Welcome @EI2030!
We’re excited to have you here, and very interested to see how this project develops - you’ve set yourselves a bold mission for a totally e-ink world by 2030… but bold missions are the best kind of missions! We’ll offer any help that we can with your pi-top experiments, and @duwudi, myself and the rest of the team are looking forward to meeting you at the next pi-top Session! It’ll be on the 26th March, details to follow…
All the best,
Wil
Welcome @EI2030!
Looking forward to hearing more about the PaperTop at pi-top Sessions, I love how things have gone full circle with pi-top 1 originally being a solar-powered laptop, and now pi-top 3 is being used to create one 
I know that at least one of our community members is interested in solar powering MCU devices (@Supernovali) and @wil has been obsessed with e-ink displays for the longest time - I think they’ll be some interesting discussions and future projects spawned from everyone meeting! 
For me, I’d love to combine robotics + e-ink + solar power into a project using a pi-top [4] if anyone has any ideas
If you need solar, I’m working on a project shortly that will do high power solar devices, such as the pi-top and peripherals such as the pi-top here in the next week. It will support USB power delivery as well as regulated output to other accessory devices requiring 5v and 3.3v. my planned power output goal is 60w maximum.
I just finished a medium power MPPT controller prototype and am waiting for my company to be finalized with business licensing before I start selling the production version.
Here is an image of what has been in the works for the last month and a half 
Hi @CAProjects! I’m Giovanni, one of the core team members of the EI2030 working group, the Low Power E-Paper OS, and e-ink is just one of the technologies we are promoting. The other main one is reflective LCD (RLCD).
Regarding lack of backlight, there are front-lit displays/components, such as Azumotech. There are also some interesting new displays that use electrowetting (EWD) instead of reflective & e-ink.
@wil Thank you and we look forward to developing this bold mission where no laptop has gone before! 
and look forward to meeting you and showing our Pi-Top mods!
Thank you again, duwudi, for the amazing history on the Pi-Top and I look forward to helping make solar power a part of it again!
Supernovali, I’ve purchased about 5 solar power managers, and the largest one that handles high power is my 6-24V Waveshare Solar Power Manager. I’ve used it briefly for testing purposes, although I have been focused on indoor light harvesters/managers such as the E-peas AEM10941 since the intended long term goal of one of my projects is to have an ambient-light powered OS & e-paper/reflective display like a solar calculator. In my link above, I’ve recently been able to power a microcontroller on a lightbulb and in the past, powered my RPi0 with a 20W & 30W portable solar pack. Your solar pack looks great though- 60W is more than I imagined for my Rpi3B+, but I can see the Rpi4 benefitting from it in cloudy conditions.
Thanks everyone for the responses, and feel free to check out our Discord server & March 7th presentation slides to find out more about our work!
This is only the beginning of a set of product lineups. Although, if your looking for that, a high efficiency charger and all passive components selected specifically for this application can increase your output and reduce costs. If you ever plan on making it into a product, I’d be happy to get you set up with a controller for it!
Thanks @Supernovali!
For a microcontroller+E-Paper, I am looking for something like: https://www.tindie.com/products/jaspersikken/solar-harvesting-into-li-ion-battery/ since I am looking to recharge a small lipo battery- 100-500mAh- enough for a few hours a day. A battery of this size would also be great for a Rpi Zero+E-Paper: https://www.adafruit.com/product/1781
Edit: I am posting as my own account, as the previous posts were written on behalf of EI2030. The laptop I am working on would use a smaller monitor in an initial prototype. I posted about it earlier today: Solar PiTop for E-Paper Display Development
Oof! 24$? That’s a bit pricy. Yes, this would be easy to make for about 8-10 after a final product is ready. The board is small and the number of components is limited. Do you need the 1v8 rail? I could reduce the cost even further if you only needed 3v3
Also, what kind of average daily watt-hours do you expect to use? Being able to optimize is for a specific amount of draw, both continuously and for peak current ratings will be important to know
@Supernovali I have found two BQ25570 power managers that may work well indoors well enough - it has a higher cold start from what I’ve read but:
https://www.arrow.com/en/products/mikroe-2814/mikroelektronika
https://www.ebay.com/itm/184366178954 are two i’ve found a little cheaper.
Unfortunately, it’s a little early for me to start my purchase for a larger order since I’m still working on the OS and I have a https://www.powerfilmsolar.com/products/solar-development-kit that I am still evaluating.
As for power draw, using an Ambiq Apollo 4 I think the range would be in the 1-2mah for the cpu , 5-15mah for RAM, and 0.6W for the display: https://www.waveshare.com/9.7inch-e-Paper-HAT.htm (would need significant driver development). Though if the refresh is partial, and and refreshes are less frequent (e.g. 1-2seconds), it may lower overall consumption.
Also, the power consumption on these smaller displays seems possibly a lower average: https://www.j-display.com/english/product/reflective.html
You’ll be able to reduce the current on the processor to almost nothing when it goes into standby, so you’re plenty good there. Are you using external RAM? May. I suggest that you use non-volatile memory so that you can reduce the power consumption to a minimum and just use the onboard 64k the processor already has? Then you can have all of your assets cached in flash and free up ram for instantaneous functions. At 96MHz, your speed will only be affected slightly, if at all even.
You will still need a 5v rail for the display. My concern is it claims a standby of .3 watts, which is 300mA. If you needed the ability to shut the 5v rail down so the epaper driver consumes no energy during standby, that is easypeasy. I’m still confident I can build you a system that would be capable of producing an 2-5 watts cheaper than those listings if you are interested. It would also allow me to add an additional product to my company specific to the niche usage case you have. And do it for less than 8 dollars USD for a production version.
My solar charger for medium power use cases currently costs 5 dollars USD for a single board and my expected sale price is approximately 15 USD. A board with the necessary circuitry to run the things you need could be built for less than 3 dollars a board. 
384KB SRAM onboard for the Apollo3, 768K for the Apollo3+, and 1.8MB SRAM (+2MB MRAM) for the Apollo4.
The Apollo3 supports external RAM, with XIP up to 96MB:
but the Apollo4 datasheet isn’t out yet. Most of my links documenting the PCB components (ideas, right now) https://github.com/EI2030/Low-power-E-Paper-OS
Ideally it will be in standby when it’s not being used, but I imagine it being like a 200mhz pentium II, it might have some updates to run, so I’m assuming it will need to run at 100% most of the time it’s on.
I have looked into non-volatile and in fact it was my first choice after seeing the freethegameboy.info project. http://www.farnell.com/datasheets/2843378.pdf is 1MB non-volatile. I’ve also seen ultra low power from Adesto (Dialog Semiconductor- ReRam or CBRAM)
I actually just looked at the pdf you sent over again. I had misread it. It states it has 1.8MB of drama. How did you find the power consumption of the ram though? I didn’t see that in the comparison chart.
Yeah, non-volatile is the way to go to free up system resources and hold your visual assets and documents. If you needed more ram, you could use an external module as a swap to transfer faster to the driver and then disable once the swap is complete and reduce power consumption.
But yeah, for your power needs, if that’s something your interested in, I’m willing to start diving into it. If you gave me a month and a half to develop and test at the power draws you need, is there any additional function you need? With most regulators disabled, you get draws that are also in the μA range, the collector setup would be enough to keep the regulators in standby at a max of 100μ draw for 4 regulators, 5v, 3v3, 2v2, and 1v8. If you don’t need one or a couple of these rails, then it would be reduced even further.
Thank you for providing a review of my components and giving this an honest look. I don’t want to overpromise anything, but it is a lot earlier than I expect it to be ready, so I apologize if I appear to be leading you to a development - I can’t say for sure when I will need it, but I am happy to research the information for you should you request it.
Also, few or none of the parts are set in stone- if I there is a monitor that uses 3.3V instead of 5V, I may opt for that for the sake of simplicity. This is more a brainstorm as a means to an end rather than an end in itself.
I base the consumption of the onboard RAM on actual Sparkfun Artemis boards, although this includes additional power like LEDs: https://www.sparkfun.com/products/15442
“Features Tab ( * 6uA/MHz (operates less than 5mW at full operation)”
The actual power consumption of the chip alone is 572uA (x3.3V) is 1.9mW. The rest must be RAM and LEDs I assume. My Artemis Nano uses less than the ATP board, since it has fewer peripherals, but in the forums it is said to use around 5-6mW.
I got the external RAM consumption figures from a couple sources:
https://www.globenewswire.com/news-release/2021/03/15/2192576/0/en/Ambiq-and-AP-Memory-Partner-to-Enable-Richer-Experiences-in-Intelligent-Endpoints.html
" AP Memory solutions offer low signal pin count (6 for QSPI, 11 for OPI), low power (standby from 20uA to 80uA, active from 3mA to 8mA), and high transfer rate options needed to meet the demanding power and space constraints of wearable and other battery-powered smart consumer devices. "
For the external memory:
based on 16MB https://www.apmemory.com/wp-content/uploads/APM_PSRAM_OPI_Xccela-APS12808L-3OBMx-v1.1b-PKG.pdf Page 22 of 26
For the above 1MB FRAM: ❐ 2.6 mA (typ) active current at 40 MHz
❐ 3.5 µA (typ) standby current
So the dilemma here is how much RAM will I need? I am just at the proof-of-concept stage. Nothing productivity in mind. If I go with one 16MB flash module, I wonder if it is enough to run the linux kernel and potentially a single Text editor app. Ideally i would add more memory. I’ve been able to get TinyCoreLinux using less than 28MB of 1GB RAM active:
If I could get that to run, I’d remove any apps I don’t need, and install a pdf viewer, and it would be like an e-reader, except solar powered.
If I added 32MB of external RAM:
https://www.apmemory.com/wp-content/uploads/APM_PSRAM_OPI_Xccela-APS25608N-OBRx-v1.0-PKG.pdf
Could I run it at 13mhz or 133mhz all the time if it is fast enough? That would be less than 20mA for 32MB, seemingly more efficient than the 14mah 16MB.
Another OS shows me using less than 30MB active.
I could try an older linux kernel (not designed for networking) and run it on 4MB of RAM:
https://en.wikipedia.org/wiki/List_of_Linux_distributions_that_run_from_RAM
There is a Raspberry Pi Port:
http://tinycorelinux.net/ports.html
So I wonder whether porting an ARM Cortex-A port to a Cortex M4F will be relatively easy compared to a direct x86 port to ARM. There is also the boot utilities and drivers needed, of which there isn’t but embedded linux companies such as https://bootlin.com/ could do it.
Oh, no, I don’t feel mislead to help you, it more of I’m currently developing a few solar powered devices and I have an idea of how it would be mutually beneficial.
And woah! That’s insanely small kernel! Makes sense though that they would have a kernel for such a purpose. What I would do is when you have it prototyped, run it on a bench power supply to measure the current draw and then plan on having an overhead of 20%. I’m not sure how long you expect the battery to last in this use case or how I would go about planning for capacity and battery size, but for my solar panel powered projects where solar cant provide 100% of the power in direct sunlight, I measure average consumption over 24 hours in a typical use case scenario. I plan on the battery lasting at least 48 hours (I prefer 72 hours) and a solar cell that can provide 3 times the continuous average power consumed in direct sunlight.
For example, if I have a project that will use an average of 1W, then I plan on a solar panel that will provide 3W. And the battery should be a minimum of 48Wh. I prefer that it would be 72Wh. This is at least the math if you plan on never having power interruptions. This takes into account cloudy days or above average consumption on occasion.
I might still develop a cheaper board, just because havent considered lower power projects that don’t consume a burst of energy on occasion.
48 hours for battery and 20% overhead sounds great. I think it is good to have have double the daily power consumption in the battery, since access to enough sunlight may not be available, but also the design of something like this is more intended as a household electronic rather than something that is typically going to be used outdoors, as idealistic I am about it- amorphous panels designed for low light could be chosen, and may even be too exposed to outdoor light to be used: https://www.nrel.gov/docs/fy12osti/51664.pdf
Also, heat and season affect efficiency: https://www.powerfilmsolar.com/education/the-horizon-blog/2019/07/19/how-does-temperature-affect-solar-panels "In the winter the amorphous panel is technically more efficient but unable to heal the effects of Staebler-Wronski as quickly. "
This panel is very expensive, but I am curious: https://www.digikey.com/en/products/detail/anysolar-ltd/SM262K10L/9990466 25% efficiency 4"x9" wired in parallel could fit on a laptop lid, and provide at least 1W of power each. Add 4 of these and is that enough to power the monitor? Then again, it’s over $320 of solar panels
Just to chip in here, one approach to providing computing access with solar power would be to retro-fit a shipping container with a bunch of low power computers (Raspberry Pi’s would be nice because of the brand and community behind it) and have a solar array built into the roof that powers the whole container. Having one solar power infrastructure powering multiple computers is much more economical than having a solar system per computer - at least that was the outcome of my project 6 years ago
These could be shipped to remote areas in developing countries and I’m sure there could be some novel things done to provide internet access (starlink, local server with periodic updates from mobile phones, mesh network between containers etc)
I like that idea! A set of 300 watt solar panels can supply enough power for 10 pi-top’s! We’d want to avoid converting anything to AC (granted starlink and the server was also powered off of DC powered adaptors), each unit could be a self contained unit! We could even use an RPi cluster to run the server, each unit in the cluster having its own hard disk and a contributed to the network in it’s work-group. Meshed together with the other work groups as an intranet.
Now there is an idea!
That would be efficient! I agree WAN “stations” would be needed in one way or another to provide internet access, and developing phones/mobile devices like a “thin client” could work to connect to a remote terminal (shipping containers) where the Rpis do the heavy lifting. In fact, @xorlof at EI2030 had that idea for a mobile device (not the shipping container though!).
https://locha.io/ uses a long-range radio, “ISM band (915 MHz in America, 868 MHz EU) for a mesh communication device” which is like a free SMS & walkie-talkie without a cell subscription.
For short nodes, it uses:
“by having multiple nodes collaborating with each other on the network, each node implements a routing protocol which makes all nodes connected to the network available through multiple hops. We use IEEE 802.15.4g radios, and a 6LoWPAN/IPv6 stack.”
Using even lower (and solar) power, I would be interested in developing that on the Sparkfun LoRa Thing Plus https://www.sparkfun.com/products/17506 https://www.youtube.com/watch?v=fTf7vAOQG7Y Though I could store the LoRa connected MCU/MPU in the Pi-top and use it like a laptop with a 2G modem/4G antenna
Edit: It would be interesting if there is a Rpi Zero cluster useful for server/network routing (although networking would need to be done via an adapter): https://thepihut.com/products/cluster-hat-v2-0
https://hackaday.com/2016/01/25/raspberry-pi-zero-cluster-packs-a-punch/ 16 Pi 0s…
@duwudi @wil @RezIN @Oli and the @pi-topTEAM
Thank you so much for hosting the pi-top session today! Alex and I are thrilled to be able to share our progress on our Working Groups and we are thankful for your support. We hope to complete the Papertop by mid-April and hope to show our e-ink pi-top.
It was also great to hear about the projects from @CAProjects & @Supernovali and the pi-top community. We look forward to supporting more e-ink and solar projects with you!
EI2030 Core Team