New blog post about the #SmolPhone (our take on #FrugalComputing): I gave a short talk about it, and decided to write the things I usually say. It's here: https://people.irisa.fr/Martin.Quinson/blog/250528/Smolphone-Magellan/
I think that the result is a nice introduction to the project. Please comment and tell us what you think of it!
First prototype of the #SmolPhone, with a screen, a keyboard and a RP2040 as main compute unit. The 4G and wifi modems will be for a future iteration.
Don't expect it to be functional yet. It's assembled in a rush since 2 days: the keyboard is connected but not recognized by the program loaded on the 2040. Only the screen is used for now, to move a little black square and its friend the white triangle around the screen.
More to come :)
And now, our software boots up on our hardware prototype of the #SmolPhone The keyboard kinda works, too. More work in the future but OK for now.
The substrate is able to display a "modern" UI with buttons, text areas, labels, checkbox and such under the RP2040 constraints (about 200k of RAM but rather OK compute power).
The goal is to allow users to build apps with lua scripts, as in #Scrappy https://jrcpl.us/contrib/2025/Scrappy Maybe before the end of the year, if we're lucky.
@bonkers The HW engineer told us that we cannot go much further with a RP2040 because our devboard has not enough pins for all the devices I'd want on a #SmolPhone, so we will use a RP2350 instead, which has twice the amount of pins.
Since we braced for of the RP2040, the new MCU seems incredible. 520kb SRAM and dual M33 cores on the chip, that's already great. But the devboard has 48 pins, a SD card reader and even 8Mb of on board RAM! That's almost too much, I'm curious about the power draw.
Today, the SW engineer got the keyboard fully working in our toolchain, despite some bad surprise (but only a Rust adept would think that dependencies are a blessing).
We won't invest much on the blackberry keyboard anyway because the HW engineer told us that it doesn't fit the design. It requires a specific voltage matching no other component of our design, so we will save a power rail and design our own keyboard at some point in the future. The #PicoCalc is inspiring on this point.
And now, the keyboard works completely with the little UI of the #SmolPhone. So does the battery (not charging the battery, but powering the board from a battery) and the little touchpad of the blackberry keyboard.
Not impressive for now, I agree, but I'm still super happy of these tiny improvements.
@emptty yeah, I wouldn't want to be a battery for this beast :)
IDK about the 2350, but 2040 is pretty power hungry. Also, there's no out of the box low power mode.
@bonkers Please define "power hungry". Both 2040 and 2350 are huge compared to typical MCUs, but from our initial measurements ( https://inria.hal.science/hal-04589322 ) I still think that we could a kind of smartphone with a battery life of about one week, using a 4.2Ah battery similar to the ones of iPhones or FairPhone 5.
The design goals and power draw computation (wandering thoughts) are on
https://people.irisa.fr/Martin.Quinson/blog/250528/Smolphone-Magellan/ In particular on slide "the hard reality"
The SmolPhone project explores the design space of modern smartphones with energy limits, increasing the battery lifetime possibly at the price of a reduced feature set. The energy consumption of classical smartphones can roughly be divided in three main parts: the processing units, the screen, and the network connectivity. This paper evaluates the practical feasibility of the SmolPhone vision, which involves heterogeneous multiprocessors and low-power technologies for the screen and network. We identify several usage modes according to the provided features, and quantify the energy consumption of the processing units in each mode, as well as the time required to switch between modes.
@emptty I measured about 15mA at lowest clock frequency, as far as I recall. That's definitely not a week of work on a phone battery.
Esp32 can go into deep sleep, so it's much easier to save power. But many boards are not designed well, so they leak a few mA anyway.
@bonkers We measured the 2040 at 37mW in light sleep and 6mW in deep sleep. I'm not speaking of a 3310 battery (under 1Ah) but of a FairPhone 5 one (4.2Ah). That amount of energy leads to a power budget of about 100mW on average (unless I'm mistaken).
This is all very hypothetical for now because we don't know for sure how often the software will allow the MCU to deep sleep, and it does not take the 4G power draw into account. Getting the POC and then the numbers is the whole project goal.
Martin Quinson ⏚
Bonkers