Saturday, September 13, 2014

We now resume our regularly scheduled programming (?)

OK, so I've been busy.  Very busy.  And I haven't posted much.  Since the beginning of the year, I've moved into a new house, become a father, worked two concurrent jobs and have somehow remained sane.

I've also acquired a number of interesting computers and parts.  I've just had no time to play with them and even less time to write about them.  Things have settled down slightly -- the move is done, I've almost figured out how this parenthood thing works, I'm basically down to one job and I'm still working on the sanity thing.  So maybe I'll have more time to post stuff?  Time will tell.  For now, here's a summary of the computer-related stuff I've added to the collection since April.

Oh, and I've moved into new digs.  I mentioned that already.  If you want to see approximately what they look like you can check out an album of pictures here.

New Stuff!!!!
The above picture encapsulates the majority of the new (old) computer junk I've amassed lately.  Here's the rundown:

PDP-11/05:
Pictured on the top left (with the magenta/purple color scheme).  I've wanted an /05 for a long time, they're the smallest blinkenlights '11 you can get (it still weighs a ton).  This one has had a rough life, it's missing the top cover, has had something dropped on it and when it arrived on my doorstep, the 8KW core memory board was just flopping around loose in the chassis.  Yuck.

Fortunately, it's survived fairly well despite the above.  The power supply tested out fine and initial powerup with just the CPU installed showed signs of life.

Installing the memory revealed the first real issue -- I was unable to examine or deposit memory from the front panel.  Any attempt to do so would hang the machine until "Start" and "Halt" were toggled. Debugging (via my KM-11 boardset, never leave home without one) indicated that the microcode was waiting for the UNIBUS and timing out.  This indicated that the memory was not responding to the request.  And indeed, on closer examination, the memory was configured to start at the 2nd 8KW boundary.  I rejumpered it (by soldering in a new lug) and now the core memory was responding appropriately.

11/05 debugging via KM-11 (the blinky panels in the rear left)

This revealed a second problem:  Normally, hitting the "Examine" or "Deposit" switches causes the address to increment on successive toggles.  This was not happening.  Further, the CPU was not executing code correctly.  Even a simple "program" consisting of nothing but NOPs would simply execute the same instruction over and over again.  The theme here seemed to be one of incrementing (or lack thereof).  I spent a great deal of time looking at the 11/05 Engineering Drawings (which are excellent, by the way) and started debugging using the KM-11 and the logic analyzer, starting with the ALU.  During a toggle of "Examine" or "Deposit" the ALU should be doing an "Add" operation to produce the next address.  The inputs looked good, and the outputs were correct given the control signal the ALU was being given, which was incorrect.  The ALU was essentially being told to pass the "A" input straight through without modification, rather than doing an Add.

More poring over schematics and documentation lead me to a pair of faulty Microcode PROMs which control the ALU.  The outputs of the PROMs were always zero, leading to the incorrect ALU behavior.  I ordered some replacement PROMs, and programmed them using my ancient Data I/O model 29.  (I had to recreate the contents of the PROMs by hand from the listings in the Engineering Drawings!  Luckily these are only 256x4bit PROMs...)

After replacing the Microcode PROMs, the front panel worked properly and the CPU can now run simple programs.  At this point I started testing out the core memory with a simple address test (see here).  For the most part, the core tested out great, which is good.  Except for a 64-word hole at address 012000, where all addresses read back as 0.  That's not so good.

After conferring with some experts, the current theory is that the drive logic is at fault.  I have not yet begun to debug this, but I'll be tackling it soon.

PDP-11/34
This is underneath the 11/05 in the picture at the top.  The /34 was one of DEC's best selling PDP-11 systems.  It's not a powerhouse but it's a very flexible and expandable UNIBUS system and I'm looking forward to getting it running.  I'm hoping to put 2.9BSD on it.

A couple of friends of mine picked it up on my behalf from a fellow in Menlo Park who was selling it for a very reasonable price.  It's a basic system (CPU, 128KW memory, and a SLU) and it came with some peripherals, a Cipher 9-track drive, a TU58 tape drive and a Fujitsu SMD drive.  I am waiting to test the 11/34 until I have time to go over the power supplies, but it's very clean and I don't expect any major problems with it.

Data General Nova 800 and 1200

These are to the right of the PDP-11s, on top of the tape drive.  These date to 1970 or so and are the successors to the original Nova and SuperNova minicomputers.  A friend of mine traded these to me a week ago.  They're currently in pieces and only one of them has a power supply but they're in solid shape and I'm hoping to get at least one working system out of the two.  I've been wanting to get into Data General hardware for quite awhile now and I couldn't turn down the opportunity for these two.  The chassis came along with a box full of boards, all in all two 800 CPU sets, two Nova 2 CPU boards, several 8KW core planes, a 16KW core plane and a single I/O board.

I'm currently waiting for a set of capacitors for the power supply to arrive and once I get the supply rebuilt, the fun can really begin.

PERQ T2 Workstation
The T2, set up on my workbench.
This machine was offered up over a month ago (on alt.sys.perq, of all places!) and I jumped on it.  It was located in Baltimore, MD but the seller was gracious enough to go through the trouble of crating it up for me while I arranged for freighting.  It includes the PERQ system, the Landscape monitor (19"), digitizer tablet, all cabling, a set of manuals, and the original software on 8" floppy.

The PERQ T2 was essentially the end of the line for the venerable PERQ line of workstations.  I also own a PERQ 1A (one of the first PERQs -- read more here!) and I honestly never thought I'd find another one -- they're just that rare.  (For reference, my 1A has serial number 259, and the T2 I just acquired has serial 2264.)

The main CPU unit.
The PERQ is a fascinating machine, a very early graphical workstation (created by ex-Xerox engineers ticked off at Xerox's lack of ambition with the Star) and a design several years ahead of its time when it was introduced in 1980 and still pretty competitive in 1985, when my T2 was built.  The PERQ is a soft-microcoded machine (meaning that the microcode store is loaded from disk at boot and can be modified) designed to run Pascal P-Code (called "Q-Code" on the PERQ) very quickly (approximately 1 million Q-Codes per second).  The T2 is essentially a facelift on the original PERQ 1's hardware.  It uses the same CPU, memory, and I/O with a few updates -- a landscape 1280x1024 display (at 60Hz, non-interlaced) and using larger capacity 5.25" MFM drives instead of the gargantuan 14" 24 MB drive in the original.

Right after the initial boot of POS.
At some point I'll write up an article about the PERQ and its history.  It's one of my favorite machines, and I liked it so much I even wrote an emulator for it.  I encourage you to check it out if you're curious to see what an early graphical workstation looked like.

This particular T2 was sold as part of a drafting package from Graphic Horizons, Inc. called "GRAPH/NET."  It cost upwards of $45,000 in 1985.  The high resolution display and graphical interface was a perfect fit for CAD applications.

An example of a graphical application on the T2 -- a simple drawing program (named Draw).


The PERQ arrived in my driveway yesterday and I spent last evening unpacking it and testing it out.  It powered right up and after giving the hard drive a small nudge it spun up and booted into POS.  The drive (a Vertex V150) is not long for this world, so I'm tracking down a replacement and working on a utility to image the original drive.

Crated PERQ

Well, that's it for now.  Here's hoping I can get back into the habit of updating this regularly.  (Heck, here's hoping I have time to work on things worthy of mentioning here...).  Until next time, don't let the sun catch you crying...