March 2-3 weekend activities

KEYCAP MODIFICATIONS
I experimented another time with the Plastaid, hoping to be able to fill in the slight grooves on keycaps from the existing legends, to allow these to serve as either blank keys or have the 1130 legends applied. It worked better this time. I will leave it on until tomorrow and then test its adhesion. If it won't flake off, it passes the test and will be used to level out etched titles on keycaps that need changing.

After drying thoroughly, I subjected it to various abuses, during which it held onto the keycap material tenaciously. Further, it was rather easily workable with Scotchbrite pads to smooth and shape the surface. The Plastaid is a winner, just what I need to erase the etched titling on keycaps that will become blank or have another, 1130-specific, lettering applied. I still haven't chosen the method and supplies to letter the keycaps that get these 1130 names - keys like Rest KB, Int Req and Erase Fld - but with all the other details of painting and shaping well in hand, I can plan out the 'production run' to clean, fill, prime, paint, letter and top coat all the keycaps that need modification, after securing some more paint and quite a bit of solvent to thin and clean up after each stage.

PUNCHED CARD READER
After picking up a fuse for the autotransformer, during a trip to pick up components for breadboarding and testing the paper tape reader and plotter interfaces, I brought up the card reader slowly then tried to run some cards through in local mode. I got many pick checks, plus the odd read and stacker checks if a card did make it through. May be a low vacuum problem or crud on the rollers or similar that a good cleaning will fix. I will open it up and start into the mechanical side study, but happy to see that the major elements work.

Inside the back of the card reader - motors and air pumps

Reader from the front - power systems underneath

The card transport path - simple and straightforward

Upon investigation with covers removed, I noticed that the picker shoe isn't returning to rest position after the first card is pulled. The picker shoe is a rubber unit with holes for vacuum. A fan blows air onto the front few cards, causing them to separate and flutter in the airstream, meanwhile the picker shoe uses vacuum to grab and hold onto the bottom card in the stack. When it is time to read, a 'pick' activates a solenoid to rotate the picker shoe inward, pulling the card up to speed and delivering it to an opposed pair of wheels that move it along through the mechanism. The rotation of the picker shoe is slow and it is sticky, staying partly rotated rather than snapping back as it should.

The manual mentions lubrication of the bearings for the mechanism that rotates the picker shoe - my working assumption is that the lubrication is dried out, gummed up or otherwise hindering the movement. I will lubricate this, work it around to ensure any solidified lubricant is removed, then check the reader another time.

The initial read on powerup pulls a card and delivers it, but the attempt to pull a second card fails because the picker shoe didn't get into place to grab the card. Thus, no card is delivered to the transport area and the reader correctly signals a read check because the photodetectors didn't see a card break the beam at the expected time. Occasionally it gives a stacker check error, which is probably part of the chain of events caused by the picker shoe malfunction, but if it is a separate problem I will work on it once the shoe is repaired.

I did remove and reseat the logic cards to eliminate the chance that any vibrated loose during transportation of the reader or has developed an oxide insulation between contacts; both possible issues would be addressed by the reseating process.

Logic cards in the card cage above stacker

After lubrication of the picker shoe mechanism, the failures moved solidly to read checks, but I also noticed that the rollers for the card transport path are moving slowly and a bit unevenly. I will have to read the manuals, figure out how this should be working and then adjust/clean/lubricate the mechanicals even further.

Assembly of the cable between the Documation card reader and the reader interface board I am building was begun. It consists of 19 twisted wire pairs carrying the 12 card row bits, the index marker strobe, plus the ready, busy, error, hopper check, and motion check status signals from the reader as well as the pick command to the device. I made this short as it will terminate in an adapter box close to the reader, from which an I2C cable will transport the traffic between the adapter and the 1130 itself. The I2C protocol on a link dedicated to this adapter has about 4-5 times as much capacity as is needed to keep up with a 600 card per minute reader. I made use of a project box I had on hand to house the adapter circuits, taking the twisted pairs directly into the housing without requiring a 38 connector plug and socket for the box.

EDAC 516 38-pin plug during cable assembly

I will leverage the input interface board I had manufactured to handle the buttons and switches of the console. The fab builds three of each board as part of an order, thus I have two spare PCBs on hand. I have just ordered the remaining chips to populate the boards as needed to handle the card reader and an interface for some other peripherals, to be chosen later. By varying the capacitance on the debouncers I can adjust the delay, or just wire through the connections on the board without using the wires when the signals are already properly debounced as is the case with the card reader. In fact, I am considering re-ordering the board design to get another three PCBs, or modifying the design slightly, to use as needed for all the remaining peripherals. If I modify it, the reason will be to provide the ability to drive outputs as well as accept inputs. For the card reader, which has only one output signal (Pick), it is fine to simply leverage a signal from the fpga for this output. In the more general case, I will conserve IOs and connections by using a multiplexor based circuit like the input board.

INPUT BOARD
The different switch types that will be used with the input interface board were all wired up and tested successfully. I can begin hooking together the final buttons and switches, put them into quick and dirty mockups,  as I did with the LED panels,until I have the final production enclosures in hand.

PAPER TAPE READER
I built an experimental circuit modeled on the GE paper tape reader that appears to use the same sensor array module as in my reader. I could find nothing about the Sensortech array online or in other sources I checked, so another situation where I need to reverse engineer to gain knowledge necessary to design the final interface. In order to work with this copied GE circuit, which required +12V and -12V power rails, I could have experimented using my bench power supply but for the final system I will leverage an PC power supply which I will interface with the necessary power on jumper and a 5W resistor load across the 5V line.This power supply was completed and used for the +12 and -12, later it will be a source for good clean +5, -5 and +3.3 volt power for the entire system.

 I had values for resistors but no transistor or diode part numbers. I built it with a pair of a common type of signal transistor, a 2N3906, and with common signal diodes of type 1N914. The results of the experiment were inconclusive at first, as I got a nearly constant -11V output with nothing I could do with light or dark causing a material change. After the fact, I discovered that I had inadvertently connected the bias diodes to the -12V rail instead of ground, but even after correction the circuit is not switching as the GE version did. The reverse bias is set by the signal diodes, with this circuit establishing a bias voltage of about 1.9V. with the common connection more negative.

Breadboard recreation of key part of GE tape reader circuit

My paper tape reader mechanism, photodiode sensor holes visible below LEDs

However, I did determine that the sensor array presents as a diode with .53V drop across it, wired as a common anode configuration. The two common ways these are used are as voltage sources or placed under reverse bias and used as current sources. As a photovoltaic source with no bias, it produces about .1 microamp of current or .23V across a 2.35M load resistor when my LEDs are illuminated and much lower when dark (non-zero because some light gets through the paper tape itself).

Back view of PT reader mechanism, part of larger device containing motor drive.

PLOTTER INTERFACE
The prototype plotter interface logic was built up on a breadboard and tested this weekend, but I am thinking of a new approach that would be more staightforward. The interface as I tested it used two shift registers (74HC194), D flip flops to synchronize signals to a clock, 74HCT04 inverter, plus 74HCT00, 74HTC32 and other chips, with a capacitor delay for setting the initial state, but I have some ideas for other ways I might like better.

Initial lab tests of plotter interface logic design

SHORT HIATUS
I am traveling to Boston and away from home for the next four days, which will restrict me to some design work while away but no construction, testing, repair of mechanical devices or other such activities. When I return, I stay just three days before driving down to southern California in another four day trip.