so i just added an assortment of shift registers, hex inverters and other ics to my 7400 series collection, and 5, 2-digit, 7 segment displays. still not enough to build anything fun.
actually i should have enough parts to build that hex keypad/display ive always wanted. which should help me debug the bus/memory i/o for my other project, my video card of doom.
the device will feature a set of 4 8-bit i/o ports. each bus can be put in input or output mode, buses set to input mode simply sample the bus periodically and update the 7 segment display with hex info (8 hex digits, which represent the data to/from the ports). output mode is fun. it will have a 20 key serial keypad with all 16 hex digits, and 4 function keys (enter, 2 nibble entry select, and another one, possibly for optional modes). this will allow you to enter data for all output ports. the data will be displayed on the display and updated on the bus with the enter key aside from the direct parallel interfaces. the serial i/o port will also allow direct interface for the keypad directly, or will allow 2-way communication with any device with a uart (i may use i2c instead, i haven't really decided). should be a nice little debug tool.
i can actually use this thing for putting test data into sram chips and the like. probibly even real time programming in machine code if one was so inclined. i can use the serial port for loading data from sd cards and writing it to memory/eeproms for example.
In all seriousness, reading about all the stuff you do is quite inspiring. Keep posting. 
id be amazed if i actually manage to accomplish something other than a crappy prototype. i have completed a couple things though. my r/c receiver kinda works, the firmware needs some work though. its really underpowered, it can only source about an amp, for servos and motors, which isnt enough to do anything big, it needs a second regulator and some heat sinks. i still need to hammer the bugs out of my wireless gamepad as well. i deadbugged the whole thing and there is probibly some noise/decoupling issues. the output signal is noisy and i havent been able to make it work with the receiver board, despite the simpler packet structure, prolly a crc glitch.
the joystick of doom works but i need to source some high quality buttons. like those big round arcade buttons. im also considering widdling a new handle out of wood, or having one 3d printed, and getting rid of that old sidewinder handle. 3d printing really doesnt give me any tweak room, where wood does allow for some dremel work. cut it, saw it in half, hollow it out, drilling holes for buttons, triggers, hats, thumbsticks. and wrap the whole think in buck skin. most manly joystick ever. cost and lack of appropriate tools are really slowing this up.
my video board got to near production when i realized that my memory was a little slow and the mcu didnt have enough time to do the layer system i had planned, it didnt even have time to page in a line of pixels during the horizontal blanking period. so no point wasting a powerful (and expesnive) mcu like the atmega1284. i decided to do composite instead of vga, using an an ad725 rgb to yuv ic, and an overclocked avr. similar configuration to a uzebox but with the hardware double buffer i had designed earlier. i also realized i want more memory. i was looking at a 120x90*8 screen resolution, and i could make it something like 256x128*8 with more ram. the new design will only use the onboard mcu to twiddle addresses, and generate sync pulses. all drawing will be done with another processor, so no hard restrictions on how drawing can be done (it could be vector/raster/mixed/whatever, you're essentially writing into ram). or i could dedicate the second mcu entirely to low level drawing calls.
anyway i decided to step back from that and do something that is less timing sensitive. ive already started my hex display dohicky in eagle cad. its a fight to make sure that i can run the whole thing with as few pins as possible. i need to get it down to 18 pins so im multiplexing the **** out of everything. im considering using 8 identical shift registers for output4 for i/o ports and 4 for the 7 segment display. the displays are 2 digit with a common anode for each digit. i realized that using the 3-8 decoder had a major drawback, i still need driver transistors, and the duty cycle for any digit would be 1/8 or less. but using 4 shift registers i only need 2 banks of 4, getting the duty cycle up to slightly less than 1/2. all these shift registers share a clock and uses an entire mcu io port, but the good news is i can use the entire io (really just the o, input is routed through a series of buffers into a common piso shift regester, so they may be sampled one at a time independant of output) port and shift in the data simultaneously. i wanted to use universal shift registers or a latched bus tranciever, so one chip covers both jobs of input and output instead of 2. , but they were kind of expensive to source from the chinese people on ebay.
ive been long considering some kind of tech blog, but last time i tried that it degenerated into a soapbox for preaching my hatred for humanity and why it needs to be destroyed.
