testing hard disks with some confidence.

Hi, I already feel I am going to regret getting involved in this.

I really don't want to do conventional drives.

A store I do flash for has somewhere between 100-200 HDD's that have been building up. Owner says they are mostly from upgrades, some could be new if a customer wanted a TB and the system originally had a 500GB for example. He has asked me what he can do with them. Obviously neither one of us want to sell anyone a bad drive, but I also don't know what I should be using to test them. Is MHDD good enough? should I test the twice? The plan is to go through the drives, and see what are sellable.

Anyone have any suggestions for testing these that isn't going to make me batty for the next two years? Is there maybe a hardware device not too expensive that is suited for this stuff? Should I not touch this request with a bargepole?

Thanks alot

I would say, lacking something better do the following :

Connect the drives one by one to a independent power supply only.

The ones that don't spin, that click, that start to burn etc you move to the pile of damaged drives. The ones that apear to be spinning ok without strange noises you move to the "to test further" pile.

Now the next step assuming you don't have specialized tools you plug one by one of the "to test" drives and check S.M.A.R.T. with something like MHDD or victoria or whatever.

The ones that have relocated sectors, pending sectors for relocation or other S.M.A.R.T. indicators that the drive is getting bad or very old you move to a pile of "not reliable" or "in need of self scan". Rest of them you move to "test further pile".

Now the next stept is a quick "scan" with MHDD that will verify surface. You do this to the drives. If they have slow sectors, red sectors or huge amout of greens or bad sectors you move the drives to the "in need of self scan" pile. If they look ok with only grey sectors and a few greens (like no more then 10 or 20) you move the drive to the pile "last paranoic test".

Now you should write to all sectors of the drive and verify again. Faster and safer way to do so (the one that worked better for me) is to run the ATA secure erase. Use the secure erase and if will 0-fill the drive and write to all sectors (be aware of buggy samsungs where it doesn't work as it should but for the other brands it should be ok). If you don't want to use secure erase use dban or even mhdd to write 0 to the drive (erase) but be aware it will be slow.

When you erase the drive completly check again with mhdd scan. If all ok check S.M.A.R.T. and see if the number of relocated sectors or pending relocation is still at 0 . If so you are good to go. If not Move the drive to "in need of self scan" pile.

Now if you don't have specific hardware like even a pirated pc3k card to help you out i would advice you to use MHDD instead of windows tools like victoria.

This is because windows might access the channel of the drive hile you are using if with the tool or the other windows apps might slow down / mess up with victoria during the tests.

If you are scanning the drive with victoria on windows connected to the motherboard and if your AV starts to scan files or if you open lots of stuff and you have resident stuff working your victoria might start to display greens and red blocks on the drive while testing not because the drive is bad but because victoria is getting slower responses due to a busy system or due to other stuff accessing the drive, etc ....

Use pure DOS, is safer and reliable.

Hope this helps.

Regards.

HaQue wrote:Hi, I already feel I am going to regret getting involved in this.

I really don't want to do conventional drives.

A store I do flash for has somewhere between 100-200 HDD's that have been building up. Owner says they are mostly from upgrades, some could be new if a customer wanted a TB and the system originally had a 500GB for example. He has asked me what he can do with them. Obviously neither one of us want to sell anyone a bad drive, but I also don't know what I should be using to test them. Is MHDD good enough? should I test the twice? The plan is to go through the drives, and see what are sellable.

Anyone have any suggestions for testing these that isn't going to make me batty for the next two years? Is there maybe a hardware device not too expensive that is suited for this stuff? Should I not touch this request with a bargepole?

Thanks alot

Haque ,
Spildit Has Nailed It Nice And Good ,Please Add The Following

1 : Open Each Drive PCB First
2 : Check + And - For a Short And Also TVS Diodes
3 : Run PCB Bare Without HSA With a Standalone Powersupply [ IC's Should Not Blow/Heat ]
4 : Clean Contacts Of Preamp/Preamp Pins Of HSA And Motor
5 : Connect Tested PCB To HSA

Follow What Spildit Said ....

Thanks a lot guys, very logical. I appreciate the detail as well Spildit. No I do not have any pirated, or real PC3K or any other hardware. I do have plenty of PCs though, so can setup any suitable environment.

I really don't want to get into Hard Disks, but this seems do-able, and a shame to waste so many disks. I didn't want to buy any hard disk tools, because you know what happens.. once you start, you just keep on... and I don't want to take away from my Flash! But if one was necessary I would consider it.

I am guessing even though DR Pros just want to get a drive to work to recover data, they would have a way to confidently asses a drive for sale?

Thanks again

All of the firmware tools like PC3K, MRT Pro, HRT have univeersal utility that can help with testing, more or less like i've described but hardware assisted. It's a little bit pointless to add the "hardware assisted" here because you want to be sure the drive will work fine even on a normal computer without the special pc3k (or whatever) card.

Stuff like Victoria for windows can do several type of tests like seek testing, when you read and write to random sectors. This tests the capability of the drive to read/write correctly and at decent speed even when it's not doing that sequentialy.

HDAT2 also have a decent set of tests that can be used to "make sure" the drive is still usable, but it might take some time to do so.

Also if you want to be really sure after all that i've stated you can copy some large files to the drive at test and compare it's md5 / crc against the original file.

Firmware tools sometimes can test other stuff depending on the drive, like head quality, servo tracks quality, etc .... but will depend on the model as it will be vendor specific.

Example :

But without investing some money i would say that the MHDD/HDAT2/Victoria testing on the lines of what i've stated will be enough, even more if you have a huge amount of drives to test and you don't want to be spending lots of time of each of them.

Unless you have more time on your hands than you know what to do with, I'd suggest zeroing the drives with Active@ KillDisk (free) and then selling them for parts/recycling to any data recovery business that's interested. Those that won't zero, just destroy the platters to protect the user data by punching a screwdriver through the inspection port on the bottom/side of the drive. KillDisk runs under Windows and will simultaneously zero as many drives as it can see, e.g., hanging off multiple SATA controllers. It still takes forever with larger drives. Selfscan also takes forever and is non-trivial to configure and run, from what little I know about it. You could buy all new drives for the value of time you will end up putting into the drives, even if you value your time at minimum wage. My advice assumes you don't really want to start doing data recovery from hard drives, otherwise you can keep them for parts and learning exercises yourself.

Thanks Spildit for the extra info. looks like software will be enough. larry, you are right I don't want to get into HDD Recovery and I have more than enough learning planned already

cheers

Even data recovery pros wouldn't want to have hard drives plugged on their pc-3000 cards testing for days and ocupying the space on the channel of the card.

I would say that even pros would move the drives to be tested to a system whithout a firmware tool and test them from there because even if they have to run tests for several hours it will be on a system that they can spare and not wasting time that could be used to do some other firmware work on a firmware tool.

Several years ago i figure out another cool way to test drives as you want.

You can 0-fill the drive with D-Ban or any other tool, and if you can use a random pattern to write to the drive (not just zeros) the better. Then you check S.M.A.R.T. and you make sure that it's still at 0 for pending and relocated sectors.

Get a copy of spinrite 6 and run it at level 4. It will read all the data for each sector, invert it, write it,chek it, invert it again, write again, confirm. For each sector. You will be sure that the each of the drive sectors can read and write acurate data and it's inverse binary data. Check S.M.A.R.T. at the end of it and make sure relocated sectors didn't increase (drive didn't swap failing sectors for spares). This is one of the few things spinrite can be used for in a good way. It will not test random writting/reading (like from random LBA) but for sequential reading/writing and surface quality it will be just great.

LarrySabo wrote:Unless you have more time on your hands than you know what to do with, I'd suggest zeroing the drives with Active@ KillDisk (free) and then selling them for parts/recycling to any data recovery business that's interested. Those that won't zero, just destroy the platters to protect the user data by punching a screwdriver through the inspection port on the bottom/side of the drive. KillDisk runs under Windows and will simultaneously zero as many drives as it can see, e.g., hanging off multiple SATA controllers. It still takes forever with larger drives. Selfscan also takes forever and is non-trivial to configure and run, from what little I know about it. You could buy all new drives for the value of time you will end up putting into the drives, even if you value your time at minimum wage. My advice assumes you don't really want to start doing data recovery from hard drives, otherwise you can keep them for parts and learning exercises yourself.

Larry Sir ,
Nice To Catch You Away From DFL Forums Also

.Spildit Already Told Zero Filling But Your Post Added a Few More Options in That .BTW Haque There Are Many Tools In The World That Can Zero Fill Many Drives At The Same Time On a Board .Assuming You Have 7 SATA Ports And 1 Is Used For Internal HDD And 1 For DVD-Writer .Then You Could Use Pending 5 For Zero filling 5 HDD's At Same Time .I Do This in Linux With Max 2 Drives

PS : I Am a Poor Man :mrgreen:

Just as a side note when HRT was develloped on the PC3K ISA days they were concentrating more in hard drive repair then data recovery itself, so HRT have some "cool" tests on the universal utility that you can run on all drives even as unatended set of tests.

Example you could define a full test with up untill 16 steps and you would chose what you would want to do on each step (what test) and then you would configure each step one by one to define what kind of test you would want to execute.

Let's say you would do a read test first, a write test as step 2, then you would verify, the you would do servo testing, whatever ....

: 1.png (10.4 KiB) Viewed 19185 times

: 2.png (13.86 KiB) Viewed 19185 times

Then even if the desired tests would run for days you wouldn't need to be near the machine as the HRT would power off/on if needed, and do all as selected by the batch. You could do a full testing and be sure that the drive would be in perfect condition if it passed all tests without further relocation but it would require huge amount of time to test larger drives for several type of tests, like writing and doing non-sequential writing/reading, etc ...

: 3.png (15.61 KiB) Viewed 19185 times

: 4.png (12.55 KiB) Viewed 19185 times

At any rate i don't think that ANY data recovery pro would let a most likely good drive connected to a firmware tool for days or even weeks just to be sure that it would be near a perfect state.

I would say just go with the regular mhdd or spinrite stuff that i mentioned on a regular machine and if some drives are found to be defected later by the client just replace them or refund.

Spildit Wowwwwwww ,
You Got a Cool Deal When Your Purchased And i Recon You Do Not Regret It a Bit ,Keep It Up

Amarbir wrote:Spildit Wowwwwwww ,
You Got a Cool Deal When Your Purchased And i Recon You Do Not Regret It a Bit ,Keep It Up

Another simple way to 0 fill is just get power supply to connect your drives and send the secure erase ATA standard commands using MHDD or whatever.

Then you can disconnect SATA/IDE port from PC/Drive and leave the drive plugged to power.

This will continue to erase the drive (write to all sectors) internally (it doesn't use SATA/IDE interface and don't require data transference from the PC.

That way you can use the same PC to initialize secure erase on the other drives and just have to keep them powered on.

This might save you time if you have huge amount of drives and few pc's to use

Very interesting, thanks alot. If I start getting sucked into the HDD vortex I will blame you guys!

I do actually find HDD recovery very interesting, but I know I don't have the time or money to do it properly, so I am going to leave it to the Pro's.

Spildit ,
Leave Haque Alone Now Lol Otherwise His Heads Gonna Blast .Haque Frank Is Still Missing :lol:

missing? I know AU is a small Island, but I haven't seen him... ;-)

drive it is better to check long smart tests (02h or 82h). (so WinDFT works)

Moltke wrote:drive it is better to check long smart tests (02h or 82h). (so WinDFT works)

Interesting

Using the internal S.M.A.R.T. testing functions to scan the drive

02h = Execute SMART Extended self-test routine immediately in off-line mode

82h = Execute SMART Extended self-test routine immediately in captive mode

So i guess it's an option as well

Run the extended S.M.A.R.T. self-test either in off-line or captive mode

Of course that if you run it on off-line mode you should leave the drive testing itself and then check the S.M.A.R.T. log.

Thanks everyone, to use a phrase I know you guys love.. I will have a play and see how it all comes together!

I've been thinking on this issue a little bit more ...

Well, the smart extended self scan can be of some use but it also have some problems. One is that it will take some time to complete and the other is that it doesn't show a graphic display of the surface/problems and as the test is done internally by the drive (like the secure ATA erase) you don't have any way to know by looking at the screen how bad the surface looks like (or not).

While using MHDD you can just read some sectors on the start then at the middle then at the end to see if they read at all, and dispose of the drives that are "slow" or showing lots of delays, the s.m.a.r.t. will not display in a graphical way like victoria or mhhd at the present LBA what is the state of the block of sectors and the speed taken to access them.

Also a Write/Read test will test the pcb and sata/ide interface while an internal self scan test will not do so as the drive is doing it's tests internally.

At the end of the day it's just a question of how much time you will want to spend for each drive and if you are willing to refund or provide another drive to a client that claims the drive you sold them is defective i think the tolerance in testing would be lower, or you can just run a surface verify and check s.m.a.r.t. and if the drive have other issues that are found later by a user the drive can just be replaced.

I think that if i were in your position i would just do as stated on my first reply, mhdd scan + secure erase + mhdd scan and check s.m.a.r.t. but even so could take some considerable amout of time.

As for the question of how to be sure that the drive is ok i think that no-one can be sure of that no matter how much testing you do. It's just a question of testing more and more and with diferent set of tests and as much time you have to waste on that the more clear it will be if the drive is in a "good" condition or more close to a "perfect condition". You can't be sure or you can't reach that perfect state where you have 100% confidence that the drive is ok, but the more you test it with different tests the more close you are to have some confidence that the drive is in "good shape".

Yesterday i've taken a look to the HRT manual and the options to scan the drive for defects, as i normally only use MHDD and Victoria for that as already stated (ok, and spinrite) and for you to have an idea there are 2 major type of tests :

Pysical test - Is when you test the rive with VENDOR SPECIFIC COMMANDS that know the exact location of C/H/S at the drive level (this is not the same as logic C/H/S that was used to access sectors on older drives and those values would be set on bios). Here we are talking about real location at the platter level, bypassing the LBA translator. This type of test requires firmware tools and are specific to brand/model as it will use VSC and it can be found on the expensive firmware tools.
Advantage of testing this way is that you can find out very quickly if one head is not reading/writing or if there is a scratch on the platter preventing one entire track or zone of tracks to read/write, etc ...
To ilustrate this is the type of testing that you would be able to do with firmware tool and specific vendor commands for instance on AVV2 IBM/Hitachi drive :

Of course you don't want to use expensive tools so let's forget about the drive internal way to address sectors and let's go for a test that can be used with universal tools or tools that use only standard ATA commands to test any drive.

We will call that a logic test. Now you have very very old drives that can't be accessed by LBA (like the 40 MB IBMs or que Conner here : viewtopic.php?f=1&t=15709 )

You are going to test those at LOGIC C/H/S instead of LBA. As stated this is drives that will not respond to LBA addressing but the addressing type of C/H/S is a logic one, not internal. You are still using ATA standard (not vendor specific) and you are still requiring to tranlate the Logic C/H/S intro physical C/H/S (the drive does that itself of course). Those are really very old drives and i doubt you will have to test any of those, i'm just leaving this here in case you have some really small capacity drives you must be aware that you will need special procedure if they are too old to support LBA (let's say any drive with more then 1GB will most likely support LBA). In any case this is how you would test it in HRT at C/H/S LOGIC (again don't confuse with pysical C/H/S) :

: aa.png (11.94 KiB) Viewed 18963 times

You can't test more then 8 GB this way but you can test very old drives that don't support LBA without a PC-3000 ISA

Finaly you have LOGIC test in LBA and that is what you will use for all of your recent drives :

Now while there is the possibility to test sector by sector like this :

The majority of tools like MHDD or Victoria will do BLOCKS of 256 sectors at a time, instead of sector by sector, but that is good enought.

Just be aware that each of those blocks on the graph of victoria or mhdd will be 256 sectors and not a single sector, but at the end of the day what counts is that all the drive is tested anyhow.

Now for the more extensive tests (instead of a simple verify) you could use the following HRT method,

1 - Write Forward (it will write to all sectors) and on each sector it will write it's position as well.
2 - Scan forware (verify).
3 - Read + Check (this reads the sector itself, send the data back to the acrd and checks if the sector number (lba) matches the sector number that it was written on the Write Forward pass.
4 - Seek + Do will do a very cool thing. You will have to disable cache so that the test is reliable (and data is not fetched from cache instead), now the heads will be positioned far from the sectors to be read/written first and only then the action of read/write will be done. This will stress the drive as the heads will be moving all over the place (even if you are writing/reading sequentially) and it will look like the drive will be accessing the platter from random points as the heads will be moving all over the place. This tests the reliability of the drive and it will heat it up as it will simulate "extreme" activity like if you were doing a random read/write to random sectors (as the head will be seeking to/from all over the platter).

This of course take a huge amount of time, but again it's allways a matter of time versus how much do you need to be close of knowing how good the drive actually is.

Now HRT have a cool concept of "Defecto-Scope" as well, in LOGIC translation.

This requires HRT card and i don't know of any alternative software (freeware or not) that will do this but as it uses ATA standard commands there must be something simmilar to this out there.

Defecto-Scope tests a range of LBA (or entire drive by default at LOGIC translation but on "Steps".

First STEP it will act as MHDD/Victoria and will test BLOCKS of 256 sectors. This is fast way to test and you can use write or verify only. Very is very fast.
Now Defecto-Scope draws a graph as it's testing the drive and shows the access speed needed to verify (or write) each block, a little bit like MHDD and Victoria but it displays a graph for the entire drive while on MHDD you can only see some squares for a portion of the drive. Testing on my old IBM right now it will look like this when the scan finish and the drive is fully tested :

Now, if the graph goes up to MAX that means that you have bad sectors. The graph shows on horizontal line the LBA space and on vertical the time needed to access blocks. If it goes up to max (500) those can be considered bad sectors or very slow sectors. On my drive just a little portion of blocks are a little bit way higher then the rest of the drive blocks, and they are still fast enought not to be considered "BAD". Yet as defecto-scope ends you have STEP 2 !!!!

Those blocks that are "slower" but are not yet bad are tested again but this time at SECTOR LEVEL.

You define a CRITICAL TIME (or use a default) and the defecto-scope will just pick up the LBAs from the first test that were considered "SLOW BLOCKS" and will pick up those LBA ranges and test sector by sector using your criteria of CRITICAL TIME to consider a sector "good" or "slow".

Here is my drive :

Sectors are now tested one by one and on my example if it takes more then 20 ms to access a sector it will be placed on a list and you can simply re-locate them and swap by spares, or even better place them on P-List and re-format the drive if HRT supports that option for the model of the drive you are testing.

But all of this is as i stated, takes a long time and if you have 1000 or more drives you might not want to have 1 drive at a time consuming space for 1 day on a slot of your firmware tool, so you will have to be a little far away from the idea of "to be certain that a drive is ok" and more near the idea of "i can verify all the sectors so i should be ok" ....

HRT and VICTORIA also do a cool PCB and BUFFER test. It checks the quality of PCB interface lines to interface RAM as it uses write and read buffer commands to send data from the tool to the drive and check if the data on RAM of the drive is still the same or got damaged :
In case of problems you will see ssomething like this (HRT) :

: cc.png (30.15 KiB) Viewed 18963 times

Again VICTORIA for windows does the same test and with this you can have more confidence on PCB ram.

At the end of the day it's a question of how much time you want to invest to be "know more" about the real state of the drive and the more you are willing to test the more close you will be of knowing (at the expense of time) of how good the drive will behave (or not) when re-used on a system

Some very good and interesting info.

I've never used Self Scan on a questionable drive with success, because I don't really know how to best configure and use it.

But having opened thousands of drives and inspected heads under a microscope, there is nothing you can really do for drives with contaminated heads, and there are a LOT of drives with minor contamination spinning -- or not.

testing hard disks with some confidence.

testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.

Re: testing hard disks with some confidence.