We really love when hacks of previous hacks show up in the tip line. It shows how the hardware hacking community can be a feedback loop, where one hack begets the next, and so on until great things are everywhere. This hacked joystick port for an FPGA Pac Man game is a perfect example of that creative churn.
The story starts with Pano Man, a version of the venerable arcade game ported to a Pano Logic FPGA thin client by [Skip]. We covered that story when it first came out, and it caught the attention of [Tom Verbeure], particularly the bit in the GitHub readme file which suggested there might be a better way to handle the joystick connections. So [Tom] took up the challenge of using the Extended Display Identification Data (EDID) circuit in the VGA connector to support an Atari 2600 joystick. The EDID system is an I²C bus, so the job needed the right port expander. [Tom] chose the MCP23017, a 16-bit device that would have enough GPIO for dual joysticks and a few extra buttons. Having never designed a PCB before, [Tom] fell down that rabbit hole for a bit, but quickly came up with a working design, and then a better one, and then the final version. The video below shows it in action with Pano Man.
We think the creative loop between [Skip] and [Tom] was great here, and we can’t wait to see who escalates next. And it’s pretty amazing how much IO can be stuffed over two wires if you have the right tools. Check out this VGA sniffing effort to learn more about EDID and I²C.
Continue reading “Two Joysticks Talk To FPGA Arcade Game Over A VGA Cable”
Did you ever wonder what your monitor and your computer are talking about behind your back? As it turns out, there’s quite a conversation going on while the monitor and the computer decide how to get along, and sniffing out VGA communications can reveal some pretty fascinating stuff about the I²C protocol.
To reverse engineer the configuration information exchanged between a VGA monitor and a video card, [Ken Shirriff] began by lopping a VGA cable in two. The inside of such cables is surprisingly complex, with separate shielding wires for each color and sync channel and a host of control wires, all bundled in multiple layers of shielding foil and braid to reduce EMI. [Ken] identified the clock and data lines used for the I²C interface and broke those out into a PocketBeagle for analysis using the tiny Linux machine’s I²C tools.
With a Python script to help decode the monitor’s Extended Display Identification Data (EDID) data, [Ken] was able to see everything the monitor knows about itself — manufacturer, serial number, all the supported resolution modes, and even deprecated timing and signal information left over from the days when CRTs ruled the desktop. Particularly interesting are the surprisingly limited capabilities of a VGA display in terms of color reproduction, as well as [Ken]’s detailed discussion on the I²C bus in general and how it works.
We always enjoy these looks under the hood that [Ken] is so good at, and we look forward to his reverse engineering write-ups. His recent efforts include a look at core memory from a 50-year old mainframe and reverse engineering at the silicon level.
[Andreas] found that his home theater PC would not boot one day. Oddly, if he disconnected his HDMI cable from his TV, it would boot fine. While most of us would have dug into the PC, he realized that it was a signal from the TV that was incorrect. Luckily, LG had included a full schematic with the TV. What he was able to figure out, using a home made snooper was that the EDID eeproms had somehow become corrupted. Not to worry, [Andreas] slapped together a full blown I2C interface and prepared to reprogram them with the correct data. He noticed, however, that the eeproms were write protected. On a whim, he decided to write to them any way and found that it was successful. He has some theories as to why they were writable, but says that he doesn’t want to pull the TV back apart to confirm.