A dealer on ebay from whom I have bought various buttons, lights and 3420 tape drive control panels to stock up for the realistic buttons and switches of the 1130 found the control panel from a 2501 card reader in his stock. It is kind of funky so he offered it for just the cost of shipping, which offer I quickly accepted. It had been spray painted a bilious yellow, but I can remove that and repaint. Now I can plan to put the Documation card reader into an enclosure that appears 2501-like, using a realistic control panel even if I can't get the hopper and stacker exactly right (and because I won't build a full size 2501 cabinet).
I am working on ideas for the enclosures that will look like the IBM 1627 (plotter) but housing the Strobe 100 mechanism, the 1055 and 1134 paper tape reader and punch but housing the Tally and Dura mechanisms, the 2501 but housing the Documation, and a card punch I just bought on ebay which I may house in a 1442 like enclosure. I am going to have to walk a fine line between recreating a system that takes up so much space I can't store it well or easily transport it, and a system that is too cut down to benefit from the attempts to maximize the deja vu for the user of the system.
I am carefully rewiring the perforated breadboard based LED matrix I built to serve as the production lamps on the display panel which stands above and behind the console printer. My matrix is the correct size to provide a 1:1 scale copy of the IBM pedestal panel. I am drawing detailed plans for that pedestal panel - allowing me to begin fabricating the box, stands, install the rotary switch and emergency power pull and other details, and of course use it with the 1130 logic from now on.
When I wired up the matrix, I connected the cathodes of the LEDs vertically in columns and the anodes horizontally as rows. However, as I built the MAX7219 based drivers which assume a common cathode 7 segment display, each logic row of eight bits is considered a digit, with the columns corresponding to the segments of an LCD display. Because of my wiring, a logic row actually displayed vertically not horizontally. It made the assignment of bits to the LEDs cumbersome in my VHDL code - I would have to build a column with, for example, bit 0 of the IAR for the first segment, bit 0 of the SAR as the second segment, bit 0 of the SBR as the third segment and so forth all assigned to a 'digit' of the 7219 chip that drove the vertical column.
It would much more straightforward to assign a digit to the sequential bits of one register and have that digit represent a row across, not a vertical column. My rewiring has reversed the connections so that I can do this - my logic will now assign IAR bits 0 to 7 to the first digit, which maps to a horizontal line of eight LED bulbs. Cleaner, easier to change and to debug. I am moving slowly and carefully, ensuring that it will last for a good while and endure some movement and abuse, unlike the first prototype that used thin wires, had exposed wires just bent to remain out of contact, and of course were wired with columns and rows transposed.
The board with the LEDs plugs into my LED driver board, which in turn plugs into the new keyboard interface board I am building. The KB interface board has a 100 pin connector that attaches to the fpga board, providing a clean path for 40 signals. Three of them are passed to the LED driver board which can control 192 bulbs over those three wires. The KB interface also hooks to the input interface board that debounces and concentrates the buttons and switches of the physical console over a pair of wires using the I2C protocol. Other pairs of wires will be attached to the PMOD connectors on the side of the FPGA, each pair supporting a different peripheral or set of peripherals. The fpga board has 28 unique signals available through those PMODs, thus 14 channels of I2C are possible each able to handle a chain of separately addressible chips.
I am designing the circuit that will interface the Strobe plotter to the 1130 adapter circuits. The 1130 will only ask for step left, step right, paper down, paper up, pen lowered and pen raised. These six signals are routed over a 2 wire I2C link to my circuit which implements two four bit circular shift registers for the stepper motor drives, local logic to implement the requests from the buttons on the plotter, and a flip flop to keep the pen on paper or raised as necessary. I will bang together a quick and dirty prototype on a breadboard, verify operation, then whip up a PCB design and send it to the fab for production.
I looked over the Tally paper tape punch a bit today and see that it has many features on it that are options on the regular Tally models for which I have documents. For example, it has a reader in line after the punch station to allow verification that the media was punched correctly. It also has a mechanical encoder on the tapeup reel that indicates the tension on the paper tape, allowing servo logic to adjust the motor to protect the tape from breaks or unspooling. However, I am still convinced that this needs a fair amount of research before I am sure it will serve the purpose and before I can design any interface logic.
My partial tests of the input board that I finished building shows all the debouncer channels I tested working properly, except for one that I will look into more tomorrow. The level shifters and multiplexors are next to investigate, but I am not getting a good level for the lower power to those chips - it is 3.3 at the bench power supply but much lower wherever I measure it on the board. This needs some further investigation. I will go over the board to check circuit continuity and freedom from inadvertent shorts, then begin checking the hot circuits.
The replacement photocells and my connectors for the card reader arrived today. I have wired the photocells into the keyboard assembly and initial checks show the replacement channels perform similarly enough to the channels with original photocells that they should work reliably with my keyboard interface circuitry. I can fine tune the resistors for these channels if they don't seem to perform properly during assembly/checkout.
I won't apply power to the card reader until I receive my autotransformer - variac - that will let me bring up the voltage slowly, allowing old capacitors to gently build up a charge and let me spot problems if the capacitors have dried out or otherwise failed from disuse and age. I expect the unit by the weekend at the latest. I will be away on business most of next week, in Boston, but hopefully I will have a productive weekend before I go. The week after next, I will spend a few days down in southern California, but as part of the trip I will pick up a card punch I bought through ebay and bring that back home for restoration and conversion into a mock 1442 card punch for the 1130.
Lawrence Wilkinson is looking into machining the Emergency Pull handles that are on the 1130 as well as on the 360/30 console he is building. His console printer interfacing is proceeding well, with the IO Selectric appearing to be fully functional, waiting only on the completion of his interface logic to drive it from the fpga at the heart of his 360 replica.
The left half of the console display light panel is rewired to my satisfaction. That is the part that displays the contents of six of the 1130's registers - IAR, SAR, SBR, AFR, ACC and EXT - each being 16 bits wide. The right half will be rewired next. That reports on the clock state, conditions and status, op code, interrupt levels, cycle counter register and other information useful to programmers and during servicing of the machine. It is also constructed of a 10" wide by 4" high phenolic breadboard into which I have inserted the LEDs in their accurate position for the 1130 display. Once I rewire this so that it has each digit of the MAX7219 chips controling a horizontal row of lights, I can adjust the VHDL for the display control logic to simply assign groups to a digit - for example the first digit, the upper left row of lights, is assigned the T-Clock states (T0 to T7), while the second digit, a row down, receives the I clock phases (I1, I2, IA, IX, E1, E2 and E3) and the X-clock X7 state.
No comments:
Post a Comment