IDE cables

[cw823]

I know the answer.....but I"ll ask the question to clarify things in my mind. The 80 pin cable, as opposed to the 40pin (IDE, sorry) cable....what are the extra 40 wires.

[Super]

They're supposed to lower the EMI.

[Wa11y]

Originally posted by Super
They're supposed to lower the EMI.

Nope. Thicker insulation would be sufficent for that. Extra wires would actually cause more EMI due to attenuation. When you bundle a whole bunch of wires together, they bleed across. I see that sometimes when our customers try to neaten up their phone closet, and just bundle all the cables they can together and wrap them with a wire tie. Signal starts bleeding over, and T1s start getting slips and frame errors, you get crosstalk on telephones, all sorts of problem.

The reason that ATA/66+ uses 80 wires versus 40 wires that ATA/33 uses is for a transmit lead, and a recieve lead. Very similar to how DDR memory can now transmit on the top of the clock cycle and recieve on the bottom, ATA/66+ has seperate transmit and recieve leads, so there's less colision, allowing for faster access times.

[Super]

Originally posted by Wa11y


Nope. Thicker insulation would be sufficent for that. Extra wires would actually cause more EMI due to attenuation. When you bundle a whole bunch of wires together, they bleed across. I see that sometimes when our customers try to neaten up their phone closet, and just bundle all the cables they can together and wrap them with a wire tie. Signal starts bleeding over, and T1s start getting slips and frame errors, you get crosstalk on telephones, all sorts of problem.

The reason that ATA/66+ uses 80 wires versus 40 wires that ATA/33 uses is for a transmit lead, and a recieve lead. Very similar to how DDR memory can now transmit on the top of the clock cycle and recieve on the bottom, ATA/66+ has seperate transmit and recieve leads, so there's less colision, allowing for faster access times.

hm... okay.
note to self: don't really trust A+ book

[Wa11y]

Originally posted by Super
hm... okay.
note to self: don't really trust A+ book

Yeah, I learned that too. I'll trust it if it's going on about stuff I don't know already. But then I try to look it up elsewhere to verify it.

Speaking of which, I'll try to russle up some info about that, just to prove my point. Not to say you think I'm lying, but it's easier to believe if I say, "Read about it here." versus "It just does!"

[Wa11y]

Grrr...CRAP!

Everything I'm finding says it uses the extra 40 wires to add grounding. Here's a ton of info on it.
http://www.pcguide.com/ref/hdd/if/ide/std.htm
http://www.pcguide.com/ref/hdd/if/ide/modes_UDMA.htm
http://www.pcguide.com/ref/hdd/if/ide/conf_Cable80.htm

However, the second link does agree with me about transmitting on both the rising and falling edge of the clock cycle, just like DDR memory does.

And that's not the only webpage I consulted about the issue. I have about 7 browser windows open about EIDE cables. I just felt that the pcguide.com pages were some of the best and most technically written, so it was the one I chose to use.

My only problem is, adding wires to prevent noise flies in the face of all wiring logic I have ever heard. I've worked with our engineers, some with upwards to 30 years on our systems, including time as field techs, which includes having to deal with wiring. I would think that adding more insulation, and not wiring would help with preventing crosstalk.

I now have a crusade. Excuse me while I obsess. Anyone who wants to help my is welcome aboard my Crazy Train.

[mrpcman]

i was going to say. i thought they were for grounding and whatnot.

[Wa11y]

Originally posted by cw823
I know the answer.....but I"ll ask the question to clarify things in my mind. The 80 pin cable, as opposed to the 40pin (IDE, sorry) cable....what are the extra 40 wires.

Just out of curiosity, what was your answer?

[Super]

Wa11y, your answer actually makes more sense to me than the book, so I'll agree. I guess the best and most accurate answer we can find, will have to be from an Engineer...

[Aedan]

Originally posted by Wa11y
My only problem is, adding wires to prevent noise flies in the face of all wiring logic I have ever heard. I've worked with our engineers, some with upwards to 30 years on our systems, including time as field techs, which includes having to deal with wiring. I would think that adding more insulation, and not wiring would help with preventing crosstalk.

Oh, adding ground wires helps avoid crosstalk all right. It's important to remember that they're NOT just bundled together. The ordering of the cable here is IMPORTANT! Each signal wire is partnered next door with a ground wires. Hence the order of the wires goes signal, ground, signal, ground, signal, ground ETC.

There's several mechanisms for crosstalk to occur. There's capacitive effects, inductive effects and radiated RF.

The capacitive effect means the wires either side of a wire are capacitively coupled. A rising voltage on the middle wire tends to induce a small rise in the wires either side. Adding a ground wire between the signal wires means that each wire is affecting the ground wire, which doesn't matter.

The inductive effect does pretty much the same thing but through inductive effects. Again, with the ground wire, this limits how well induction occurs.

Lastly, radiated RF - This is much more of a problem at 133MHz than at 33MHz. Each wire is an omnidirectional antenna. Putting a groundwire right next to it means that RF being thrown out sideways is absorbed by the ground wire rather than being transmitted into the next signal wire. The ground wires effectively notch out the RF emission either side of the transmitting wire.

Remeber, when you have customers making their wiring nice and neat, they're not ordering their cables nicely with ground wires. If they were, then they'd not suffer crosstalk either.

AidanII

[SteveI]

Aidan,
I use rounded ATA100 cables, does that have a negative impact ?

[Wa11y]

Originally posted by Super
Wa11y, your answer actually makes more sense to me than the book, so I'll agree. I guess the best and most accurate answer we can find, will have to be from an Engineer...

Bear in mind, these are telephone system engineers, not computer engineers. My engineers don't work with ribbon cable.

Since all the evidence points to me being wrong, I may have to accept that. AidanII makes some good points as well.

Honestly, at this point, I'm kinda befuddled, but leaning towards the grounding thing.

[Aedan]

Originally posted by SteveI
Aidan, I use rounded ATA100 cables, does that have a negative impact ?

In a word - possibly! It depends on how good the cable is, and how it's rounded. Once you get to questions like this, the answer is generally "suck it and see"!

People who have done ad-hoc tests claim there is no difference.

AidanII

[Aedan]

Originally posted by Wa11y


Bear in mind, these are telephone system engineers, not computer engineers. My engineers don't work with ribbon cable.

Since all the evidence points to me being wrong, I may have to accept that. AidanII makes some good points as well.

Honestly, at this point, I'm kinda befuddled, but leaning towards the grounding thing.

The grounding is very important. If they were left unconnected they would not be as effective.

The fact that it is ribbon cable and the fact that the order matters make it very different from the usual telephone cabling, where the multicore is arranged in twisted pair fashion with relatively little attention paid to the ordering.

My suspicion is that it's a little overkill just to keep system manufacturers out of trouble when they get too close to the limits with El cheapo cable. Not having anything over ATA33 (and not using IDE!) makes it difficult to test what would happen if you got a 40pin cable operating at ATA100.

AidanII

[SteveI]

Originally posted by AidanII


In a word - possibly! It depends on how good the cable is, and how it's rounded. Once you get to questions like this, the answer is generally "suck it and see"!

People who have done ad-hoc tests claim there is no difference.

AidanII

Just to let you know, I ask out of curiousity. I use 2 rounded cables (store bought) and haven't had problems that I know of. I suspected data integrity problems a while back, but I blame high FSB for that. I dropped FSB to 142.5 for the last month, and not a hiccup.

As always, thanks for your insight.

[mrpcman]

if you cut your cable up to round it wouldnt you increase interference then?

[Jazz]

It all depends. If the signal is powerfull enough yea you can get crosstalk between leads. But i suspect that the signal is so low voltage that crosstalk is not much of a issue. More like a bit of prevention is valualble. And i dont think they are grounded together so its pretty much pointless. I had one of the cables in my old case get rubbed bare and i tested the bare wires and i found only one or two went to ground.

[Aedan]

The ATA100/133 spec calls for those extra cables to be grounded. If your cables haven't been grounded then the cables/connectors don't meet the ATA100/133 spec! (Or have been damaged)

AidanII

[Wa11y]

Originally posted by AidanII
The grounding is very important. If they were left unconnected they would not be as effective.

AidanII

But where are they grounded at? I understand they're connected to the connector, but what does it connect to beyond that? Just having it terminate in the connector doesn't make it grounded, as far as I can tell. It would have to connecto to something else to ground it, like a ground lead.

[Jazz]

Well i think one of the pins in the ide port is ground. On both ends it should be grounded. the drive can use either the chassis or the molex power ground.