MHDD Stripes - Samsung HN-M750MBB

[rguy]

I currently have a brand new (made November 2011, in China) Samsung HN-M750MBB hard drive - SATA, 2.5 inch, 750 GB.

After testing the drive out in a USB enclosure, I got a little concerned when the SMART was reporting a 1 for the write error rate. I put it into a computer properly and ran MHDD.

MHDD has reported a fair mix of <3ms blocks and <10ms blocks with occasional <50ms blocks. The number of blocks that are <10ms are roughly 10% of the number of <3ms blocks.

However, what is concerning, really, is the pattern. Although in some sections of the drive, a pattern not entirely unlike a chess board does show up, in others, there are "streaks," or "stripes" - in some a single stripe just goes straight down for several rows. In others, they are "diagonal" (and a bit broken up), pointing off to the lower left or right. The blocks making up the lines are <10ms blocks amidst <3ms blocks, with occasional <50ms blocks in there.

There are also occasionally large, thick stripes of nothing but <3ms blocks in there (pointing towards the lower right) amidst these stripes.

The machine it is in for testing with MHDD is a Dell Inspiron 1720, with an Intel Core 2 Duo processor, 6 GB of RAM and an nVidia 8600GT circa summer of 2008.

In short, should I return this drive or not? I'd include a screen shot but the resources I have available right now make that prohibitive. I am, however, running out of time to return it, and this Samsung may be one of the few new Samsungs left before Seagate ate them. Any thoughts would be greatly appreciated!

[labtech]

How long has this drive been in use?

What are the values for the reallocated and the pending sector entries?

[rguy]

It's brand new. It has less than 200 hours of running time (don't recall the exact number), but it's been out of the package less than two weeks.

It also has 0 sectors pending and 0 reallocated. The only SMART error it has that I've seen thus far is the write rate error ticked up by 1 - though it's possible that that's due to using it in an enclosure for some of the testing (although the enclosure is good, I suppose I could have bumped the USB cable or something). That said, I can't recall if such a thing may have happened or not.

[rguy]

As a further update, one scan showed:

<3ms blocks: 4961184
<10ms blocks: 784178
<50ms blocks: 321
<500ms blocks: 1

Another scan showed NO <500 ms blocks.

The ratio of <10ms blocks to <3ms blocks increased the further it went on the scan, and the average distance between <10ms blocks (with the gaps being made of <3ms blocks) decreased. Speed roughly halved, coming in at around 90 MB/sec at the beginning and somewhere around 40 MB/sec at the end.

The software I've used has been MHDD, Passmark DiskCheck, HDDScan (which reported some groups of blocks - it wasn't too specific - taking anywhere from a bit less than 100 ms to nearly 6000 ms to respond), HDDTune, HD Sentinel, chkdsk, and smarttools for Win32.

Passmark Diskcheck DID detect a drop of the Raw Read Error value from 252 to 100 VERY early in the drive's life, though other (non-Samsung) "healthy" drives I've worked with seem to have that for this value, and it seems to have happened around the time a few other indicators repeated that exact pattern, so I'm thinking it was the result of either the very first scan I ran, or the format I ran on the drive early on. Almost seemed like initialization behavior of the SMART system.

It DID run several degrees hotter than a Hitachi I was also testing (which I am taking back due to a reallocated sector) and an internal WD drive I got readings on at the same time. The enclosure it was in was one of those "toaster" type enclosures, wherein you slip in the drive on the top like a video game cartridge. The enclosure has worked well for me for several years so I don't think there's a problem with it.

[labtech]

Well, it is a relatively healthy drive.

If you were to try to RMA this drive, you would have a hard time proving the drive is faulty.

I would say keep using it, but make sure you have back up.