I'm personally interested in both. Making and or using motherboards for modules (end users) and making modules (module maker for the lack of a better term).
Following pt's suggestion I making two issues one for end user (this one) and one for module makers (here).
For end users I see two probably very distinct users. On the one side the maybe more "advanced" users who are interested in making PCBs, this would be more akin to Feather Wings. A PCB that houses a module and can serve a very specific purpose. Here I see a lot of different reasons why you would want that.
- Size, a Feather is small compared to older form factors but for many applications still too big (especially wearables). There exist smaller dev boards standard but at the same time many applications can't afford to loose so many pins for size.
- Pins, even if you don't care too much about size. any dev board with 65 pins is just enormous. Such a module can offer similar amount GPIO/BUSes as a Arduino Mega form factor board that right now would be way too big for most applications if it were used as a module within an application.
- IO Flexibility, very often when you make your own boards with a MCU (or FPGA), you most of the time only really care about being more flexible with which IO's are exposed and what connectors they are terminated to or need more control over the form factor, which where traditional dev boards can become difficult because they make a lot of assumptions or are just too large to be flexible enough. The M.2 connector offers enough pins to break out complete chips, giving you access to absolutely everything available to any specific piece of silicon. This would make it much less necessary to re-implement a MCU every time you simply want control over connectors or form factor.
Why should I implement a SAMD51 on my badge or wearable for the 100th time if I can just pop a tiny module in there and concentrate on the stuff I really care about and also be able to easily switch to a different MCU if I think I need wireless after all.
This to me personally is the biggest reason why I find this interesting. I would design a lot of my boards very differently if I had such an ecosystem at my disposal. It gives a lot of design freedom and flexibility without loosing in function or size like I would with a traditional dev board.
The other user group would be more the beginner who probably does not wanna make a motherboard PCB but is probably interested in using motherboards from other people and being able to switch brains of these boards to whatever suits their needs. This is more close to what Makerdiary did for their keyboard. It's very attractive to a user to be able to essentially do a CPU upgrade to their fancy keyboard. Want Bluetooth in the future? No problem. No need to buy a whole new keyboard.
You are upgrading your keyboard? Take the brain with you from your old keyboard and save some money with the upgrade. This can make certain products a lot cheaper in the long term if you don't have to sell the same chip to your users over and over if your actual product is something else and the brains is just the vehicle to that, in turn making it easier to sell a new generation because the base cost comes down.
This is also a train of thought I followed with my Tablet ( https://hackaday.io/project/164845-dlt-one-a-damn-linux-tablet ).
It's not M.2 but the ability for an end user to buy and use different brains with the same periphery is very attractive and makes the hardware more sustainable and long living while also giving more feature freedom to the user.
And one last point albeit very niche and tbh not too close to my own interests atm but I very much see the use case; The M.2 connector is rated for really high-speed signals. Something you can't really achieve with good old pin headers. A board that can be used with a RISC-V softcore as a more MCU style board in one motherboard could serve in a different motherboard as an HDMI output. Instead of PMODs you could just have purpose made more optimized tiny motherboards where you plug your FPGA module into. This enables not only more high-end applications but also makes all of them a lot smaller compared to a board with tons of PMOD connectors to serve the same amount of pins.
This is something no other standard so far on the market could offer.
I'm personally interested in both. Making and or using motherboards for modules (end users) and making modules (module maker for the lack of a better term).
Following pt's suggestion I making two issues one for end user (this one) and one for module makers (here).
For end users I see two probably very distinct users. On the one side the maybe more "advanced" users who are interested in making PCBs, this would be more akin to Feather Wings. A PCB that houses a module and can serve a very specific purpose. Here I see a lot of different reasons why you would want that.
Why should I implement a SAMD51 on my badge or wearable for the 100th time if I can just pop a tiny module in there and concentrate on the stuff I really care about and also be able to easily switch to a different MCU if I think I need wireless after all.
This to me personally is the biggest reason why I find this interesting. I would design a lot of my boards very differently if I had such an ecosystem at my disposal. It gives a lot of design freedom and flexibility without loosing in function or size like I would with a traditional dev board.
The other user group would be more the beginner who probably does not wanna make a motherboard PCB but is probably interested in using motherboards from other people and being able to switch brains of these boards to whatever suits their needs. This is more close to what Makerdiary did for their keyboard. It's very attractive to a user to be able to essentially do a CPU upgrade to their fancy keyboard. Want Bluetooth in the future? No problem. No need to buy a whole new keyboard.
You are upgrading your keyboard? Take the brain with you from your old keyboard and save some money with the upgrade. This can make certain products a lot cheaper in the long term if you don't have to sell the same chip to your users over and over if your actual product is something else and the brains is just the vehicle to that, in turn making it easier to sell a new generation because the base cost comes down.
This is also a train of thought I followed with my Tablet ( https://hackaday.io/project/164845-dlt-one-a-damn-linux-tablet ).
It's not M.2 but the ability for an end user to buy and use different brains with the same periphery is very attractive and makes the hardware more sustainable and long living while also giving more feature freedom to the user.
And one last point albeit very niche and tbh not too close to my own interests atm but I very much see the use case; The M.2 connector is rated for really high-speed signals. Something you can't really achieve with good old pin headers. A board that can be used with a RISC-V softcore as a more MCU style board in one motherboard could serve in a different motherboard as an HDMI output. Instead of PMODs you could just have purpose made more optimized tiny motherboards where you plug your FPGA module into. This enables not only more high-end applications but also makes all of them a lot smaller compared to a board with tons of PMOD connectors to serve the same amount of pins.
This is something no other standard so far on the market could offer.