• Please review our updated Terms and Rules here

Question about 3270 Coax Protocols

C

cfsimmons

Member
Joined
Nov 9, 2017
Messages
47
Location
Huntsville AL
Hi all,
I've been trying to learn about the protocol IBM used between their coax terminals and the controller such as the 3174. I've found some documentation on the Type B terminals like the 3277, but I'm more interested in the Type A implementation, used for the 3278/3279. I'm not looking for a detailed explanation of the 3270 protocol since I know it can fill multiple books, but rather just hoping someone can point me in the right direction.
Pretty much everything I know at the moment comes from the CHIPS 82C570 data sheet. Given the amount of ISA emulation boards and clone terminals made I would think this information must be out there somewhere.
 
Thanks for the links, although from what I've gathered the documents there only describe the connection to a 3277 terminal, which is of Type B (or is it Category B).

Well, my other option is to get a 3174 running, but as people have pointed out the microcode is hard to find and harder to duplicate. But I was wondering if it would be easier to image and emulate the 20mb fixed disk drive. It appears to be a ST506/412 drive in a specialized caddy. Has anyone tried this?
 
Just curious about what you are wanting to achieve. Are you wanting to connect a 3278/3279 to a modern system and use it as some sort of console?

BTW, from what I understand the disk drive in the 3174 was just a standard MFM disk drive of that era. Someone with such a working drive from a 3174 should be able to image it for you if they are willing.
 
cfsimmons said:
Thanks for the links, although from what I've gathered the documents there only describe the connection to a 3277 terminal, which is of Type B (or is it Category B).

Well, my other option is to get a 3174 running, but as people have pointed out the microcode is hard to find and harder to duplicate. But I was wondering if it would be easier to image and emulate the 20mb fixed disk drive. It appears to be a ST506/412 drive in a specialized caddy. Has anyone tried this?
Click to expand...

The microcode is easy to duplicate, but the 2.44 floppy disks can be hard to find. Some one else has used the project here http://www.pdp8online.com/mfm/ to replace the disk.

Finding a working drive is hard. I have three, but none work reliably. I think they suffer if not properly parked.
 
MossyRock said:
Just curious about what you are wanting to achieve. Are you wanting to connect a 3278/3279 to a modern system and use it as some sort of console?

BTW, from what I understand the disk drive in the 3174 was just a standard MFM disk drive of that era. Someone with such a working drive from a 3174 should be able to image it for you if they are willing.
Click to expand...

The 3174 will talk TN3270 so can connect to Hercules. I have used it that way
 
MossyRock said:
Just curious about what you are wanting to achieve.
Click to expand...
Ultimately, I'd like to be able to talk to these terminals. Connecting one to Hercules is an end goal. The only terminal I have at the moment is a 3270 PC, but I'd love to get a 3279. I won't bother though until I know i can get it working.
I'm glad to hear that someone has emulated the drive, although the problem of getting the microcode still remains.

My knowledge of mfm drives is pretty limited. Would a typical pc controller be able to read the raw data from the drive, or can the 3174's disk controller be implemented differently?
 
Al Kossow said:
Is much known about the Memorex/Telex 1174, like
https://www.ebay.com/itm/191672782519 ?

I bought one a while ago because it used SCSI drives and commodity ICs.

Haven't had a chance to do much more than image the drive from it.
Click to expand...

Al,
I did have a couple of 3.5" floppy disks from a memorex telex controller but no idea if they are any use. I will have a try at imaging them if I can find them
Dave
 
cfsimmons said:
Ultimately, I'd like to be able to talk to these terminals. Connecting one to Hercules is an end goal. The only terminal I have at the moment is a 3270 PC, but I'd love to get a 3279. I won't bother though until I know i can get it working.
I'm glad to hear that someone has emulated the drive, although the problem of getting the microcode still remains.

My knowledge of mfm drives is pretty limited. Would a typical pc controller be able to read the raw data from the drive, or can the 3174's disk controller be implemented differently?
Click to expand...

Your end goal is what I was trying to do, but I was going to use an actual 3174. Things were going well until I bought an IBM 3278 on eBay but discovered that it had no keyboard when it arrived. Long story short, my mistake. Now I can't find a keyboard anywhere for less than $2,000, and I'm not going to take out a loan for this.

Unless you have good contacts in the vintage computing community, finding a working, complete 3279 is going to be very difficult (and probably costly).

The 3270 protocol is a insanely complicated beast. I can't imagine trying to emulate it from scratch unless you are an electrical engineer AND a low-level programming whiz. Unless I'm misunderstanding something here, just emulating a MFM drive from a 3174 is only a fraction of what is required as you have none of the required infrastructure hardware for a 327x terminal to work. To go the normal route, you're going to need something like an IBM 3174-23L (with the ethernet option) to be able to telnet into Hercules' console port to get the console displayed on a coax terminal like the 3279. If the 3174 has an option for a hard disk, THEN you can install the MFM emulator in it with the appropriate IBM microcode (I think it is Configuration Support 'C' version 6.2 or higher).
 
MossyRock said:
Unless I'm misunderstanding something here, just emulating a MFM drive from a 3174 is only a fraction of what is required as you have none of the required infrastructure hardware for a 327x terminal to work.
Click to expand...
I should have been more clear. I'm not interested in getting a 3174 but I very well might if its the only solution. Emulating the mfm drive is just a way of getting that setup working.

From what I was able to find on the type A coax protocol (http://bitsavers.org/pdf/dca/82C570_Data_Sheet.pdf), the protocol doesn't appear to be too complex. The complex bit is the connection between the 3174 and the host. I reckon a board (perhaps ISA) could be made quite simply with a DP8340 and DP8341 which do most of the electrical work.
 
cfsimmons said:
The complex bit is the connection between the 3174 and the host.
Click to expand...

Yes, that's where the transport protocol transformations from synchronous line, ethernet, or token ring to the polled bisync protocol of the 3270 happen.
As far as I know, there is no open-sourced code around that does that, which is why the people I know who have it working with a real terminal just buy
a 3174-23L with ethernet or token ring.

The reason I brought up the Telex was that was an alternative to the 3174 which has options for modem, ethernet or token-ring connectivity,
but there is even less known about them. At least they have off the self parts in them like 68000's and DP834x.
 
You can think of the IBM coax protocol as a layer 1/2 transport using a bi-phase protocol and . In 1987 Capstone Technology did an ISA development board for National Semiconductor's then new DP8344 BCP (biphase communications processor). Here is one of their boards with a Engineering Sample of the chip from my collection:

BCP.jpg



As other pointed out earlier in the thread the actual display on the screen was carried by the 3270 data stream (think layer 3) which is well documented.

This paper suggests that the layer 1/2 is not well documented (I don't have the actual journal paper):

https://www.sciencedirect.com/science/article/pii/014193319591861W

but you can find the National BCP datasheet here:

http://www.bitsavers.org/components...2_DP8343_Biphase_Local_Area_Network_Mar88.pdf
 
Kiwi said:
You can think of the IBM coax protocol as a layer 1/2 transport using a bi-phase protocol and . In 1987 Capstone Technology did an ISA development board for National Semiconductor's then new DP8344 BCP (biphase communications processor).
Click to expand...
That looks nicer than the old DP8340/8341 chips (separate chips for transmit and receive). As I recall, those were pretty much glorified serial/parallel converters. The TAC / DCA IRMA boards used an 8X305 CPU to do the actual protocol work. The ISA IRMA board had an 8X305 CPU to handle the coax side of things. The IRMAprint / IRMAlink added an 8085 CPU for the user side, since there was no PC to handle that. I have the 8085-side source code around here somewhere, and probably still have the listings for the 8X305 side. [I did NOT want to deal with that CPU, but needed the listings to see what was going on, from the 8085's point of view.]

NS's reasoning for separate TX/RX chips was that mainframe-side controllers might need many more receivers than transmitters. I don't know if they ever got any substantial design wins on the mainframe side, though.

Later IRMA boards ditched the code ROMs for the 8X305 in favor of RAM, and used a DCA-labeled part instead of the 8340/8341 set. The DCA-labeled part may well have been an 8344.
 
Kiwi said:
You can think of the IBM coax protocol as a layer 1/2 transport using a bi-phase protocol and . In 1987 Capstone Technology did an ISA development board for National Semiconductor's then new DP8344 BCP (biphase communications processor). Here is one of their boards with a Engineering Sample of the chip from my collection:
Click to expand...

Do you have any of the BCP development tools (pretty much just the assembler) or their sample code?

They show up in a couple of companies terminals. Telex used a later DP83445 which appears to have integrated the CRT controller as well.
Information on that part doesn't seem to exist.

National sold an MPA ISA card. I just bought this one and am planning on dumping the code from it and the code in the Telex terminal.

ns_mpa.jpg
 
Kiwi said:
This paper suggests that the layer 1/2 is not well documented (I don't have the actual journal paper):

https://www.sciencedirect.com/science/article/pii/014193319591861W
Click to expand...

I spent the $35 for a pdf (all 5 pages of it). Pretty much an echo of the data sheet, other than them saying they used
a coax packet sniffer to build a printer interface. They didn't describe anything in detail.

The best source outside of IBM is the National 1990 IBM Data Communcations Handbook, which is under
http://bitsavers.org/components/national/_dataBooks
 
Last edited:
Al Kossow said:
It would have been obvious if there were two high speed proms or rams attached to it for the 8344 code.
Click to expand...
No, it just appears to be a re-spin of the older designs, since it still has the 8X305 and 3 2KB RAM chips. I'll attach 3 pictures (none taken by me) of what appear to be 3 generations of the card design.

I'd be interested in the code you might have. I'd like to try turning some IRMAprints into IRMAlinks (which are MUCH more rare)
Click to expand...

I think the standalone converter was the IRMAline and IRMAlink was the name of the file transfer software that ran on the PC-based IRMA boards. That part of my memory does not have parity protection, though. :p

The IRMAprint came in serial and parallel versions (PC/3287-S and -P, respectively) with a different logic board inside. -S used a Signetics 2651 USART and the -P used a 7406 and 2 7407s. A universal board would have been nice.

I modified both the IRMAprints (to speak Dataproducts protocol instead of Centronics) and IRMAlines (in conjunction with local mods to MS-DOS Kermit, to have a full additional line for the status display instead of bouncing the image up and down the screen by one line). The binaries at least should be on my old "ROM-burner PC", if it still works after 25+ years. It should also have the last version of the 8X305 PROMs - the earlier versions had a nasty problem where a 43xx CPU with integrated 3278 controller would fail to IMPL if the IRMAline was powered on and connected to the coax at the time. Arguably IBM's problem, but given the relative sizes of IBM and DCA, DCA fixed it on their side. I also remember that the power supplies for the IRMAline tended to fail if there were brief hits on the 120VAC line.

IRMA-1983.jpg

IRMA-1986.jpg

IRMA-1987.jpg
 
Ok, so I happen to have a small quantity (EDIT: 9) of some generic C&T 82C570 cards that have an EPROM on them; would a dump of that EPROM be of any help to anyone? I don't have one in front of me right now, but I'll take a pic when I get a chance later today.

Pic:

1537478339379-2010646007.jpg

EDIT: I have dumped the ROM. Here's edited output of 'strings' on the binary:
Code: 
J1 Coax Version 1.25
Copyright (c) 1989,91 J1 Systems, Inc.
Version 1.25-J1
ICOT ShortCUT Terminal Emulation  ROM 
Qrom E
Quadram MainLink II   ROM 
...
J1 Coax Version 1.25
Unable to load micro code.
$QMIC3.02
...

That 'Unable to load micro code.' string leads me to believe that this ROM has uploadable microcode for the 82C570 embedded.
 
Last edited:
I spotted a National MainLink (same C&T chipset as your boards) on ebid, and picked it up
National bought Quadram, so I'm guessing the firmware will be the same, given the Qrom
reference in the prom

The C&T chip isn't as well documented as the National DP8344, particularly the instruction
set of the embedded processor.
 
You must log in or register to reply here.
Back
Top