Data General NOVA4/X recovered from Bakersfield

A collection of photos of the machine, in no particular order, many without captions.

Here are a few photos from the trip to retrieve it, 17 Jan 2005.

Here the system sits in the middle of the floor in my lab, taking up a lot of space, while I renovate it. It'll live against the wall (not visible) when done and I can actually work on something besides the Nova.
It looks like I'm keeping it as a two-rack system so that I can use both drives. Not visible in the right cabinet is the DGDAC 4300 Digital and Analog I/O system. I'd love to trade these two 6021/6023 vacuum column tape drives for simple side-by-sides; two would fit in the space of one drive with room to spare!
Tape drive, door open, tape mounted. You can (barely) see the tapes loaded into the vaccuum columns, but you can't hear the roar of the vaccuum-cleaner sound sucking the tapes into place.
Data General made their own tape drives. It looks it. That's not a put-down; it's simply a very straightforward design, very easy to maintain, operate and repair, it's all standard parts (except of course custom castings etc). Those big reel motors have torque; they will break your fingers if you argue with them. That's what it takes to accelerate a 4 pound reel of tape to 75 inches/second.
Most of the mechanical and logical guts are mounted on the front panel casting, which is hinged on the right. There's the big interface cable and the grey vacuum hose visible. The junk around the inside of the cabinet is the power supply and the giant opamps that drive the reel motors.
Slightly better look at all the inner junk.
The disk drive, also made by Data General, is fairly large. It contains two 14" double-sided platters; one fixed, one removable. It lives pushed into the rack cabinet. You shut down the drive (front panel "LOAD"), then unlatch and slide it out.
It pulls out about 24" and weighs a ton. The only reason the rack doesn't fall over is the tape drive is even heavier. Here you can see the removeable cartridge where it lives in the drive.
The cartridge installed in the drive, with the dust cover lifted off for the photo.
The disk cartridge lives in this hole. Under the flat aluminum cover is the fixed platter. The heads live retracted in the center opening in the cylindrical wall; you can see the four heads (two platters, top and bottom surfaces for each) ready to go. On the right a brush swings out of the larger opening to sweep dust from the surfaces right before head load begins. The brown ring in the middle of the cavity is the magnet that holds the disk platter; the index pulse sensor is visible in front of it.
The head assembly is in the center of the photo; the board on upper left is the linear motor servo system, and barely visible on the right is the logic board, which is about 8" high, 24" long, and lives vertically along the right edge. The trianglular box with what looks like a toothbrush is the platter brush.
It's a bit hard to see, but this is the head assembly looking from the left of the chassis. This is the view for cleaning heads. There's a lucite cover over the heads, you can see a keyhole and screw. Loosening the screw allows the cover to slide back enough to get a cleaner in there. When retracted (as shown) the heads lift, by rubbing on a wedge cast into the surrounding housing. The lifting wedge is only on one side, so the metal holder the head (circular whitish disk on the right, with a spot of orange paint on the back) is severely twisted. You'd think it would ruin head alignment (roll) but they were parked for 20 years with no harm done it seems.
Hmm, no photo of the closed-up cartridge. Use your imagination; it looks like a UFO. The handle has a slide latch that lifts the magnetic latch, letting the dust cover come off. The mag latch also holds the platter to the rotating spindle in the drive.
Here you can see the bottom surface of the removable platter; this is the crashed cartridge, otherwise I wouldn't be waving it around outside, open.
This is the Read/Write Board, component side, that I suspect is the source of my excessive read errors. It's quite a botch job, covered in mods (including milled-out access to a middle layer!), doesn't match the documentation, and has a trimpot kludged in.
This is the Read/Write Board, solder side, lots of trace cuts and jumpers visible, as well as the cutout for middle layer access.
Pretty clean inside; there's power wiring built into the rack, a single 20-amp circuit breaker down near the floor. The big fat cables are 100-conductor jobs for the tape and disk. The disk is on the bottom, you can see the folding cable assembly all folded up.
This is a Data General D410 terminal; it did not come with this system, which was headless, since it was essentially an embedded controller. I got this with a bunch of other D.G. terminals from a business in Redondo Beach that recently (!) upgraded computer hardware. You can easily see the burned-in menu on the CRT... The keyboard is a little crunchy but it works fine otherwise.
Nice keyboard layout, except for the odd LF and CR keys. Some DG programs terminate lines with LF, some with CR. Ahh, the good old days. 1967ASCII was only two years old when the first Nova computer shipped. The keyboard is made out of cast metal!
If you're not familiar with traditional 1/2" magtape, here's a few photos for you.
There's a latching cover over the tape; there's a hook for hanging on a tape rack, and a place for a label.
The yellow ring on the back of the tape reel is a removable insert called the "write ring" that is removed to write-protect the tape. A mechanical sensor pokes it when you mount the tape; the poker retracts so it doesn't drag on the reel.
The tape, when mounted, is held under tension by either spring-loaded arms or vacuum columns, as in this tape drive.
The backplane is symmetrical, with no wiring except for I/O and the occasional jumper.
There was really no need for the 16-slot backplane; salesmammal probably wanted a bigger commision, or maybe the factory had a lot of them laying around and they got it cheap.
Some Einstein dymo-labelled the Nova 4 CPU so the programmer wouldn't get confused. That's proper fake-woodgrain Dymo, too, very 1970's. The blue tape on the left is holding the front panel on until I take the chassis out of the rack to adjust the hold-down clip.
Nova cards are 15 inches square. Some, like this BASIC CONTROL card, the 4007, have nearly nothing on them; the rest is "options" like EIA (vs. current loop; this card has it) and the multi-speed option (the little daughterboard). Just enough TTL to make a UART.
OK, this board is high density. It's 4 or more layers, it's got 15 AMD 2901's, does floating point arithmetic and makes a lot of heat. I hope I never have to debug and repair it.
This is the hard disk controller, basically a bunch of registers and a data channel, aka "DMA".