Rounded IDE cables are bad?

[Ichigo]

According to one of the stickies, they don't follow IDE specifications or something? Any truth to that?

 

[UsandThem]

Originally posted by: Ichigo
According to one of the stickies, they don't follow IDE specifications or something? Any truth to that?

Never had a problem with them. Most higher end motherboards come with rounded cables now.

 

[Shadowknight]

Yeah, the don't fit spec. According to Ralf Hutter at SPCR they cause data corruption, and he's seen this on several customers computers he's worked on that used round cables. The drive drops the transfer rate down a notch and resends the corrupted data, but Windows doesn't tell you this. So without your knowledge your IDE channel will drop all the way to PIO, which is significantly slower than DMA5. So you have to manually check to see if this has happened then delete the IDE channels from your hardware window and have Windows reinstall them to get the drive back to DMA5. You can't just change the setting without deleting the channels for some reason.

 

[MichaelD]

Technically speaking; no, they don't follow the IDE specification.

Realistically speaking; I've never had a problem with them. Now, I've had cables go bad from too much use/bending and stuff. But those have been both regular flat AND rounded IDE cables.

Use the rounded cables if you want and don't worry about it.

 

[avi85]

ATA133 Specifications are a max cable length of 18"
most rounded IDE cables are 24" long although you can definitely fnd 18" ones.
other than the length I don't think there are any other problems.

 

[Ken90630]

Reportedly one of the probs with rounded HD cables is that with the 80-wire type, they don't have ground cables right next to their signal cable counterparts the way that flat ribbon IDE cables do. As you prolly know, each of the 40 signal wires is supposed to have a ground wire next to it to prevent crosstalk. I've heard that rounded IDE hard drive cables don't provide for that.

I also read an interview awhile back in CPU magazine, with the founder & chief guru of Voodoo PC (whose name escapes me at the moment), and he also said his company had found some sort of problem with them. I'm sorry I don't remember the specifics, however.

My guess is they're prolly fine for FDDs, but I only use flat ribbon IDE cables for my HDs. Obviously if you have serial HDs, it's irrelevant.

 

[JM Aggie08]

you can't expect something bundled up to work properly.

 

[computer]

Other than the length problem of being longer than 18" some others have pointed out, I have also heard about data corruption with them. I used some a few years back and had problems with them, but that could have changed by now. I don't take any chances using them though, I use flat IDE cables then round them myself with split-loom tubing.

 

[Pens1566]

Some are harder to route/hide for cable managment. I tend to like IDE ribbon cables that have been folded/rolled and then sleeved. DFI MBs come with these.

 

[jonnyGURU]

The problem is with the lack of shield wire in between each wire as there is in the case of a typical 80-conductor, 40-pin IDE cable.

Although I have heard some round cables are ATA133 complaint, I'm not sure how they do that. Better shielding on the wires themselves?

 

[Jeff7]

Originally posted by: JM Aggie08
you can't expect something bundled up to work properly.

Like, perhaps, twisted-pair CAT5 cable?

 

[RallyMaster]

Haven't exactly had problems with my cheapo Directron cables myself...but I only use it for my DVD burner...so it doesn't matter that much to me. I'm using a flat one for HDD though.

 

[Johnniewalker]

Originally posted by: Jeff7

Originally posted by: JM Aggie08
you can't expect something bundled up to work properly.

Like, perhaps, twisted-pair CAT5 cable?

Or, perhaps like the ones holding up a certain bridge between san fran and oakland

 

[Fallen Kell]

Originally posted by: Ken90630
Reportedly one of the probs with rounded HD cables is that with the 80-wire type, they don't have ground cables right next to their signal cable counterparts the way that flat ribbon IDE cables do. As you prolly know, each of the 40 signal wires is supposed to have a ground wire next to it to prevent crosstalk. I've heard that rounded IDE hard drive cables don't provide for that.

I also read an interview awhile back in CPU magazine, with the founder & chief guru of Voodoo PC (whose name escapes me at the moment), and he also said his company had found some sort of problem with them. I'm sorry I don't remember the specifics, however.

My guess is they're prolly fine for FDDs, but I only use flat ribbon IDE cables for my HDs. Obviously if you have serial HDs, it's irrelevant.

There are ways around the issue of crosstalk. First way is use better insulation/semi-conductor around the individual wires, TFE (teflon) is actually one of the best insulators in existance. The second important thing is to make sure that the wires are properly weaved/twisted. There is a reason why network wires are twisted the way they are, it is so the electromagnetic fileld generated by the electrical current moving along the wires is always oriented 90degs out of phase with the closest wires, which effectively makes the interference incapable of affecting the data on those closest wires (people who have any electrical physics background will know of the right-hand-rule and the forces of how the fields/current will interact and will see that the vectors will be wrong for a proper interaction to occur).

Again, there are ways around this, but you need to correct cables. Yes, they do not follow spec simply because the spec is specifically for a flat cable. But in reality, the spec should be for the overall tolerances on signal to noise, delay, and integrity of the data, not on the physical layout of the cable itself, which is what the IDE spec does (it say cable will be flat with 80 pins, with data line alternated next to ground wires, etc.,), but it shouldn't say anything other then what the pin-in/pin-outs on the cable in terms of physical spec, the rest should just be the tolorance levels of the signal.

 

[GalvanizedYankee]

One would think that if there were an issue with round IDE cabling, the web would
be filled with complaints



...Galvanized

 

[Ken90630]

Originally posted by: GalvanizedYankee
One would think that if there were an issue with round IDE cabling, the web would
be filled with complaints

...Galvanized

Only if those afflicted actually knew that the cause of their probs was in fact the round cabling.

For the heck of it, I decided to dig up that old issue of CPU magazine I mentioned before. In the March 2004 issue, page 108, there's a one-page interview with Voodoo PC's President and Chief Technology Officer Rahul Sood. Towards the end of the interview he states:

"Another thing is that I don't agree with rounded cables. I never have. SATA cables are great, of course, but rounded cables are a different story because there is potential for noise. Any benchmarks that I've run on any of the rounded cables that we've tested show either errors generating over time or they're slower than good quality, flat IDE cables."

So take that for whatever it's worth. I can't think of any reason for him to lie, or have any ulterior motive for disliking round cables -- after all, why would he care? -- so if he says their benchmarks show errors or slower speeds, I'm inclined to take him at his word. And obviously the "noise" he refers to would be crosstalk.

Having said this, Fallen Kell brought up an interesting point about insulation. Theoretically, I'd guess, if the insulation is good enough, that could mitigate the need for the ground wires in flat IDE hard drive cables. Anyone know of any high quality round cables that use the kind of better insulation (Teflon or comparable) Kell talks about?

 

[computer]

"Another thing is that I don't agree with rounded cables. I never have. SATA cables are great, of course, but rounded cables are a different story because there is potential for noise. Any benchmarks that I've run on any of the rounded cables that we've tested show either errors generating over time or they're slower than good quality, flat IDE cables."

I saw the exact same thing, that post reminded of it. I did some HD benchmarks with flat and round IDE cables and the rounded cables showed lower results. I think some website needs to test a load of rounded cables against each other (AND against flat cables) and see what they come up with.

 

[computer]

Originally posted by: Fallen Kell

Originally posted by: Ken90630
Reportedly one of the probs with rounded HD cables is that with the 80-wire type, they don't have ground cables right next to their signal cable counterparts the way that flat ribbon IDE cables do. As you prolly know, each of the 40 signal wires is supposed to have a ground wire next to it to prevent crosstalk. I've heard that rounded IDE hard drive cables don't provide for that.

I also read an interview awhile back in CPU magazine, with the founder & chief guru of Voodoo PC (whose name escapes me at the moment), and he also said his company had found some sort of problem with them. I'm sorry I don't remember the specifics, however.

My guess is they're prolly fine for FDDs, but I only use flat ribbon IDE cables for my HDs. Obviously if you have serial HDs, it's irrelevant.

There are ways around the issue of crosstalk. First way is use better insulation/semi-conductor around the individual wires, TFE (teflon) is actually one of the best insulators in existance. The second important thing is to make sure that the wires are properly weaved/twisted. There is a reason why network wires are twisted the way they are, it is so the electromagnetic fileld generated by the electrical current moving along the wires is always oriented 90degs out of phase with the closest wires, which effectively makes the interference incapable of affecting the data on those closest wires (people who have any electrical physics background will know of the right-hand-rule and the forces of how the fields/current will interact and will see that the vectors will be wrong for a proper interaction to occur).

Again, there are ways around this, but you need to correct cables. Yes, they do not follow spec simply because the spec is specifically for a flat cable. But in reality, the spec should be for the overall tolerances on signal to noise, delay, and integrity of the data, not on the physical layout of the cable itself, which is what the IDE spec does (it say cable will be flat with 80 pins, with data line alternated next to ground wires, etc.,), but it shouldn't say anything other then what the pin-in/pin-outs on the cable in terms of physical spec, the rest should just be the tolorance levels of the signal.

Very interesting. What do you think would happen if one took a flat IDE cable and instead of just rounding them as I do, twisted the cable before rounding it with tubing? I wonder if that would help/hurt anything?

 

[avi85]

Originally posted by: computer
...I think some website needs to test a load of rounded cables against each other (AND against flat cables) and see what they come up with.

I definitely second this, maybe someone should send an email to THG or something I'd bet that they would be interested in it (they seem to have a lot of time for benchmarking everything that just might be remotely connected to technology).

EDIT: They should also see how this affect optical drives cause if I'm not mistaken there are less crosstalk issues with ATA66 (but correct me if I'm wrong)

 

[Ken90630]

Originally posted by: avi85

Originally posted by: computer
...I think some website needs to test a load of rounded cables against each other (AND against flat cables) and see what they come up with.

I definitely second this, maybe someone should send an email to THG or something I'd bet that they would be interested in it (they seem to have a lot of time for benchmarking everything that just might be remotely connected to technology).

EDIT: They should also see how this affect optical drives cause if I'm not mistaken there are less crosstalk issues with ATA66 (but correct me if I'm wrong)

Yeah, I'll 'third' that. I don't recall ever seeing any test reports or benchmarks comparing rounded IDE cables vs. flat ones. Would be interesting.

Maybe Anandtech.com could do a comprehensive test report on rounded cables vs. flat ones. Hint, hint. Anyone watching? Mods? Anand? Anyone at all?

 

[TrevorRC]

If I get bored sometime I might look into doing it.

No promises, though

--Trevor

 

[DidlySquat]

rofl @ IDE

 

[JEDIYoda]

Concerning rounded cables I hear all these stories about rounded cables benchmarking slower than ribbon cables.
I also heard horror stoires about data being corrupted.

There are a few things you need to remember...
Unlike stereo wires where its been a proven fact that monster cables($$$) are not worth the money when compared to a good set of regular stereo cables.

When dealing with rounded cables if you buy the cheapest rounded cables you can find just like buying the cheapest PSU - IMO you are asking for trouble.

I use rounded cables on both my riggs and have run tests on my gaming rigg with ribbon versus rounded sad to say-- no discernable difference in anything!

 

[avi85]

Originally posted by: JEDIYoda
Concerning rounded cables I hear all these sroties about rounded cables benchmarking slower than ribbon cables.
I also heard horror stoires about data being corrupted.

There are a few things you need to remember...
Unlike stereo wires where its been a proven fact that monster cables($$$) are not worth the money when compared to a good set of regular stereo cables.

When dealing with rounded cables if you buy the cheapest rounded cables you can find just like buying the cheapest PSU - IMO you are asking for trouble.

I use rounded cables on both my riggs and have run tests on my gaming rigg with ribbon versus rounded sad to say-- no discernable difference in anything!

So which companies do you recommend?

 

[imported_Lucent]

Antec makes a premium rounded cable.