Setting up a DECnet network with a Pro380 and Windows client?

[stanp]

Anyone have any experience setting up a DECnet network between a PRO and a MS-DOS/W95 computer? I recently acquired a DECnet card for my PRO. I also purchased a twisted pair transceiver. I have a W95 machine with a network card. I am hoping to get both talking over DECnet to make file transfer easier. Some of my questions are:

1. Can I run DECnet over the same switch that I use for my IP network or do I need a separate network? A switch operates at layer 2, so I assume this is possible.
2. Does DECnet use something like ARP at layer 2?
3. I understand that I need to install Pathworks on the W95 machine. Will installing Pathworks prevent the W95 computer from speaking IP over the network interface. I understand DECnet will change the adapter MAC, which seems to indicate that it is one stack or the other.
4. Does the W95 version of Pathworks automatically take over the network card on start-up? I would prefer to manually start DECnet services if both IP and DECNet cannot run concurrently. Is it better to install the MS-DOS version?
5. Is a license required for the client if it accesses a PRO? I assume the Pathworks client can talk to a PRO, or is that a wrong assumption?

 

[czunit]

It will run on the same network, different protocol and probably different frames (I don't think it used EtherNet_II, and Appletalk was what used SNAP frames)

I believe it uses a broadcast type format but I did have to type in the name of my pdp11 to get it to work.

As for the rest, I can hook up my 380 and test stuff if you would like. So glad to have another person with a Pro Ethernet adapter out there....

 

[stanp]

I managed to acquire a DECNA-K card. It came with the original box, documentation and was still wrapped and sealed in the antistatic bag. I placed it into my PRO 380 and the startup diagnostic flagged the card with code 10. I tried different slots with the same results. The maintenance disk recognizes the card as DECNA and reports the DECnet address info, but flags the card as a failure. The documentation says to replace the card when the error code is less than 110. Has anyone seen this condition? I'm wondering if the card has older firmware and is not 380 compatible. I ran into a memory card that I could not get to work in the 380, but another did. Could be a bad memory card. When I ran the DECnet system test, it crashed the system with a list of octal codes, only two with a value of 300 and 4 and all others 0. Hence, I'm thinking something is not compatible with the 380. The latest date on the chips is 16th week of 1984. Any suggestions?

 

[czunit]

Hi there!

Here are some thoughts from a working DECNA in my 380 (tested this morning)
1) Put the Decna in the back most slot. I have had problems putting it into other slots in the past.
2) Make sure you have a loopback connector plugged in or a 10bt connector plugged into a working switch that can handle 10mbps.
The diagnostics require this. When I took my loopback off this morning I got an error, then a crash report. Here's pics


With the jumper in I get a nice little:

 

[czunit]

Other questions:
1) Do you have any other cards in your 380? Try removing everything possible to simplify.
2) How much memory do you have? The DECNA has an additional 128kb of memory because the original Pro/350 had 512k of memory which was not enough to load all the Decnet stuff. However I think the 380 came with 512k on the board and one daughter card that could handle another 512 (or 128 if you use a really old Pro/350 memory board). Hm, that should not be a problem.

Can you post pics of what you see? When your 380 comes up do you see the extra 128k memory?

I never thought to try the card in a 350 but I could do that and see if anything works (I have one now but it's in the closet). I'd defer to others but I thought the CTI bus stuff had ROMs on each card which contained specific diagnostics so I don't think it's a 380 rom issue.

If you would like you could send the card to me and I can try putting it in my known good 380 here. That would at least isolate the problem to either the board or everything else

 

[stanp]

Hi there!

Here are some thoughts from a working DECNA in my 380 (tested this morning)
1) Put the Decna in the back most slot. I have had problems putting it into other slots in the past.
2) Make sure you have a loopback connector plugged in or a 10bt connector plugged into a working switch that can handle 10mbps.
The diagnostics require this. When I took my loopback off this morning I got an error, then a crash report. Here's pics
View attachment 1277765
View attachment 1277766
With the jumper in I get a nice little:

View attachment 1277767

Thanks for powering up your system and helping me with some testing. I do have the loopback connector installed and the green LED is on. I initially tried the DECNA card in slots 3, 4, and 5 because I have the RTI card installed in 6 and the cable is screwed in. However, at your suggestion, I removed the RTI card and installed the DECNA card into slot 6. I also removed all other cards except for the hard disk controller and floppy controller (slots 1 and 2). Unfortunately, I get the same error. Whereas you get error code 111, I get error code 10. The documentation that came with the card says an error over 110 means it is not the card.

 

[stanp]

Other questions:
1) Do you have any other cards in your 380? Try removing everything possible to simplify.
2) How much memory do you have? The DECNA has an additional 128kb of memory because the original Pro/350 had 512k of memory which was not enough to load all the Decnet stuff. However I think the 380 came with 512k on the board and one daughter card that could handle another 512 (or 128 if you use a really old Pro/350 memory board). Hm, that should not be a problem.

Can you post pics of what you see? When your 380 comes up do you see the extra 128k memory?

I never thought to try the card in a 350 but I could do that and see if anything works (I have one now but it's in the closet). I'd defer to others but I thought the CTI bus stuff had ROMs on each card which contained specific diagnostics so I don't think it's a 380 rom issue.

If you would like you could send the card to me and I can try putting it in my known good 380 here. That would at least isolate the problem to either the board or everything else

I had the same thought and removed the other cards, which were memory cards. The manual does say to place all memory cards in lower ordered slots below the DECNA card. I guess they want the DECNA memory in high memory. As for memory, the 380 has 512K by default and I have the 512K daughter board also installed. Attached are the diagnostic screen shot and the memory command result. I'm not very familiar with the memory display.

I now have three cards that are not functional. This DECNA card, a memory card that looks physically different than those that work because it has wirewrap jumpers, yet the part number is identical and this is common to see the rework on some of these cards, and a serial line card that I recently acquired, which is not recognized since the tab code does not come up on the diagnostic, just 177776. I might take you up on your kind offer to test the cards in your 380 or a 350. But first, I am going to try replacing the socketed chips on the DECNA card, with the exception of the ROM. They are the Intel ethernet controller and another ethernet chip. I should get those parts early next week.

I'm open to more suggestions if you have any. Anyone have experience repairing DECNA cards? Unfortunately, these are hard to find.

 

[stanp]

Thanks for powering up your system and helping me with some testing. I do have the loopback connector installed and the green LED is on. I initially tried the DECNA card in slots 3, 4, and 5 because I have the RTI card installed in 6 and the cable is screwed in. However, at your suggestion, I removed the RTI card and installed the DECNA card into slot 6. I also removed all other cards except for the hard disk controller and floppy controller (slots 1 and 2). Unfortunately, I get the same error. Whereas you get error code 111, I get error code 10. The documentation that came with the card says an error over 110 means it is not the card.

Attached are images of my card. Do the part numbers on both sides match yours? Notice the rework jumper on the back in the center of the picture.

Given I also have a problem with one memory card, is it possible that the memory chips on both the DECNA and the problematic memory card are too slow for the 380? Did they upgrade memory chips when they released the 380 and maybe older cards don't work in the 380?

 

[MattisLind]

You could try to load the contents of the Diag ROM into SIMH and disassemble it. The code is supposed to be position independent but a PRO loads them at octal 2000. Then try to figure out what error code 10 really is about. I disassembled the RAM card diag ROM and commented it (https://github.com/MattisLind/DECPROMEM/blob/main/mem-orig.asm). I guess the idea behind the different Diag ROM are very similar; Do a test and then return an exit code if failing.

I wonder why it need the DECNA card to be the last module? I could guess that it has to do with how it configures the base memory address. Each memory card is given a base memory address and returns the new base memory address which is old base plus size. They probably want physical main memory space to not have any holes. The Pro initalizes each board in order from first to last slot.

 

[stanp]

You could try to load the contents of the Diag ROM into SIMH and disassemble it. The code is supposed to be position independent but a PRO loads them at octal 2000. Then try to figure out what error code 10 really is about. I disassembled the RAM card diag ROM and commented it (https://github.com/MattisLind/DECPROMEM/blob/main/mem-orig.asm). I guess the idea behind the different Diag ROM are very similar; Do a test and then return an exit code if failing.

I wonder why it need the DECNA card to be the last module? I could guess that it has to do with how it configures the base memory address. Each memory card is given a base memory address and returns the new base memory address which is old base plus size. They probably want physical main memory space to not have any holes. The Pro initalizes each board in order from first to last slot.

Wow. I did not know about this project. Your disassembly might help me figure out what's wrong with the memory card that doesn't work. The board you show does not have the rework leads. The two boards that work for me do not have rework. The one with rework does not work. The strange thing about the memory card is that it passes the startup diagnostic test, but the memory does not show up in the system and the maintenance disk does not recognize it as I recall. I think it realizes there is a card present, though.

Where are the jumpers on the memory board that you mention?

How did you download the ROM? Did you pop the chip into a reader or did you write a program to read it?

The DECNA document says that it can go into any slot, but you should move all memory cards to a lower slot number. This is consistent with the field guide also.

 

[MattisLind]

The Diag ROM from the memory board is in the Xhomer project. I used a EPROM programmer to read the one I had and it was identical to the one in the Xhomer project.

The tech manual talks about the size jumper and a fully populated jumper. I could guess that the size jumper is not implemented on the 256 k boards. Probably meant for an upcoming 1meg board.

The current diag firmware is likely to be buggy since it processes the size jumper in a weird way. It will max out at 512 k.

You could also create a small assembly program that writes to the baseaddress of the board + 2 and then reads from the base address. Each read will give you the next byte.

I have a dump of the DECNA ROM here:

DECNA firmware

 

[czunit]

Interesting stuff. My reason for the last card was that when I tried to stuff my Pro/380 like a turkey I found that the Decna would throw errors if it was not in the last slot. I thought it might be due to the CTI extension bus not having all the wires for all the devices bussed.

A similar issue seems to happen with my TMS board: It really wants to be in one specific slot and not others. So I leave it there and lore is created (wonder if anyone has TMS+the daughtercard and console). Does anyone have the disks for Pro/Communications 3.2? The disks in the POS install set have a missing/improperly duplicated disk.

If it helps I can pull the DECNA and take a picture of the front and back. But I'm pretty sure the memory speeds are the same as the rest of the boards: Unibus, Q Bus, and probably CTI were asynchronous anyway thus you could run anything from core memory to static memory on the systems.

Interested about the "size" jumper on the 256k memory cards: There is one on the daughtercards of course, and those can be upgraded with 256k chips. I have idly wondered if it was possible to swap out the 64k DRAMS and put in 256k ones. But then again I have heard the 350/380 can only address 2mb of memory.

 

[MattisLind]

I think the Pro can handle 3 meg. That at least the maximum amount of memory the Diag ROM will configure. I have tested my patched firmware in the Xhomer somulator and it gives me 3 meg in P/OS. Of course the simulator my differ from the real thing but I cannot find anything in the tech manual indicating that it is limited to two meg.

On the other hand Venix refused to boot if it had more than 2 meg in Xhomer. It thought it had 65 meg! BSD refused to boot with any size of expansion memory in Xhomer. Not sure if this is a Venix/BSD thing. P/OS and RT-11 both reported 3 meg.

 

[AndrewZ]

Is there a source for the larger DRAM chips? I read in the FAQ about desoldering the smaller DRAMs and resoldering with the larger ones. I have a spare RAM card and would like to give that a try.

 

[MattisLind]

The small daughtercards on the PRO 350 (are those the same on the 380?) can easily be changed to 512 k modules by replacing the 64kbit chips into 256 kbits. I don’t remember if you need to change a jumper or so as well. I did this mod 30 years ago. I think it is possible to find 256 kbit chips on Ebay or possibly someone here has a bunch of them.

I don’t know about the CTI bus memory expansion. The firmware has a bug so that it will never handle more than 512 k.

 

[stanp]

The small daughtercards on the PRO 350 (are those the same on the 380?) can easily be changed to 512 k modules by replacing the 64kbit chips into 256 kbits. I don’t remember if you need to change a jumper or so as well. I did this mod 30 years ago. I think it is possible to find 256 kbit chips on Ebay or possibly someone here has a bunch of them.

I don’t know about the CTI bus memory expansion. The firmware has a bug so that it will never handle more than 512 k.

While searching for RAM memory speed, I discovered the daughterboard schematics and notes are in the PRO 350 print set. There are two jumpers on the 350 daughterboard for 128K and 512K. I don't know if the 380 daughterboard is identical, but it seems likely with just the jumpers set.

 

[stanp]

The Diag ROM from the memory board is in the Xhomer project. I used a EPROM programmer to read the one I had and it was identical to the one in the Xhomer project.

The tech manual talks about the size jumper and a fully populated jumper. I could guess that the size jumper is not implemented on the 256 k boards. Probably meant for an upcoming 1meg board.

The current diag firmware is likely to be buggy since it processes the size jumper in a weird way. It will max out at 512 k.

You could also create a small assembly program that writes to the baseaddress of the board + 2 and then reads from the base address. Each read will give you the next byte.

I have a dump of the DECNA ROM here:

DECNA firmware

Thanks. I disassembled the file with the base at 2000 octal in SIMH. I briefly stated to annotate it, but there is much reverse engineering needed. It looks quite different than the memory card ROM, even the entrance and exit.

 

[stanp]

Interesting stuff. My reason for the last card was that when I tried to stuff my Pro/380 like a turkey I found that the Decna would throw errors if it was not in the last slot. I thought it might be due to the CTI extension bus not having all the wires for all the devices bussed.

A similar issue seems to happen with my TMS board: It really wants to be in one specific slot and not others. So I leave it there and lore is created (wonder if anyone has TMS+the daughtercard and console). Does anyone have the disks for Pro/Communications 3.2? The disks in the POS install set have a missing/improperly duplicated disk.

If it helps I can pull the DECNA and take a picture of the front and back. But I'm pretty sure the memory speeds are the same as the rest of the boards: Unibus, Q Bus, and probably CTI were asynchronous anyway thus you could run anything from core memory to static memory on the systems.

Interested about the "size" jumper on the 256k memory cards: There is one on the daughtercards of course, and those can be upgraded with 256k chips. I have idly wondered if it was possible to swap out the 64k DRAMS and put in 256k ones. But then again I have heard the 350/380 can only address 2mb of memory.

I would appreciate the card photos if you can do that. On the memory, it looks like DEC used three sources of memory: Fujitsu, Hitachi, and NEC. They are 200 ns access time.

 

[HoJoPo]

Is there a source for the larger DRAM chips? I read in the FAQ about desoldering the smaller DRAMs and resoldering with the larger ones. I have a spare RAM card and would like to give that a try.

Jameco has 41256 DRAM chips in stock, from the schematic above it looks like the existing chips are 4164s, so swapping to 41256 for a 4x capacity upgrade makes sense.

 

[MattisLind]

Thanks. I disassembled the file with the base at 2000 octal in SIMH. I briefly stated to annotate it, but there is much reverse engineering needed. It looks quite different than the memory card ROM, even the entrance and exit.

Ahh. I should have said that the Pro loads the code at octal 2000. It probably doesn’t matter much since it is position independent.

It indeed look very different. A possibility is to add it to xhomer and see what it does. At least you will get sure about the entry and exit of the code since it will fail very quickly.