Bare CPU Chassis Frame
As with the PDP-8 replica project the PDP-1 replica was also constructed mostly out of materials found out my local hardware stores, with some specialty items bought from electronics surplus stores in the Silicon Valley and certain hard to find items (like switches) mail ordered from surplus joints across the country.
We see here the bare naked skeleton of the main chassis in all its simplicity, constructed of 2x4 and 2x2 lumber and metal framing brackets. The gaps that would be visible once the doors were mounted were filled in with putty and sanded down to give a smooth, continuouss appearance once the paint was applied.
View of Chassis Interior From Rear
This is a shot of the interior of the CPU chassis viewed from the rear looking forward. Some of the front panels have been mounted.
You can see the upright supports that also act as framing for the side doors are assembled using metal 'T' brackets.
The platform at the bottom of the picture is where the simulator PC sits.
View of Interior Chassis From Front
Looking towards the rear of the chassis internally, we can see more of the support members used to provide stability to the chassis. The angle brackets in the corners of the frame provide horizontal stability, as do the cross-members between uprights (bottom of photo).
The interior was painted for a cleaner overall appearance and to avoid exposing any bare wood to viewers.
The white blocks against the back panel are actually affixed (glued) to the back panel. These are the mounting blocks for mounting the back panel to the chassis frame. Holes were pre-drilled and countersunk, and the panel is then fitted and fixed into place with wood screws.
Interior View of Door Frame Joint
Another view of the interior frame assembly. Painting all surfaces whether exposed or not helped complete the illusion of reality.
Side Door Construction
The side doors were constructed of 3/4" particle board. The edges were rounded with a file and sandpaper to give them the same contour as on the original doors.
The hinges are simply pieces of 3/8" diameter steel rod cut to length. A hole is drilled on the top and bottom edge of the door and the rod was pushed inside.
The top and bottom of the CPU chassis was lined with sections of angled aluminum (the original PDP-1 used stainless steel). These serve as decorative trim as well as mount points for the door hinges.
Holes are drilled on either end of the bracket. The door is then set in place and hinges are inserted from the top throuh the bracket and into the hole drilled into the door. The holes were drilled at such a depth to ensure the hinge pins were flush with the top of the bracket.
As on the original, the brackets are affixed to the chassis with screws on either end.
Door Magnet Latch
Magnetic latches where then affixed to the door and chassis so that the doors would stay firmly closed.
Top of CPU Chassis
This top view of the CPU chassis shows a better view of the trim. The trim is sectioned as with the original PDP-1, one sector for each set of doors.
The top panels are 1/8" press board fixed in place to mounting blocks attached internally to the CPU chassis.
External View of Frame Without Panels
This view exposes the front of the CPU chassis. This is the corner section where the front panel is mounted. The table top has supports that slide into internal mounting points on frame.
Each sectional panel on the front of the chassis is framed with aluminum trim. The trim is a smaller guage aluminum angle bracket which is affixed to the outer trim.
Bottom View of Front Panel Table Top
This view of the underside of the table top shows the supports fixed into the chassis frame.
Table Top Support Anchor
Internally, the end of the table top support is anchored to a piece of aluminum bracket fixed to the chassis frame.
The white blocks to the right are the mounting fixtures for the panel below the table top.
Table Top Support Mount
The table top support is inserted into a notched piece of 2x4, which is bolted to the chassis frame.
Rear of Paper Tape Punch Panel Assembly
This is the rear of the paper tape punch panel. On the original PDP-1, this area is where the paper tape punch mechanism would be mounted, and the slot is where the punched paper tape would be ejected.
The metal chute was fabricated from sheet metal and glued in place. The white mounting blocks are glued to the panel and then fixed to the frame with countersunk screws.
Front Panel Mounting Assembly
This view shows more of the mounting assembly for the front chassis panels. The top panel is a paper tape punch status display. The white cylinders seen are the back end of the crimp connectors used to simulate the lamp lenses (more on this below).
The console desk was a bit more simple to construct. The base was fabricated from 1" square steel tubing welded together. As with the CPU chassis, the front panels are made from 1/8" press board, but the side panels are constructed from sheet metal. The table top was outsourced to Pavl Zachary.
The CRT mounting pedestal is made from different sections of sewer pipe. The CRT enclosure was fabricated to specifications from 1/8" aluminum at a metal shop.
The IBM Model B typewriter is simply a typewriter, re-painted to match the rest of the replica. The actual PDP-1 console typewriter has an additional section on the bottom of the unit containing solenoids and sensors for actuating or detecting typewriter keypresses respectively.
Console Desk Status Panel
The underside of the console desk shows the CRT status panel. The lenses are the ends of butt splice connectors. The panel has been silk-screened with the labels as copied from the status panel on the actual PDP-1.
The table top is anchored to the desk frame with screws all along the side and center supports of the desk frame.
Interior and Side Views of Console Desk
This view shows how the side panel is mounted to the frame using metal screws. Holes were drilled into the desk frame and threaded to accept the mounting screws.
On the inside we can see the front panels mounted to the desk frame using the same scheme.
Plywood shims were glued between the desktop and the desk frame to provide a better mounting base for the screws and to also ensure the desk was at the proper height relative to the frame.
Internal View of Status Panel and CRT Mount
A direct view of the interior of the status panel, which also shows the bottom end of the CRT mount.
Bottom View of CRT Mount
The CRT mount is constructed from sections of sewer pipe. The base itself is a drain mount for a toilet. A 3" coupling is then inserted into the drain mount and fixed on four sides with bolts on the other side of the internal lip which will prevent the pipe inserted from the otherside from falling all the way through.
Top View of CRT Mount
This shows the CRT mount from the top. An appropriate sized hole was cut into the table top to accomodate the drain mount. The screw holes in the mount, which would have normally been used to affix the mount to the floor, were filled in with a steel epoxy, sanded, and then painted over with a chrome paint to give the illusion of metal.
CRT Mounting Assembly
This is an "exploded view" of the CRT mount assembly. On the left side I'm holding the base mount which is the toilet drain fixture. Moving right, the thin section of 4" sewer pipe is used to fill the gap between the inside diameter of the toilet drain fixture and the outside diameter of the 3" coupling. The 3" coupling is then affixed through the 4" pipe to the drain mount with bolts.
The last section on the right is the stem fixture that mounts to the inside of the CRT enclosure and is inserted into the mounting assembly on the desktop. It's made from a different type of toilet drain mount with a piece of 3" sewer pipe fitted inside. The 3" piece then mates into the mounting on the desktop, forming a stable stand for the CRT.
CRT Mount Stem
This is a photo of the bottom half of the CRT enclosure with the mount stem fixed to the base with screws.
Checking fit of CRT Inside Enclosure
In this photo I was checking the fit of the CRT after I had done some serious hacking on its corners in order to make it fit into the hexagonal enclosure that tapers towards the back.
On an actual PDP-1, the CRT is a round tube with a 14" diameter. While it is theoretically possible to find a similar tube and then adapt it to work on a PC, this would have been a project in and of itself. Instead, I chose the sane option of using a modern day computer CRT. Unfortunately, the aspect ratio of your typical CRT is always approximately 4:3. The width will always be larger than the height. This is not a problem, as the from will be covered with a round portal. However, the because of the aspect ratio, it is not possible to find a monitor with a 14" height that would also fit inside the enclosure with (following the 4:3 ration) an 18.7" width. The dimensions of the enclosure can only allow for a 14" width.
I found the largest CRT that could still reasonanbly fit inside the enclosure and have the tallest height possible. That still only left me with a 12" vertical, so I decided to make the portal only 12" in diameter instead of 14". Only the most die hard PDP-1 hackers would know the difference anyway, so this was a reasonable compromise.
Modifying the CRT Housing
Still, I had to make some major modifications to the CRT in order to make it fit. After removing the plastic shell of this Sony 21" CRT, I found that it was enclosed in a metal shielding. I spent many hours cutting and hammering away at the shielding until I finally pushed the corners in enough to clear the dimensions of the enclosure.
Modifying the CRT Housing
I did a much better job on the left side than on the right. The whole time I was cutting and hammering, I was afraid I would do serious damage to the CRT, from frying its electronics to making a wrong move and imploding it. In the end, the modification was a success and, although it was still a tricky fit, I made it work.
CRT Mount Points
The CRT has some fortunately placed mount points on the steel frame that I used to fix it to the inside of the enclosure. I fashioned brackets from the aluminum angle bar which mounted to the enclosure base and then to the CRT. To isolate the CRT from the enclosure and forstall the possibility of any potential shock hazard, I fashioned isolation shims from a nylon cutting board. These also propped up the CRT to an appropriate height inside the enclosure. The bolt attaching the anchor bracket to the enclosure base is further isolated with a nylon washer.
Rear CRT Mount Point
I constructed a similar mounting fixture for the back of the CRT. In this case I fashioned a long block from the nylon cutting board on which to mount the aluminum angle bracket and bolted the back of the CRT to the bracket for a sturdy mount.
Re-routing the CRT Cables
The monitor originally had its power switch and settings controls mounted on the front of the unit. I routed the control strip wires to the back of the CRT housing and mounted the control strip to the back panel.
The video and power cables run down the CRT mounting pedestal. The power switch wires were spliced to the new location of the power switch.
Control Settings Panel Relocated to Back of CRT
The control panel was actually cut from the old plastic CRT casing and mounted to the back of the replica CRT enclosure.
CRT Mounting Completed
Overall it turned out to be a very neat job and everything worked great. Unfortunately, the CRT turned up dead when it got to Japan and a last minute re-work had to be implemented. More on this below.
View Inside Front CRT Portal Panel
The front panel portal was mounted to the base of the CRT enclosure by way of an anchor stick. The anchor stick was glued to the base of the portal panel and then fixed to the CRT enclosure base with screws.
Interior View of Front Panel Assembly
The front panel is shown mounted to the CPU chassis. The wiring comes through the opening and connect to the front panel controller which is mounted directly above.
Front Panel Wiring
This close-up shot of the front panel wiring hints at the amount of work required to make all the solder connections. There are 119 "lamps" (actually yellow LEDs) and 44 DPST switches (of which only 24 were implemented).
The ribbon cables are standard 40-pin IDE cables that have been stripped and soldered to the cathode of an individual LED. The anode of the LEDs were then soldered to a common ground bus (in this case standard 24 guage telephone wire) which was routed to a ground bar affixed to the CPU chassis.
Whereas the original PDP-1 used lamps, I chose yellow LEDs for a number of reasons, not the least of which was reliability and ease of sourcing. Also, yellow LEDs, once filtered through the lens, gives off a color similar to the lamps on the original PDP-1, which were run at a low voltage to preserve their life cycle and as a result were yellow in appearance. The lenses on the actual PDP-1 are 5/16" diameter opaque covers over the bulbs.
The lamp lenses on the replica are actually clear nylon butt splices for electrical wire. The top of the butt splice is a perfect cylinder, 3/8" in diameter, and almost completely flat on the top edge except for a small nub left over from the manufacturing process. The bottom half is flared out, and by sheer coincidence perfectly accomodates a round LED clamp. The clamp is inserted into the bottom of the butt splice, then the LED is inserted in the clamp and a dab of hot glue is applied over the LED and clamp to lock them in place.
Front Panel Controller Board
The front panel controller was designed by my friend Andre Lamothe of Nurve Networks.
It's a simple design consisting of 16 8-bit cascaded shift registers for the lamp output and 8 8-bit cascaded shift registers for the switch input (128 outputs, 64 inputs). It interfaces to the PC by way of the parallel port.
The desired status of each LED is shifted into the shift registers, which then latch. To read the switch settings, you simply shift in the input shift registers.
The board puts out enough current to drive the LEDs at ample current for a good, bright display.
Each 40-pin header configured horizontally are the output pins, numbered sequentially from 0 to 127 (top to bottom). The last 6 pins on the last header are shorted together and are the return ground. The vertically oriented 40-pin headers are the input pins, again numbered sequentially from 0 to 63.
This board also has the capability of being cascaded to another similar board by way of the dual 10-pin headers at the top of the board. One header is in and one is out, allowing the boards to be daisy-chained.
Front Panel Diagnostic Display
The front panel labels were accurately reconstructed based on measurements and placement on an original PDP-1 front panel. The label mask was created using Visio. The font is a near 100% spot-on match, and the grouping lines were painstakingly measured and positioned to within a millimeter of accuracy.
The control switches at the base of the front panel are surplus telecom leaf switches. The handles have been removed in this particular view.
The panel itself was stamped from 1/8" thick aluminum with a CNC press.
Here the front panel is being run through a diagnostic loop where the LEDs are made to flash in sequence across each section.
Front Panel Mounted to CPU Chassis
The front panel is shown here, complete and mounted to the replica CPU chassis.
Paper Tape Reader Mock-Up
Mounted above the front panel is the mock-up of the paper tape reader mechanism, with the mock-up paper tape punch chute in view as well.
The paper tape mechanism was fabricated using mainly sheet aluminum and items found in the plumbing section of the local hardware store. The paper tape hoppers, the front cover and the square cover (in the middle of all the fixtures) were manually cut and formed using 1/16" sheet aluminum. The large roller underneath the square cover is the drainage cap for a dishwasher overflow spout. The two spindles on either side are faucet handles. The roller on the left side is a 1-1/4" copper coupling with a rubber strip wrapped around it; the tapered cap is part of a showerhead fixture with a standard metallic hole cap fixed onto the end to close the hole. Finally, the "paper tape guide" mechanism on the right hand side was fabricated from scrap bits of press board shaped and glued together, then painted chrome.
Actual PDP-1 Paper Tape Reader
For comparison, here is the real PDP-1 paper tape reader. The replica didn't capture all of the little detail, but enough to effect the illusion.
Completed Replica at National Science Museum in Tokyo
A photo of the completed replica freshly installed in its exhibit space in the National Science Museum of Toyko, Japan.
The simulator PC is installed inside the CPU chassis and Spacewar! can be seen running on the display.
Full View of Actual PDP-1
Compare and contrast the real PDP-1 in the restoration lab at the Computer History Museum in Mountain View, California. This is the PDP-1 after which the replica was modeled.
Every edge was measured and every feature recorded in exceptional detail in order to produce a near exact replica.
Close-Up View of Replica CRT
During transit from the USA to Japan, the CRT suffered some manner of electronic damage that I could not successfully repair. I spent hours trying to determine the problem before giving up and replacing the big, bulky and heavy CRT with a nice and simple flat panel LCD display. In hindsight, I should have gone with a flat panel LCD in the first place, but it never occured to me because a converted CRT seemed like the natural choice. Unfortunately, all the work that went into fitting the CRT was lost. Oh well.
The LCD panel was removed from its original housing and mounted to the front panel portal of the CRT enclosure. The voids on the top and bottom were exposed areas of the LCD frame that were covered with black construction paper.
Here, the simulator PC has booted into SuSE Linux and is awaiting a login.
The Baddest Case Mod You Will Ever See
Now taking orders!
Replica Playing Spacewar!
Here, the replica is running Spacewar!, with the player ships visibile on the screen (click the photo for a larger view).
The simulator employed as the soul of the replica is the excellent, multi-platform and multi-target SIMH, written by Bob Supnik. The Type 33 graphics simulator code was developed by Phil Budne. Both Bob and Phil were extremely gracious in providing me with lots of generous assistance with getting the simulator package compiled and running.
The version of Spacewar! playing under simulation is the original 1960s MIT code executing unmodified within the simulated PDP-1 environment on the PC.
PDP-1 Spacewar! Controllers
The last bit of hardware to construct was the Spacewar! controllers. This set is loosely based on the design of the originals as depicted in a drawing from the August 1981 issue of Creative Computing (which was composed from memory by J.M. "Shag" Graetz). I took additional hints from an old photo showing Dan Edwards (left) and Pete Samson playing Spacewar! at MIT back in the early 1960s and using their original controllers.
These were made from off-the-shelf aluminum project enclosures, using heavy duty switches for ship control and a big red fire button to launch torpedos. The top switch is for rotate (left and right), the switch on the side is for acceleration (pull down) or hyperspace (push up), and the red button is of course for firing torpedos.
The switch contacts were wired into unused inputs on the front panel controller board and then OR'd (in software) with the correlated switch on the front panel. This is somewhat similar to the way the controllers were wired in to the actual PDP-1. It is useful to know that the original game was played with the 4 switches on either end of the TEST WORD row on the front panel. One player used the set of four switches on the left side and the other player used the set of switches on the right. The switch functions (in order) were rotate left, rotate right, accelerate, and fire. Hyperspace was effected by flipping both rotate switches simultaneously.
Last Minute Details
Here I am in my hotel room in Tokyo the evening before the installation of the replica PDP-1 at the National Science Museum making some adjustment to the controllers. The Tokyo skyline can be seen outside the window of the 43rd floor of the famous Asakusa View Hotel.
The Real Deal
This is a photo of the original PDP-1 after which the replica was modeled. This PDP-1, shown in the Restoration Lab at the Computer History Museum, was fully restored and made 100% operational by an amazing team of Museum volunteers.
For more information on this project, visit the PDP-1 Restoration Project website.
A Rewarding Endeavor
This final photo shows the completed replica in the completed exhibit at the National Science Museum in Tokyo.
This project was extremely rewarding. Over 300 hours went into its actual construction, plus several hours taking notes and making drawings of the original PDP-1, not to mention countless hours spent in the aisles of the local hardware store, staring at materials and wondering how they could be made to look like a piece of a computer.