovery environment.

Point well taken. If I was working at this for a living, I would certainly be far more concerned about the environment. My attitude right now is very experimental. I don't want to damage the drive farther, but if I do it's no big deal.

[Old Tech]

Oh, I mean by the pcb itself

You mean a self-diagnostic? Hmmmm....more to think about.

[niddo]

Hi Old Tech

I seem to remember some paper I read about the lubricant used on platter surfaces being particularly sensitive to most solvents, I think this may cause more problems than it solves.

From experience the platter in an old style Maxtor is normally fine, clicking is almost always a fault within the head assembly itself.

Alignment issues will lead the drive to click away and not do an awful lot else so be extremely careful when mounting the new heads, they have a tendency to come together making it very difficult to seperate them and get them on the outer edge of the platter.

Just my thoughts on it, my advice; if you have done head changes before and are confident in your abilities (and have access to the correct spare) then go for it, its the only way to learn, we all started somewhere, normally with a few damaged drives under our belts before we achieve any level of success

Last edited by niddo on March 22nd, 2006, 16:22, edited 1 time in total.

[Old Tech]

Hi Old Tech

We use Isopropyl Alcohol to clean head sliders, I seem to remember some paper I read about the lubricant used on platter surfaces being particularly sensitive to most solvents, I think this may cause more problems than it solves.

Thanks for info, niddo. Maybe I have it wrong. Do the modern heads slide on the surface like the floppy drives? I assumed it was still air pressure holding them up. There appears to be a lubricant on the surface of the platters. I don't have a professional clean room and I got a speck of dust on the surface that would not blow off. I used isopropyl alcohol and it left a residue that did came off with a soft material.

From experience the platter in an old style Maxtor is normally fine, clicking is almost always a fault within the head assembly itself.

If you are prepared to try I would suggest sourcing an identical drive and conducting a head change, this is very difficult without special equipment, at least in cases where there are 4 or more heads, normally a head change comb or some kind of platter mounted head seperators would allow you to safely remove the heads from the outer platter edge.

Right now, I am taking my time, and I will only try a head change as a last resort. I want to be sure the preamp is working first and I need to finish tracing the circuit. I have not run the drive since the clicking started. I am looking at a way to disable the voice coil mechanism so I can power up the drive without the heads seeking. Unfortunately, the spindle motor seems to use a sensing circuit which checks for the magnetic field on the motor itself.

For that reason, I don't want to power up the board without it being attached to the spindle motor. Without a load, the board will probably shut down, but worse still, I might blow something running it with no motor load. I thought about using resistors as a load, but if the circuit is testing for the back EMF produced by the motor magnetic field, that wont work.

Alignment issues will lead the drive to click away and not do an awful lot else so be extremely careful when mounting the new heads, they have a tendency to come together making it very difficult to seperate them and get them on the outer edge of the platter.

thanks for that advice. I have read Stanislav's article and he used flattened plastic straws, which I thought was a brilliant idea. When I worked on drives, long ago, they used an oscilloscope to read a pattern on the drive. The modern drives seems to align themselves. Is that right?

The reason I don't think it is a mechanical failure on the heads is that the original problem began after a power failure. I am attempting this data recovery for a family member and they had the AC power running straight off a wall outlet rather than through a surge protector. The plug was accidently pulled, and that's when the problems began.

There's a possibility that the drive wrote to the disk while it's logic was out of kilter. Then again, the drive worked for a while afterwards but it was very slow and Windows was having trouble reading. It was reporting an inability to read certain files that were there. Also, it reported difficulty with different files each time.

Just my thoughts on it, my advice; if you have done head changes before and are confident in your abilities (and have access to the correct spare) then go for it, its the only way to learn, we all started somewhere, normally with a few damaged drives under our belts before we achieve any level of success

When I say that I did head changes, it was many years ago on much smaller drives. Still, the head mechanism was under considerable pressure and it was nearly impossible to pry the heads apart with one's fingernails. Also, the heads were rounded and did not sit on the disk surface. They were activated from a linear motor and loaded when required.

I appreciate your encouragement and I'm looking forward to trying. If I did not have data to be recovered, I'd be a little more reckless. After your advice, and reading Stanislav's article, I feel confident in my skills as far as attempting it.

As I said, I don't have a professional clean room. I need to work in my home on a table. I'm looking into ways to build a small tent over the table using plastic sheets and using a hepafilter to control the dust. Of course, I'll use a mask, gloves and hair cover, like a surgeon.

With respect to alignment, do you know if it's possible to remove the platters?

[Old Tech]

Hi there OldTech,

If you want to determine whether the fault lies within the head assembly Maxtor make it very easy to test as they have the test points marked out on the PCB. I'm assuming with your level of experience that you have access to a storage scope, just use a differential probe on the RDN and RDP test points to determine whether there is a good signal coming from the preamp. You'll need a storage scope with a fair bit of bandwidth but it is doable.

I said my scope was 10 Mhz, it's actually 20 Mhz, but it's not a storage scope. I have done some research and there are devices available today for a few hundred dollars that capture the data and use a PC to display the data. Of course, the storage ability is based on the amount of RAM and disk space you have. Also, I don't know what kind of bandwidth you'd get with the cheaper units.

I imagine what you're getting at with your reference to RDN and RDP is checking the overall signal for a read on each head. I assume the ticking is caused by the drive seeking for a read then parking itself. That would mean you'd need a lot of storage that would have to be reviewed. Also, you'd need some software that would allow selection of each head. That would not be difficult to write in assembler and I'm sure there is software available to do that.

One thing concerning me is running the drive with the heads ticking. I don't imagine it would hurt for a few minutes, but not knowing the cause of it directly, I am concerned about trying it.

When I repaired computers, we used the Basic language to write small programs to make the heads seek. We could select the heads individually. Of course, Basic is one of the few languages that allowed direct interaction with a target. It ran in two modes. In one mode, you could write a normal program, compile it, then execute it. In the other mode, you could write a program on the spot and have Basic run the target in real time. Of course, in those days, you could write directly to an OUT port. You still can if you use DOS, or one of it's offspring....or Linux. I have Linux Fedora on another partition but I'm not that good using it.