How are GND & data lines pared in parallel I/O cable❓

[L0uis.m]

Hello 0010 all,

Can someone, please, tell me where to find documentation on the twisted pairs in parallel data cable?

I would like to know which ground wire forms a pair with a data wire.

In some documentation it is mentioned they are twisted pares, data wires are numbered, but GND wires are not.

[leepivonka]

They aren't doing differential signaling & all the GNDs are connected together at the ends, so there is no particular pairing of data lines & GNDs.

[BigDumbDinosaur]

L0uis.m wrote:

Can someone, please, tell me where to find documentation on the twisted pairs in parallel (printer) data cable?

As Leepivonka said, there are no twisted pairs in a Centronics (parallel) printer cable.  All active circuits are unbalanced to ground.  Also, the DB-25 connector used on PC-compatible hardware was not the standard connector as used by Centronics when they developed the interface.

Incidentally, the parallel port implemented in PC-compatible hardware is a bastardization of the original Centronics design.  Some of the signals, such as Paper Out and /SelectPrinter, are not part of the Centronics standard.  They were added by IBM (working with Epson as a “partner in crime”) in an ultimately-futile effort to prevent third-party printers from working with PeeCees.  A functional interface to any parallel-port printer only needs D0-D7, signal ground, /ACK and /STROBE.  Contrary to popular opinion, Busy is not needed and. in fact, was not part of the original Centronics standard.

[L0uis.m]

Hello 0010 all,

leepivonka wrote:

They aren't doing differential signaling & all the GNDs are connected together at the ends, so there is no particular pairing of data lines & GNDs.

Well, actually it's not supposed to be a Centronics "low speed printer cable", that's why I wrote (printer) in brackets, it's supposed to be a "real" parallel cable, DB25 male to DB25 male, with eight separate data return wires, intended to connect a PC's parallel I/O to a measuring device, of more than €1000.00, which has differential I/O.
Quote from a well known reliable source: [··] lines with differential receivers will reject noise regardless of whether the signal [source] is differential or single-ended, [··]

The supplier, Conrad, says "We don't know, but you can get a new cable." and that's exactly where it goes wrong again, the cable that came with the apparatus didn't work at all. They send me a replacement cable with the same result, just an other wrong (printer?) cable.
I even sent the apparatus, with clearly described complaint, back to Conrad for "repair", it returned with a message from Conrad "everything OK, works well", but it doesn't. I also got three different versions of software packages from them, but nope. Most probably they didn't bother to connect it to a PC, but only switched the device on and viewed the startup sequence, which then gives an "OK, ready" result.

According to the device manual it can be connected to a normal parallel port, but I am pretty sure with sugar on top I need something with "twisted pair" data lines and, similar to a "Null-modem cable", some of the other lines "crossed". After I figured out which wires to be "crossed", I tried to make such a cable myself, so now I established at least some communication between PC and device, but as soon as I cut the end of a cable, the wires become a rat's nest that makes determining which (colour coded) wires are twisted pairs impossible, wires not paired correctly result in unstable/indecisive measuring results.

Still hoping for some advice to gain good result . . .

P.S.: Conrad sucks

[John West]

The PC printer port was standardised as IEEE 1284. If you don't want to pay for the specification, some of the information in it can be found in https://web.archive.org/web/20061115015 ... _ports.pdf. There's a table on pages 9 and 10 showing which signal pins each ground pin acts as return for. There are multiple signals for most grounds, so twisted pairs aren't going to be possible. The best you could do would be a ribbon cable with each ground between or at least near its signals.

And there is no way any lines need to be crossed - that's what you do with serial ports when you're connecting devices of the same type. That's definitely not the situation here. Your device, whatever it is, was designed to be plugged into a PC.

How do you know that the problem is the cable? I would be looking at the PC end. What are you plugging it into? Most PCs haven't had parallel printer ports for decades. If you're using a cheap USB adaptor, I'd be surprised if did work. The printer port was originally designed to send data from a computer to a printer, while this is going to be working in the opposite direction. If you are using an adaptor, you'll need to find one that supports this mode. Or find an old PC with a real printer port in it.

[L0uis.m]

Hello John,

Please let's drop "Printer port/cable", just call it what it is, a Parallel port and cable. Yes, I know it's meown fault, mentioning (printer), but it was meant as a mere indication, "port also used for", therefore between brackets.

John West wrote:

And there is no way any lines need to be crossed - that's what you do with serial ports when you're connecting devices of the same type. That's definitely not the situation here. Your device, whatever it is, was designed to be plugged into a PC.

I am pretty sure with sugar on top there are "crossed" wires needed, see modification I mention below.

John West wrote:

How do you know that the problem is the cable?

Again, I am pretty sure with sugar on top, because there was no communication at all, until I figured out which wires had to be "crossed" and made a modification according.

John West wrote:

What are you plugging it into? Most PCs haven't had parallel printer ports for decades. If you're using a cheap USB adaptor, I'd be surprised if did work. The printer port was originally designed to send data from a computer to a printer, while this is going to be working in the opposite direction. If you are using an adaptor, you'll need to find one that supports this mode. Or find an old PC with a real printer port in it.

No fancy USB S**T or wacky adaptor, first I tried a vintage Dell and then a good old thoroughbred HP, both with genuine Parallel Port, of which I'm sure they are Two Way, because after the above mentioned modification, communication in both directions between device and PC was established, but measuring results (data from device) still are faltering of which my instinct (I'm a learned industrial technician) says me it's most certainly distorted data due to wrongly connected twisted pairs!

Please realize how many combinations are possible to make two times eight connections out of twenty-five wires. . . , so what I really like to know (pretty please with sugar on top) is the standard colour coding of 25 core cable with twisted pair data lines, such as used for (two way) parallel data connections (I know it existed in my days, but I can't find it anyware any more).

[dmsc]

Hi!

There is not enough grounds in the DB25 interface to cover all signals, so the IEEE 1284 standard simply reuses grounds, and the standard explicitly says that you need a 36pin connector at the peripheral side to get the full speed (of more than 2MB/s).

See https://www.ti.com/lit/an/scea013/scea013.pdf for details and wiring between connectors.

Have Fun!

[BigDumbDinosaur]

dmsc wrote:

There is not enough grounds in the DB25 interface to cover all signals, so the IEEE 1284 standard simply reuses grounds, and the standard explicitly says that you need a 36pin connector at the peripheral side to get the full speed (of more than 2MB/s).

It wouldn’t matter how many grounds there are.  The Centronics port is unbalanced and the device the OP is trying to connect apparently has a differential input.

[BigDumbDinosaur]

L0uis.m wrote:

Well, actually it's not supposed to be a Centronics "low speed printer cable", that's why I wrote (printer) in brackets, it's supposed to be a "real" parallel cable, DB25 male to DB25 male, with eight separate data return wires, intended to connect a PC's parallel I/O to a measuring device, of more than €1000.00, which has differential I/O.
Quote from a well known reliable source: [··] lines with differential receivers will reject noise regardless of whether the signal [source] is differential or single-ended, [··]

Perhaps you need to be more specific in your posts about your application.    You said “printer” and posted an image of a PC-compatible parallel port.  So anyone reading your post would have been right to assume this had something to do with a printer.

The applications in which such a cable might be used would be single-ended, either TIA-232 or Centronics, if talking about a PC.  In any case, the grounds that are on a PC-compatible parallel port are all you get.  I am not aware of any 25-conductor cable terminated on DB-25 connectors that has twisted pairs.

If your measuring device has differential inputs, then a direct connection to a PC parallel port isn’t going to work.  You would need some sort of adapter to make the transition from single-ended to differential, which is something that you should be able to build if you are skilled in the art of electronics.

Assuming you can take care of the transition from single-ended to differential, the problems you will likely encounter in getting things to work are several:

1. Parallel ports have all-but-vanished from modern PCs.  You may be able to use a PCI adapter to add a parallel port to a PC that lacks one.  However, that may lead to problem #2...
2. PCI devices are not mapped to fixed I/O ports and IRQs the way the original serial and parallel ports were.  That complicates using a PCI adapter outside of Windows, Linux or another operating environment that knows how to enumerate the PCI bus and detect a parallel port card...which leads to problem #3...
3. All versions of Windows since NT will not let you directly access parallel port hardware.  Whatever application you might use to communicate with your measuring device would have to have a Windows-compatible parallel port driver that understands the protocol used with your measuring device.  If your application doesn’t have such a driver and instead communicates directly with the parallel port hardware, the application will be fighting with Windows for control.

Good luck!

[BigEd]

Hi L0uis.m
A tricky situation. Do you have an original cable which might possibly have a clue in the colours of the individual wires? For example red paired with red/white.

Failing that my first guess would be to pair the 8 grounds with the 8 signal lines in the order of the connections - north to south. (18 with 2 … 25 with 9)

Or, maybe there’s a way to measure the electrical characteristics at the peripheral end??

[L0uis.m]

Hello 0010 all,

Mea culpa, mea maxima etcetera, mentioning "printer" was my mistake (no Centronics connector to be seem within miles) !

It's all about an MS-9150, Function Generator/Frequency Counter/Digital Multimeter/Multiple Power Supply (image below, from now on named the Thing).

No fancy USB S**T nor any wacky adaptor needed, the user manual says the Thing should be connected to a standard parallel I/O port "using the supplied cable".

As I mentioned before, Conrad (twice) supplied an (obviously wrong) straight wired DB25 to DB25 cable with which no data transfer was established at all. Than (by reverse engineering and comparing the results with info gained by searching the web) I discovered the Thing needed some signals "crossed", i.e. a Null-Modem cable.

I did not and never will nor ever wished to connect the Thing to a modern PC, but instead to a vintage HP (from the same era), which has a parallel I/O port as required, I also tried a Dell of roughly the same age, which also has a suitable parallel I/O port, it produced the same results, working but wonky !

I have also tried out several versions of driver/software, of which the most resent generates a few extra features, but further no difference, and I use the appropriate version of Windows, as prescribed in the Thing's user manual.

Now (with the modified "Null-Modem" cable) the Thing works, but if the measured signals fluctuate fast (but with the measured signal still well within specs), resulting in fast changing data, data transferred to the PC is distorted.

N.B.: the Thing has a startup/self checking sequence, which ends with No Errors, everything OK, device ready to be used, so there seems nothing to be wrong with it, therefore I suspect there is still something not quite right within the configuration of the cable (twisted pairs, resistors ? in data current return lines to prevent shorts between the Thing's (balanced ?) data outputs and the common PC "GND" ?).

[barnacle]

You may be able to get information from the US distributor: https://www.tequipment.net/MetexMS-9150 ... escription (phone number at end of page). I also found a possible circuit diagram on Scribd but as I'm not a member I see only the first page, which also appears elsewhere but doesn't itself answer your question: https://www.scribd.com/document/4278728 ... Ms-9150-Sm

There's a more extensive manual here https://verleih.fhstp.ac.at/upload/Mult ... 239070.pdf but it refers to a USB connection, so may be a later variant.

Neil

[dmsc]

L0uis.m wrote:

Hello 0010 all,

Mea culpa, mea maxima etcetera, mentioning "printer" was my mistake (no Centronics connector to be seem within miles) !

The IEEE-1284 standard is not a "printer" standard, is a bi-directional parallel communication protocol between PC computers and other devices.

Did you read the PDF I linked to, with the cable connection diagrams and ground returns?

And about your cable, back more than 25 years ago, there were 2 ways to connect computers via parallel port cables that used the DB25 male connector in each side: the 4bit/transaction cables (compatible with old parallel printer ports) and the 8bit/transaction cables (compatible with the IEEE-1284 ports).

See here for both cable diagrams: https://github.com/torvalds/linux/blob/ ... de-0-cable

Note that both are non-standard, and assume that you have a PC device at both sides of the cable, could be that your device acts like a PC.

Have Fun!

[BigDumbDinosaur]

You sure that “parallel port” is not actually an IEEE-488 port?

[SamCoVT]

Does the port on the back of your device look like this, and it is labeled RS-232C?

Serial port on back of MS-9150

If so, that's a serial port (sometimes called a COM port) and not a parallel port. The cabling is different for those ports and there indeed is a "crossover" cable available for that kind of port.

Most computers that have RS-232C ports have the 9-pin variant, so you'd need a 25pin-9pin adapter if your cable is 25pin-25pin.

Some photos of the ports and cables you are using might help clear things up.