PDP-11/04 – Restoration | Hacker News Reader

PDP-11/04 – Restoration

datormuseum.se

91 pointslelf19 comments

Anyone who calls themselves a hacker needs to read this to calibrate their definition.

This guy - he's a hacker.

Technically he's a DEC field engineer :-) But this is one of the reasons I really enjoy the older DEC gear, there is so much documentation on how it works (or is supposed to) that you can debug it to the chip level.

DEC field engineer? That's handy, my waffle maker is broken. ;)

Be careful stripping the wires...

Speaking as a former Data General field engineer these old machines (DEC/DG, hell even old IBM mini's like the S/36) were a joy to work on; and you're right, with a copy of the right diagnostics, the relevant micro-fiche pages and a scope you could fix pretty much anything on-site if you had to.

The documentation for these old machines was truly amazing.

I'm currently working on an 11/23 here: http://bensinclair.com/category/pdp-11

It's not quite as involved as this 11/04, but still a lot of fun. I'm not having much like right now though!

I should finish my PDP-11 simulator one day: http://i.imgur.com/6RuI41d.jpg

The javascript PDP-11 emulator is pretty cool and works a lot better than my own: http://pdp11.aiju.de/

Ask any DEC Field Service Engineer for a wire-wrap tool and they all reached for their back pockets. Used to amaze me to see them sitting on the floor, a step-by-step guide next to them, wire-wrapping some field service change into the backplane.

Plus the banter: "You done yet?" "Two more. Got a screwdriver?" "No. How 'bout a rum and coke?" "What's this power switch do?" "Haha, very funny."

It's cool seeing this technology being used. My dad had a PDP-11/23 in the his basement along with three 5 MB RL01 hard disk drives and 256K RX01 8" floppy disk drive to with it. Is there anything useful/interesting to do with this technology today?

Warning, 40 megabytes of images.

If the author reads this, consider turning those photograph PNGs into JPEG. There is no reason not to. Also resize them so they are not resized by the browser but shown 1:1.

> The CPU has no crystal to provide clocking, instead DEC used a delay line in a feedback loop.

Were oscillators back then expensive, unreliable, or it's another reason for not using one?

No, but fooling around with the delay line is one way to implement single-step or really slow step.

So either your next clock pulse is when a humanoid hits a button or when a delay line times out.

Can't just gate in a clock using a "AND" because of jitter and syncing is more trouble than its worth.

I never worked on this model so this is just semi-educated guess.

Makes me sorry I had gotten rid of my LSI-11.

Very cool! I started thinking in Octal again a bit while reading this...

Reminds me I need to get back to working on my VAX-11/730. Nethack isn't going to play itself...

I don't know about the 730, but the 780 had a PDP-11 (/03 I think?) in the cabinet which was in charge of the system console, and bootstrapping the VAX processor with code from an 8" floppy disk.

Funny how their (then) new system needed its predecessor to get it up and running...

That's actually pretty common in larger systems. You don't want your big, expensive machine bogged down with all the simple, stupid housekeeping stuff, so most big iron machines have a frontend processor of some sort to handle that. The DECs used a PDP/11 or an Intel 8080, depending on model, IBM tended to use PCs or Thinkpads running special versions of OS/2, etc.

Even a modern mainframe such as a Cray requires a management computer to get it started. In this case of a Cray today it's a generic PC.

The 780 was the first of the line of DEC 32-bit minicomputers. Of course it had a front end processor to manage bringing it on line; IBMs always had this and newer generations of Intel processors have and adjunct processors to do similar things.

The 730 was the lowest end of the line, designed for folks who couldn't afford a full 780 (or 750, software developers often). It's often cited as the slowest real 32-bit production machine.

Reply on news.ycombinator.com