02-05-2021, 03:02 PM
(This post was last modified: 02-05-2021, 03:31 PM by EternityForest.)
[quote pid='89016' dateline='1612558952']
These systems keep popping up everywhere, and they have a ton of potential, but the education-focused companies don't really take advantage of the full possibilities of the system. They flood everything with gimmicky sensors for throwaways projects, but they're missing a lot of what makes the whole thing special.
I think there should be a new line of compatible systems, designed for building truly commercial quality products in a modular way.
Right now, modular systems just aren't that great for anything but educational use. I never see anyone actually building anything with them, even on a DIY level.
But there's only a few changes needed to make them totally usable, none of which should add more than a few dollars to the price. The whole thing might even be able to replace or supplement other ideas like the PineCom, by just letting people build their own device for everything that the PineTab and phone aren't suited for.
Lego Mindstorms already proved the concept. Modular Electronics, even at the consumer level, is totally possible and practical, but only when it's actually modular and doesn't depend on adding in a ton of ugly non modular hacks.
Here's what I've noticed has room for improvement in the existing modular space:
* Modular power. Define a new connector for unregulated 3-20v, which has a Vin bus and a main battery bus. Any module can supply power or take power from either one with a diode.
In this way, things like robots that need 12v for motors can be totally modular.
For maximum compatibility, the connector could be a 5-pin JST-PH, having ground, unregulated 0-24Vin, 0-24Vbat, and regulated 5v, and 12v, all busses that any module can take from or supply to.
Almost every project has that one part that needs one of those voltages, with no way to supply it.
Power input modules should also provide power to the VBat bus, but batteries should not provide to Vin, to avoid charging a large battery from another battery.
And everything should be reasonably protected against input transients as might occur from cable inductance.
* Lower Power/better batteries
Although they're terribly hard to find in America, LTO batteries are cheap and last 20k cycles, and can be charged like capacitors, with a nice voltage curve that makes it easy to check the level. To really let people make high quality devices, we should be using batteries that don't explode and don't wear out.
So many of the worlds "Throw an Arduino at it" projects use almost no power. A huge number could in fact only be a mA or two.
A mainboard should be able to run from USB or the 5V bus, but also should be able to at least run in very low power trickle charge mode from the unregulated VBat bus.
There's so many projects you just can't do without building a ton of stuff from scratch, because nothing is really designed for this kind of thing. Everything just has WAY too much idle power draw.
* Fewer Modules
At most, I think there should probably be an ESP32-S2 and an RFM69 or LoRa based module, a USB-C power input module, maybe a 12v battery, and maybe a motor/LED driver.
Most of the cheap modules are easy and trivial to make yourself, or else they're already covered by STEMMA or the like. Just use the STEMMA standard connectors, give us a top notch base, and the rest will be pretty trivial with a basic breakout board for the connector.
Adding 800 different kinds of main processor chip is just fragmentation.
* Higher Quality modules
Things should be designed with all proper protections. A motor driver with current limiting and I2C diagnostics for short/open can handle a whole crapload of use cases. A giant relay is cheap, but needs support component and can't do certain things.
If there's an ADC module, it needs all appropriate ESD shielding you would want in an industrial grade part.
* Case Friendly mechanical design
We should be able to 3D print a case for this stuff, in a way that exposes the USB port, and maybe even a power switch and a button or two for things like easy pairing and setup. The usual modular boards have tiny buttons that are not at all friendly to enclosures.
Perhaps a standard back panel layout could even be designed, that would be common to all mainboards, with a USB, power switch, and a button and LED or two.
Any additional addons should also be designed with consideration for enclosures.
* Standard connectors
Grove is cool and all, but Adafruit's STEMMA uses standard JST-PH connectors that are cloned basically everywhere. Might as well use something like that.
* Sold as fully complete kits?
One way to avoid just making a bunch of random useless junk would be to start by focusing on complete kits that resulted in something usable for everyday use.
Imagine a PinePlug modular smarthome plug that you could customize with your own firmware, or add a motion sensor to the externally-exposed connector, etc.
Or a PineBot robot, or even a PineLight flashlight. Maybe a PineStation weather station, or a PineCalc graphing calc/universal remote/etc type device that could take over a lot of the PineCom's use cases.
Door and gate sensors that run forever on tiny solar panels would be amazing. Tiny E-Ink displays would be cool.
It's easy to see how a half-dozen modules or so could be used to make most of these things, with so many consumer gadgets only having at most one custom bit of tech. A modular system like this would make it very easy to add new products without the risk of making e waste trash.
And a lot of potential product suggestions would, IMHO, be a lot better as modules than full products.
[/quote]
These systems keep popping up everywhere, and they have a ton of potential, but the education-focused companies don't really take advantage of the full possibilities of the system. They flood everything with gimmicky sensors for throwaways projects, but they're missing a lot of what makes the whole thing special.
I think there should be a new line of compatible systems, designed for building truly commercial quality products in a modular way.
Right now, modular systems just aren't that great for anything but educational use. I never see anyone actually building anything with them, even on a DIY level.
But there's only a few changes needed to make them totally usable, none of which should add more than a few dollars to the price. The whole thing might even be able to replace or supplement other ideas like the PineCom, by just letting people build their own device for everything that the PineTab and phone aren't suited for.
Lego Mindstorms already proved the concept. Modular Electronics, even at the consumer level, is totally possible and practical, but only when it's actually modular and doesn't depend on adding in a ton of ugly non modular hacks.
Here's what I've noticed has room for improvement in the existing modular space:
* Modular power. Define a new connector for unregulated 3-20v, which has a Vin bus and a main battery bus. Any module can supply power or take power from either one with a diode.
In this way, things like robots that need 12v for motors can be totally modular.
For maximum compatibility, the connector could be a 5-pin JST-PH, having ground, unregulated 0-24Vin, 0-24Vbat, and regulated 5v, and 12v, all busses that any module can take from or supply to.
Almost every project has that one part that needs one of those voltages, with no way to supply it.
Power input modules should also provide power to the VBat bus, but batteries should not provide to Vin, to avoid charging a large battery from another battery.
And everything should be reasonably protected against input transients as might occur from cable inductance.
* Lower Power/better batteries
Although they're terribly hard to find in America, LTO batteries are cheap and last 20k cycles, and can be charged like capacitors, with a nice voltage curve that makes it easy to check the level. To really let people make high quality devices, we should be using batteries that don't explode and don't wear out.
So many of the worlds "Throw an Arduino at it" projects use almost no power. A huge number could in fact only be a mA or two.
A mainboard should be able to run from USB or the 5V bus, but also should be able to at least run in very low power trickle charge mode from the unregulated VBat bus.
There's so many projects you just can't do without building a ton of stuff from scratch, because nothing is really designed for this kind of thing. Everything just has WAY too much idle power draw.
* Fewer Modules
At most, I think there should probably be an ESP32-S2 and an RFM69 or LoRa based module, a USB-C power input module, maybe a 12v battery, and maybe a motor/LED driver.
Most of the cheap modules are easy and trivial to make yourself, or else they're already covered by STEMMA or the like. Just use the STEMMA standard connectors, give us a top notch base, and the rest will be pretty trivial with a basic breakout board for the connector.
Adding 800 different kinds of main processor chip is just fragmentation.
* Higher Quality modules
Things should be designed with all proper protections. A motor driver with current limiting and I2C diagnostics for short/open can handle a whole crapload of use cases. A giant relay is cheap, but needs support component and can't do certain things.
If there's an ADC module, it needs all appropriate ESD shielding you would want in an industrial grade part.
* Case Friendly mechanical design
We should be able to 3D print a case for this stuff, in a way that exposes the USB port, and maybe even a power switch and a button or two for things like easy pairing and setup. The usual modular boards have tiny buttons that are not at all friendly to enclosures.
Perhaps a standard back panel layout could even be designed, that would be common to all mainboards, with a USB, power switch, and a button and LED or two.
Any additional addons should also be designed with consideration for enclosures.
* Standard connectors
Grove is cool and all, but Adafruit's STEMMA uses standard JST-PH connectors that are cloned basically everywhere. Might as well use something like that.
* Sold as fully complete kits?
One way to avoid just making a bunch of random useless junk would be to start by focusing on complete kits that resulted in something usable for everyday use.
Imagine a PinePlug modular smarthome plug that you could customize with your own firmware, or add a motion sensor to the externally-exposed connector, etc.
Or a PineBot robot, or even a PineLight flashlight. Maybe a PineStation weather station, or a PineCalc graphing calc/universal remote/etc type device that could take over a lot of the PineCom's use cases.
Door and gate sensors that run forever on tiny solar panels would be amazing. Tiny E-Ink displays would be cool.
It's easy to see how a half-dozen modules or so could be used to make most of these things, with so many consumer gadgets only having at most one custom bit of tech. A modular system like this would make it very easy to add new products without the risk of making e waste trash.
And a lot of potential product suggestions would, IMHO, be a lot better as modules than full products.
[/quote]
