PCB BIOS and FLASH swap

[zvit]

The head is not bad, the surface is scratched. Can you make the difference?

Yes, I make the difference. Head crashes on disk for whatever reason and surface gets scratched. I assumed that a head touching the platter and scratching it would be called a bad head. But you're correct, I should have said how much data was recovered from a platter with a scratch.

Normally it takes less then 24 hours to recover the data from such a drive by the way proposed by @Northwind..

I know. I didn't have the donor stuff yet and we decided to take the risk without it.

[pepe]

even you pepe

thx

I like helping here, my problem was that people needed to tell you the same things over and over again coz you didn't listen. You still ask questions about U14 despite of it's been told there is nothing interesting in there for you. Just forget about U14. If you understood the purpose of these chips you would not ask how these 2 can have the same PN while have different function. That's why i said 'do your homework' and understand these things yourself, coz we can't make you understand apparently. Moreover, if you want to get into DR, you need to think a lot more analytically and try to find out basic stuff on your own making use of the internet. We are at the 5th page of this thread, for such a simple problem and there goes a huge resource sucking here (time). You ask things that you should find out yourself from resources available to all of us.

I don't want to let you down but about the 30% of that 500G data is probably corrupt due to skipping one head. R-studio is not aware of the areas you skipped, so it saved files without good data in them, in fact, if you haven't zero-filled the image drive prior the operation, then these files actually contain previous data on the image drive.

edit: just saw you imaged it to a file, which should eliminate the zero filling issue...

pepe

[zvit]

I like helping here, my problem was that people needed to tell you the same things over and over again coz you didn't listen. You still ask questions about U14 despite of it's been told there is nothing interesting in there for you. Just forget about U14. If you understood the purpose of these chips you would not ask how these 2 can have the same PN while have different function.

I totally understand what you mean by "not listening" but let me explain that I actually DID listen but because your knowledge is so much above mine, that to you it's ABC, so you didn't really answer what I was asking. Sometimes it's good to try and understand not only the question but WHY the question was asked - what's bothering the asker.

I only asked about the donor U12 and donor U14 once. At first, I only had one question: Why MY U12 said flash and donor one didn't. Your initial answer didn't answer my question. You just explained, once again, the difference between U12 and U14. I already understood that there was a difference, I was asking why my U12 said flash and the other didn't. You finally answered that on my third attempt. But until now, you haven't answered why donor U12 and donor U14 both have identical casing and PN. All you say is "If you understood the purpose of these chips you would not ask how these 2 can have the same PN while have different function." which doesn't help my curiosity. Well, let me ask you, do YOU understand the purpose of a CPU? Wouldn't it be strange if Intel was selling two types of CPU chips but they both had the same version\model number on them? Meaning, on your own, there would be no way to tell the apart just by looking at them. Why would any company make two different kinds of chips with identical casing and PN but are different on the inside are NOT interchangeable? How would they even know which is which at the manufacturer? Get my point?

That's why i said 'do your homework' and understand these things yourself, coz we can't make you understand apparently. Moreover, if you want to get into DR, you need to think a lot more analytically and try to find out basic stuff on your own making use of the internet.

Part of online homework IS asking on forums. I research first so I know what to ask but ultimately, forums are going to be a big part of my research. To you, U12 is ABC. I have been doing occasional logical DR only for the past 15 years but I had other main jobs and didn't want to make DR my main job (still don't) so I never researched into chips - didn't have to. When I decided to take this case a little further, I saw about swapping chips in donor PCBs but no one mentioned U12 or U14. They just said BIOS or ROM. When my donor board came with a different U12 casing, I came here to ask about it. I only except people who have the time to answer to answer.

We are at the 5th page of this thread, for such a simple problem and there goes a huge resource sucking here (time).

I think I already answered this in my previous post. No one is forced into spending time reading or answering here. We are at the 5th page because the posts varied between subjects. Some were questions, some were progress updates, some were support from maximus for a product I purchased, and some were to explain why I asked a question a second time, like now. One last time, if you don't have time to read my posts - I give you my full permission not to, just don't accuse me of wasting your time if you do.

With love and respect - Zvi.

[pepe]

I am sorry if i was rude or something, it wasn't my intention.
I meant to search for datasheets of those 'bios' chips and find out their purpose, it is pretty much common knowledge and does not need much explanation. Dmitris and I made it clear in our first posts that you need to swap U12 if you decide to swap pcb. After that info you should not ask much more, just check the datasheets on the internet and draw the conclusions that they are the same chip if you are interested in this matter.

If we want to help we need to read your and others posts which takes a bunch of time. And if we need to read 5 pages, picking the important information from that 5 pages is tremendous, I do not run a 90091e server in my head yet, so can't classify information that fast. It would be desirable that you think on the problem first and try to find out yourself and only ask only what's completely beyond you. Then, if you get an answer, even if it's not clear why you got that exact answer, try to figure it out yourself, take your time to look up documentation of the chips, etc, not just ask immediately. This is what i called spoon-feeding. We take the time to help you and from this side it looks like it's only us who make efforts to solve Your problem.
It also looked like it's not your drive and you are risking somebody else's data which didn't look very good.

with respect,
pepe

[maximus]

I am amazed at how much data was extracted with a bad head.

The head is not bad, the surface is scratched. Can you make the difference?

Normally it takes less then 24 hours to recover the data from such a drive by the way proposed by @Northwind..

Everything I saw in the logs provided indicate that there is a totally dead head. Maybe the file system (MFT) is all in the good heads, but there will likely be some files or chunks of files that may seem recoverable, but are either damaged or blank if any part of them was in the bad head. This drive would almost certainly require a head swap to get any data from the dead head, assuming the platter surface is not already destroyed from the current recovery attempt. But yes, a pro would have got to this point much faster, and hopefully inspected the heads before making a decision to thrash on the drive without a head swap.

[fzabkar]

I would have thought that a head crash would be audibly obvious, especially after such a long recovery.

[pepe]

It's probably not a head crash, just surface damages which affect data integrity, or the head is weak, ie it reads servo (so no clicking) but not user data.
i didn't see the logs, but from the screenshots i would think the head can still read user data but the drive gets stuck at damaged sectors where resets can come handy or change config to reduce retries.
But anyway, data read from those areas will be corrupt, nothing to do about it.

pepe

[maximus]

The drive had a complex issue, both the slow issue and a bad/dead head. The drive was old enough that the slow issue could be fixed with HDDSuperTool. Only then could it be seen that there was also a bad/dead head. From what log data I have I can see that the head is not reading any data at all. That is MY assessment based on data available to me.

[pepe]

https://forum.hddguru.com/viewtopic.php ... 75#p285375

I might have made a wrong assumption about the colors based on PC3k map coloring. But here he has green and other colors mixed, not large blocks of green, then large blocks of something else. Or this part is from a good head albeit damaged surface...

[maximus]

https://forum.hddguru.com/viewtopic.php?p=285375#p285375

I might have made a wrong assumption about the colors based on PC3k map coloring. But here he has green and other colors mixed, not large blocks of green, then large blocks of something else. Or this part is from a good head albeit damaged surface...

Yes. The upper part of that is where the slow issue was present. Then you can see a pattern related to a head issue. But with the last image the OP has selected the show good data option in HDDSCViewer, and the green dot inside the cells indicates that there was data read in that area. Everything in the bad head area has no green dot, no data read.

[zvit]

Thanks pete, no worries, all is good. I wasn't offended in any way.

After fixing the slow issue with HDDSuperTool and starting again, the green started from the top.

Can't find on Google a reason how a surface gets damaged without a head crash or some physical shock to the drive, which didn't happen. (Maybe some static electricity discharge?) If the head is just weak or bad (like maximus saw in the logs) then I understand that the head can't read data, but since there is no noise or clicking, what else could cause surface damage (and put the drive at risk to further surface damage during a scan?) Perhaps a piece of dust broken off of the bad head? If a bad head is just not accessing the data but no surface damage, wouldn't it be safe to scan without further damage? (Or perhaps a bad head can break off a piece of dust at any time?)

Attached screenshots from last 40 minutes of scan. One is a time-lapse (in case it's interesting to anyone)

[Lardman]

Then you can see a pattern related to a head issue. But with the last image the OP has selected the show good data option in HDDSCViewer, and the green dot inside the cells indicates that there was data read in that area. Everything in the bad head area has no green dot, no data read.

Here's a good example of how physical damage to the platter tends presents, You can see the heads struggling for the first 10% and again around the 75% mark but some data is able to be read in the area and the rest of the surface is fine.

Attachments

[BGman]

Everything I saw in the logs provided indicate that there is a totally dead head.

Every atempt the drive to read from a bad head will end with a svr_xxxx error and the drive will STOP responding. Maybe you have a different definition for "bad head" but in general "bad head" means head that cannot read servo marks .

[pepe]

I should probably read the manual for HddSuperclone, coz one thing is not clear: a green (or whatever color) block means one sector or a block of sectors of arbitrary count? First I thought it is one sector now i tend to believe it is a block of sectors. The whole map would be more readable if it was sector based or if it was possible to select how many sectors they represent at least (maybe it is... ?)
pepe

[zvit]

I should probably read the manual for HddSuperclone, coz one thing is not clear: a green (or whatever color) block means one sector or a block of sectors of arbitrary count? First I thought it is one sector now i tend to believe it is a block of sectors. The whole map would be more readable if it was sector based or if it was possible to select how many sectors they represent at least (maybe it is... ?)
pepe

In the viewer, you can choose how many sectors a block will be (for visibility purposes.) You can also play with the points (block size) for display purposes. There is a nice video showing how to use the viewer here: https://www.youtube.com/watch?v=jD_tKUh8npM

[pepe]

Ok, then what was your setting for map display block size?

pepe

[zvit]

Ok, then what was your setting for map display block size?

If you watched the video, you'd see a trick to play around with the sizes and window size, to get the bad head section lined up nicely in the graph. Since this would change every few hours (depending where it was scanning) I changed the sizes a few times, depending on what I wanted to see and how I wanted to line up the graph. So I can't remember what setting I had exactly at the time of the screenshots (and gif) but I started with: 1 dot = 12K and dots size 8pt.

[pepe]

that video is a demo and it is not your case, i was curious about yours. I wanted to see if there are any good sectors in the marked areas, or they are completely bad or they were skipped. I don't care about how nicely the pattern lines up, i wanted to get a picture about what's going on there. Which i can't see coz the block represents multiple sectors, which in fact can be a pretty large area.
The only log file i saw was from the very beginning of your process, i might have missed a larger log?
pepe

[maximus]

Everything I saw in the logs provided indicate that there is a totally dead head.

Every atempt the drive to read from a bad head will end with a svr_xxxx error and the drive will STOP responding. Maybe you have a different definition for "bad head" but in general "bad head" means head that cannot read servo marks .

In this case my definition of bad head is one that is not reading any data at all. When attempting to read data from that head, the drive appears to stay busy until some form of reset is sent.

[maximus]

I should probably read the manual for HddSuperclone, coz one thing is not clear: a green (or whatever color) block means one sector or a block of sectors of arbitrary count? First I thought it is one sector now i tend to believe it is a block of sectors. The whole map would be more readable if it was sector based or if it was possible to select how many sectors they represent at least (maybe it is... ?)
pepe

HDDSCViewer does not have a set block size, but instead has a way to set how big the map is. The number of sectors in a block would be determined by the size of the drive in sectors divided by the size of the map. But I never intended it to be an exact size, it is meant for being able to visualize the recovery. The user can click on a block and in the terminal it will display the first sector of that block, but that does not help someone like you asking absolute questions. BTW, only a sith deals in absolutes

In this case, the option to highlight good data has been selected. That option would put a little green dot in any area that had been attempted and had any good read at all, but was in a state other than fully read (non-trimmed, non-scraped, bad...). The non-trimmed areas are in grey, and if any of those would have had a good read, there would be a green dot in the middle of them. I have examined the logs that were available, and at no time did I ever see any data read from what would be considered the bad head. If you want to know the size of the head reads, I can get that from one of the logs, but it won't be exact since phase 2 has not been run and there will be data that was skipped and not read up to the back edge of the bad head, if that makes any sense to you.