Momentus 7200.4 ST9500420AS - strange SMART

[michael chiklis]

Hello,
my customer gave me this drive with bad pcb.
I replaced the pcb and did rom swap, all data are recovered but i think smart attributes are not good (or at list contradictory).
Hd Tune writes "warning", R-studio say "good".

Someone of you can tell me if there is really a problem?
Thanks.

[seed.helper]

could you please use hard disk sentinel professional trial ?

[michael chiklis]

HD Doctor SMART

[michael chiklis]

Thank you Spildit, i will tell to customer to not trust anymore this drive.

[fzabkar]

"Airflow Temperature" is a misnomer. In fact the attribute is reporting the same values as the Temperature attribute (C2). "Airflow Temperature" should be more appropriately termed "100 - Temperature". All it does is convert the values of the Temperature attribute so that they can be viewed in the same way as a normalised health score.

For example, a value of 66 for Airflow Temperature corresponds to an actual temperature value of 34C (= 100 - 66).

HD Tune is complaining because at one time the temperature reached 60C (= 100 - 40). The threshold value of 45 corresponds to 55C.

I believe the reason for the comma in the raw value of the "HD Doctor SMART" report is that it is being used as a separator for two data values. In fact the author has misinterpreted this particular attribute, and the raw values are incorrect. AIUI, the numbers "33,8481" actualy reflect 3 temperature values, namely 33C, 33C, and 33C. That's because 8481 = 0x2121, and this is broken down into 0x21 and 0x21. These temperatures would probably be the current, minimum, and maximum values for the current power cycle.

Notice also that the Load Cycle Count is displayed as "25479,4" whereas R-Studio displays it as 0x46359. HD Doctor appears to be interpreting this as two separate values, namely 0x4 and 0x6359 (= 25433). That's just stupid.

The Raw Read Error Rate, Seek Error Rate, and Hardware ECC Recovered attributes are also incorrectly interpreted. For example, the lower 32 bits of the SER are actually a lifetime seek count whereas the upper 16 bits are a lifetime seek error count.

HD Tune reports the SER as 17296133767.

17296133767 = 0x000406EE0E87

HD Doctor appears to be ignoring the uppermost 16 bits. Instead it displays ...

63354, 1776

In other words, HD Doctor breaks down the raw value of 0x000406EE0E87 into 0x06EE (= 1774) and 0x0E87. It ignores the leading 0x0004. Once again, that's stupid.

In short, "HD Doctor" is one of many tools that insists on interpreting the raw SMART data rather than merely displaying it, and it does so incorrectly.

Here is my attempt to understand the meaning of the RRER, SER, and HER attributes:
http://www.users.on.net/~fzabkar/HDD/Se ... R_HEC.html

[michael chiklis]

I checked defects lists of this drive in terminal with commands V4, V40, V80.

I found only one reported sector in NRG list (V40 command).
I did also a surface scan in MHDD, there are any bad sector (only 2 slow sectors, green square).

I think all smart are wrong about "Reallocated Event Count" but it might be right about high temperature, what do you think?
Maybe attribute "Reallocated Event Count" increased due to the pcb fault?

Because i can't see any sector issue unless 1 reported in NRG list

[fzabkar]

R-Studio is reporting the temperature as 0x02952F1A0022.

I suspect that this raw value is structured as follows:

0x02952F1A0022 --> 0x0295 0x2F 0x1A 0x0022

I believe that this is telling us that the max/min/current temperatures for the current power cycle are 47C (= 0x2F), 26C (= 0x1A), and 34C (= 0x22).

HD Doctor appears to report this attribute as 34,12058 (= 0x22, 0x2F1A).

I don't know what the leading value of 0x0295 means, but at a wild guess I'm wondering whether it could reflect the number of times that the temperature has exceeded the threshold.

The Reallocation Event Count reported by R-Studio is 0x28E000001A95. I suspect that the lower 16 or 32 bits may be the actual event count, but once again I don't know what the uppermost 16 bits mean. AISI R-Studio appears to be reporting 6805 reallocation events, whereas HD Doctor is reporting 6806. Therefore the drive is recording some activity in this attribute at the present time, not just in the past. As for how "reallocation events" relate to the various defect lists, I can only imagine that all the pending defects have been cleared by subsequent retesting of the affected sectors during the reallocation process. AIUI, if a pending sector passes the test, it is returned to service, otherwise its LBA is reallocated to a spare.

BTW, most users would be very disturbed at the very high Load/Unload Cycle Count.

One other thing to note is that current Seagate models sometimes use all 64 bits of the raw value, not just the lower 48 bits. IIRC, HDDScan reports all 64 bits. For example, I believe that bits 49 - 63 of the Power On Hours attribute record the number of seconds between each increment in the hours count. That's why the raw value appears to be ridiculously high.