Because I had six PC05's in my lab, there was the seldom opportunity to see the evolution of PC05.
The repaired PC05's date from a time range of about 1970 to 1975. There were not two identical device, all had minor differences in construction.
The sampling of data bits in the reader: M705 vs. M7050.
The biggest evolution was a change in the way how data bits are read. While the tape moves over the photo transistors, the point in time were the paper holes are converted to data bits is crucial. The electronic signal which indicates a "valid bit" is called "data clock".
Earlier PC05 models with M705 "Reader Control" module use a signal from the stepper motor as data clock. That motor drives the dented wheel, which moves the tape. So when the tape motor is driven into a certain position, it is assumed the holes are right over the photo transistors and data bits can be read in. The rotary angle of stepper motor and attached tape wheel must be precisely adjusted then. The M705 is also used in the PDP-8 (together with M715 and M710), read here about the paper tape in PDP-8/L.
On later models with M7050 reader control module, the central "feed hole" on the paper strip is also sampled by photo transistors and is used to generated the data clock. This is more reliable, since the feed hole is always aligned to the data holes at punching time. No further adjustment needs to be done.
Detecting End of Tape
Related to the change in data clock is the way how the presence of the tape is detected in the reader. On (almost) all models the photo array scans not just 8 data bits and the feed hole, there is also a 10th photo cell outside all holes. So if there's no tape, this cell get lighted and gives an "end of tape" signal. This is how the M705-based PC05 work.
But the G918 photo amplifier has only 9 channels it can amplify. Since the M7050-based PC05's need the 9th channel for the feed hole, they can not also use the tape sensor photo cell. The M7050 recognizes the "end of tape" condition dynamically by a missing feed hole pulse after a tape step.
Power supply
Apparently DEC had difficulties in building a good power supply: on early regulator boards a 18V rectifier overheated. Field Service replaced it and mounted a bigger rectifier onto the chassis for cooling.
I've seen this picture on three PC05: all had the same burn marks on the PCB. The position of the replaced rectifier was very indivdual, Field Service was creative here. Later models had an reworked regulator board and a bigger rectifier.
The reader's light source
It is important that the light bulb gives same brightness to all photocells.
On early PC05's an adjustable lens was used to focus light. This was sub-optimal, as irregularities of the tungsten filament were projected onto the photo cells.
 |
 |
| Old model | New model |
On newer PC05's the light is bundeled by a light conductor which shines precisely onto the 9 photo cells. This gives better light yield and makes the reader insensitive to ambient light changes.
The photo transistors are most sensitive to near infrared at 850nm, that's the reason why replacement with visible LEDs will not work.
Puncher: Tape depressor
In the puncher, the tape is also moved by a dented wheel. The puncher always punches a feed hole, and the wheel grabs the holes and pulls the tape.
 |
 |
| Old model | New model |
The tape must be pressed against the wheel. Early PC05's had a transparent plate, while newer one's have a sturdy metal clip there.
Definitely the transparent plate is much nicer to look at: in presentations you can show people much better how the holes are produced by the puncher.
Puncher: universal paper feeder
The puncher normally processed only fan-folded paper tape. One PC05 had a (crude) upgrade so he could also punch tape from a reel.
See the metal bar guiding the tape? It looks quite handmade, but the idea is good: Beeing able to punch both fan-folded paper and paper from a reel is a big advantage, since fan-folded paper is quite rare and expensive today.
Photo amplifier ECOs
The 9-channel photo amplifier on module G918 was changed a lot. Apparently DEC Field Service tinkered with several ECOs (Engineering Change Order) until a stable design was found.
The job at the G918 is to find proper amplification levels, so the raw photo signals give stable logic signals.
The raw photo signals are influenced by everything:
- lamp base brightness (varying between bulb changes and as bulb ages)
- value of the variable light bulb series resistor
- lamp focus and lens position
- ambient light level
- sensitivity of the photo cells
- exact position of photo cells
- alignment of stepper motor angle
- color range of the paper tape.
Gulb! Improper adjustment leads to unreliable data ... and errors may occur at very low frequency. I spent hours and hours running XXDP ZPCA while trimming G918.
All changes on G918 are related to the adjustment procedure.
One G918 had individual potentiometers (my favorite!), so photo cells could be adjusted independent of each other. Nice idea, but too much work in the field. All other G918 have only one trimmer for all signals.
Later models produce also a "sum" signal, which is the addition of all 9 data holes. The image from the manual shows the "sum" signal for 0xFF byte (all holes punched) and for a 0x00 byte (only feed hole punched):
The sum signal finally allowed a fast and secure threshold alignment.
M7050 Reader Control board revisions
The M7050 "Reader Control" has an 8-bit register to save the data bits for read out by the host computer. This register is realized with flip-flops. On M7050 revision C the Motorola chip "MC4015" is used. Production of MC4015 was ceased in the early 70ths, so DEC used a 74175 chip instead and made a new board revision "D". If you have an 7050-C with defective MC4015, you have to craft a MC4015 with a 74175 and some wire.
Punch control voltage
The punch solenoids are powered by driver stages located on M044 modules. On newer PC05's the solenoids are driven by 18V voltage, but there were two (older?) models which use 36V. This is still within specification for the solenoids, but double voltage means four times more heat for some resistors on M044 ... too much.
To fix DEC reduced the voltage, and I scrapped away the ashes and planted in bigger resistors. The left resistors are sitting vertically above the right ones, they got extra heat by convecting air ... and more tan :-)
Bottom line
These examples show how step-by-step Evolution corrects initial (Un-)Intelligent Design. Even for man-made objects!