GPIB: 24-pin Centronix connectors?

[Joerg]

Digging out my copy of IEEE Std 488-1978, the answer appears to be yes.
As a footnote, it calls out Amphenol (or Cinch) "Series 57 Microribbon"
or AMP "CHAMP". (An Allied or Newark catalog would give a good overview
of that old stuff).

I have a 1991 AMP catalog 2ft from my eyes. Hardcover no less. But it
functions as a stand for a really heavy CAD monitor because its color
blends in nicely ;-)

So I'll just get a few of those 24-pin Centronix connectors. But solder
cup, I do not like ribbon cables, they spew out too much EMI.

 

[Joerg]

Joerg,

Remind me again why you're looking to build your own GPIB cable rather than
just use one of the big "hoses?" No 3-4 foot GPIB hoses in your junk drawers?

The hoses are so unwieldy. After one of them recoiled and flung a coffee
mug off the bench I just had it. The guy who spec'd in those monster
cables should .... nah, I won't say it, it's Christmas.

Theroretically, if you make sure the GPIB is quiet during measurements
it should be possible to even string 10ft-20ft of ribbon and push on a
connector everywhere you need one. And at the very end the Prologix.
I'll probably go CAT-5 though.

--
Merry Christmas, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

 

[Rich Grise]

The hoses are so unwieldy. After one of them recoiled and flung a coffee
mug off the bench I just had it. The guy who spec'd in those monster
cables should .... nah, I won't say it, it's Christmas.

Theroretically, if you make sure the GPIB is quiet during measurements it
should be possible to even string 10ft-20ft of ribbon and push on a
connector everywhere you need one. And at the very end the Prologix. I'll
probably go CAT-5 though.

I've found that making every other conductor ground helps a lot with noise
immunity, crosstalk, etc.

Thinking back, the HPIB cables I've seen had piggy-back connectors - i.e.,
a male on the front, with a female piggy-backed on it, to make daisy
chains.

Do you plan on making them as well? Or just one custom cable for the lot?

Thanks,
Rich

Good Luck!
Rich

 

[Joerg]

I've found that making every other conductor ground helps a lot with noise
immunity, crosstalk, etc.

Thinking back, the HPIB cables I've seen had piggy-back connectors - i.e.,
a male on the front, with a female piggy-backed on it, to make daisy
chains.

Do you plan on making them as well? Or just one custom cable for the lot?

No, just one long daisy chain. With enough slack or extra connectors
that minor reshuffling in the equipment rack doesn't cause some gear to
become orphaned.

I just don't want a maze of garden hoses back there.

 

[Rich Grise]

No, just one long daisy chain. With enough slack or extra connectors that
minor reshuffling in the equipment rack doesn't cause some gear to become
orphaned.

I just don't want a maze of garden hoses back there.

Well, heck; if you can do GPIB/HPIB in four pairs, then go for it! I'd
kind of have a tendendency to look around for shielded cat5, just 'cuz
( ^^^^^^^^^^^ - oopps! Looks like the Xmas eggnog is kicking
in!)
I'm superstitious that way. ;-)

Have Fun!
Rich

 

[Jasen Betts]

That's my straw man for any connector layout. Unfortunately, that's
not what GPIB used.

The best example of alternate signal/ground that I know of is the
IDE connector. They have a version that uses 80 wires rather than
the normal 40. It's necessary for the high speed modes.

I haven't looked at the details. Does anybody know how they get
80 wires connected to 40 pins? I assume there is some magic in the
connector that knows where the grounds are and connects them and
the 40 extra wires together.

there's only 39 pins for a start (one is a key).

yeah there's magic in the connector (I expect blue black and grey
connectors are each different too)

I'll see if I can find a cable to experiment destructively on.

 

[Spehro Pefhany]

there's only 39 pins for a start (one is a key).

yeah there's magic in the connector (I expect blue black and grey
connectors are each different too)

That is correct, they are slightly different inside.

I'll see if I can find a cable to experiment destructively on.

Best regards,
Spehro Pefhany

 

[Archimedes' Lever]

there's only 39 pins for a start (one is a key).

It is a 40 pin connector and all pins are used at the connector, even
if they are not all used in the male header.

yeah there's magic in the connector (I expect blue black and grey
connectors are each different too)

No. They are all the same. The master/slave differentiation has to do
with the connector location on the cable. Color is simply color, and
nothing else is different.

I'll see if I can find a cable to experiment destructively on.

Try simply using a bit of common sense. Oh... that's right...
nevermind. This post proves that you do not have any.

 

[Archimedes' Lever]

That is correct, they are slightly different inside.

Absolutely not.

 

[Spehro Pefhany]

Absolutely not.

Absolutely so.

http://en.wikipedia.org/wiki/AT_Attachment

See info on pin 28 & 34.


Best regards,
Spehro Pefhany

 

[Joerg]

Well, heck; if you can do GPIB/HPIB in four pairs, then go for it! I'd
kind of have a tendendency to look around for shielded cat5, just 'cuz
( ^^^^^^^^^^^ - oopps! Looks like the Xmas eggnog is kicking
in!)
I'm superstitious that way. ;-)

I have a certain reckecklessness in that respepect. Whoops ...

 

[Archimedes' Lever]

Absolutely so.

http://en.wikipedia.org/wiki/AT_Attachment

See info on pin 28 & 34.

UDMI cables are not using cable select features, idiot.

Remember how the selection is made AT the drive? Learn to read ALL of
the article you reference, or it will bite you in the ass.

 

[Jasen Betts]

It is a 40 pin connector and all pins are used at the connector, even
if they are not all used in the male header.

some cables and most connectors have pin 20 missing.

No. They are all the same. The master/slave differentiation has to do
with the connector location on the cable. Color is simply color, and
nothing else is different.

Try simply using a bit of common sense.

why would all manufacturers of 80-conductor ATAPI cables stock and use
connectors in three different colours when doing so it only costing
them money? (over using the same colour in each place)

 

[Spehro Pefhany]

some cables and most connectors have pin 20 missing.

"Pin 20" is optionally a molded-in "key" on the female (or it could be
a contact like most of the other signal lines). There's no hole at all
on the molded key type. These things are NOT intended to be
general-purpose connectors.

why would all manufacturers of 80-conductor ATAPI cables stock and use
connectors in three different colours when doing so it only costing
them money? (over using the same colour in each place)

Blue is the system board connector, pin 34 will be NC
Black is the Device 0, Primary connector
Gray is the Device 1, Secondary connector, pin 28 will be NC

For original references, see the work of T13 Technical Committee, and
of course the relevant connector data sheets.

I've designed a product using these components... when you need a few
dozen controlled-impedance lines and the BOM can't afford individual
coax cables...


Best regards,
Spehro Pefhany

 

[Archimedes' Lever]

some cables and most connectors have pin 20 missing.



why would all manufacturers of 80-conductor ATAPI cables stock and use
connectors in three different colours when doing so it only costing
them money? (over using the same colour in each place)

Fucking retard. UDMA REQUIRED placement of the master at the ENED of
the cable.

IF you had actually read the wiki, you would have noted and remembered
that. With regular ATA it was not a requisite.

This is NOT due to a difference in the cable (we still select master of
slave manually), it is due to the higher data rate failing due to
reflections if there is a stub hanging there.

The different colored connectors were done so that savvy IT pros could
remember thru association, which connector was the master.

IF the cables were indeed "cable select", the we would HAVE TO set the
jumper on the drive to "cable select" and we do not.

Do you get it now?

 

[Archimedes' Lever]

Gray is the Device 1, Secondary connector, pin 28 will be NC

Not true. It is NOT a "cable select" cable.

 

[Jasen Betts]

Not true. It is NOT a "cable select" cable.

Why should anyone believe this claim of yours?

 

[Jasen Betts]

On 26 Dec 2008 09:33:48 GMT, Jasen Betts <[email protected]> wrote:

Fucking retard. UDMA REQUIRED placement of the master at the ENED of
the cable.

incorrect,
1 either drive can be put on the end of the cable.
2 drives are drive-0 or drive-1 not master ans slave.
(and a single drive can be either and still function)

IF you had actually read the wiki, you would have noted and remembered
that. With regular ATA it was not a requisite.

This is NOT due to a difference in the cable (we still select master of
slave manually), it is due to the higher data rate failing due to
reflections if there is a stub hanging there.

if you had actually read the wiki you would have learned that it is a
cable-select cable.

The different colored connectors were done so that savvy IT pros could
remember thru association, which connector was the master.

IF the cables were indeed "cable select", the we would HAVE TO set the
jumper on the drive to "cable select" and we do not.

that is false, cable select is optional with cable select cables.

 

[JW]

That would make it even worse, by about 1" in extra length. I just want
to solder cable onto a solder-cupped 24-pin Centronix and come out
sideways, then 3-4ft to some location where the Prologix can comfortably
reside.

I wonder if the twists in CAT5 would mess up the signals when using each
pair for two data lines. We'll see. After all, GPIB is really slow.

Sorry... Had a "duh" moment. The problem I was speaking of is that the
shell of the prologix device often interferes with the case of the
instrument. I had thought that that was what you were having trouble with.

 

[Archimedes' Lever]

Why should anyone believe this claim of yours?

Look, fucktard... You STILL select 'master' or 'slave' when you install
a drive in a UDMA interface.

That means that it is NOT 'cable select'.

Using a 'cable select' cable would REQUIRE that the drive be set to
'cable select' as well, and we are NOT doing that, therefore, we are NOT
using 'cable select' cables. BASIC common sense.

Can you really be that fucking stupid?