The University of Iowa's DEC PDP-8

Restoration Log

Part of the UI-8 pages
by Douglas W. Jones
THE UNIVERSITY OF IOWA Department of Computer Science



This is a chronological log of the progress restoring the University of Iowa's PDP-8 computer. Entries are added at the end as work progresses. Click on any thumbnail image to see full-sized image.

Jan. 7, 2014, TTY Work, ADC Rear Plenum Panel

  thumbnail image  
TTY Reader
Bug 7: There was no problem mounting the TTY reader on the repaired reader frame. Final alignment will need to wait until the cover is repaired and put in place. The switch handle on the reader remains to be re-attached; it is wedged into a spot in the reader lid. Gluing it will be a bit tricky.

  thumbnail image   thumbnail image  
The ADC rack rear plenum door
Bug 18: The large hole in the scrap panel described on Dec. 12, 2013 was patched with a chunk of aluminum cut to the size of the hole and clamped in place by multiple screws. This repair minimizes the need to modify the panel, since it is, itself, a relic of the University of Michigan's PDP-7A, serial number 120. The patched-up panel fit perfectly in the space in the rear plenum door, although it is held in with inappropriate screws (common hardware-store 10-32 screws have the wrong head style to match those that originally came with the rack.

The stickers on the replacement plenum door panel are transcribed here, since they are a physical record of at least part of the configuration of the University of Michigan's PDP-7 -- parentheses surround guesses about the content of these stickers lost to the saber saw used to slice the hole in the panel in its later days at the University of Michigan:

07-1003-0100 Type PDP-7A-120

07-1003-0300 Type 444B

07-1003-4700 Type 426

07-10(03-???? Type 75D)

07-1(003-???? Type ???)

07-100(3-???? Type 649)

07-1(003-???? Type ???)

07-10(03-???? Type 647)

07-100(3-???? Type ???)

07-1(003-???? Type ???)

07-1003-1200 Type 149

07-1003-4400 (Type)  

The guesses at the type numbers in the above list are based on material from the Digital Equipment Corporation PDP-7 Options List.

Jan. 23, 2014, Introduce Students to Project

6 students have signed up for 22C:199 (old numbering) or CS:5990 (new numbering, how nice), officially titled Indiv Research or Programming Project, with the goal of getting the machine up and running.

We held an orientation in the lab, showing work that needed doing, and including an introductory lecture on the theory of the power supplies used in the PDP-8 -- transformers, diodes, and capacitors. leading up to a discussion of how electrolytic capacitors are constructed, a bit of the underlying electrochemistry, and the whole idea of how they are reformed.

The lab now contains (on loan) a tool box containing a soldering gun, a small soldering iron, a spool of solder, a small set of screwdrivers, and an old Radio-Shack VOM. Also on loan, a variable 1.5 amp DC power supply good from 0 to about 18 volts. This should sufice for reforming the 10 and 15 volt capacitors in the PDP-8. We purchased five 520 ohm resistors for use as current limiters during capacitor reforming, along with a stock of clip leads.

  thumbnail image     thumbnail image  
Capacitors to reform
Bug 10: The basic proposal for reforming the capacitors is as follows:

  1. Unsolder one terminal of each of the big electrolytic capacitors in a bank.
  2. Tack-solder a resistor to the free capacitor terminal.
  3. Use clip leads to connect the resistors to the appropriate power supply lead.
  4. Clip the other power supply lead to the still-ganged side of the capacitor bank.
  5. Turn up the voltage, monitoring the voltage drop across the resistors as a proxy for measuring the leakage current. Do this in stages until we reach the operating voltage of the capacitor (or at least the operating voltage of the supply it is in).
  6. Unsolder the resistors, for reuse reforming other capacitors.
  7. Resolder the successfully reformed capacitors into place.
It should not be necessary to actually remove any capacitors, except those that have gone bad (either short-circuited or dried out -- indicated by having inadequate capacity). With 5 series resistors, we ought to be able to reform capacitors 5 at a time, unless some have gone bad.

Bug 5: While there, we lowered the leveling feet on the PDP-8 CPU rack so that it can no-longer be rolled around on its casters. This should significantly reduce the risk of pulling the CPU out from the rack -- the rack is stable with the CPU fully extended, but only just. Removal of the power supply or would not be wise without first installing a good solid prop under the extended CPU -- something resembling a sawhorse that would reach up behind the front panel and support the weight of the CPU's steel frame would be ideal, until or unless we can recreate a table like the ones that DEC sold with the machine.

  thumbnail image     thumbnail image  
Preliminary maps
Bug 19: We need to do the following:

  1. Map and inventory the boards in each half-backplane of the computer. This involves comparing the boards in each slot of the backplane with the maps in the maintenance manuals in order to determine what board is in each backplane slot. In most cases, the maintenance manuals and the backplane boards will match, but: We have two versions of the maintenance manual, and they do not always agree -- there seem to have been variations in the details as successive models were built, and, worse, in the years the machine was idle, some boards may have been pulled for "showing off" and reinserted in the wrong slots.
  2. Pull all the boards (the map is crucial to putting them back correctly).
  3. Dust the backplane.
  4. Dust the boards.
  5. Re-insert all the boards, each board going precisely where it was pulled. Some boards (time delays, for exmple) are customized to specific backplane slots), so boards with the same number are not always interchangable.

See David Gesswein's PDP-8 backplane restoration web page for a description and photos of a similar effort.

Jan. 24, 2014, More Introductions

In addition to introducing a student to the project who couldn't make it the day before, we got some grubby work done: Paper towel and spray cleaner work on the grime on the CPU cabinet. Cleaning just the very top turned one sheet of paper towel the most disgusting shade of grey, but the top is much less disgusting now. There's plenty more grime that needs removal, but that's a start.

Also, the set of tools on loan to the lab tool box has grown by the addition of long-nose pliers, nippers (long nose pliers with wire cutting edges at the tip of the jaws), and a rather grubby but still functional automatic wire stripper.

Feb. 2, 2014, PDP-8 Props and Cleaning

  thumbnail image  
The prop
Bug 5: A prop was made from scrap 2×4 lumber and plywood to support the weight of the computer when fully extended from the rack. This will allow us to remove the side skins of the rack and part or all of the power supply with no fear of the rack falling. We're planning on leaving the prop in place with the computer extended from the rack until we finish with the power supply and with cleaning the backplane.

Bug 2: The loose cards on the sides of the core memory have been plugged back in, although they are likely to be removed again during backplane cleaning.

The students working on the machine have done considerable cleaning, so the mud splashes around the bottom of the machine are gone, as is most of the dust and grime on exposed surfaces.

Feb. 10, 2014, Disconnect Power Supply

  thumbnail image     thumbnail image     thumbnail image  
Power supply wiring harness
Bug 10 and: Bug 20: We had hoped to be able to access the capacitors in the PDP-8 power supply for reforming without removing the power supply from the rack, but aside from the one row of capacitors that are mounted vertically, all of the others look difficult or impossible to access without at least partially disassembling the power supply. In fact, the more we looked at it, the more difficulty we had seeing how the supply was assembled in the first place. So, we decided to remove the supply from the rack, creating a new problem, the need to reassemble the supply and reinstall it in the rack. Matt Adamczyk's cellphone takes very good close-up photos.

We disconnected the wiring harness at the back of the power supply; this involved pulling several dozen Faston connectors, both the power supply output connectors ranked across the back top of the supply, and the AC power connectors located in the left top rear corner of the supply. We took photos of all the connectors before and after disconnection.

Two of the male Faston connectors had broken from their colored plastic insulated mountings and had been insulated with medical-style adhesive tape. One, connected to the green leg of the inhibit thermistor, was still routed through the broken plastic housing of the Faston connector. The other, part of the power panel, was simply routed around the panel. We had to remove the tape in order to disconnect these two connectors.

  thumbnail image     thumbnail image     thumbnail image  
Wires to front circuit breaker
One cable went from the wiring harness at the back of the rack forward under the power supply and up to the circuit breaker in front, where the power cord was also attached. This was clamped to the bottom left edge of the supply by three screws. This also appears to be a poor example of design for manufacturability -- we had to remove (and tag) the cover over the back of the front circuit breaker and then unbolt the power cord and this cable from the breaker. We concluded that the only way to remove the ground connector from the power cord was to get a socket wrench, deferring this step a day. The ground wire was bolted to one of the capacitor clamps in the supply.

Bug 16: A long trailing cable with a cut end was found among the tangle of wires in the bottom of the PDP-8 rack. One end included several wires that joined the power harness at the back of the PDP-8 power supply. Once untangled, the other end was found to be cut, exactly matching the cut power wires previously noted in the ADC rack. This cable was tied into a coil and tied to the bar on the back door of the PDP-8 rack to keep it out of the way. It needs tagging.

Feb. 11, 2014, Power Supply and TTY Work

  thumbnail image     thumbnail image     thumbnail image  
Detail photos of TTY disassembly
Bug 25 and Bug 26: Harrison Picket has taken on the job of Teletype cleaning, removing the keyboard and typing platten so he can get to the bits and pieces that are in need of degreasing. The Teletype case is now very clean. By the end of the day, he had disassembled the machine into its major components - keyboard, typing unit, call control unit (local power supply), etc. He took these photos of his work.

In the process of disassembling the Teletype, Harrison identified several parts that need replacement. In the following list, part and page numbers are references to

The parts in question are:

  thumbnail image     thumbnail image  
Decayed foam and vibration mounts
Bug 21: The foam rubber pad under the typing unit was decayed and crumbled when disturbed. In the photo, Harrison shows the fingerprint he left in the pad by poking it. This is part number 181138, pad, pages 25 (drawing) and 32 (parts list).

Bug 22: The rubber grommets that serve as vibration isolators were decayed and crumbled when disturbed. This is part number 181109, mount, vibration, pages 25 (drawing) and 32 (parts list).

Bug 23: One teletype sub-base mounting bolt was missing. and two more appear to have been inappropriate substitutes. We have not yet located this in the parts list.

Bug 24: One call control unit mounting screw was not installed (left center screw). We have not yet located this in the parts list.

  thumbnail image     thumbnail image     thumbnail image  
The supply out of the rack
Bug 20: Matt Adamczyk and Douglas Jones removed and set aside the I/O bus cables from the computer and then removed the last wire connected to the power supply. Two people working together could just handle removing the supply. We could have used a third person! What we did is remove the 2 screws on each side holding the supply to the front rails of the rack, and then the 3 screws on each side holding the supply to the rear mounting ears. This allowed us to lower the supply below all obstructions and then back it up and remove it through the open side of the rack. Matt photographed the empty rack and power supply work.

With all the cables out of the rack bottom, we removed the cooling fan, cleaned the dirt out of the air filter, cleaned the grime off the bottom of the rack, and then re-assembled the closure around the cable holes and re-attached the fan.

  thumbnail image     thumbnail image  
C12 — C14 dewired and reforming
Bug 10: With the supply on out of the rack and on a work table, we set to work reforming capacitors, starting with the three capacitors that are most accessible, C12, C13 and C14, filtering the red (+10V) line in the supply. These are 35,000µF 25V capacitors. (See schematic, Page 10-8 in the 1966 Maintenance Manual) We used a 100W solder gun to unsolder the red wire from the positive tab of each capacitor, and then tack soldered a 510Ω resistor to that tab. We then clipped the negative output of a variable DC supply to the black (ground) wire joining the negative poles of the capacitors and clipped the positive output to the free ends of all 3 resistors.

First, we turned the voltage up to 1/2 volt and checked the voltage across the capacitors. It rose rapidly to 1/2 volt across all three capacitors in under a minute.

Then, we raised the voltage to 1 volt and checked the voltage across the capacitors. It rose less rapidly to 1 volt across all three capacitors.

  thumbnail image     thumbnail image  
The capacitor reforming setup
The photos here show the complete reforming setup we are using, an unregulated DC power supply with voltmeter and ammeter, with clip leads to the capacitors being reformed. Zoom in on the overall photo, and you can see that the power supply is reading about 12 volts. A separate voltmeter is shown, set to it 0.6 volt scale, (a Radio-Shack VOM) is shown, set to it 0.6 volt scale, being used to monitor the voltage drop across one of the resistors. Zoom in on the photo, and you can see that it reads 0.3 volts, indicating that the capacitor being monitored is currently leaking about 0.6mA.

We continued raising the voltage in small increments, monitoring the voltage drop across the resistors in order to keep the current under 1mA. Within 15 minutes, we had the voltage up to about 12 volts, and we left the power on for the night. At our final check of the current, it was still under 1mA at each capacitor. Judging by this, our first 3 capacitors are in good shape. We need to measure their capacity by discharge through known resistances after they've "cooked" for a while at their working voltage.

Feb. 12, 2014, Power Supply Check

Bug 10: A brief visit to check on the state of the capacitors after 24 hours under power. With the voltage hovering at 14 volts, the leakage currents on the 3 capacitors were roughly 0.04mA, 0.01mA and 0.01mA -- excellent values, at the lower limit of what we can detect on the 0.6V scale of our voltmeter when connected across a 500 ohm resistor. We turned the voltage up to 15 to let them cook for another day.

Feb. 13, 2014, Power Supply Disassembly

Bug 10: The capacitors were comfortable at 15 volts, with leakage currents under 0.02mA, very hard to measure with any accuracy. On the most sensitive current scale on our Radio-Shack VOM, the current slowly fluctuated by tens of microamps, sometimes even reversing polarity.

Discharging the capacitors through the 510 ohm resistor, the discharge rate was under 1 volt per second, suggesting that their capacities are at or greater than the rated 35000µF. This may be because they have only reformed to 15 volts, so the voltage was worked up to 24V, the rated working voltage of the capacitors, and they were left to cook for another day, which will probably thicken the dielectric layer and reduce their capacitance to near the rated value.

  thumbnail image     thumbnail image  
Pop-rivets and their removal
Bug 20: It is not at all obvious how the power supply was assembled, but the fastest path to accessing the terminals of the capacitors mounted on the back panel of the supply seems to be to remove the back panel. All of the other pieces of the supply are screwed together, but the back panel is attached with pop rivets, 5 on each side. (An additional 4 pop rivets on the left side attach a block of 8 backplane slots, used for the G-series flip-chip boards in the power supply. These are not part of the problem here.) All of the left side pop rivets are visible in the first photo.

The pop rivets were drilled out with a brace and countersink bit. After the heads of the pop rivets were removed, the countersink bit was used to countersink the rivet hole in anticipation of eventually replacing the rivet with a flat-head screw. The countersunk holes with the rivets removed can be seen in the second photo. We will probably use swage nuts press-fit into the holes of the bent ears of the back panel.

Feb. 13, 2014, Power Supply Check

Bug 10: After a day at 24V, C12, C13 and C14 were drawing less than 0.05mA, measured by the slight voltage drop across the 510Ω resistors. The capacitors were discharged to 15V through the 510Ω resistors in order to note the voltage drop per second at 15V. It was a bit under 1 volt per second on all three capacitors, so they were fully discharged and then marked with dated green striped tags indicating the voltage to which they had been tested.

Bug 20: With care to avoid stressing the wires, the back panel of the supply was eased out. This required loosening some screws so that one side panel of the supply could be eased sideways to clear a rivnut, and it required tilting the panel to clear the backplane connector block on the left side. Once loose, it was clear that the slack in the wires would not let the rear panel tilt, even with many of the wire ties clipped. However, if the supply is lifted an inch or two, it looks like the rear panel can be swung under the supply as it is tilted. This will have to wait until we have props.

Feb. 17, 2014, Power Supply Progress

Bug 20:
  thumbnail image     thumbnail image  
Spreading the supply sides
to drop the back
We were eventually able to lower the back of the supply to the table top so the tops of all the capacitors were exposed. This required the following:

  thumbnail image     thumbnail image  
Drilling to clear swage nuts
With the rear lowered, we drilled out the holes in the side of the supply that used to hold pop rivets to clear 8-32 screws, and we drilled out the corresponding holes in the side flanges of the back of the supply to the correct size for the 8-32 swage nuts we located. These were salvaged from old equipment so are of unknown make but they look identical to the Flexithread™ SWX-832 non-locking swage nuts. The photo shows Cale Bierman drilling a former pop-rivet hole to clear a number 8 screw.

To press in the swage nuts, we used a steel screw, with a square nut and stack of washers on it. The screw (with nut and washers in place) was pushed through the hole into which the swage nut was to be anchored, and then screwed into the swage nut, pulling it into the hole. When the knurled part of the swage nut began to dig in, we used a wrench on the square nut in order to pull the swage nut firmly into its hole and lock it in place.

The procedure for pressing in the swage nuts sounds simple, but in several cases, the clearance between the hole in which the nut was to be swaged and the capacitors mounted on the supply back was too small for a finger. After fumbling with long-nose pliers, we eventually found that we could stick a swage nut to the side of a cotton swab by pulling out a small a tuft of cotton and screwing the nut onto it. Holding the stick of the swab in a pair of needle-nose pliers, we could position it behind its hole and screw our installation tool (the screw-nut-washer stack) into it.

  thumbnail image  
Soldering Lessons
With the back of the power supply down, a bank of 6 capacitors is exposed. These are C6 through C11, filtering the blue (-15V) line in the supply. These are 35,000µF 25V capacitors. (See schematic, Page 10-8 in the 1966 Maintenance Manual) As with the previous bank of capacitors, we used a 100W solder gun. This time, we unsoldered the blue wire from the negative tab of each capacitor, and then tack soldered 510Ω resistors to 3 of the tabs.

None of the students involved in the project had any soldering experience, so everyone present had a turn at unsoldering a capacitor. The photo shows Cale Bierman taking his turn. Once the unsoldering was done, the soldering lessons were extended to include tack-soldering the resistors used during reforming and re-soldering the wires previously removed from C12-14.

Bug 10:
  thumbnail image  
At work while
capacitors reform
After the soldering lesson, we began reforming 3 of the capacitors. We will arbitrarily refer to these as C9, C10 and C11. The circuit we used was essentially the same as the one used for C12-14, but with the opposite polarity: We clipped the positive output of the variable DC supply to the black (ground) wire joining the positive poles of the capacitors and clipped the negative output to the free ends of all 3 resistors. We then turned the voltage up until the maximum current we measured was 1mA (measured by an approximately 0.5V drop across the 510Ω resistor).

We installed more of the swage nuts while we watched the current through the capacitors slowly drop. The photo shows Matthew Adamczyk at the drill while the clip leads powering the reforming capacitors are visible above the rear of the power supply. Within a few minutes, the current had dropped, so we turned up the voltage. One capacitor drew considerably more current than the others, so we monitored it most closely as we raised the voltage slowly to 5 volts, where we left it to "cook" overnight.

Feb. 18, 2014, Power Supply Check

Bug 10: After a day at 5V, C9, C10 and C11 were drawing less than 0.05mA, measured by the slight voltage drop across the 510Ω resistors, so we turned up the voltage over an hour to 12V, never letting the current exceed 1mA. While waiting for the current to drop after each voltage increment, we did some housecleaning and finished up various little tasks that weren't noteworthy.

Feb. 19, 2014, Power Supply Check, TTY Work

Bug 10: After a day at 12V, C9, C10 and C11 were drawing less than 0.05mA, measured across the 510Ω resistors, so we gradually turned up the voltage to 18V, keeping the current under 1mA, and then an hour later, turned it up to 25V keeping to the same limit.

Bug 7: While waiting for the current to fall in the capacitors, there was time to work installing the replacement control lever on the Teletype's paper-tape reader. This is Teletype Part Number "183045 lever" according to Section 574-124-800 (Issue 1, April 1964) of Bulletin 1184B — Parts, Model 32 and 33 Page Printer Set.

Wayne Durkee sent us a replacement lever. In the limited time available, we got the old lever out and installed the new lever, but did not complete the job of re-installing the contact block into the paper-tape reader.

Feb. 20, 2014, Power Supply, TTY Work

Bug 7: The job of re-installing the contact block into the paper-tape reader was finished. A jeweler's screwdriver was needed to carefully poke each contact wire into the bottom of the correct sensing pin. Once everything was put back together, the mechanism was tested mechanically by rocking the armature of the advance mechanism with a screwdriver. The tape advance sprocket rotates correctly, the hole sensing pins rise to sense the tape, and all the contact wires make or break contact with the common contact bar as they should. Further testing will require powering up the Teletype.

Bug 10: After capacitors C9 to C11 in the PDP-8 power supply had spent a day at 25 volts, all were drawing under 40 microamps, so we tested them by discharging to 15 volts and then measuring the rate of discharge through the 510Ω resistor. All were good, so we labeled them and began work on C6 to C8, working up to 7.5 volts while keeping the current under 1mA.

We need a bigger soldering iron. The 140 watt soldering gun doesn't put out enough heat to resolder the 4 heavy wires that go on each capacitor terminal in this bank.

Feb. 21, 2014, Power Supply, TTY Work

Bug 10: The voltage on C9 to C11 was raised to 15 volts, ready to bake for at least a day.

Bug 17: Teletype work continues. We purchased a plexiglass window to replace the broken window in the Teletype cover, and Doug Jones planed the paper-tear edge to a sharp edge using a block plane. The planing was done before removing the protective scratch-guard sheets from the window, with the window clamped in a vise between two thin softwood boards (old cedar paneling) to control the angle of the planed surface. (Plexiglass planes well with the plane set to take off the thinnest possible shaving.) After planing, the original window was clamped to the replacement and used as a template for drilling the mounting holes and planing the square edges of the new window. Aside from the lack of polish on the planed tear edge, the new window is a good match for the original.

Bug 25: Harrison Picket has disassembled and degreased the entire keyboard, wiping down all of the code bars with GooGone™. The grease in the mechanism was very sticky -- apparently, whatever volatile content had formerly made it slippery had long-since evaporated.

Feb. 24, 2014, Power Supply, TTY Work

Bug 10: The voltage on C9 to C11 was raised to 20 volts, ready to bake for another 24 hours. As usual, we limited the current to 1mA as the voltage was raised.

Bug 17: We purchased a bottle of Plastruct Bondene solvent cement, in order to repair cracks in the ABS parts of the Teletype. While waiting for the capacitors to charge up, we used this to bond several cracked ears for different parts of the Teletype.

Feb. 25, 2014, Power Supply, TTY Work

Bug 10: The voltage on C9 to C11 was raised to 25 volts, ready to bake for another 24 hours. As usual, we limited the current to 1mA as the voltage was raised.

Bug 17: The broken surfaces glued on the 24th were inspected and reinforced with additional ABS material (1/8 inch Plastruct channel) glued to surfaces where it would not be visible from outside the Teletype and where it would not interfere with any of the mechanism inside. We cannot find the broken-off piece of one finger; we may need to manufacture a replacement for it.

Feb. 26, 2014, Power Supply

Bug 10: After a day at 25 volts, C9 to C11 had leakage currents under 20µA, as measured by the barely perceptable voltage drop across the 510Ω resistors. The capacitors were disconnected and their capacity estimated by observing the rate of discharge when shorted by 510Ω as the voltage crossed 15V. The rate was about 1 volt per second, on target, so the capacitors were declared good and the reforming rig was moved to C1 to C4.

C1 ro C4 are 50 volt capacitors, significantly smaller capacity than the ones we have reformed up until now. For these, we will limit the current during reforming to 0.5mA and move the voltage up 7 volts at a time. We left them at 7 volts.

Feb. 27, 2014, Power Supply, TTY Work

Bug 10: C1 to C4 all had imperceptable leakage after a day at 7 volts, so the voltage was gently raised to 15V.

  thumbnail image     thumbnail image  
Repairs to the Teletype lid
Bug 17: The Plexiglass window on the Teletype cover was held on by 9 posts that fit through holes in the window and were then melted to rivet the window on. Whatever the accident was that broke the window also broke all 9 posts, leaving stumps of varying height. The replacement window made on Feb. 21 was clamped over the nubs of the old posts to serve as a drilling template. Each location that had formerly sported a post was drilled out to form a flat-bottomed hole approximately 1/32 inch deep (carefully! the lid is only about 3/32 thick here). Where the former post had broken off leaving a concave scar, the drilling was done with a specially modified 1/8 inch drill chucked in a pin vise. This drill was ground to have a flat nose like an end mill. Where the former post protruded from the base, a regular 1/8 inch drill was used to form a shallow concavity first.

New posts were fabricated from 3mm ABS rod, with the cut end filed flat so the replacement posts were stable standing vertically on end on a table top. These posts were then glued into the holes in the cover with Plastruct Bondene solvent cement.

Feb. 28, 2014, Power Supply, Cleanup

Bug 10: C1 to C4 all had imperceptable leakage after a day at 15 volts, so the voltage was gently raised to 22V.

Bug 3 and Bug 11: The decayed foam on the PDP-8 cabinet sides and one side of the cabinet door were cleaned out. The foam strip on the other side of the door is harder to get at.

Mar. 2, 2014, Power Supply, TTY Work

Bug 10: C1 to C4 all had imperceptable leakage after a day at 22 volts, so the voltage was gently raised to 30V, keeping the current under 0.5mA.

The power supply we were using to reform the capacitors had a 25V limit (It used 18V RMS, which is about 25V peak). Getting above 25V therefore required a power supply rebuild -- adding a switch to select a different transformer rectifier configuration to extend the range to 50V (using the second 18V winding on the transformer to make a 36V RMS input to the rectifier).

Bug 17: The Teletype lid had a broken ear, the rear ear of the left lid pivot; a replacement was fabricated from laminated 3mm ABS rod (scraped flat for lamination). The replacement was then filed to shape and to mate closely with the stub of the old ear, and then glued in place Plastruct Bondene solvent cement. The glue joint will need reinforcing before any stress is put on it.

Mar. 3, 2014, Power Supply, TTY Work

  thumbnail image     thumbnail image  
Checking the repaired lid
Bug 17: The fit of the repaired lid with the teletype cover looks good. The replaced ear is sound, and it snaps into place with just a little less effort than is required to snap in the unbroken ears on the other side.

Bug 10: C1 to C4 all had imperceptable leakage after a day at 30 volts, so the voltage was gently raised to 36V, keeping the current under 0.5mA.

Mar. 4, 2014, Power Supply, Foam Strips

Bug 10: C1 to C4 all had imperceptable leakage after a day at 36 volts, so the voltage was gently raised to 43V, keeping the current under 0.5mA.

While this was in progress, we started work disconnecting one leg each of C15 and C16 (950µF 75V capacitors) and C5 (35,000µF 25V). The first two are output filters on the regulated power supplies (maximum output voltage, 40V). Careful reading of the power supply schematic was needed to work out which leg of each capacitor to disconnect in order to allow reforming in circuit with a floating supply. We will try to reform these capacitors in parallel until we reach 25V, at which point, we will disconnect from C5 and continue raising the voltage on C15 and C16 toward 50V. Reforming these to their full rated 75V will not be possible with our 50V supply, nor will it be necessary with their 40V maximum working voltage in this circuit.

Bug 3: We also measured the crumbling remains of the foam strips on the cabinet sides. The strips sit in a metal channel that is 3/4 inch wide (inside dimensions) and with sides that stand out 3/4 inch from the rack frame. The inside face of the side-skin of the rack is 7/8 inch from the rack frame, suggesting that the original foam strip was 3/4 inch wide on its adhesive face, standing out 1 inch from the rack and compressed to 7/8 of an inch when the rack side skin is in place.

We can get 3/4 inch wide by 1/2 inch thick foam weatherstrip, two layers of this will work in a pinch. Given that we have 4 strips to replace, about 6 feet each, 48 feet of this would work, or 5 10-foot packages. It would be no more durable than the foam DEC originally used, and possibly worse.

Mar. 5, 2014, Power Supply

Bug 10: C1 to C4 all had imperceptable leakage after a day at 43 volts, so the voltage was gently raised to 50V, keeping the current under 0.5mA. A day at that voltage should suffice to declare them reformed.

While raising the voltage, the soldering gun we have been using was fixed. The copper tip was badly oxidized, preventing electrical contact with the transformer primary. Removing the tip, filing off the oxide from the contact areas and reinstalling it gave it new life.

With that done, we unsoldered the easier to access of C15 and C16 (950µF 75V capacitors), leaving the other for later, and we unsoldered one leg of C5 (35,000µF 25V).

Mar. 6, 2014, Power Supply, Visitors

Bug 10: C1 to C4 all had just measurable leakage after a day at 50 volts. After disconnection, we discharged them through the 510Ω resistors to 15 volts and then noted that all were discharging at under 5 volts per second. This is acceptable, so this batch of capacitors was declared good and labeled as such.

With that done, we finished preparing C15 and C16 (950µF 75V capacitors) for reforming, as well as C5 (35,000µF 25V). We wired them in parallel with tack-soldered 510Ω resistors in series with each, and then started reforming, keeping the current under 0.25mA as we raised the voltage to 7.5V. They will cook at that voltage for a day.

Bug 14 and Bug 15: Two matched pairs 6-pin connectors were acquired (from an E-bay dealer), Cinch Jones P-306-CCT (male) and S-306-CCT (female) acquired to repair the cut Teletype data cable and replace the inappropriate 12-pin that someone had put in the cable. We need to propose a sensible "standard" for this connector data cables between teletypes and DEC computers that incorporate a reader-run relay along with full-duplex data transfer.

Mar. 7, 2014, Power Supply

Bug 10: C15 and C16, plus C5 all seemed happy after a day, so we cranked the voltage up to 15V.

Bug 26: Later, Harrison Pickett reassembled the degreased Teletype keyboard. It still needs some lubrication. The manual calls for KS7470 oil -- note that part numbers starting with KS are Bell System part numbers. Various federal databases give the Federal supply number 9150-00-448-5009 and describe this as "lubricating oil, instrument", or "An oil primarily prepared for lubrication of aircraft instruments, gyro instruments, electronic equipment, gyros in torpedoes and aerial bombs, and calibrating and measuring instruments." Other cross references are to TP88970. The specific gravity is 0.8828, the boiling point is over 400F (204C) The viscosity at 100°F is supposed to be between 39.70 and 41.90. The viscosity units not specified, but are probably centistokes, cSt; if so, this is midway between 10 weight and 20W20 motor oil; Wilson-Jones Paper Shredder oil has a higher density and a higher boiling point.

Some Teletype restorers have recommended using 3-in-1 oil, but there are others who note that 3-in-1 oil tends to leave a gummy residue after a while. We need to investigate this further.

Mar. 9, 2014, Power Supply

Bug 10: C15 and C16, plus C5 all seemed happy after over a day at 15V, so we cranked the voltage up to 20V.

Mar. 10, 2014, Power Supply

Bug 10: C15 and C16, plus C5 all seemed happy after over a day at 15V, so we cranked the voltage up to 25V.

Mar. 11, 2014, Power Supply, TTY Work

Bug 10: C15 and C16, plus C5 all seemed happy after over a day at 25V. This is the rated voltage for C5, and 10 volts higher than its working voltage, so C5 was tested by discharging through the 510Ω resistor and measuring the discharge rate at 15V. This was around 1V/sec, so we declared C5 to be good and added a dated sticker to it saying so.

The voltage on C15 and C16 was raised gently to 33 volts, keeping the current under 0.5mA. The current was monitored by checking the voltage across the 510Ω series resistor between the capacitors and the power supply. Judging by the leakage current, one of these two capacitors is reforming normally. The other draws no measurable current at all, suggesting that it might be bad. We will have to check this capacitor carefully.

  thumbnail image  
Soldering reformed
capacitors back
into place
Bug 20: We re-soldered the connections to C5, after testing the resistors that had bridged the capacitor to make sure they were undamaged by the desoldering.

We also set to work re-soldering the connections on C6 to C11. These capacitors had 4 heavy wires soldered to each connection tab, an arrangement that crowds the tabs to the point that reconnecting them this way seemed unreasonable. Therefore, we opted to crimp and solder just one wire to each tab, with the second wire threaded through the hole in the tab and the final two crimped to the second wire. This arrangement proved relatively easy to solder. The photo shows Jacob Acord working on this.

  thumbnail image     thumbnail image  
Typing unit and keyboard
  thumbnail image  
Call control unit
Bug 26: We took the time to photograph some of the subassemblies of the Teletype, the typing unit (platten and print head), the keyboard and the call control unit.

Bug 27: While photographing the keyboard, we noticed how thoroughly degraded the keys are. Whether this is due to reaction with oil from people's fingers, cleaning solvents used at some time past, or just age we cannot tell, but the outer grey plastic is has split, almost shattering in places, and come free from the white plastic cores of the keys. If we could fabricate new keys, this would be useful.

Mar. 12, 2014, Power Supply

Bug 10: after a day at 33 volts, the voltage across C15 and C16 was raised to 42 volts.

Mar. 13, 2014, Power Supply, TTY Work

Bug 10: after a day at 42 volts C15 was confirmed to be bad. It has almost no capacity, discharging to under ten volts in a matter of seconds when loaded by a voltmeter on the 60V scale. In contrast, C16 seems to be in good shape.

A replacement 980µF 75V electrolytic capacitor is needed. This should fit in a capacitor bracket 1.25" diameter (no larger than 1.375"), and it should be no more than 3.5 inches high, with an insulated can.

We turned the voltage up to 50V to complete the reforming of C16.

  thumbnail image     thumbnail image  
The repaired Teltype cover
Bug 17: We mounted the new Teletype cover window in its space using ABS washers solvent-welded to the protruding ABS studs instead of the thermal riveting that was used in the original manufacture. With that done, we reassembled the lid and cover, declaring the cover to be finished. There is one "dog bite" out of the lower left side of the cover, but it is purely cosmetic and will be left as evidence of honest wear and tear.

Mar. 16, 2014, Power Supply, TTY Work

Bug 20: C16 tested good, so we completed the re-soldering the connections on C6 to C11 and to C16. We also began planning the re-connection of C1 to C4. We will not reconnect them the way they were originally wired because the mechanics of opening up the supply for access to these capacitors was too awkward. Instead, we will make all outside connections to the bottom two capacitors so that tilting the back panel of the supply up will pose a minimal need to bend and stretch wires.

Bug 10: All connections to C15 (the bad capacitor) were un-soldered, and C15 was removed from the supply. Candidate replacements were located on Ebay, 980µF 75V electrolytic capacitors are hard to find, but we found a 1000µF capacitor that had nearly identical dimensions but was seriously overpriced and a 950µF capacitor the same diameter but much shorter (short enough to be difficult to work with) at a very good price. We'll see what we can get.

There are two electrolytic capacitors in the Teletype Call Control Unit (Power supply); we started reforming them, leaving them to idle at 5V for a day.

Mar. 20, 2014, Power Supply, TTY Work

Bug 10: Stopping in to turn up the voltage daily, the electrolytic capacitors on the Teletype's Call Control Unit are now up to 30 volts. That is the rated voltage of one of them, the other is rated at 75V, so we will work our way up to 50V on that one before stopping with the reforming.

  thumbnail image  
C15, new and old
The new capacitor for the PDP-8 power supply arrived -- purchased on Ebay. It's rated at 1000µF 75V, replacing one rated at 950µF 75. (The difference is insignificant; the actual capacity of aluminum electrolytics frequently varies by 10% from the rated capacity.) The vendor said it was tested, but we connected it into the reforming circuit with the usual 510Ω resistor in series to check its leakage. We connected it across the 30 volt supply, and it showed evidence of reforming (the voltage across the reforming resistor dropped exponentially to around 0.2V and then held steady for a while before dropping down to closer to 0.02V). Therefore, we'll leave it in the reforming circuit while we ramp it up to 50V.

The photo shows the original C15 and the new one. The original has a date code of 6532 (32nd week of 1965). That's fully consistent with what we know about the early 1966 delivery date for this machine.

Bug 15: We worked at reverse engineering the cable from the W070 board (photos, board schematic, interface schematic) to the Teletype's terminal strip. This required working out the pinouts of the old inappropriate 12-pin Jones plug, half of which was smashed. This led us to the following recommendation:

Pin #
Pin label
Signal Pin #
Jones Plug
Pin #
- RDR RUN +orangereader-run +
from W050
1 6 orange red
- RDR RUN - blue reader-run -
2 4 blue black
4 4 red keyboard +
3 7 red orange
3 3 green keyboard -
shift reg in
4 3 white green
7 7 black typing unit +
from w050
5 5 black white
6 6 white typing unit -
750Ω to -15
6 2 green blue
yellow may be used instead of white
gray may be used instead of green
column UI - wire colors for our W070, as received
column RICM - wire colors for RICM W070

The plug we are using is Cinch P306 CCT (plug, Teletype cable) and S306 CCT (socket, PDP-8 cable). Later, around 1970, DEC standardized on an 8-pin Mate-n-Lok connector for this purpose (see Teletype Modifications, Drawing LT-33, 1970). Using the Mate-n-Lock connector on our 1966 vintage machine would be anachronistic.

The rightmost two columns above show the wire colors on our W070 board, as received, and on the W070 board on the Rhode Island Computer Museum's PDP-8/S #1. The wire colors used on the latter bear no relationship to the colors shown in the DEC documentation, while the wire colors used on our board are largely as documented, except that the white and green wires have been exchanged. When we rebuild this cable, we will bring it into conformity with the documentation.

There is logic in both the original DEC pin assignments in the Mate-n-Lock connector and in the proposed pin assginments in the 6-pin Jones plug. In both, positive terminations are grouped on one side, and negative terminations are grouped on the other, minimizing the likelihood that shorting adjacent pins will cause damage. Given that the Teletype is a passive device, from the perspective of the communications line, this means the Teletype connector, while the PDP-8 is active, providing +10 and -15 volts on some signal lines, this also means that the male pins should be on the end of the cable attached to the Teletype, while the female connector should be on the PDP-8 side.

Mar. 24, 2014, Power Supply, TTY Work

Bug 10: After a day at 43 volts, the 250µF capacitor in the Teletype's Call Control Unit and the replacement for C15 in the PDP-8 power supply appeared good, so we turned up the voltage to 51 volts. Immediately after increasing the voltage, the leakage current in the CCU capacitor was about 0.2mA, while the leakage in C15 was about 0.05mA.

Bug 19: Tony Andrys completed an inventory of the boards of the backplane. One discrepancy immediately came to light: The board in location PA36 (that is, the processor side of the PDP-8, top row, endmost board) should be a W501 (schematic, data sheet) according to the module maps in both versions of the PDP-8 maintenance manual. Instead, the board here was an R302 (schematic, data sheet, photo ). We have exactly one W501 board in our spare parts stock. Since this board is the most accessible board in the entire computer, it is highly likely that someone pulled it while investigating the machine at some time during the machine's 30 years in mothballs, and then put back the wrong board.

Mar. 27, 2014, Power Supply, TTY Work

Bug 10: We declared the Teletype capacitors and the replacement for C15 to be reformed. The 250µF capacitor in the Teletype's CCU has somewhat higher leakage than we would like, about 0.2mA after a day at 60 volts, but this is not high enough for us to replace it. The supplies in the ADC rack remain untouched, and will remain so until we decide to attack that rack, most likely after we get the basic PDP-8 up and running.

  thumbnail image     thumbnail image
C15 and C1-4 rewired
  thumbnail image
C17 resonating capacitor
Bug 20: We completed the rewiring of C1 to C4, connecting them with wires long enough that the power supply can be unfolded (and refolded) without removing any wires. This required rotating the wiring harness 90 degrees and replacing two wires with wires about an inch longer. As part of the same job, we wired the freshly reformed replacement for C15 into place. Except for the screw terminals, it is the near twin of the capacitor it replaced.

We removed C17. This is the resonating capacitor for the constant-voltage transformer. It had a sticker reading "Danger, High Voltage" on it, on very brittle paper with failed adhesive. The engraved markings on the capacitor add weight to the warning; they read:

65-13 60CY

That is, this capacitor was made by General Electric in the 13th week of 1965. It is rated at 6µF at 660V, and designed to operate at 60 cycles. This is not an electrolytic capacitor, but rather, it probably has an oil and paper dielectric separating the foil electrodes. Given the date of manufacture, the oil is almost certainly formulated with Polychlorinated biphenyls, since the ban on these chemicals was phased in gradually starting in 1972.

David Gesswein has had some of these old oil-filled capacitors fail, and one of them burst. He recommends replacing them all with new capacitors that a) do not contain PCBs and b) include short-circuit protection features.

Note added April 18: Michael Thompson commented on this section, noting that "all of the resonating capacitors in the PDP-8 systems at the Rhode Island Computer Museum failed, some spectacularly. We also recommend that you replace them."

  thumbnail image     thumbnail image
Broken DC-202 parts and drawing
As noted on Feb. 10, some of the male Faston connectors on the power supply are broken. In the photo, you can see two broken DC-202 connector shells along with a T-202 terminal that fit into one of these shells.

We researched the DC-202 connector, determining that these connectors are based on U.S. Patent 2,931,006 issued on Mar. 29, 1960 and assigned to Heyman Mfg Company. That is the company that now goes by the name Heyco. They are still very much in business, so we serched their on-line catalogs and spoke with a company representative. They still make the T-202 contact blade that is at the heart of the broken connectors, and these are distributed by several companies. Unfortunately, there is no US source for the nylon DC-202 bushing that holds this contact. We did locate a British source, Anixter Component Solutions of Dorset, that still lists the part, in either black or white -- they may not have the full rainbow of colors used by DEC. We took complete measurements of the connector housing, enough that it should be possible to manufacture new ones using a 3-D printing service. A quick consult with the 3-D printing staff at M.C. Ginsberg led us to conclude that it would cost about $100 to make anywhere from 1 to several hundred replacements for the DC-202 bushing (a $50 set-up charge and about $100 for a single print run -- the number of bushings made per run is limited by the number that will fit on the bed of their 3-D printer. The resolution required is comparable to the resolution of their stereolithography process, and they have photopolymers that give final products comparable to ABS or polycarbonate.

Bug 14 and Bug 15: After determining that our Teletype data cable was 24 feet long before it was cut, and that DEC sold 2 length, a 12 foot and an 18 foot cable, we concluded that the best way to deal with the scuffed cable sheath of our cable was to cut off the bad parts, leaving a total of about 18 feet of good cable, with one break where we will put the new 6-pin Jones plug. So, we cut the cable and Harrison Pickett set to work soldering the cable to the W-070 circuit board and the new Jones plug.