Rounding IDE Cables
mike
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
Rod Speed
> Rounding IDE cables this looks as if it could
> cause electrical problems on the PATA cables.
While that is theoretically true, it does work most of the time.
> Is it really as safe as he suggests it is?
Nope, its easy to fuck it up.
If you must have round cables, it makes a hell of a lot more sense to go SATA instead.
> http://cpu-central.com/Articles.asp?article_id=7249&decor_int=27
Yousuf Khan
Well, there's a lot of theoretical thought out there that this will
negatively affect their electrical performance. In practice, there is no
difference. If you want the rounded cables, it's cosmetically better to
go with pre-rounded cables, and safer too, as you won't likely cut one
of them.
If you've got to cut your own, then I'd say that the author's suggestion
to cut only every 4th strand is a good choice. I'd go even further and
say you should only cut every 8th strand, as an IDE cable is an 80-wire
cable, so you'll only have 10 sections to worry about here. If you had
to cut through all 80 wires separately, you'll very likely make a mistake.
Yousuf Khan
Grant
I've done it before the switch to SATA, slit at every eighth wire,
easy to stuff up by cutting a conductor or having ribbon that
doesn't want to split cleanly and exposes conductor.
Benefit is that it's easier to route the wires, improve ventilation,
electrically I'd not expect a difference.
Grant.
--
http://bugsplatter.id.au
Arno
> 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
--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: ar...@wagner.name
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans
Igor Batinic
Vince wrote:
> On 23:10 9 Nov 2009, Arno wrote:
>
>> mike <n...@here.invalid> wrote:
>>
>> 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.
>
> Ready made rounded PATA/IDE cables arrange the wires (signal and
> ground) as twisted pairs.
>
> Don't know how much difference this makes.
Actually, a lot. Therefore, creating "home-made" ATA cables from
ordinary ATA cables is not a very good idea.
Best regards,
Iggy
Rod Speed
> Arno wrote
>> mike <n...@here.invalid> 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.
> 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.
Vince
My impression is that the 80-way PATA cables were a big improvement
on the old 40-way cables. Does anyone know of any data or tests
which shows how much improvement they gave?
In this thread Igor Batinic says twisted pair is quite an
improvement over untwisted. Does anyone know of any tests or
comparisons for this?
Arno
Look into the standard. They made UDMA speeds of 66MHz and above
possible by adjusted impedance and inter-wire shilding. It is
not so much about a better cable as one adjusted to what you
intend to send over it.
> In this thread Igor Batinic says twisted pair is quite an
> improvement over untwisted. Does anyone know of any tests or
> comparisons for this?
That is historic by now. Look into any undergraduate text on
signal transmission.
Igor Batinic
It is hard to find exact tests, cause we'll need separate tests for all
possible combinations - new cable is defined in standard, maybe you can
find some tests via Google.
> In this thread Igor Batinic says twisted pair is quite an improvement
> over untwisted. Does anyone know of any tests or comparisons for this?
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
Rod Speed
>>>>> http://cpu-central.com/Articles.asp?article_id=7249&decor_int=27
>> Some do, some dont.
>> Not much.
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.
nobody >
>
> 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.
Igor Batinic
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. :o)
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
With best regards,
Iggy
Tri Cutter
That look good.
Presumably the twisting doesn't create extra problems so (apart
form cost) why aren't PATA cables like that?
Rod Speed
Essentially because the traditional 80 wire cable is good enough.
They did go for a much smaller serial cable with SATA.
Grant
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
Proteus IIV
> >> 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
>
> intereference.
>
> ground) as twisted pairs.
>
>
> - Show quoted text -
TROLL
I AM PROTEUS