Shield your sata cables!

[picbits]

At 12" it's all in your imagination.

That was just so tempting to put as my signature ...

 

[Frosty_47]

You made the claim, can do you have any hard data? Surely you're not the only one ever to notice this huge increase in speed?

I will be happy to provide some if you tell me the details of the "hard data".

If by hard data you mean simple proof showing distinc diference, than I can try and post a youtube video that will show unshielded vs. shielded cable performance based on data obtained in HD_Speed.

BTW, my system loads under 30 seconds now. So I don't even need lab equipment to tell the difference.

 

[Frosty_47]

Rip up photo of SATA cable is attached
if the shield strip is broken perhaps it might misbehave.Perhaps external shielding might have helped Frosty_47.

That's an interesting figure.

However I don't see how that shield will work unless it's somehow grounded...

I suspect the shield inside the cable is not grounded due to presence of plastic connectors. However I could be wrong in that the shield can be connected to a GND pin inside the cable.

 

[blueroomelectronics]

There is HD benchmarking software freely available why not post the results of both cables. Tell us the brand of shielded SATA and we can try to replicate the results.

My unshielded SATA based PC boots in about 30 seconds with a relic of a SATA drive. Also make sure you're running in DMA mode as a bad cable can force the OS to switch to a non DMA mode.

 

[blueroomelectronics]

This is an interesting article
Seagate Barracuda 1.5 Tb MOD | TechwareLabs

 

[zagam]

Standard SATA shielding is insufficient

The slow down is due to detected errors that are correctable. What about all the uncorrected errors causing file system corruption!

Internal SATA cables are bad and must be routed separately and only cross other cables at right angles. These are a big PC support issue. Very rarely do the actual SATA drives fail.

I originally found this out with a Sun X2100 server. I did the experiment and results were conclusive. It was the cables. It is still a problem with our user desktops. Our servers have since been switched over to FC.

See experiment here here. (I'm an underclocker and very aggressive with power management under GNU/Linux. Get some thermals in reserve by staying well below TPD so I can burst up on demand when the computer actually has work to do.)

Ideally cables should be well shielded and well grounded at both ends for potential equalisation so that signal at each end is within common mode range.

Serial Attached SCSI may be better at picking up these very frequent errors.

(This is similar to the UTP Ethernet problem. Use overall foil and ground at each end. Also run earths at least 6 mm² between frames along with the horizontals.)

 

[Hero999]

Old thread but who cares? It's still interesting.

I have an old computer which uses IDE. The cables aren't shielded.

I've recently upgraded from a 60GB Maxtor POS (which went bad) to a Western digital 120GB drive. Because my new dirve spins at 7200rpm it's faster, the speed increase was more noticeable than doubling the RAM from 256MB to 512MB and upgrading the graphics card.

I will some upgrade again, to my bother's old PC which is still newer than mine. At first, I'll move the hrad drive, then I'll consider upgrading to a solid state SATA drive.

 

[Sceadwian]

One thing that has never been addressed with solid state drives is the read/write cycle limits.

 

[Hero999]

Wear leveling has mitigated that to a large extent.
Wear levelling - Wikipedia, the free encyclopedia

From what I gather other issues such as asymmetric read/write times and lower bulk transfer speed are more of a problem.

 

[Sceadwian]

Thanks for the link Hero, I didn't realize that it was transparently implemented on a lot of solid state drives. Good to know, that's always been a nagging thought at the back of my mind.

 

[zagam]

Ide

Old thread but who cares? It's still interesting.

I have an old computer which uses IDE. The cables aren't shielded.

I've recently upgraded from a 60GB Maxtor POS (which went bad) to a Western digital 120GB drive. Because my new dirve spins at 7200rpm it's faster, the speed increase was more noticeable than doubling the RAM from 256MB to 512MB and upgrading the graphics card.

I will some upgrade again, to my bother's old PC which is still newer than mine. At first, I'll move the hrad drive, then I'll consider upgrading to a solid state SATA drive.

IDE did not need to be. There were some issues with the last 80 way ATA.
Though SATA is 3 Gbaud the actual data rate is 300 MB/s because of the encoding.

I call them SATA/300.

This is fast enough for SSDs. All they need to do is copy the eSATA cable specs.
I have rescued SATA drives destined for warranty or the skip only to find they were okay.
When the cables were sorted and the drives formatted again they were as good as new.

40 way IDE uses parallel logic level lines without ground.

The 80 way ATA introduced grounds, but these were commoned up on a few pins and caused probs at the highest speeds to UDMA 6 (133).

It works for UDMA 5 (100). The WD Scorpios I bought recently are only this speed.

The use of parallel logic lines could be very energy efficient if these were at modern logic levels such as LVDS.

Embedded systems still have Compact Flash Plus (which is basically a 1" UDMA SSD).
I presume because failures, power and cost are very low.

 

[Hero999]

I was wrong, my IDE cables are shielded.

Still as far as I'm aware, all SATA cables already have shielding inside and they wouldn't work without it so there's no need to add more shielding.

 

[zagam]

SATA shielding

I was wrong, my IDE cables are shielded.

80-way ATA have grounds for each signal line where as 40 way IDE have common grounds.

Still as far as I'm aware, all SATA cables already have shielding inside and they wouldn't work without it so there's no need to add more shielding.

Internal SATA cables have less shielding than that in external eSATA.

The additional shielding in eSATA is to limit interference of radio by the eSATA, but it also stops interference of the eSATA by RF.

The digital electronics in a PeeCee generates enormous amount of broad spectrum RF and will interfere with anything if it is not hardened.

The errors due to noise are unacceptably high in my opinion.

They only become usable with careful routing or by using better grade cables.

The problem is that you can get ones that meet the spec, but they do not work.