- Log entires for 2013
- Log entires for January-March 2014
- Apr. 4, 2014, TTY Work, Power Supply, Fans
- Apr. 14, 2014, Misc
- Apr. 20, 2014, TTY and Tally Reader Cleaning
- Apr. 25, 2014, TTY reader-run relay, Tally motor
- Apr. 29, 2014, Tally motor, TTY, PDP-8 backplane
- Apr. 30, 2014, TTY reassembly, 779 power supply
- May 1, 2014, TTY cable and reader-run relay
- May 5, 2014, Rack repair, power connectors
- May 6, 2014, Power supply reassembly
- May 12, 2014, TTY reassembly
- May 27, 2014, Tally reader reassembly, cables
- June 2, 2014, Tally reader reverse engineering
- June 13, 2014, Tally reader reverse engineering
- Log entries for July-December 2014
- Log entries for January-February 2015
- Log entries for March-June 2015
- Log entries for July 2015-December 2016
- Most recent log entires
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.
The Teletype print head would not move sideways on its carriage, so Harrison
Pickett and Doug Jones
removed the platten and then extracted the print head and carriage from the
print mechanism so that the frozen ball bearings could be removed.
All of them were stiff with congealed grease, and
several would not rotate at all. Repeated soakings in WD40 removed all
of the congealed grease so that the bearings spun freely. This was very
|TTY print mechanism pieces
With the print head removed, it is easy to see the cluster of vertical code
bars that extend side-to-side under the carriage. These
rise and fall to convey the ASCII code of the letter to be printed.
The bottom sized of these bars have a system of notches and tabs that make
up the "stunt box" that decodes control characters such as CR, LF, and BEL,
while the heavy horizontal bar across the front of the machine actually
powers the carriage, causing it to raise and rotate the print wheel to
position the right character for printing, and then driving the hammer against
the print wheel.
The replacement for C17 arrived
(See log entry for Mar. 27, 2014).
The new capacitor was significantly smaller than the old one, so we had
to re-size the capacitor clamp (a simple matter of bending some U-bends into
the sides of the clamp and tightening the curves at the ends), and then Nick
Becker drilled a new hole in the power supply frame to hold the shortened
|Mounting the new C17
|All of the capacitors
With the new C17 in place, all of the capacitors in the supply have been
replaced or reformed. The only job remaining to be completed on the power
supply is that of fixing the broken connectors and, of course, testing the
Nick Becker and Jacob Accord removed the fan assemblies from the bottoms
of the two relay
racks and set to work cleaning the fans and air filters. As is usually
the case with fans, they were dirty, and the air filters were filthy.
The stickers on the air filters that gave the cleaning instructions fell
off immediately when the filters were removed, but their instructins were
clear enough. In the photo, Nick Becker is cleaning gummy deposits off of
the fan with Goo Gone.
The filters are E Z Kleen filters made by Reserch Products Corporation.
This company is still around, doing business as Aprilaire. The filters are
designed to be coated with a product (still available) called
RP Super Filter Coat. U.S. Patent 2,865,466 explains this coating as
a tacky self-emulsifying petroleum product. The idea is that it acts as an
adhesive when the filter is in use, grabbing dust particles, but it emulsifies
in warm water, releasing the accumulated dirt and rinsing off of the filter.
On close inspection, we found electrical tape around one fan's power leads,
and on removal, we found a poorly done splice. We cut that lead and used a
crimp splice to attach a replacement wire. Both fans were tested and work;
as fans go, they are reasonably quiet. We also concluded that the plug
on the PDP-8 line cord was unsafe both because it is almost impossible to
safely terminate fine stranded 12/gauge wire on the screw terminals in the
plug and because the cardboard insulation over the termination can easily
allow individual escaped wire strands to escape the connector shell.
|An unsafe plug
Tally paper-tape reader
has been removed from the rack and partially
disassembled for cleaning. The photos, by Jacob Accord, show the arrangement
of the wires before disassembly. We have no evidence concerning the relay
socket to the left in the image of the main reader-cable connector. Note,
also, the splicing of the blue wires to the power switch. Evidently, the
power switch in the bottom center of the front panel is a local addition
to the Tally reader.
|Tally Reader Disassembly
We got a box of 100 screws and lock washers, indistinguishable from DEC's
original screws. The correct specification is
10-32 Stainless Steel Truss Head Screws with
#10 External Tooth Stainless Steel Lock Washers.
All missing screws have now been replaced, and we have many spares.
We researched the KS7470 that
the Teletype manual calls for, and concluded that it was
equivalent to Starrett 1620 Tool and Instrument Oil, which we then
The code bars under the Teletype print mechanism are caked with grease
residue. To get access to these for cleaning, we had to remove the paper-tape
punch and platten. With these removed, the base of the Teletype printing
unit looks rather naked.
|The Teletype punch|
and print unit base
In the photo of the Teletype print unit base, the code bars run vertically.
These carry the ASCII code of the character to be printed, being "read" by the
carriage from above. The horizontal bars below the center portion of the
code bars are the "stunt box" that decodes the control characters to actuate
carriage return, line feed, bell and other non-printing functions. The shaft
extending vertically between the motor and the code bars is the main cam shaft,
while the selector magnet is at the bottom center. The latter actuates the
code bars, with a cluster of cams on the camshaft used to determine which
code bar is being addressed at each instant.
The Tally 424 paper tape reader mechanism was rather grimy, so we decided
to take it apart and clean it. This allowed some nice photos of the
many differentials in the mechanism.
|Tally 424 capstan and take-up reel drive
|Motor Run Capacitor
In disassembling the reader, we realized that, as with the PDP-8 power
supply, this reader includes a motor-run capacitor that is probably PCB-filled
and devoid of internal protection. As these are prone to explosion and
constitute an environmental hazard, we will replace it.
The engraved markings on the capacitor read:
|MADE IN USA
That is, this capacitor was made by General Electric in the 34th week of
1962, and it has a capacity of 3 microfarads at 370 volts.
We scrounged up an ancient Dataprobe CL/232 converter box, never used, still
in its original wrapping. This box converts from RS-232 voltage levels to
the 20mA current loop signalling conventions required by the Teletype.
This will allow us to test the Teletype without relying on the PDP-8, and it
will allow us (with different internal jumpering) to substitute a modern
computer for the Teletype as a console device on the PDP-8.
|An RS-232 converter|
The CL/232 converter was made in the 1980s and never used until this year,
so its power supply capacitors were in as much need of reforming as those in
the PDP-8. Its power supplies are all simple
linerar supplies, so we hooked it to a Variac and slowly raised the
line voltage over a period of a week in order to reform the power supply
capacitors. We also built a cable for it to connect to the serial port
of a Raspberry PI computer. We still need to construct the cables needed
to connect it to the PDP-8.
Having obtained a working salvaged reader-run relay from Wayne Durkee, we
decided to install it, following the instructions in DEC's
Teletypewriters LT33 drawings,
with supplementary explanations given by
Harris Application Note HM005.
The documentation was straightforward, but DEC's material did not
provide the explanations that Harris provided. In sum, our Teletype had
already been modified to operate on a 20mA current loop -- half of the DEC
modification had to do with that. What remained to do was install the
reader-run relay board and wire the cable appropriately.
Bug 29 and
While starting work at replacingt the Tally paper-tape reader's motor run
capacitor, we discovered that the insulation on the motor leads was in bad
shape. While the wiring in other parts of the reader seems to have modern
plastic insulation, the motor wires appear to have been insulated with
natural rubber, and like much old rubber insulation, it tends to crack under
stress. Almost an inch of insulation was gone from one motor lead, and several
cracks were evident. Therefore, we set out to replace the motor leads with
The photo shows motor rebuild in progress, with one of the old wires
set across the end to display the condition of the old insulation.
The most time-consuming part of the motor rebuild process was removing the
old lacings on the end of the motor winding and re-lacing the windings.
This was necessary because the crimp connectors connecting the motor leads
to the windings were held in by the lacings, and any strain on the leads was
taken by the lacing. We used new waxed lacing tape and did our best to
duplicate the origina lacing pattern. Note that the crimp connectors we
replaced were more compact than the new ones, but tight lacing holds the new
bigger connectors down so they just fit inside the motor end caps.
Bug 29 and
The motor rebuild was completed, oil was added to the bearings, and the
motor, and the new motor run capacitor were installed on the tape-reader
chassis shelf, wired up and tested. While disassembling the motor, all
leads were tagged as to where they had been connected. The wires were
replaced one at a time and the tag transferred to the new wire. This allowed
us to rewire things with some assurance that the wires went to the right
Before reassembly, all wires were tested statically for
continuity through the motor with an ohm-meter, and after reassembly, it
was verified that there was no conductive path from the motor windings to
ground. Then, as shown in the photo, the motor was powered up with
clip leads, using baggies over all clip-lead to wire connections for
safety. The drive gear on the left end of the motor is a blur in the
photo because it is spinning. The motor ran almost silently, and after
running a few minutes, it was warm but not hot. With an AC voltmeter,
the motor housing measured about 20 volts, but this was at a very
high source impedance, almost certainly the result of capacitive coupling
to the motor windings, as DC testing of the motor after installation
confirmed again that there is no DC path from motor windings to housing.
Cale Bierman and Matt Adamczyk began reassembling the Teletype after
degumming was complete.
Tony Andrys has completed the backplane inventory for the PDP-8.
We have confirmed 3 serious discrepancies where it appears that the wrong
boards are in the wrong slots, probably the result of "tourists" pulling
random boards to show them to people and then putting them back in the wrong
We will need to do some reverse engineering of the wiring for these
slots in order to verify whether they are wired for these anomalous boards
or wired in conformance with the documentation.
When the Tally reader was removed from the rack, we found that two of the
original rivnuts in the rack had been removed and new holes drilled to
accomodate the mounting screws for the Tally reader. Since our plans are
to put the reader back in a different rack locaiton, we decided to replace
the missing rivnuts with new ones.
|Installing a rivnut|
Rivnut installation tools are widely available for $20 or more, but an
improvised tool can be made from a stack of washers, a hex-head or allen-head
screw, a nut and a pair of wrenches. We made such an improvised tool and
used it to replace the missing rivnuts.
Cale Bierman and Matt Adamczyk continued reassembling the Teletype.
The focus was on the selector cam cluster and codebar clutch. Reassembly
was straightforward until it came time to install the codebar clutch drum
onto the clutch shoe assembly. No matter how we manipulated the pieces,
the drum would not fit over the two shoes. Referring to drawings in the
Teletype manuals, we found that the clutch shoe lever had to be pushed
all the way to the stop lug against the tension of the clutch shoe lever
spring in order to bring the clutch shoes into disengaged position before
slipping the cutch drum in place.
Bug 10 and
The Type 709 power supply was pulled from the ADC rack in preparation for
reforming the capacitors there. The first three photos here were taken by
Jacob Accord. They document the original wiring of the supply. Note,
particularly, the connection of the red wire to the common terminal of the
right side of the supply. This wire was simply pushed into the terminal!
It was obviously a local addition, apparently the ground wire for the
home-made mystery power supply mounted above the 779 supply. If we
reconnect that supply, we will have to do a better job of connecting it.
|Type 709 supply and odd wiring|
|Supply wiring details|
Bug 14 and
The shorter half of the old Teletype cable, terminated with a new 6-pin
Jones plug, was wired to the terminal block at the back of the Teletype's
call control unit, with two wires continuing forward to the reader-run relay.
All of this was done following the instructions in DEC's
Teletypewriters LT33 drawings.
The cable routing used was based on the very poorly reproduced photo in
the center of sheet 1 of drawing 7505038-0-0 Rev. C (the first drawing sheet
the document). Because of the low quality of the photo, we had to guess some
of the cable route, but the drawing (and the length of the wire in the salvaged
reader-run relay) make it clear that the cable had to be routed along the
left side of the the call control unit, very close to the moving parts of the
Teletype. Our photos are much better than the original DEC photo and should
make our cable routing clear. Also note that we added a disconnect point in
the wiring to the reader-run relay so that the external portion of the
Teletype cable can be removed without the need to redo the wiring all the
way to the Reader Run relay.
See the log entry for March 20
for the terminal strip connections in the Teletype and the pinout on the
6-pin Jones plugs in the cable.
The top French doors on the front of the ADC rack had no doorstop to prevent
their pressing inward against the wire-wrap pins on the backplanes they
conceal. Close inspection of the bar that forms the bottom threshold of this
pair of doors showed a pair of mounting holes, 3 inches apart, centered in the
door opening. These evidently held a doorstop that mated with the magnetic
latches on the lower backs of the doors.
The magnetic latches extend to almost an inch above the bottom threshold and
are almost six inches apart.
When the doors are held so that their front surfaces are coplanar, the
latches are about half an inch forward of the line through the two doorstop
mounting holes. From this, we inferred that the doorstop had been a piece
of 1" angle iron 6" long. We fabricated a new stop and installed it, as
shown in the photo. The iron is unfinished, with
the front surface polished to a high shine using a wire brush. It might be
appropriate to paint it, but the few poor images we have found of
similar DEC racks show a shiny metal doorstop.
Anixter Component Solutions of
Dorset, England provided free samples of the DC-202 connector bushings,
in black, and Michael Thompson and the
Rhode Island Computer Museum
provided salvged DC-202 bushings with H-202 connectors in the correct colors.
The photo shows both, before unsoldering the salvaged connector clusters
so that individual connectors could be reinstalled in the
Type 708 power supply.
See the log entry for March 20
for additional information about these power connectors.
After replacing the broken DC-202 connectors facing upward on the back of
the PDP-8 power supply, we loosened the sides of the supply and then
Cale Bierman, Matt Adamczyk and Doug Jones carefully tilted the back wall
of the supply up and screwed it in place.
Recall that the back wall of the power supply was tilted down on
to allow reforming of the capacitors mounted on the back wall. This
involved drilling out the pop rivets that held the back wall in place. On
we installed swage nuts to allow eventual reassembly. Today, we used those
swage nuts. Finally, with the back wall of the supply in place, we replaced
the final broken DC-202 connector. The first photo, taken by Matt Adamczyk
shows Doug Jones soldering the connections to the final DC-202 connector
before pressing it into its place. Zoom in on the back corner of the
supply and you can see the screws that replaced the pop rivets visible
in the photos from
|Repairing the Type 708 supply
The final step in reassembling the supply was to replace all the
we cut in order to gain access to various capacitors for reforming. All of
the original zip ties are white nylon, while our replacements are black.
We have changed some wire routings as a result of our approach to swinging
the back of the supply down during repair, but most of the cable routings
|Replaced zip ties
Cale Bierman, Matt Adamczyk and Doug Jones lifted the re-assembled supply
into place in the computer rack, and finally, reattached the power cord,
with careful attention to the photos taken on
The photo, taken by Matt, shows Cale Bierman and Doug Jones tightening the
screws that hold the supply in place.
Barring our discovery of errors in our work, this completes the reassembly
of this power supply.
|Putting it back in the rack|
The next stage in this project will be to test the reassembled power supply.
This must be done before attaching the supply to the PDP-8. Ideally, we will
need to attach dummy loads for each of the supply outputs, and then turn on
the supplies and measure both the DC output voltage and AC ripple voltage on
each output. Only after we confirm that all of these are correct should
we risk connecting the power supply outputs to the computer.
Cale Bierman, Matt Adamczyk and Doug Jones reassembled the Teletype's
typing unit and re-attached the paper tape punch to it. During the process,
we noticed a label we had not previously recorded on the inside of the right
carriage support pillar:
|Typing unit reassembly|
|REG. U.S. PAT OFF.
REG. CAN PAT. OFF.
OTHER PATENTS PENDING
Assembly took hours. This was despite careful disassembly,
with photos, notes, and the various diagrams in the manual,
and despite tagging each piece during disassembly. All the small bits were
in baggies with labels, and large bits were tagged with string and a
label. Despite all these precautions, we kept encountering
little bits of Teletype and saying "what's this?" Some parts (such as
the right carriage motion pulley mount) could be put on two different ways,
and were initially put on the wrong way around. You can see this mistake
in Matt Adamczyk's photo of the Teletype label.
Several times, we thought we had the job done, only to pick up the Teletype
and find a spring on the table under it, or a screw. Eventually, we ran out
of parts and, turning the motor or moving the carriage by hand, everything
seemed able to work. We will need to carefully go over all of the adjustments
in the manual, and oil or grease all the lubrication points noted in the
instructions before we power anything up.
We bid on a Tally 424 tape reader we found on eBay. Their buy-it-now
asking price was $100, we offered them $15 plus shipping and won.
The reader was shipped our way in an extraordinarily battered box held
together by more duct tape than I've ever seen on one box before.
Nonetheless, it arrived in good shape, which is to say, less than fully
operational but still useful as a source of spare parts. What we're most
concerned about are all the nylon gears in the differential and the slip
clutches on the drive capstan. Matthiew Biger took the photo.
Reassembly of the Tally paper tape reader went smoothly until the
power switch was screwed in place in the front panel. As nearly as
we can tell, this switch was added by the U. of Iowa Psychology department,
judging by the handwritten label, on adhesive tape, saying "turn off when not
in use" and the Dymo OFF label. I removed both in the process
of replacing the switch, but added a metal
"on-off" label salvaged from an unused switch at the bottom of the ADC rack.
The switch allows users to turn off the motor when the reader is not in use.
The switch the Psychology department used has terminals that protrude to
the rear, and when we put the switch in place, the motor mount shorted to
one of the terminal tabs. The back side of the switch is visible in the
exact center of the photo, with the shiny connector lugs protruding back
under the motor. A short circuit here could connect 110
volt power wires to the frame of the reader.
I reduced the danger by carefully bending the tabs so
they would not contact the motor mount, but we should go further and
slip insulating sleeves over these tabs.
Michael Thompson of the
Rhode Island Computer Museum
reported that they have a box of black flexprint cables in good condition, and
wondered how many cables we need and how long. So, we pulled one of the
gummy cables for a more complete description. In the process, we discovered
that GooGone lives up to its name. Wiping the cables removed the
gummyness. A complete degumming would require disassembling the cables, and
if we go that far, replacing the gummy wires would be sensible.
There are 6 cable sets connecting the two backplanes of the PDP-8.
Each set is terminated, at both ends, with a W034D Flip Chip connector
paddle. Stretched out on a work table, the cable sets measure 23 inches
from card-edge connector to card-edge connector. There are two flex-print
ribbon cables in each assembly, one about 21 inches long, one about 19 inches
long. The final 1/8 inch of each cable has had the mylar stripped away so that
the foils protrude for soldering.
Each flex-print ribbon has 19 conductors, spaced at 18
conductors per inch, making each cable about 1 3/36 of an inch wide.
These are apparently copper plated onto one sheet of mylar, and then etched
to make printed-circuit conductors before being coated with an insulating
layer to make the finished cable. I speculate that the goo is the result
of the depolymerization of the insulating layer.
We finished reassembling the Tally reader and checked the electrical
continuity of all of the wiring, making reference to the schematics in
Figure 9 on
page 8 of the Tally manual.
Aside from the omission of the end-of-tape switches and capstan commutator
options, we noted the following discrepancies. For each, we provide a new
drawing; in all drawings, the connector is shown as seen looking at the pins of
the rear-facing plug on the back of the Tally reader. This is a
Continental Connector Corporation
Series 20 34-pin connector.
The power wiring is as indicated in the original Tally schematic,
except for two changes: First, the motor has three wires coming out of it,
black, red and yellow, instead of two pairs of wires as shown in Tally's
schematic. The common connection between the wires (to the yellow lead)
is inside the motor, instead of an outside connection, as shown in the
schematic. When we rewired the motor, we were careful to retain the
original wire colors, so it seems that this was a discrepancy introduced
in the Tally factory.
Second, the a power switch was added by cutting the light blue wire between the
capacitor and the fuse and then splicing in longer dark blue wire to connect
to the switch. In the process of reassembling the tape reader, the cut
fragment of light blue wire on the back of the fuse was removed and
the dark blue wire was reconnected to the switch.
The wiring to the capstan escapement coils was significantly changed, although
the wire colors attached to pins M, H and C of the connector are exactly as
indicated in the Tally manual. The Forward coil is the only coil present,
and it is wired across the two terminals documented as the reverse coil
terminals in the manual. The terminal strip is not shown at all in Tally's
wiring schematic, but it is shown in parts diagram for the
capstan drive assembly, Item 11, Figure 25 on
page 26 of the Tally manual.
|Tape Advance Wiring|
The Tally manual discusses the use of a
or spark suppression circuit on
but it does not say where to mount it physically.
On our reader, the snubber was mounted on the terminal strip. In fact, given
the length of the cable between the escapement coil and the driving circuitry,
the snubber inside the Tally reader will not generally suffice. Additonal
snubbing must be provided at the point where the drive pulses are actually
The Tally manual documents the use of a
snubber but what we found is either a
or a more sophisticated varistor-based snubber (we did not look up the device).
Finally, we checked the wiring between the connector on the back of the Tally
reader and the 8 switches that mechanically sense the presence of lack of holes
in each channel of the tape. This wiring corresponds exactly to that shown in
the schematic in the Tally manual, but relating that schematic to the
hole positions on the tape and to the connector wiring is sufficiently
difficult that we have drawn
a new diagram here. There are 8 clusters of 3 pins, one cluster for each data
channel on the tape. Each cluster is wired identically, only one is shown.
The Tally reader needs lubrication.
The lubricaiton instructions are in Figure 12 on
page 11 of the Tally Manual,
with accompanying text in Section 5.1 on
The 350 centipoise Silicone oil required appears to be equivalent to.
Dow Corning DC200, 350CS;
A quick check shows that this oil is available on eBay in 4oz bottles.
The SAE-20-grade oil requirement should be satisfied by the
KS7470 oil recommended for
The Tally pivot grease A required for the escapement armature lever fulcrums
is more problematic. Perhaps KS7471 grease (or equivalent) will suffice.
The Tally reader interface on the PDP-8 ADC rack needs to be reverse engineered.
What we have determined from the reader itself is that this reader has a very
simple interface, but unfortunately, that means that the interface logic
on the PDP-8 end must be somewhat different from that anticipated by DEC.
Bug 12 and
Working backward from the Tally reader, I took an ohm meter to the cable
that is clearly marked "Tally Reader" and began tracing out the wiring.
It immediately became apparent that this is not the cable for this paper
tape reader. Comparing the wiring of the plug that connects to the
reader with the wiring of the reader itself (see
June 2), the wiring for the
power connector is correct, as is the wiring to the movable contacts on
the data sensing switches, but the wiring for the tape advance
escapement coil is wrong, as is the wiring to the fixed contacts
on the data sensing switches. Looking at the paper-tape-reader connector,
there is solder on some of the pins that are currently unused, making it
clear that this cable has been modified from its original wiring.
The engineering of the cable is suspect. The data rate on this cable
is 60 pulses per second. At this data rate, there is absolutely no
reason to use coaxial cable for each data line. DEC routinely used
ribbon cable for such interfaces, and we have a new, unused
ribbon cable attached to a W021B connector, still in its factory original
packaging from 1968. It has only 2 feet of ribbon cable, but we can get new
ribbon cable that exactly matches the original.
A second troublesome feature of the Tally cable involves changes made to
the W021 connector at the computer end. As made by DEC, each coaxial cable
is supposed to be connected to consecutive tie-points on the connector paddle,
with just 9 data conductors and 9 (or 10) grounded shield connections
tied together by ground traces on the paddle.
As we got it, the ground traces connecting these tie points have been
removed (peeled off of the printed circuit board), and instead,
the shields of the coaxial cables have been soldered together in a glob
at each end.
Nov 22, 2013,
we determined that the Type 750C High Speed Reader interface was
part of the Type 34D Oscilloscope Display package, with the cable
to the reader plugging into slot B32 of the display interface backplane.
It is also clear that the high-speed reader interface has been
modified. Looking at the backplane wiring for slot B32, we found
pull-down resistors soldered directly to 8 pins on the connector.
As was the case with the paper-tape reader end, none of the backplane
wiring can be reconciled with the wiring of the cable itself. The plug on
this cable that connects to the Tally reader is almost certainly the
original, but this cable has been reworked to serve some other purpose!
Therefore, without doubt, we must build a new cable. The most conservative
option, from the point of view of future maintainability of the system,
would be to undo what were apparently local modifications to the paper-tape
reader and return it to its factory original wiring, and also redo the
wiring of the backplane so that the connector can be used in more or
less the way it was intended. It should be possible to interface to the
Tally reader using just 9 wires, although half-amp tape advance pulses
should be isolated from the backplane as much as possible.