Talk:Parallel port

Photo needed

PUT UP A PICTURE!!! PLEASE!!! —Preceding unsigned comment added by 68.255.69.19 (talk • contribs) 20 Nov 2004

• Working on it... 8-) ComCat 01:43, 29 Nov 2004 (UTC)

Someone had added one, but it was a slightly modified version of the one at es.com/nozomsite/parallel.htm, which doesn't have any copyright or licensing notices so is probabably a copyright violation. So I removed it. But a similar picture would be nice... --Rick Sidwell 2 July 2005 23:21 (UTC)

Due to the lack of copyright info, that image was removed on 9 April 2006 by OrphanBot.—mjb 23:19, 24 May 2006 (UTC)

New pic required

Just that. —Preceding unsigned comment added by 59.92.242.21 (talk • contribs) 13 April 2006

DuLithgow added a new image today.—mjb 23:19, 24 May 2006 (UTC)

Merging with Parallel communications and Parallel transmission

Hi:

Regarding whether this should be merged with Parallel Communications/Transmission

First of all, I clicked on "(discuss)" and ended up here. IF I am at an inappropriate place with this comment, my apologies.

Now, I often send people to Wikipedia to research various portions of Computer Hardware, Software, IT, etc. They need to end up at very specific portions of those subjects (such as Parallel Ports, Centronics, and IEEE-1284). IF this portion is merged with other sections, then I hope that the person's ability to get to exactly the pieces of information that they need, will not be impeded by having the information buried amongst tonnes of other information.

Thanks in advance

Ross Tucker Canberra, Australia —The preceding unsigned comment was added by 150.101.125.246 (talk) 21:15, 9 December 2006 (UTC).

Disagree. Do not merge the articles here. The concept of parallel communication is quite distinct from its particular implementation in the parallel port found on older PCs. Other examples of parallel communication are other buses such as GPIB and computer arcitechtures such as the 64-bit address bus. On the other hand, merging parallel communications and parallel transmission (and their serial variants) may be worthwhile. --IanOsgood 18:10, 11 December 2006 (UTC)

Disagree Do not merge. They are completely different things.--Energman 17:37, 17 December 2006 (UTC)

Disagree Do not merge. They are completely different things. Also, flight should not be merged with neither penguins nor flings... ntg (talk) —Preceding undated comment added 12:05, 17 June 2012 (UTC)

Disagree Do not merge. LPT applies to a specific device driver which implements IEEE-1284 interface on a Microsoft DOS/Windows platform. The EEE-1284 parallel port has been around long before that and is also utilized on other (non MS Windows-Windows based) platforms. We are talking apples and oranges here ..

External links section removed

External links were filling up with commercial and advertising sites and were removed. Please add new links that conform to WP guidelines and refrain from adding commercial links WP:EL. - Thanks Calltech 16:52, 18 December 2006 (UTC)

Merger?

All our information about the parallel port is split into various disconnected sections in at least four different articles, none of which is very useful. They need to be merged into one comprehensive article. --88.110.235.235 22:18, 13 January 2007 (UTC)

The problem here is the various versions of the parallel printer interface:

• First, there is the Centronics style. This is discussed in the Centronics article from a historical view, as the Centronics spec is obsolete.
• Then there is the HP Bitronics port, a variant of the Centronics interface.
• These converged into the IEEE 1284 specification.
• There is also the Dataproducts parallel interface, available in the M50 (Winchester) [1] and DB50 [2] connectors. (Thankfully I have not had to deal with this interface in quite a while)
• DEC, SUN, Amstrad, Atari, Commodore and others used a variety of pinouts for Centronics parallel.
• IBM set the defacto standard for LPT ports, setting the IRQ and other specs.

So: I would recommend this article and an overview of the parallel printer port and linking to the other articles. --Gadget850 ( Ed) 20:54, 15 January 2007 (UTC)

That doesn't require separate articles. None of the separate articles are substantial enough that they couldn't be made into sections of this article. --88.109.83.157 23:11, 15 January 2007 (UTC)

It has been three weeks and only the two of us have discussed this. While looking at this, I see some some basic issues with this article. First is the scope and the name. The scope of the article deals with the printer interface, whereas a parallel port can refer to a great many things. Parallel printer port would be a better title. Currently the article is PC centric- as noted above there were a number of variants and connectors. The article is also Centronics/IEEE-1284 centric as it does not note the Dataproducts paralle interface- a completely different standard. --Gadget850 ( Ed) 15:34, 9 February 2007 (UTC)

I don't agree with the above suggestions. IEEE 1284/Parallel Port/LPT pretty much cover the same thing, with different names. I'd just ensure each name permutation is covered in the merged article, which I'd expect would be done automatically. - MSTCrow 05:57, 6 March 2007 (UTC)

Centronics should retain its own article focusing on the company; the information on the parallel port should probably be merged in though. Fourohfour 18:37, 18 March 2007 (UTC)

The Centronics article should retain all of the history on how the interface was developed and how it became the defacto standard for printers. This interface was a significant and historical development in the computer industry. The details of how the Centronics interface was developed; such as why a 36 pin Amphenol connector was selected, belong in the Centronics article not in the Parallel Port article. -- SWTPC6800 16:28, 30 April 2007 (UTC)

Lots about IBMoid-PCs; not enough about the interface, itself

This article seems to be about the IBMoid PC implementation of the Centronics-style parallel port, rather than the port itself.

While this information can be useful, and may very well bear inclusion (along with other implementations), the article should discuss the particulars of the interface itself—including signaling, and the 36-pin micro ribbon connector that was originally used at both ends of the cable, before IBM began substituting the DB-25 at the host end.

And the DEC DB-37 connector, TI 46-pin card-edge connector and Data General 50-pin AMP connector. Tandy used a header connector, but I don't have a reference on that. --— Gadget850 (Ed) ^talk - 19:42, 31 October 2007 (UTC)

Port addresses

If there is an unused LPTx slot, the port addresses of the others are moved up. (For example, if a port at 0x3bc does not exist, the port at 0x378 will then become LPT1.) The IRQ lines, however, remain fixed (therefore, 0x378 at LPT1 would use IRQ 7). The port addresses assigned to each LPTx slot can be determined by reading the BIOS Data Area (BDA) at 0000:0408.

Should this in fact read as follows? I'm not sure enough to make the edit.

If there is an unused LPTx slot, the port addresses of the others are moved up. (For example, if a port at 0x378 does not exist, the port at 0x278 will then become LPT1.) The IRQ lines, however, remain fixed (therefore, 0x278 at LPT1 would use IRQ 6). The port addresses assigned to each LPTx slot can be determined by reading the BIOS Data Area (BDA) at 0000:0408.

The current sentence implies that there will be no change, as 0x378 is LPT1 as standard.

Be bold. This needs fixing and a citation added to nail down the right addresses, ports and interrupts.Cuddlyable3 (talk) 05:57, 23 May 2008 (UTC)

The problem is that the table lists the addresses checked in the wrong order. The sequence (according to chapter 14 of The Undocumented PC by Frank Van Gilluwe) is 3BC,378 and finally 278. The 3BC address was used by the parallel port found on the MDA video card while the 378 & 278 were originally found on ISA parallel port cards and later integrated into motherboard chipsets. I will update the table to reflect this. TheGiantHogweed (talk) 11:24, 7 June 2009 (UTC)

Diagram

A diagram like this would be nice. — Omegatron 22:19, 14 January 2008 (UTC)

There is actually an infobox for this: {{Infobox Connector}} --— Gadget850 (Ed) ^talk - 02:47, 14 March 2008 (UTC)

The diagram and table seems to contain errors. As they currently stand (July 6, 2015), I believe the following corrections should be considered:

1. (1) "ACK" (pin 10) should have a line over it as it is active low (just like STROBE).
2. (2) "BUSY" (pin 11) should NOT have a line over it as it is active high.
3. (3) "Error" (pin 15) should have a line over it as it is active low.
4. (4) "RESET" (pin 16, aka. "Initialize Printer") should have a line over it as it is active low.

I've made a revised diagram with the above suggested changes here.

For the table mods, I'd suggest if "overline" text can't be entered, use the +/- prefix syntax instead, e.g. -STROBE, -ACK, +BUSY, etc.

Sources for these corrections:

1. "IBM PC XT Technical Reference Manual" (Part #1502237) specifically the Pinout on pp.1-112 and signal description on pp 1-86 thru 1-89,
2. "The Indispensable PC Hardware Book, 2nd Ed" (ISBN 0-201-87697-3) specifically the Table 29.2 on pp. 846,
3. "PC Intern System Programming" (ISBN 1-55755-145-6), specifically the table on pp. 564,
4. "Upgrading and Repairing PCs, 3rd Ed" (ISBN 1-56529-467-X), specifically the Figure 8.14 on pp.518

I'd make the changes myself, but I'm not sure I should without some peer review, as the original table + image are heavily referenced. Greg ercolano (talk) 00:36, 7 July 2015 (UTC)

needs info on speed

I agree this would be nice to have. AFAIK there is no specification since the port can deliver much faster than a printer, the originally intended peripheral, requires. However the speed is relevant for other uses of the port such as

• input sensing e.g. some data logging uses
• data exchange between PCs

It may be difficult to find verfiable sources about speed rather than just OR. Cuddlyable3 (talk) 13:32, 24 May 2008 (UTC)

PC variant

how exactly did the PC variant differ from the original? Plugwash (talk) 03:20, 29 December 2008 (UTC)

Centronics pins 15, 18, 34 and 35 are missing

What is this pins used for? Are they "reserved"? --MrBurns (talk) 08:51, 6 April 2009 (UTC)

The table is a mess; among other things, it does not list the specification. Per the Centronics Model 101 Interface Specifications (Parallel) and IEEE 1284-1994: Standard Signaling Method for a Bi-directional Parallel Peripheral Interface for Personal Computers:

Pin	Centronics	PC (IBM/Epson)	IEEE-1284
1	DATA STROBE	Strobe	nStrobe
2	Data 1	Data 1	Data 1
3	Data 2	Data 2	Data 2
4	Data 3	Data 3	Data 3
5	Data 4	Data 4	Data 4
6	Data 5	Data 5	Data 5
7	Data 6	Data 6	Data 6
8	Data 7	Data 7	Data 7
9	Data 8	Data 8	Data 8
10	ACKNLG	Ack	nAck
11	Busy	Busy	Busy
12	PE (paper empty)	Paper Empty	PError
13	SLCT	Select	Select
14	SS (hardware alarm)	AutoFdXT	nAutoFd
15	OSCXT (100kHz clock)	not defined	not defined
16	0v	Signal Ground	Logic Ground
17	Chassis Gnd	Chassis Ground	Chassis Ground
18	5v	not defined	Peripheral logic high
19	Twisted Pair Gnd	Signal Ground (Strobe)	Signal Ground (nStrobe)
20	Twisted Pair Gnd	Signal Ground (Data 1)	Signal Ground (Data 1)
21	Twisted Pair Gnd	Signal Ground (Data 2)	Signal Ground (Data 2)
22	Twisted Pair Gnd	Signal Ground (Data 3)	Signal Ground (Data 3)
23	Twisted Pair Gnd	Signal Ground (Data 4)	Signal Ground (Data 4)
24	Twisted Pair Gnd	Signal Ground (Data 5)	Signal Ground (Data 5)
25	Twisted Pair Gnd	Signal Ground (Data 6)	Signal Ground (Data 6)
26	Twisted Pair Gnd	Signal Ground (Data 7)	Signal Ground (Data 7)
27	Twisted Pair Gnd	Signal Ground (Data 8)	Signal Ground (Data 8)
28	Twisted Pair Gnd	Signal Ground (Ack)	Signal Ground (PError, Select, nAck)
29	Twisted Pair Gnd	Signal Ground (Busy)	Signal Ground (Busy, nFault)
30	PRIME RET	Signal Ground (Init)	Signal Ground (nAutoFd, nSelectIn, nInit)
31	PRIME	Init	nInit
32	FAULT	Fault	nFault
33	Light Detection (video lamp inop)	Signal Ground (Select In)	not defined
34	Line Count	not defined	not defined
35	Line Count Return	5v	not defined
36	not defined	Select In	not defined

• Inverted signals are indicated by use of overline or the prefix n depending on the specification.

IEE-1284 pin 18 is often used to power an external device such as a network print server. Depends on the parallel port and how much current it supplies.

--—— Gadget850 (Ed) ^talk - 13:58, 6 April 2009 (UTC)

Logic Levels

Maybe it is important to note, that modern PC extension cards, like PCI Express parport cards use a logic high level of 3.3V. I've seen pages linking to this Wikipedia-Topic: "but they say its 5V". Just to avoid problems, which I've stumbled. — Preceding unsigned comment added by Jedzia (talk • contribs) 11:43, 30 March 2014 (UTC)

Pin 25 not connected?

The article contains the following information (apparently added in 2011 by Uwezi ):

Pin 25 on the DB25 connector might not be connected to Ground on modern computers.

Well, this IMHO just does not make any sense why something like this should become the norm. I can, however, see one reason for this to occur, but that is by a design error only and should be regarded as a bug. If someone knows of any legitimate reason for the pin 25 to be left unconnected or knows a particular example of common hardware with this property, please feel free to add a citation and/or further clarification. I have just verified the parallel port wiring on a couple of motherboards and on 1 LPT card (some of which could hopefully be considered modern by the 2011) and found none having the pin 25 not connected to ground. Meanwhile, I'm thus marking this information with {{Dubious}}.

As for the above hinted possible reason for the pin 25 to remain unconnected, I have recently discovered a possibility for this to occur with Gigabyte motherboards. Gigabyte motherboards seem to be using a slightly modified pin-out for the LPT header (pin 24 swapped with key/pin 26) which, if used with a generic-wired (and unkeyed) bracket cable, would leave the pin 25 unconnected. Unfortunately, it seems even Gigabyte themselves are selling brackets with the generic wiring only that are not compatible with their own modified pin-out. I have recently discovered this myself (Czech Republic, as of 2015 – unsuitable bracket bought from one e-shop and the very same being confirmed by another e-shop to be the only available; in both cases the bracket was described along with Gigabyte's part numbers), but this has already been reported as early as in 2009 on Tom's Hardware. However, as noted above, this is a design error in the bracket cable and the pin 25 would be correctly connected to ground with the appropriately wired cable, so this might not have been the reason for putting this information in the article.

And if this was the original reason for putting the information there, the statement should be at least rephrased to reflect the fact this is by a wiring error only. (Which I might do myself some time in the future, if no objections arise...)

--JITR (talk) 15:58, 7 October 2015 (UTC)

@JITR I am inclined to agree that this is a fault / wiring error and not the norm.

My big issue though is why a 25-pin connector had so many pins connected to Ground when these pins could have been put to better use, for example, as additional data lines making the parallel interface capable of 16-bit transfer when it began being used for more than just a printer interface. 81.108.186.222 (talk) 12:55, 6 March 2022 (UTC)

Missing history info - transition from really dumb printer to ASCII input?

There is a significant gap in the history section between the earliest use of this connector, which per the text here (it's news to me) required the host computer to convert character data into dot matrix bit patterns, and the slightly later usage wherein the 7 data bits in the connector were used to send ASCII characters and the character generation was done in the printer. Certainly the first printers I dealt with that used this interface (on CP/M machines) just took ASCII data. More sophisticated dot-matrix printers did have a "binary" mode that could be used for graphics. Jeh (talk) 14:24, 9 October 2016 (UTC)

Looks like Maury Markowitz (talk · contribs) has changed the text to the effect that the original Centronics-interface printers did take ASCII data. Thanks. Jeh (talk) 18:53, 9 October 2016 (UTC)

Lots more to come... Maury Markowitz (talk) 20:26, 9 October 2016 (UTC)

Thnaks veru mach Dian Nikolow (talk) 11:06, 6 March 2024 (UTC)

DataProducts interface

It might be worth mentioning that the DataProducts interface was more different than just a different connector. The handshaking was different. I built a couple of interfaces to these back in the 70s. I can probably dig out details on the differences. Jeh (talk) 21:13, 9 October 2016 (UTC)

Writing Convention for Inverted Signals

The table "Bit-to-pin mapping for the Standard Parallel Port (SPP)" uses the tilde (~) to specify inverted signals.

In the good old days of computers, the slash forward was usually used in writing, when the line above was not feasible, e.g. /Strobe, /Busy.

The table designates pins, not signals, but it might nevertheless be more common to use the slash instead of the tilde.

--Stefanhanoi (talk) 18:12, 30 September 2021 (UTC)

Parallel port IRQ

What was the idea behind the original IBM hardware assigning an IRQ to the parallel port?

Does a parallel port/printer combination actually generate interrupts, and if so, under what conditions? There's only a few lines like ACK and PAPER OUT that go back into the port in the computer and surely these don't generate an interrupt? Valgrus Thunderaxe (talk) 17:09, 27 June 2022 (UTC)

Biostar

Is this manufacturer important enough to be included in this page? Their Wikipedia page has no parallel ports mentioned anywhere. 2A01:112F:90C:BD00:5E44:E0F9:CA35:8335 (talk) 14:53, 21 March 2024 (UTC)

The manufacturer is important, but the sentence makes no sense. I'll remove it. ;) --Eliza Winterborn (talk) 16:28, 12 May 2024 (UTC)

Operation description

The host sent ASCII characters to the printer using seven of eight data pins, pulling them high to +5V to represent a 1.  When the data was ready, the host pulled the STROBE pin low, to 0 V.  The printer responded by pulling the BUSY line high, printing the character, and then returning BUSY to low again.  The host could then send another character.

That description is inaccurate.  My old hardware reference book from the 1970s indicates that the BUSY signal was not part of Wang’s original interface design.  In those days, handshaking connections were always low-true, not high-true, mainly due to the limitations of totem-pole outputs in 74LS logic devices.  Hence a handshaking signal such as BUSY would not have been implemented.  BUSY was an IBM “enhancement” for their version of the Centronics interface.

The two handshaking signals that are essential to making the interface operate are /STROBE and /ACK.  Following the toggling of /STROBE, the printer would toggle /ACK to indicate that it had accepted the data.  As soon as /ACK went high, the host could send another byte.  Hence the interface can be made to work by connecting the data lines, /STROBE, /ACK, and ground.  BUSY et al are not necessary.

216.152.18.132 (talk) 05:26, 5 October 2024 (UTC)