This isn't about building something that is just slapping an
SOC or FPGA but more like LPC instead of having a whole ISA
bus in a system.
So, just to call this out: busses like LPC require a state machine to decode them, and far as I can tell there aren't simple glue chips you can buy off the shelf to turn LPC into something that you can easily use to build some kind of homebrew peripheral. The bus was designed to go from a Northbridge to one (or rarely more) legacy-providing SuperIO chips (and sometimes a specialized LPC flash ROM), and all the glue necessary to use the bus exists inside the ASICs. So... unless someone sells off the shelf chips to interface with a bus like this and break it out to something like GPIO pins it's basically useless for homebrewers unless they intend to use it solely as an interconnect between programmable logic? There are verilog implementations of LPC you can download easily enough, sure, but... let's get real here, if you're already up to FPGA-level tech what is the point of having a bus between two FPGAs? Once you get off the bus what you really want is "GPIO", and if GPIO is what you want then why not just slap a microcontroller with plenty of GPIO pins down to handle that? (And do networking with an even *simpler* wiring arrangement to the central control node, assuming you even need it?)
I dunno, this just seems like it's a non-sequitur to me if we're talking about "hobby fun time"... which let's be frank, is not about maximum efficiency in the first place, but... okay, let's pretend we *do* care about efficiency and compare this to a dumb parallel bus? In our scenario here we're going to say we have some kind of "hobby computer", and we want to whip up a peripheral to do some kind of real-world interfacing. For this scenario let's say we want a card that drives some relays to control, say, a sprinkler system. IE, a pretty brain dead application. So let's consider some options and figure out where a "new bus" fits into this:
1: It's 1980, or you're pretending you're living in 1980, and you're looking at a simple dumb parallel bus on a computer with an 8 bit CPU. To interface your relays to this you need an address decoder on the parallel address lines, a latch on the data lines to hold the on/off state for your sprinkler relays, and an open collector buffer or opto-isolator to drive the relays. The most complicated thing here is the address decoder, but that's going to be just a few chips even it you don't use something like a PAL, right? (Unless you're using a bus that has slot enable signals, like an Apple II, in which case you pretty much just need the latch.) So... in this scenario your simple peripheral card is going to need like 4-6 TTL chips, and an ape can solder it together on perfboard. The resulting board lets you toggle the sprinklers on and off with a POKE or OUT command in BASIC, so there you go, off to the races. It's stupidly oldschool, but it seems to me like it's still a pretty reasonable and practical solution to this problem.
2: Your computer is a Raspberry Pi running Linux. This board, which you bought for $35 (unless it's a zero, in which case it was $15), has a header on it which has GPIO pins on it. So you sit down for an hour with the interwebs, read the howtos on how to configure some of those GPIO pins into an output port, and also run into a bunch of grumbling about how preassembled relay boards (which you were thinking of using because they're dirt chip) usually expect 5v logic inputs, but also find a million different ways to handle that. So, boom, you're wetting your lawn in python in no time and you still didn't need more than the most basic soldering/breadboard skills to do it. Easy Peasy.
3: Your computer is an Arduino-style microcontroller board bristling with GPIO ports, because it occurs to you it's dumb for you to have a keyboard and monitor on a thing that's running your sprinkler. So... it's basically just option 2 but lower stakes because your whole computer basically costs nothing. ($3 verses $15.) Hook up the USB cable, bang up some C code in the IDE, and again, problem solved without needing a "bus" at all.
4: You have either 2 or 3 for your host computer, but before you get to work breadboarding your design using parallel GPIO pins to directly map to each relay you discover
you can buy a four-channel relay board that already has an i2c bus chip on it for only $13. Okay, fine, forget the breadboarding, just do that; I can control 8 of these boards with just two pins. Man, at this point this project is getting so easy I'm not sure it's even fun anymore...
5: FINALLY: You have this weird "hobby computer" that has a variant of LPC on it as a "general purpose bus", but no GPIO already broken out. Hrrrm. LPC doesn't have any breadboard-friendly discrete bridge chips that will decode and break out some port lines for you, so I guess you're going to get to work surface-mount (or BGA) soldering an FPGA or CPLD onto a custom board to be able to interface to this bus connector, programming that device with a payload that has the bus/docoding logic, along with an addressable latch for the port, and break out some pins to buffer/voltage shift for your relay module. So... congratulations, we did the job, but the CPLDs and custom circuit boards and precision soldering just completely broke the skill curve compared to the all the other options, in a profoundly negative direction.
I think the point is made here: where is the need for with some *new* kind of "hobby bus" that requires nonexistent bridge chips to actually break it down and doesn't do anything you can't already do better with existing options? (And one of those options that's already better is, in fact, a brain-dead parallel 8-bit bus. It's easy to understand, it doesn't require any specialized chips to interface with it, and can be breadboard friendly.) If you want a modern modular hobby computer there's a huge ecosystem of SPI and i2c peripherals, including options for breaking down these busses into parallel GPIO pins for breadboarding, and you can drive these with nearly any modern microcontroller or SOC, there's no need to invent some other bus. And conversely, the microcontroller model of just slapping a cheap ARM core with GPIO ports in front of whatever you want to control basically negates the need for busses at all. Wifi/Bluetooth-equipped ESP32 microcontrollers with a decent amount of GPIO sell for less than $5; if you want a modern modular system all the pieces are there right now and you don't even need wires. The lack of a "modern bus" just does not seem like a serious problem.
This all seems like a weird soap box to get up on in a thread that's specifically about building a card for the S-100 bus. I mean, it kind of goes without saying that anyone reading this likely *to* specifically be interested in this old hardware? I mean, sure, we can argue all day about whether S-100 is a "good" bus or not (objectively, it's not), but it is certainly historically significant and there are people that want to keep S-100-based computers "alive" in a physical sense, and that's going to mean building things to fit the slots despite all the warts. I'm personally not a *huge* fan of "software defined cards" (while completely being aware that they can serve a legit purpose for some people) but I can certainly see why *someone* might want it.
