Rounding IDE Cables

Vince wrote
Rod Speed wrote
Vince wrote
Arno wrote
mike wrote

Rounding IDE cables this looks as if it could cause
electrical problems on the PATA cables. Is it really
as safe as he suggests it is?

http://cpu-central.com/Articles.asp?...9&decor_int=27

Depends. With UDMA 3 or before you can get data corruption.
I had this with a burner that did not support data checksums
on UDMA 3. Most HDDs do suuport these checksums, but
the standard does not require it before UDMA 4. For all
UDMA levels you can get command corruption, checksums
on commands are not present before SATA.

It also depends on lenght. My experiences are (If I remember
correctly): 30cm - works, 45cm - data corruption with the
burner, 60cm - command problems with HDDs, 90cm - basically
unusable with HDDs being dropped by the kernel within
minutes. Cable quality can influence that in both directions.

The home made mod doesn't seem to have as much protection from intereference.

True, but neither did the original 40 wire cables either.

Ready made rounded PATA/IDE cables arrange the wires (signal and ground) as twisted pairs.

Some do, some dont.

Don't know how much difference this makes.

Not much.

My impression is that the 80-way PATA cables were a big improvement on the old 40-way cables.

Only because the controllers refused to use the faster modes unless an 80 wire cable is used.

Does anyone know of any data or tests which shows how much improvement they gave?

It wouldnt be that hard to test, just make up a 40 wire cable that
pretends to be 80 wire as far as the controller is concerned and
monitor the SMART error data. Dunno if anyone has bothered.

In this thread Igor Batinic says twisted pair is quite an improvement
over untwisted. Does anyone know of any tests or comparisons for this?

There have been plenty on the general concept. Presumably
someone has done that with the round IDE cables.

Igor Batinic wrote:


A plenty of that, also Google a little bit. Therefore, whenever you have
any kind of high-speed copper connection, twisted pair must be used to
reduce the noise in the cable. You can find, also, twisted ribbon cable
(pretty standard cable in SCSI).

Best regards,

Iggy

"twisted pair must be used" isn't totally accurate.
TP technology has been the design of choice for over a century, but it
doesn't lend itself to two constraints on cabling for computers.. cheap
and compact.

The sig-gnd-sig-gnd layout of an 80 wire EIDE cable does work well.

I've seen/used twisted-pair ribbon cables, but I didn't have to buy them.

Hi!

nobody wrote:
Igor Batinic wrote:

A plenty of that, also Google a little bit. Therefore, whenever you
have any kind of high-speed copper connection, twisted pair must be
used to reduce the noise in the cable. You can find, also, twisted
ribbon cable (pretty standard cable in SCSI).

"twisted pair must be used" isn't totally accurate.

It was more a figure of speech than a technical law. )

But, of course, it is in specific areas.

TP technology has been the design of choice for over a century, but it
doesn't lend itself to two constraints on cabling for computers.. cheap
and compact.

The sig-gnd-sig-gnd layout of an 80 wire EIDE cable does work well.

Of course it does, if you keep it the way it is supposed to.

Which means, if you don't cut it and try to create "rounded" cable of
it. Then you can expect some problems. Therefore good rounded cables
have completely different cable schematics.

I've seen/used twisted-pair ribbon cables, but I didn't have to buy them.

It was almost a non-written standard in all latest SCSI implementations.
Of course you will not need it for ATA implementation (and I doubt you
can find something like that).

For instance, IBM FRU PN 37L5558:

http://www.aykat.com/ebay/kabel/kabe...5x_68pin_3.jpg

With best regards,

Iggy

On 13:40 12 Nov 2009, Igor Batinic wrote:

Hi!

nobody wrote:
Igor Batinic wrote:

A plenty of that, also Google a little bit. Therefore,
whenever you have any kind of high-speed copper connection,
twisted pair must be used to reduce the noise in the cable.
You can find, also, twisted ribbon cable (pretty standard
cable in SCSI).

"twisted pair must be used" isn't totally accurate.

It was more a figure of speech than a technical law. )

But, of course, it is in specific areas.

TP technology has been the design of choice for over a
century, but it doesn't lend itself to two constraints on
cabling for computers.. cheap and compact.

The sig-gnd-sig-gnd layout of an 80 wire EIDE cable does work
well.

Of course it does, if you keep it the way it is supposed to.

Which means, if you don't cut it and try to create "rounded"
cable of it. Then you can expect some problems. Therefore good
rounded cables have completely different cable schematics.

I've seen/used twisted-pair ribbon cables, but I didn't have
to buy them.

It was almost a non-written standard in all latest SCSI
implementations. Of course you will not need it for ATA
implementation (and I doubt you can find something like that).

For instance, IBM FRU PN 37L5558:

http://www.aykat.com/ebay/kabel/kabel_scsi_5x_68pin/
kabel_scsi_5x_68pin_3.jpg

That look good.

Presumably the twisting doesn't create extra problems so (apart
form cost) why aren't PATA cables like that?

Tri Cutter wrote:
On 13:40 12 Nov 2009, Igor Batinic wrote:

Hi!

nobody wrote:
Igor Batinic wrote:

A plenty of that, also Google a little bit. Therefore,
whenever you have any kind of high-speed copper connection,
twisted pair must be used to reduce the noise in the cable.
You can find, also, twisted ribbon cable (pretty standard
cable in SCSI).

"twisted pair must be used" isn't totally accurate.

It was more a figure of speech than a technical law. )

But, of course, it is in specific areas.

TP technology has been the design of choice for over a
century, but it doesn't lend itself to two constraints on
cabling for computers.. cheap and compact.

The sig-gnd-sig-gnd layout of an 80 wire EIDE cable does work
well.

Of course it does, if you keep it the way it is supposed to.

Which means, if you don't cut it and try to create "rounded"
cable of it. Then you can expect some problems. Therefore good
rounded cables have completely different cable schematics.

I've seen/used twisted-pair ribbon cables, but I didn't have
to buy them.

It was almost a non-written standard in all latest SCSI
implementations. Of course you will not need it for ATA
implementation (and I doubt you can find something like that).

For instance, IBM FRU PN 37L5558:

http://www.aykat.com/ebay/kabel/kabel_scsi_5x_68pin/
kabel_scsi_5x_68pin_3.jpg

That look good.

Presumably the twisting doesn't create extra problems so (apart
form cost) why aren't PATA cables like that?

Essentially because the traditional 80 wire cable is good enough.

They did go for a much smaller serial cable with SATA.

On Sat, 14 Nov 2009 04:08:13 +1100, "Rod Speed" wrote:

Tri Cutter wrote:
On 13:40 12 Nov 2009, Igor Batinic wrote:

Hi!

nobody wrote:
Igor Batinic wrote:

A plenty of that, also Google a little bit. Therefore,
whenever you have any kind of high-speed copper connection,
twisted pair must be used to reduce the noise in the cable.
You can find, also, twisted ribbon cable (pretty standard
cable in SCSI).

"twisted pair must be used" isn't totally accurate.

It was more a figure of speech than a technical law. )

But, of course, it is in specific areas.

TP technology has been the design of choice for over a
century, but it doesn't lend itself to two constraints on
cabling for computers.. cheap and compact.

The sig-gnd-sig-gnd layout of an 80 wire EIDE cable does work
well.

Of course it does, if you keep it the way it is supposed to.

Which means, if you don't cut it and try to create "rounded"
cable of it. Then you can expect some problems. Therefore good
rounded cables have completely different cable schematics.

I've seen/used twisted-pair ribbon cables, but I didn't have
to buy them.

It was almost a non-written standard in all latest SCSI
implementations. Of course you will not need it for ATA
implementation (and I doubt you can find something like that).

For instance, IBM FRU PN 37L5558:

http://www.aykat.com/ebay/kabel/kabel_scsi_5x_68pin/
kabel_scsi_5x_68pin_3.jpg

That look good.

Presumably the twisting doesn't create extra problems so (apart
form cost) why aren't PATA cables like that?

Essentially because the traditional 80 wire cable is good enough.

They did go for a much smaller serial cable with SATA.

And, differential signalling on twisted pairs instead of the old
parallel signalling like PATA has. The SCSI TP cable is for
differential signalling. TP cable pairs don't help with PATA
single ended signalling, nor the source terminated method that
creates reflections back down the cable.

Grant.
--
http://bugsplatter.id.au

On Nov 10, 2:41*am, Vince wrote:
On 23:10 *9 Nov 2009, Arno wrote:





mike wrote:
Rounding IDE cables this looks as if it could cause electrical
problems on the PATA cables. Is it really as safe as he
suggests it is?

http://cpu-central.com/Articles.asp?article_id=7249&
decor_int=27

Depends. With UDMA 3 or before you can get data corruption. I
had this with a burner that did not support data checksums on
UDMA 3. Most HDDs do suuport these checksums, but the standard
does not require it before UDMA 4. For all UDMA levels you can
get command corruption, checksums on commands are not present
before SATA.

It also depends on lenght. My experiences are (If I remember
correctly): 30cm - works, 45cm - data corruption with the
burner, 60cm - command problems with HDDs, 90cm - basically
unusable with HDDs being dropped by the kernel within minutes.
Cable quality can influence that in both directions.

Arno

The home made mod doesn't seem to have as much protection from
intereference.

Ready made rounded PATA/IDE cables arrange the wires (signal and
ground) as twisted pairs.

Don't know how much difference this makes.- Hide quoted text -

- Show quoted text -

TROLL

I AM PROTEUS