Jump to content

Command Timeouts caused by Cables too close?

jeffreywhunter

By jeffreywhunter
in General Support

Recommended Posts

jeffreywhunter Collaborator

jeffreywhunter

Posted
Posted

I'm getting a very high command timeout value on one of my drives.  I've run preclear on the drive 3 or 4 times, no error.  Also ran smart diags, no reported problems.  Here's a pic...

 

http://my.jetscreenshot.com/12412/20150621-kof2-260kb.jpg

 

In trying to have a tidy interior of my system, I used heat shrink and cable mesh to combine cables in a neat bundle. 

 

Could it be possible that the alignment of the cables in this manner induce currents in the other cables?  No other drives reporting problems.

 

I've checked cable connections and they all look good.  Ideas?

Link to comment
RobJ Mentor

RobJ

Posted
Posted

Please ignore the Command Timeout count, as it's not useful.  I've proposed some changes to the SMART notifications, and part of that includes dropping the Command_Timeout count.  The current SMART notifications are the first version, and just need tweaking.

Link to comment
bonienl Veteran

bonienl

Posted
Posted

Please ignore the Command Timeout count, as it's not useful.  I've proposed some changes to the SMART notifications, and part of that includes dropping the Command_Timeout count.  The current SMART notifications are the first version, and just need tweaking.

 

As a temporary measure you can disable notifications related to command timeout (188).

 

Go  to settings -> notification settings -> click advanced view -> uncheck 188

 

Link to comment
Frank1940 Veteran

Frank1940

Posted
Posted

Routing SATA cables together is a bad idea.

 

Nonsense.

 

Look up LVDS.

They were engineered to handle it and have the appropriate shielding to allow it.  Sata cables and standard power cords do not.

 

Isn't the point that SATA does in fact use LVDS http://www.interfacebus.com/Design_Connector_Serial_ATA.html

I stand corrected.

 

Problem is that GOOD SATA cables are shielded and you can do pretty much what you want with them.  The cheap ones that most people use are not shielded.  They are susceptible to crosstalk and it is best not to cable them together.  (Shielded ones cost about 5X what the cheap unshielded ones do!  The cheap ones are flat and the shielded ones are rounded.) 

Link to comment
SSD Mentor

SSD

Posted
Posted

Never seen a round sata cable. Premium round IDE cables proved to be much worse than the cheap ribbon variety, so cost and cable quality are not always correlated. I do believe crosstalk can happen but have seen some mighty tidy case interiors with cables wrapped tighter than a hooker in spandex and run just fine, so it's way way down on the list of causes for drive issues for me. Only the CRC count is really related to cabling I believe, but command timeout is not an attribute I often see growing so it may also be impacted by cabling. Things like reallocation and pending sectors have nothing to do with cabling. Note that once most attributes are incremented, they will never reduce. So if you have some, even a lot, but they do not increase, it is probably an already corrected issue. If it continues, you can run a fresh cable to that drive and see if it stops increasing. If it keeps increasing even after that, so long as drive is performing and not seeing other issues, I'd not rush to replace it, but it is odd and I would continue to research possible causes.

Link to comment
Frank1940 Veteran

Frank1940

Posted
Posted

Never seen a round sata cable. Premium round IDE cables proved to be much worse than the cheap ribbon variety, so cost and cable quality are not always correlated. I do believe crosstalk can happen but have seen some mighty tidy case interiors with cables wrapped tighter than a hooker in spandex and run just fine, so it's way way down on the list of causes for drive issues for me. Only the CRC count is really related to cabling I believe, but command timeout is not an attribute I often see growing so it may also be impacted by cabling. Things like reallocation and pending sectors have nothing to do with cabling. Note that once most attributes are incremented, they will never reduce. So if you have some, even a lot, but they do not increase, it is probably an already corrected issue. If it continues, you can run a fresh cable to that drive and see if it stops increasing. If it keeps increasing even after that, so long as drive is performing and not seeing other issues, I'd not rush to replace it, but it is odd and I would continue to research possible causes.

 

This:

 

    http://www.monoprice.com/Product?c_id=102&cp_id=10226&cs_id=1022602&p_id=5121&seq=1&format=2

 

as compared to this:

 

  http://www.monoprice.com/Product?c_id=102&cp_id=10226&cs_id=1022602&p_id=8782&seq=1&format=2

Link to comment
garycase Grand Master

garycase

Posted
Posted

Crosstalk is far less of an issue in LVDS systems, but it IS still something that happens.  I can't find a specific study on crosstalk with SATA connections; but Texas Instruments did one for Cat-6 cables, and a paper was presented at EMC '09 in Kyoto that studied the issue in PCB's, which have a lot of LVDS interconnects.  The latter concluded "... that a PCB structure with parallel overhead traces in two adjacent layers must be absolutely avoided due to the high level of crosstalk."    I'd think similar caveats would apply to SATA cables tightly bound together.

 

[The full paper is here: http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&uact=8&ved=0CDoQFjAE&url=http%3A%2F%2Fwww.ieice.org%2Fproceedings%2FEMC09%2Fpdf%2F22Q1-1.pdf&ei=AyiIVfq4KIqayAT1uYPwDg&usg=AFQjCNFM_TBFJagFHRRDaU1CF6HG1T4UEA&bvm=bv.96339352,d.aWw ]

 

Not as detailed, but specific to SATA, is a brief discussion of SATA integrity issues here, which includes some discussion on crosstalk:  http://blog.asset-intertech.com/test_data_out/2014/06/common-sata-signal-integrity-issues.html

 

 

Bottom line:  Crosstalk is alive and well ... even in LVDS systems.  It's clearly not as bad as it once was with earlier signaling designs; but is still a factor.    I'd err on the side of caution and not bundle cables together.  Neat is okay; but tied together and "... wrapped tighter than a hooker in spandex  ..." is not  :)

 

 

Link to comment
jumperalex Rising Star

jumperalex

Posted
Posted

Good information indeed. I suppose the one saving grace at least with SATA cables is that even in spandex, the conductor distance (on the order of a mm or more) is a fair bit larger than between layer-to-layer PCB traces (small fractions of a mm). It could be made even more with some spacers at the points where the cables are wrapped. There is even a "trick" used by car guys for organizing spark plug wires to maintain good air gap between sparkplug wires.

 

0910st_10_+msd_super_conductor_wires+zip_ties.jpg

 

Still it IS something to keep in mind, and check for if you see problems. But a LOT of people run with reasonably bundled SATA wires (I know I do) without trouble. In my case they are tied at a few points to keep them all moving in the direction I want together, but they also have varying air gaps between the wraps (of which there are as few as possible), the cable are as short as possible, and they diverge as quickly as possible on the drive end.

Link to comment
Frank1940 Veteran

Frank1940

Posted
Posted

Well, I think this must put the entire conversation to bed.  I was looking at the LimeTech Unraid box they sell preconfigured.  Check out the gallery with the way they arrange Sata cables!

 

http://lime-technology.com/avs-104-server-gallery/

 

Must not be an issue eh?

 

BUT the pictures of the SATA data cables that I saw look like they are shielded!  Look at how they are not flat but the edges are rounded indicating that they are much thicker than the flat unshielded cables. 

Link to comment

Archived

This topic is now archived and is closed to further replies.

Go to topic listing
×
×
  • Create New...