Designing new peripheral interfaces, general work progressing

I have completed the prototype for the keyboard interface PCB and am testing it with the keyboard mechanism, verifying that it will detect and decode the key entries, operate the keyboard unlock solenoids properly and will work reliably.

There is enough variability in resistance between photocells that I am a bit concerned. I need a value for the resistor that reliably swings the voltage below 1V when illuminated, the threshold for both the MCP23017 and the 74HCT14 chip, and keeps it above 2.65V when dark, presuming a 3.3V supply to the photocells.

My initial measurements lead to a resistance bridge, 5V to photocell to logic node to 100K resistor to ground. The voltage at the logic nodes is well above the high signal minimum but not below the threshold to register as a low signal when illuminated. I experimented with reducing the resistance, with one photocell working well at 20K, but another was still above the 1 V target.

I tried all the photocells sequentially with 20K and 10K resistances, ensuring a high quality connection which my quick and dirty test was not guaranteeing initially. This gave me good results for photocells 2 thru 9, but both 1 and 10 were erratic, not just different resistance but often went open circuit. After investigation, photocell 1 had a bad solder joint, apparently good but susceptible to movement. I resoldered it; it works perfectly now. Photocell 10, however, seems damaged. Its surface is cloudy, unlike the other nine, and I suspect that the connection inside the base is loose. I believe that my current design will work fine with only nine active photocells, as long as I don't implement the Alpha key, since it is the only key that needs photocell 4.

Based on that, my approach is to relocate the working cell from position 4 to position 10, then go with the 20K resistors on the board.

I moved on to check the operation of the relay and solenoid when the 74HCT14 is triggered by a photocell going dark. The relay cleanly fires the solenoid to restore the keyboard, and the signal is cleanly, reliably decoded from the Rest KB key on the keyboard, but I have some minor problem in the driver transistor to the relay coil, or there is something wrong with the relay. This means a bit more work tomorrow morning just to certify that it will work when wired properly and using the right values.

In this way, I set up the final prototype board with the chosen values, and verified that all worked as intended. Thus, the design will be finalized tomorrow and I will commit the board design files to the manufacturer, OSH Park, a community service site that combines PCB designs from many people into large single sheets that are cut apart and mailed to us after fabrication. There is probably a bit of a backup in PCB manufacturing because many of the fabs are in Asia and closed during their New Year period. I am expecting them back in about three weeks, but will have a better estimate when I place the order.

The Documation M600 reader has a very simple and straightforward interface. TTL levels, reports ready, busy, and three error conditions. Is triggered to read by the pick signal set on. Emits a index pulse for each card column when the data is ready on the 12 row signals. I am working out the mapping of 2501 card reader states to the M600, then will build the interface logic into the fpga and build some level shifting circuits on a board to drive and accept the TTL logic levels. The reader has a 38 pin connector that I have on order; once I have the connector, I will check out the reader to ensure it is in good condition and reading properly, then test a prototype of the shifting circuit board.

Physically, I will think about whether I can make this appear more like a 2501 or 1442 device.

The Tally paper tape punch I received is also conceptually quite simple, although it expects different voltage levels to drive the punch mechanism. The manuals I have access to concern different models of the device, most working at 48V or 24V, but I will need to work out the differences carefully to understand the electrical connection requirements. Linking this to the PT punch adapter logic of the 1130 should be straightforward - I will do that mapping as I sort out the electrical needs. I see a couple of large electrolytic capacitors in the unit, which I should carefully address and reform them to be sure they are good, lest I get a dramatic failure on first powerup.

The plotter will take more work to interface, because I have no documentation at all on the Strobe plotters. I have an interface board, labeled Strobe S100 parallel interface, designed to connect to a S-100 bus, which is well known so that at worst I can mimic an S-100 to interact with the unit. However, the meaning of the various signals, locations and commands is totally opaque. Some comments in old articles in hobbyist computer magazines hint at the protocol, but it will be some time before I can figure out the best way to drive this unless I can find someone with the manuals.

I found a person selling the same plotter, new in its box, which also has the manual. I am certainly not interested in buying another for $499 just to get the documentation, but perhaps I can work out a deal where the seller might photocopy or let me or someone else photocopy the manual for a reasonable price. I will also check with the classic computing mailing list to see if any others have access to a manual for this or the related plotter models (Strobe 200 or 260 manuals would be fully usable).

While I am designing the paper tape punch, I need to finish designing and building the prototype of my paper tape reader interface. I have the physical unit in hand but need to whip together a prototype electrical interface and then some adapting logic.

I do have a box of 2000 blank punched cards, useful for the card reader once I get it operational. I need to get some blank paper tape stock to make use of the PT peripherals. Michael Albaugh has a source for this and expects to be making a run to pick up some supplies later this spring; he has offered to help acquire some media on my behalf when he is taking that trip.

There are currently two DEC RK-05 disk drives listed on ebay, one in 'not working/for parts' status with no clear picture of what is inside the covers, and another listed as used but fully operational. The price of both is high for what they are - $400 for a box that may be empty or unrepairable inside is too high, and $2,800 is pretty steep for the working drive. I made a lower offer to the owner of the nonworking version, but could not reach a mutually satisfactory price. I mentioned that if he posts or sends pictures that show enough of the inside to know what kind of gamble I am taking, I would increase my offer, but he has not responded at all.

While the RK-05 and its disk cartridges look very much like the IBM ramkit used on an 1130, they do not make the same grunting noise during seeks. I have to ponder this to see if there is a price at which I would consider buying the 'working' model. The way these drives operate and the documentation available online makes it likely that I could get this working as the 1130 disk drive with some luck and a lot of hard work.

Monday - the base resistor to the relay driver transistor changed to 1K, but most importantly the wiring of the sample relay was based on an erroneous pin layout in the package. This was a relay on a bargains wall at the local electronics supply shop, with a hand drawn diagram of the connections put in the plastic bag along with the relay. When I discovered that the pins labeled NO had a 43 ohm resistance but the pins labeled coil were an open circuit, I realized the diagram was wrong. Once I rewired, it snapped along just fine.

I finished the design file and uploaded it to OSH Park, which estimated that it would be part of the large panel going to the fab on February 27th, which should get them back to me in about two weeks and a couple of days.

No comments:

Post a Comment