-Do you have a picture of it of a first prototype?
I added a sketch (thanks to MS Paint):
Update: To show you what someone really cool did in building a laptop, check out this:
My next picture is not really trying to compete with that. It is just using all off-the shelf- products.
The cost of the chassis is remarkably lowered by using pre-built components like the Apache 1800. The most convenient feature is the hinge, which would probably cost more to manufacture elsewhere.
I think the least expensive e-reader with 80% of the app functionality of a Raspberry Pi is the Onyx Boox Poke 3 ($189).or the Poke 2 Color ($279). I could connect my bluetooth keyboard and use the Poke 2 as a typewriter. It has several Google Play apps that can be sideloaded. I think it’s a nice looking tablet too. My Apache 1800 cost me $13, and as of yesterday it was only $10 at Harbor Freight. Combined with an e-reader that allows blue-tooth keyboards, I would be hard pressed to built a cheaper laptop. That said, there is elegance in a simple, but featured design like a lightweight linux running on an e-ink display. This e-ink is the 10th generation Kindle with front light, and I paid $45 like new. That means the $10 bluetooth keyboard pictured doesn’t even work with it, lol, but it looks like a laptop, right? 
It even supports landscape mode, but not on the experimental browser. There are so many preferences when it comes to building a laptop that I partly believe it could be kept an open standard, but at the same time sold as a ready to use PC like the Raspberry Pi 400 as an option, with much more design already picked out.
I think for the most modular approach, replacing that Pelican 1150 LCD screen in that youtube link with an e-ink screen would be the simplest way to build an e-ink laptop. It would be low-cost, and accessible, rather than proprietary. The main downside is that it wouldn’t nearly as thin as other laptops. I consider the Pelican/Apache size so versatile that it appears to be the de-facto open standard (ATX-like) of SoC based laptop cases.
My first “prototype” was a 5 Watt panel. A large laptop, typically 15.6 or 17" could fit a 10Watt panel, although there may be more efficient panels that can fit 21 watts. I would like to stick to this size, although fold out panels are available too. I have added a couple pictures here: but I have more here: https://www.raspberrypi.org/forums/viewtopic.php?f=62&t=289595#p1753787
-Any plans to commercialize it, or more like a hobby project?
Currently it is a hobby project, but I welcome anyone to contribute ideas/sample boards/displays. I have built many PCs with the ATX and microATX form factor, so I am partly leaning towards having a laptop that can be partially open-standard, although this isn’t a requirement for a prototype, only because the first assembly of any self-powered laptop is likely to come from bringing together three companies that are not necessarily already working on some integrated product- that is a solar panel manufacturer, a SoC maker, and a display maker. I have read that the Intel IoT Edge Mote picture:
uses a solar panel on an IoT, so there may already be some specialized companies that are capable of bringing the solar panel tech to the circuit board and may just not have adapted it to an e-ink capable of a larger resolution, and one that can run linux.
"Have you tried this?
I have not, although I have heard of similar projects: https://hackaday.com/2020/12/21/solar-pi-zero-e-paper-photo-frame-waits-for-the-right-moment/
Also, I found this RTC: https://ambiq.com/rtc/
As a practical proof of concept for a solar laptop, I would like to update with my research from today and cover three topics: Screen, solar, and SoC.
Screen:
Considering: A screen that would be considered e-reader or netbook sized: 5-7." A smaller screen wouldn’t be considered really practical, even if it is still “proof of concept”.
I checked a handful of Waveshare eink panels for some of the slowest refresh rates from 26s to <1s refresh, from slowest to fastest:
600x448, 5.83inch E-Ink Display HAT SKU: 14732
Grey level: 2
Full refresh time: 26s
Refresh power: 26.4mW(typ.)
Standby power: <0.017mW
880×528, 7.5inch 3-color E-Ink display HAT for Raspberry Pi: SKU: 17960
Grey scale: 2
Full refresh time: 21s
Refresh power: 50mW(typ.) (power increase seems due to more pixels)
Standby power: <0.017mW
5.65inch ACeP 7-Color E-Paper E-Ink Display Module, 600×448 SKU: 18295
GREY SCALE 2
FULL REFRESH TIME 15s
REFRESH POWER 50mW (typ.)
STANDBY CURRENT <0.01uA (almost none)
640x480 5.83inch E-Paper E-Ink Display HAT SKU: 14597
GREY SCALE 2
FULL REFRESH TIME 5s
REFRESH POWER 26.4mW(typ.)
STANDBY CURRENT <0.01uA (almost none)
The above----compared to SKU 14732 it seems to have >5x faster refresh with similar resolution. This may just mean that the refresh rate is using 26.4mw every 5 seconds rather than every 26 seconds. Assuming any of these panels can even be hacked to run LXDE linux and not just a slideshow 
It is a $52 panel, and I am curious how it would run
800x600, 6inch E-Ink display HAT for Raspberry Pi 800x600, 6inch E-Ink display HAT for Raspberry Pi SKU: 15852
Greyscale: 16
Operating voltage: 5V
Interface: USB/SPI/I80
Resolution: 800 × 600
Full refresh time: <1s
Total refresh power: 0.6W(typ.) [600mW]
Total standby power: 0.3W(typ.)
I like the one above because it still has relatively low power use among E-ink. The higher resolution, the more board system power.
However, there doesn’t appear to be a driver that can use this for a GUI. It appears to be only for displaying a set of images or text.
Pros: relatively fast refresh. 1s is definitely not fast enough for video, but could be used for browsing and typing. Without solar panels being more than 22% efficient and the power consumption being in the microwatt range for refresh, limiting the refresh rate to 1s-2s is currently one of the only ways to keep the power consumption as low as possible.
Since it is a $52 panel with no known LXDE capability, a development panel I would like to see worked on is the $29. It includes a header for the Pi.
3.7inch e-Paper e-Ink Display HAT For Raspberry Pi, 480×280. SKU: 18057
GREY SCALE 4
OPERATING VOLTAGE 3.3V/5V
INTERFACE 3-wire SPI, 4-wire SPI
RESOLUTION 480 × 280 pixels
PARTIAL REFRESH TIME 0.3s
FULL REFRESH TIME 3s
REFRESH POWER 50mW (typ.)
STANDBY CURRENT <0.01uA (almost none)
18057
Placing a hard limit on the refresh like 3s could help some beginners /power user from running the battery out much faster than the solar panels can recharge the laptop. The alternative is to have a larger battery, one that can last for 2-3 days, or as many days as the user believes they will be out of the sun. The problem with this however, is it’s not really my ideal of a solar laptop- it is one that can turn on with a single 60W equivalent light bulb and not be drained even in a stress test. Some features, like video, could even be disabled to prevent crashes and wear and tear on the screen. But future models could include it if it able to handle video on ambient light power.
Another display of a very similar resolution is the Adafruit SHARP Memory Display Breakout (400x240) $44.95
“The Adafruit 2.7” 400x240 SHARP Memory Display Breakout is a chonky cross between an eInk (e-paper) display and an LCD. It has the ultra-low power usage of eInk and the fast-refresh rates of an LCD. This model has a gray background, and the pixels show up as black-on-gray for a nice e-reader type display.
“The display is ‘write only’ which means that it only needs 3 pins to send data. However, the downside of a write-only display is that the entire 400x240 bits (13.5 KB) must be buffered by the microcontroller driver. That means you cannot use this with an ATmega328 (e.g. Arduino UNO) or ATmega32u4 (Feather 32u4, etc). You must use a high-RAM chip such as ATSAMD21 (Feather M0), Teensy, ESP8266, ESP32, etc. On those chips, this display works great and looks fantastic.”
(This appears to only have one gray, although the refresh rate is great for a low power display- i do not know how much power it actually uses though)
Waveshare has confirmed on both their 1872×1404 7.8" and 10.1 e-ink screens that use HDMI:
“Dear, Reading documents, around 5V 400mA-550mA，playing video, around 550mA-1100mA. The power consumption is mainly related to the refresh rate and the number of pixels that are refreshed.”
5Vx1.1A is 5.5W. If one toggles browser tabs or windows on a laptop, a refresh will happen every several seconds and use I estimate, 2Whr. These 2 screens are nice for reducing eye-strain, but ambient light and even sunlight is nowhere near capable of powering these two without several hours of charging in daylight.
I am curious if these e-paper displays can work the Ambiq Apollo 3 Blue/Apollo 4 Dev boards, which has SPI pins, and which I will return to later in this post.
I will return to displays soon but I wanted to cover a little on solar.
Solar:
There are a lot of misconceptions about amorphous solar panels, and even I was confused. I used the TI-30Xa Solar as an example, but I realize it doesn’t use anywhere near the power as a Pi Zero. That said, there are some advantages to using amorphous because it is more efficient in low-light (i.e. indoor). I found a discarded solar light that also is able to charge an AAA battery using just 3 light bulbs. It also uses amorphous panels. The benefit is that it will charge with a VERY low threshold compared to crystalline. I am a little surprised I didn’t think of amorphous earlier. Even if it costs a little more, the amount of system power design limit (TDP) being used will be lowered so that it can handle it, and so that not a ton of panel area is needed.
from https://www.redarc.com.au/poly-vs-mono-vs-amorphous-know-the-difference :
Amorphous cells offer higher efficiency than the other two. They are your most efficient cell in the market today, although they do require twice as much surface area for the same power output as a monocrystalline blanket or panel. However, they are more flexible and can handle higher temperatures better.
"Amorphous cells are constructed from a fine layer of silicon, which enables solar panels to be more flexible and therefore lightweight.
Amorphous cells can withstand higher temperatures without output being affected, compared to poly or mono crystalline cells.
Amorphous cells perform better in low light conditions compared to even the most efficient monocrystalline panels. This is because they can absorb a wider band of the visible light spectrum due to the uni-solar triple junction cell technology."
from: https://news.energysage.com/types-of-thin-film-solar-panels/
Advantages and disadvantages of amorphous solar panels
Amorphous solar panels have many advantages over their solar panel counterparts. For one, companies don’t need to use a lot of toxic materials to build amorphous silicon panels – this is not always true with some other types of panels. Additionally, they require much less silicon than conventional solar panels. Amorphous solar panels are also bendable, and less susceptible to cracks than traditional panels constructed from solid wafers of silicon.
However, there are also some disadvantages to amorphous solar panel technology, the primary challenge being their efficiency: compared to conventional silicon solar cells, amorphous solar cells are typically less than half as efficient. Most types of amorphous solar panels hover around 7 percent efficiency, while mono- or poly-crystalline solar panels on the market today can have efficiencies of over 20 percent.
Cadmium telluride (CdTe) solar panels
cadmium telluride solar panels
Cadmium telluride (CdTe) panels are the most popular type of thin-film solar technology used in installations today. These panels are made up of several thin layers: one main energy-producing layer made from the compound cadmium telluride, and surrounding layers for electricity conduction and collection. One of the most well-known manufacturers of CdTe panels is First Solar, an American company headquartered in Tempe, Arizona.
Advantages and disadvantages of cadmium telluride solar panels
One of the most exciting benefits of CdTe panels is their ability to absorb sunlight close to an ideal wavelength, or shorter wavelengths than are possible with traditional silicon solar cells. Simply put, shorter wavelengths mean higher energy, which is easier to convert to electricity. Plus, cadmium telluride panels cost less to manufacture and install than other types of solar panels.
However, one of the biggest concerns with CdTe panels is pollution. Cadmium is a toxic heavy metal – one of the most potent ones at that. Cadmium telluride, the compound used in these panels, also has some toxic properties. Importantly, CdTe panels are not harmful to humans or the environment as they generate electricity on rooftops, and companies take proper health precautions when handling the materials during the manufacturing process. However, the disposal of old CdTe panels continues to be a concern.
Also, like amorphous panels, cadmium telluride panels come in at lower efficiencies than other types of solar panels. Sitting around 10 to 11 percent, CdTe panels are above the efficiencies of amorphous panels, but still don’t come close to the average efficiencies of standard silicon panels.
Copper gallium indium diselenide (CIGS) solar panels
CIGS solar cells are made from a compound called copper gallium indium diselenide (try saying that five times fast!) sandwiched between conductive layers. This material goes on top of different types of layers, such as glass, plastic, steel, and aluminum. Some types of CIGS panels use a flexible backing, and the thin layers allow for full-panel flexibility.
Advantages and disadvantages of CIGS solar panels
Unlike most other thin-film solar technologies, CIGS solar panels offer competitive efficiencies to traditional silicon panels. With efficiencies exceeding 20 percent in laboratory tests, there may be a place for high-efficiency CIGS panels in the global solar panel market.
Like CdTe panels, many CIGS cells also use the toxic chemical cadmium. However, CIGS technologies use a lower percentage of cadmium, and therefore are a more environmentally-friendly choice as far as thin-film solutions go – even better, some models exchange the use of cadmium for zinc. [Considering these CIGS panels may be on, instead of in laptop or a rooftop, it would appear that the Zinc replacement might be the better non-toxic alternative. Is it called ZIGS? ;)]
The biggest disadvantage of CIGS panels is their price. While CIGS solar panels are an exciting technology, they remain very expensive to produce, to the point where they have a hard time competing against the more economical silicon or CdTe panels."
So I think the key points to consider while reading this all is that that many laptops are used primarily indoors, and while the lighter weight of thin film amorphous cells is definitely a bonus, the primary reason for researching alternatives to crystalline is their ability to produce more power in low light- indoor light included. A little background on charging via solar power indoors: https://www.axionpower.com/knowledge/charge-solar-without-sun/
Now, onto the SoC:
The first thing that popped out to me when I was skimming the Apollo 3 Blue SoC boards is that they can run SPI 3-wire 4-wire. Now, I don’t know if they can run the E-ink displays that use 3/4 Wire, but they have have 74 GPIO pins, but it is a technology that is already out, unlike the the Apollo 4, which is being released next year. It is likely that USB uses too much power for the board and might need to be dropped (from the sketch). Having uSD might work though.
At this point, I am curious of assembling three core components:
-
The system on a chip that uses the lowest power consumption of any linux-capable ARM M or ARM A series (likely ARM M) with at least 1MB of RAM. Current leader (correct me if i am wrong): Apollo 3 Blue Plus, with 576 microAmps at 96mhz TurboSPOT or 2.9mW. I do not know if it can run linux, but I would like to know if it can run. Some ARM M operating systems like MBED OS seem like they can run a limited GUI.
-
The E-Paper or Rdot display that uses the least average power consumption and can still display an LXDE environment, starting from 400x280 and up. From the above Waveshare displays, that appears to be something like the 640x480 5.83inch E-Paper E-Ink Display HAT SKU: 14597 with 26mW refresh power and 1-5s refresh. That uses about 26mW of power, every 5seconds. With a 2s refresh, it seems like 52mw would be barely usable (although I can’t say for sure, because I can’t change my monitor to 0.5 hz, though I would be curious to see what that is like). The question is, how much power can a 22% solar cell produce on a 10x15 lid, and whether it is enough to power a system running approximately 60mWh.
-
The solar panel capable of collecting the most power on a typical 15.6 or 17" laptop lid. It could even be designed to have solar on its entire chassis, like around the display and keyboard and the backlid, but that might appear a little too tacky. According to Arcadia, the average solar panel produces 15 watts per square foot. However, indoors, that might be zero with an all-or-nothing voltage threshold in crystalline panels. Therefore, a solution using amorphous panels might be able to collect a little more wattage, perhaps more power than is used by an ARM M series and moderately refreshable e-ink.
It’s not clear why the 800x600, 6inch E-Ink display HAT for Raspberry Pi with 600mW has such a huge jump in power usage, but it appears the refresh rate could be as fast as 0.25s and the 16 greys(as opposed to 2-4) has something to do with it
I will try to sketch a more elaborate prototype of a more ideal laptop. So far I have run the Pi Zero on 21 watts (approx 12"x19" when folded out without a battery in sunlight) I don’t even have an on/off switch for my Pi Laptop, but that is really the last thing I really need to build it.
