NODAF/NORD-10.5 progress log

From NDWiki
Jump to navigation Jump to search

(originally on nodaf.no, rescued from archive.org)

This page is in English because non-Norwegians might very well be interested. A project page in Norwegian is at NORD-10.5

NORD-10 serial number 5 restoration progress log

If you have any suggestions or comments, please contact me [user:toresb]. Thank you!

19/10/2006 - Acquired machine

Finally picked up the machine.

20/10/2006 - Performed initial research

Have talked to Jonny Oddene. Semiconductor and core memory is *not* trivially interchangeable. The serial number is not in his records.

Have done research on the drum storage. It is a Vermont Research 310 series system. Apparently, the machine's hermetically sealed drum might be made to be pressurized with helium at 2 PSI! It might also be the free-air version. Will investigate. Have docs for this drum, including beautiful service info.

Did not succeed in finding someone to let me into the IFI building this evening. Will try to go there tomorrow, for further inspections.

21/10/2006 - Further inspections

Drum seems to be of the air type, and not the pressurized helium type. Puh!

The microelectronics labs at the Institute for Informatics, IFI, were open today. Goodie! This gives me access to the measuring equipment I needed. If this lab is open every day, or some sort of access could be arranged, I could use this lab during the whole restauration! This would certainly be an immeasurable aid.

Tested the power supplies under load. Man, I could use a camera! Unwired the cut cords - 5V cords from the right (from the front of the rack) PSU, and the broken 220V ones. Jotted down how the chocolate bar wiring was. Will try to replicate in digital format for the log.

First and second time, the left PSU tripped the AC mains breaker. Tried a second time (as the first time could be attributed to inrush). Added a resistor on the five volt line, though no visible current arose on the (admittedly slow) voltmeter either of the two first times. Applied mains again. No DC current on output.

Tested the second power supply. This one made a sound! Something like "Click-ticticticrrrr". And that was all. There was measurable current on the DC line. Smells funny. Reassembled the PSUs back to its original wiring, except for the 5V wire stubs. There was so much dirt on the outside, that I suspect it may well have leaked into the inside of this PSU. This has worrying ramifications for the other power supplies, though none were so exposed to dirt as this one.

I am recommending use of two ordinary ATX power supplies, and have been lent some, for power delivery to the NORD-10 CPU. The ferrite core memory power supply unit will of course also need testing, and is probably far less trivial to replace (As are the components it powers!) - IIRC, the PDP-7 utilized a range of voltages for this, including +65 volts.

If someone feels comfortable enough around electronics to look these power supplies over, I shall gladly hand them over. If so, please contact me. Maybe one could acquire the schematics somehow.

I also have yet to test the drum power supply. Gave the drum a brief check. The fan seems in acceptable, if suboptimal condition. The core and CPU fans are not near acceptable. I am unsure if they will turn at all.

22/10/2006 - Further work on the machine

(I need to write this up, soon!)

25/10/2006 - The designer visits!

Today the designer of (amongst other things) the NORD-10, Bård Einar Sørbye, came by! And he brought schematics for the machine!! Also, he brought some other very interesting documentation, and also some very illuminating internal memorandums. I shall make an effort to and catalog these, and also to scan in the smaller ones. There is also a lot of very interesting promotional material. I sorted them roughly into three categories - technical information not immediately relevant to this project, corporate stuff (marketing, internal memos, yearly reports, press, etc), and technical information relevant to this project.

I moved the machine out of the way for Dagen@ifi. The NORD-10 was placed near the second floor exit. The NORD-1 was temporarily placed in the bomb shelter in the basement.

We looked the machine over, and we seem to have come to the conclusion that it was in decent shape, and that a restoration is indeed a feasible, yet perhaps a more substantial project than initially believed.

We also agreed that for a long-term exhibit with a running machine, a NORD-10/S would be vastly preferable to this system, for countless reasons. (Speed, reliability, flexibility, spares availability, both new and from other NORD-10/S'es we have).

We both agreed that the ATX power supplies, even as a temporary solution, delivered insufficient or only marginally sufficient current. Pål Spilling is footing the bill for a 5V/40A PSU, and Bård Sørbye has agreed to buy one more if this current is not enough.

26/10/2006 - More work. The panel lives!

Arrived around 11 AM for the free Dagen breakfast, yum. Continued to work on the machine.

I discovered a well-hidden, cut cable, which was the 5 volt supply for the panel! I'd traced the cables going out of the panel. The 3.6v cable coming from the Teletype supply was bundled together with the 5v supply, which was very hard to spot.

Applied 5 volts to the input. Beautiful lights! The level counter only responds to the +/- buttons intermittently. It will, however, sometimes happily count away - sometimes all the way to 19. Hmm. Quite impressive, for a 4-bit counter!

Reviving the panel gave me a glimpse into the problems we'll be facing with regard to contact problems. Jiggling one of the panels will toggle the red lights and button lights seemingly at random. The panel initializes to random values. When power is applied to the card cages, though, the random values change to other random values. The panel might be using an AC clock from a transformer as a timing mechanism (Which obviously is not present, since I'm fuelling these things with bench lab supplies).

A priority should be getting the two small PSUs on either side of the AC contactors working again. To do this, I will need to perform more research on the wiring.

We seem to currently need isopropanol, to clean the contacts. (Wash it with water, with an isopropanol chaser?)

31/10/2006 - Cards cleaning - Got a real PSU!

Finally, I managed to order the PSU and installed it. Gah! The freaking contact mount broke off. Impressively awful build quality. Though, I have to hand it to them - packing 40A at 5V into such a small package really is impressive. I fully was expecting to have to lug this thing. The damn PSU weighs nothing. It's a fraction of the size of an ATX PSU!

I've asked for a return, and they've ordered a new one. Problem is - this PSU might not be sufficient. I may have to order another one, and this is the last one they have in stock.

See, measured at the terminals of the PSU, the voltage was 5.02V. Perfect. Measured at the CPU, the voltage was 4.52V. Hrmph. Measured at the I/O cage, it's ... 1.14V? Okay, we need another PSU. There are simply not enough terminals on the PSU to wire up the remaining power supply wires going down into the cage.

So I think I'll buy another PSU. Bård Sørbye has graciously offered to sponsor that one. If it turns out not to be necessary, I will return it anyway.

My current plan is thus twofold: Buy the second power supply, and wire that up, measure the voltage in the card cage. Then, remove all the cards, of course noting whence they came, and clean the bus slots and card edges with isopropanol. Also, possibly removing the panel connector and cleaning it (and/or checking for the OSC and CC signals.)

01/11/2006 - Wiring up the PSU

Today, I wired up the PSUs. My current configuration consists of three power supplies. One 40A@5V PSU serving the I/O cabinet, one serving the CPU, and one of the ATXes (seemed the simplest to wire up, and I'd cut the Molex wires to stubs anyway) serving the panel (approximately 1A load). For the CPU PSU, I measured the voltages at the PSU's terminals. 5.00 volts. Perfect. Measured in the cage. This time, the rack measured 4.85 - which *is* within specification, albeit relatively barely, and not something we'd want in a situation like this. I might adjust the PSU upwards a bit, if this turns out to be a problem.

Moved the machine down into the basement. Wrote down what cards are where. Spoke with Wiggo at Telemuseet and agreed to visit the storage hall, and cleared borrowing a Bus Extender card with Anne Solheim.

Borrowed a storage scope from the microelectronics group at Computer Science. It's a really old one, and there are some really bad contacts, but it looks like it might work.

02/11/2006 - Visited Telemuseet, basic debugging

Visited Telemuseet's storage facility. ...wow! It's ...impressive. NORDs of every age of variety. From SAM 2/FLINK through the earliest NORD-1s, NORD-5s, NORD-10s, and NORD-100s, through the late-model TpServer Rallar. Found the bus extender card.

(Must write more about it - docs and media.)

Did only rudimentary debugging on the machine - mainly to try to figure out how to work an oscilloscope. The 4MHz clock from the Panel Control 1 reaches the Panel Driver, so this is our first sign of life. Will write email tomorrow or thereabouts to try to get a hold of a logic analyzer.

03/11/2006 - Cleaning, cleaning, cleaning.

Today, I removed all the boards, cleaned them thoroughly with isopropanol and q-tips, cleaned the bus backplane, and reseated the cards. This was very, very necessary. However: The cards were incredibly hard to get out of the case the first time. They'd somehow gotten very badly stuck. After cleaning, they were far more cooperative. However: I cut several traces on two boards while trying to get the boards out. I have drawn arrows on the boards where the traces are cut, and I've confirmed this with a multimeter. I will have to resolder them, or get somebody more experienced to do so.

The affected boards were the 1003 STATUS and 1033 PANEL DRIVER boards.

Postponed the power-up due to possibility of isopropanol residue inside the machine.

While reseating the cards, I discovered to my delight that this machine has a floppy disc controller! The 1111C and 1118C cards which I had been unable to identify simply had their names printed on the solder side of the PCB. The boards are obviously cramped for space!

I am unsure of what kind of future work will be possible without a logic analyzer. I will ask around for one. I may possibly be allowed to borrow one from University on Monday, but this is, alas, quite doubtful. I'll see what happens.

I have updated the card list to reflect this change - also wikified the list.

06/11/2006 - Back to work!

Weekend's over. Talked to the microelectronics group at IFI. I am allowed to borrow a logic analyzer within office hours. A more flexible way to solve this - either with an L.A. of our own, or a way to lock in the LA myself - would be very nice.

No work done on the machine as I only had 30 minutes of "LA time".

07/11/2006 - Working with the LA!

Adjusted the voltage on the CPU cage PSU to compensate for the attenuation by the power cables from the CPU. Highest voltage (measured at the edges of the cage) is 5.00v. Voltage in the middle is probably around 4.90 or thereabouts, though I haven't measured this yet.

Today was my first day with the NORD-10 and a logic analyzer. I managed to program it to trigger on CC from Panel Control 1. Collected one complete roll of the Panel Bus, all 16 states. All numbers are octal:

States 0, 2, 4, 6, 10, 12, 14, and 16 yield 0200, which is supposed to be flashed onto the lights. Isn't. As the signal to do this is coming correctly timed from the Panel Driver in the CPU card cage, I can only suppose that the panel cards themselves are bad, or some connection between them. Perhaps it might be an idea to attempt getting a hold of a printer for the logic analyzer so that I can print out some interesting results when I encounter them.

08/11/2006 - A field trip!

I went to the museum of telecomms today. That's a pretty impressive sight! Their ND collection is in a word, impressive. Kåre Trøym and Jonny Oddene have done a great and important job in preserving not only the hardware, but software, and documentation. Even the source code archive has been preserved!

09/11/2006 - Acting upon data

Removed the panel cards and cleaned them. Will check some traces.

Source

This article was originally a copy of the NODAF wiki article NORD-10.5_progress_log in 2007, via archive.org snapshot.