Smartwatch 10 Years Ultra-Long Battery Life

Details

This project is an effort to design a smartwatch wearable with 10 year battery life on a single charge.

The secondary goals are to learn/advance knowledge low-level design of embedded full stack circuitry and devices for wearables, ultra-low power, maintenance-free, battery powered, wireless, app-driven, and off-the-grid.

Developing a smart watch from scratch would be a great learning experience to advance my knowledge of low-power, self-contained powered devices. Hence, we learn about watches, smart watches and wearables in general by designing one.

This is not a commercial effort. I doubt that we are making anything new per se, however, we are trying to advance the art a lot further than others into the low-power domain.

My background should allow me to filter out the noise and focus on the salient points. My points of view below.

Features:

Ultra-long life. Years, if not decades. The last watch you'll ever need.
Maintenance free. Never needed to open and fix anything inside.
Avoid buttons. Buttons open up the watch to maintenance. Dirt. No moving parts. No bolts, no screws, no mechanical parts. Sealed shut. Control via smartphone? Bluetooth?
Tough. Bang it on a rock or use it to crush rocks. Metal, steel? Titanium? Aircraft aluminum?
Under water. Deep ocean depth grade watch.
Solar / RF powered may be in the future? No battery change.
Multiple time zones. Automatic time adjust / set up.
Other features? Compass, altitude, GPS? Atomic clock sync up? Accelerometer? NFC, BLE? What are other watches using?
Sapphire crystal. Seems other watches are big on sapphire. Diamond? Too brittle? Mineral crystal? Acrylic?
Anything else?

Work to do... Please check the logs for daily progress!

https://hackaday.io/project/193509-smartwatch-10-years-ultra-long-battery-life/log/225387-smartwatch-automata-third-assembly

Project Logs

Collapse

Smartwatch PCB Alpha R5
VALENTINE • a day ago • 0 comments

Lunch time, and I redesigned the PCB, as well as added some extra stuff.

Skipping R4 for manufacturing, will go straight to R5.

Fun fact: The original iPhone PCB went through about 10,000 revisions, imagine that… Picture attached at the end.

I am estimating my hardware to go about ten revisions, until we get to Beta R1. Does that mean I'm three orders of magnitude better than them? Stay tuned!

Check out the first iPhone alpha mockup board, very educational.

Cheers!
Inflection Point :: Polyfurcation
VALENTINE • 2 days ago • 0 comments

Captain's log. Smartwatch date 2023.92.

Oh, wait. That's from another project, he he.

This project is entering an inflection point where the complexity increases exponentially and the tasks polyfurcate which necessitates strict planning and task management to avoid total feature collapse and morale sublimation.

In simple words, this is the moment where most self-driven DIY projects die.

Unless the projector is aware of the projected phænomeni, able to jump over the forks, keep track and overcome the myriad of small decisions and inter-connected project events which in a normal project are handled by a large team of project manager, focus group, feature designer, hardware expert, radio engineer, software designer and industrial design and manufacturing liaison, as well as supply chain coordinator, sometimes more than one... well.

Unless managed, this will turn into a quagmire of endless mistakes and re-iterations, and we must be cognizant of the lattice decision making and ladder-iterative waterfall design process; or else.

In plain words, we are a one-man orchestra carefully playing all instruments at once. Better not trip!

Cheers!
LCD Display Power Circuit
VALENTINE • 4 days ago • 0 comments

Project log.

The LCD will not display unless the voltage is around 2.69, let's say 2.70, therefore, we need a high efficiency separate booster for the display to keep it on. As a side benefit, having a regulated booster is really beneficial for the other auxiliary devices that require 3+ volts to operate.

TPS610985 or TPS610981 seem like good candidates.

Therefore, we feed the NRF52 directly with the power thus bypassing the booster, and use the booster only for aux devices. That would improve the efficiency. Note to self, need to add perhaps an extra circuit for the harvester, tbd.

Also, some decent low power level shifters to go with this

NTS0104GU12,115 / C2677363

NLSX4378ABFCT1G / C604195 (0.5uA best choice)
LSF0204DYZPR / C2677496
Energy Harvester
VALENTINE • 4 days ago • 0 comments

Some quick notes on harvesters:

BQ25504RG
LTC3105
AEM10941
ADP5091 / ADP5092

https://www.digikey.com/reference-designs/en/energy-harvesting

TPS61098x (not really a harvester but an ultra-low quiescent / high efficiency booster for the display, could double-duty as a solar cell booster)

More to come later.
Battery Consumption Experiment Real World
VALENTINE • 5 days ago • 0 comments

The fully assembled watch you see is actually using a rechargeable Panasonic ML2020 battery, which is about 40mAh.

As of today, the voltage is at exactly 2.70V, which corresponds to 10% of the capacity based on the published material spec from Panasonic.

Based on the experimentally estimated 1mAh/day draw of the un-optimized hardware and code, we are currently looking at about 1000 days on a 1Ah battery, or about a little less than 3 years, give or take. The watch is not optimized for the rechargeable ML2020 battery, which performs best under a different use case, therefore I estimate a little over 1000 hours even on an un-optimized design.

Regardless, we are currently at 3 years on a charge, so that's a massive win for my original postulate that a 10-year battery watch life is wholesale possible.

On a side note I also proved that the watch would run on ML2020 which is extremely important use case one we add a solar cell, where we will need to use a rechargeable battery.

To put this in perspective, the longest life on a single battery charge smartwatch is about a month, so we are way ahead. Of course we are also talking a pretty thin functionality so that comparison is probably (most definitely) unfair, however, it points out that with some effort we can hit the 10 year mark on a certain set of smartwatch functionality that strikes a good balance between battery life and creature features.

Cheers!
Day 4 :: Hike
VALENTINE • 6 days ago • 0 comments

It was really cold and windy, and at 4000 meters things look a little strange. Hope my pics were good.
Day 3 / Hike and Bike
VALENTINE • 7 days ago • 0 comments

Took the assembled watch with me on a week-long excursion. Mountain hiking, biking, driving, and some urban exploration, elevation 4km. Will post pics later. So far the smartwatch is performing... well, like a smartwatch. Which I guess is a good thing.
Flex ribbon programmer/debugger connection OK
VALENTINE • 11/22/2023 at 17:13 • 0 comments

Good news, the ribbon programmer connector I designed works fine, not sure where the problem was with the first one, all connections looked OK, I traced all, soldering OK, no heat shrink breaks. My educated guess would be, the actual connector on that is defective, where the female pressure leads do not make correct contact with the male leads. That's a new one, I guess at such scale and my manual work with the small parts introduces a mechanical error. That's actually very good to know as the design is sound, but I might have to choose a less-error prone method for programing, pogo pins comes to mind. To do list later.
Smartwatch: Automata :: Third Assembly
VALENTINE • 11/21/2023 at 21:43 • 0 comments

If your prototype is not held together by Kapton tape, sheer will and determination, you are holding it wrong.
I'm from the school of better get a working hack today than a perfect future design never.
After a whole hour of soldering and name calling it's here! I'm gonna wear this over thanksgiving and see if its a pterodactyl or a turkey.
Stay tuned for the next release.
Cheers!
Smartwatch: Automata :: Second Assembly
VALENTINE • 11/21/2023 at 01:36 • 2 comments

The PCB works! Yeah! Revision 2 is a go!!!
I can program it and see the BLE on the app, the watch displays correctly, so the antenna and programmer connector and display connector are all correct. The PCB is dabomb!
Minor setback, the flex-cable for the external programmer doesn't connect for some reason, I guess there is some subtle manufacturing conflict of a flex cable, 80 micron trace and the wave solder process. I will look into it tomorrow, as I'm extremely tired, doing this after regular work hours is a recipe for disaster. The PCBs and components are so minute and fragile it's trivial to destroy a $200 PCB and a month of work with the slightest wrong move with my fat fingers.
I'll see if I could resolve the programmer connector tomorrow, there must be a simple reason, nothing a microscope and an X-Ray can't solve.
We are one step closer.
Cheers!

View all 44 project logs

Discussions

Ale o co chodzi wrote a day ago

Is possible create a tamagotchi version?

or old fasion clock pocket? for example with solar panel? https://hackaday.com/2022/03/30/solar-harvesting-is-better-with-big-capacitors/ or ..... peltier power https://hackaday.com/2018/07/05/a-flashlight-powered-by-your-hot-little-hands/

What You think about PixelQ screen?

(I dream about small fuzix or linux device with week working time, no sound, bigest as oqo, pocket vaio P or ELLO2 https://www.crowdsupply.com/yellow-beak-computer/ello-2m , irda as ethernet slow cpu , keyboard

but today clock is very good starting point)

Are you sure? yes | no

VALENTINE wrote a day ago

Tamagochi: Yes.

Solar: Yes.

PixelQ screen: Never heard of it. Part number/manufacturer?

Fuzix: Yes.

Please understand those are hypotheticals. I will not be pursuing them.

Cheers!

Are you sure? yes | no

Ken Yap wrote a day ago

😉

Are you sure? yes | no

powiadam.ci wrote 13 hours ago

https://en.wikipedia.org/wiki/Pixel_Qi

clock or tamagotchi with irda and my own code is great idea!

but in my opinion better is OTP system

this is simple generate pseudo random number using time and salt

https://en.wikipedia.org/wiki/One-time_password

Are you sure? yes | no

VALENTINE wrote 9 hours ago

Ha, you are talking about the company PixelChi (or was it PixelQi?), oh that brings memories, yeah their office was just a little down from my office across SFO. They don't exist anymore, last thing I heard was 10 years ago they disappeared. They did those ubuntu laptops with rabbit ears, I had one, that was a piece of art. Great idea though. You can find so many cheap transflective displays nowadays, I'm not sure why you need to go back in time to those. Technology has moved forward so far, there is a reason those old displays disappeared. I'm sure there is some niche market but I just don't see people paying hundreds of dollars for a low-quality/low resolution transflective LCD for a laptop.

Didn't a Taiwanese company acquire the technology? That was so long ago I vaguely remember something like that.

https://www.hannspree.eu/uploadfiles/202304/20230427163724-8qlvc230td.jpg

https://www.hannspree.eu/product/21.5PaperDisplayTransflective/

Are you sure? yes | no

Ale o co chodzi wrote 7 hours ago

thanks @powiadam.ci

Yes I talk about this screen.

I have nothing against the new displays. If they consume less power have better contrast in the sun then why not. ;D

It is important that the system allows you to write your own scripts. I see huge popularity of the expensive device https://flipperzero.one/ precisely because you can write your own script.

Are you sure? yes | no

VALENTINE wrote 6 hours ago

No scripts. This is a fully custom ROM/firmware geared singularly towards low power consumption, with the PCB designed to match the code and vice versa. There is no OS, or scripting, or anything like that. Think low level C code with some assembly here and there. Think of having a CASIO watch, not something running RTOS or whatever, but with all the smart features of a modern smartwatch.

Are you sure? yes | no

nephrita wrote 3 days ago

Beautiful work!

Are you sure? yes | no

VALENTINE wrote 3 days ago

Thank you!

Are you sure? yes | no

Patrick Lepoutre wrote 11/23/2023 at 12:59

Nice work. As a pebble watch nostalgic, I really loved the time-line view that showed next events for the day. Could this be implemented? Even with no button?

Are you sure? yes | no

VALENTINE wrote 11/23/2023 at 15:10

Thank you. I keep hearing about this pebble thing but havent been able to find (havent looked hard either) much information, I guess that was before smartwatches got popular. What are the features? You got a captive audience here, care to elaborate please?

You could implement any controls you want, lack of buttons doesn't mean lack of controls. Next iteration has an IMU (accelerometer/gyro) and a magnetometer, so any 5-d motion could be translated/assigned to action. By 5-d I mean (+/-) X-Y-Z + North-South + Time.

Think double-tap on the right, single-tap on the left, taps on the edges, orientation, etc.

Cheers!

Are you sure? yes | no

Patrick Lepoutre wrote 11/23/2023 at 20:05

I found some screen shots and designs that could help understand. Pebble worked in color and monochrome. Here is the link http://alexanderkirov.com/works/pebble-os

Are you sure? yes | no

VALENTINE wrote 7 days ago

Thank you, checking. There is no guarantee certain action and features are possible due to this watch having from-scratch custom code, there is no rt-os or anything like that. It's coded from scratch with power savings in mind. The hardware is also designed from scratch around power saving, anything consuming more than a few uA is a cut.

Are you sure? yes | no

Yaroslav wrote 11/22/2023 at 19:18

Can you share mfg code of the screen or a datasheet? Thanks!

Are you sure? yes | no

VALENTINE wrote 11/22/2023 at 19:42

LS013B7DH05

https://d7rh5s3nxmpy4.cloudfront.net/CMP7377/files/LS013B7DH05_15Jun15_Spec_LD-27503A.pdf

https://www.digikey.com/en/products/detail/sharp-microelectronics/LS013B7DH05/5799456

Are you sure? yes | no

Yaroslav wrote 11/22/2023 at 20:38

Thanks! Any reason to not choose an epaper? (ghosting, UV degradation, etc..)

Are you sure? yes | no

VALENTINE wrote 11/22/2023 at 20:44

epaper has high power consumption, wrong use case for a watch.

Are you sure? yes | no

fdufnews wrote 11/14/2023 at 08:45

I just discovered this project and it is in my opinion a super exciting one. Battery life is precisely what keeps me from using “smart” watches.

If it works on a mini solar panel, it is even more awesome.

Are you sure? yes | no

VALENTINE wrote 11/14/2023 at 18:11

The mini solar would be a feature to be added once we confirm and test the first design iteration.

Are you sure? yes | no

Ken Yap wrote 11/13/2023 at 01:12

I'm surprised that a certain forum participant hasn't asked you yet if you can run Fuzix or mesh networking on it, since you satisfy the long battery life condition. 🤣

Nonetheless I applaud your goal. If you achieve 10 years, then maybe the battery doesn't have to be chargeable and can be return to depot to change the battery.

Are you sure? yes | no

VALENTINE wrote 11/13/2023 at 19:24

Ken:

I'm not sure about anyone asking, I'm new to this platform.

Yes, the design could run Fuzix and/or mesh networking. In fact you could run full-scale Doom, with a TFT screen and sound. However, this would immediately drop the battery life from 4000 days to about that of a standard off-the-shelf smartwatch of 3 to 5 days, give or take (I actually did the experiment and the battery died really quickly). Which is the principal problem I'm solving, a smartwatch with 10 years life. Else you could find plenty of other designs, I'm not keen on reinventing the wheel. To paraphrase a well known quote, I want to do something no one else had done before.

The downside is that the firmware is 100% custom from scratch and the hardware I've also designed from scratch with discrete components to solve the battery life. I've already proved that the watch can live 10 years on a single battery, now I'm only slowly adding all the features. The fact that the watch successfully runs on a single calculator photocell, with Bluetooth and graphics kinda proves my point.

In any respect thank you for looking at the project, and commenting, it's interesting. I appreciate any further input you got, as I said, that's why we are here.

Cheers,

Valentine

Are you sure? yes | no

Ken Yap wrote 11/13/2023 at 20:50

It was just a little dig at that particular member who has Fuzix on the brain. Hang around a bit and you will see their comments on other µC projects.

Are you sure? yes | no

VALENTINE wrote 11/13/2023 at 22:59

I see. Reminds me a little of the Beowulf cluster comments from the late 90s on Slashdot.

Are you sure? yes | no

trialexhill wrote 11/14/2023 at 03:59

Ha!

Are you sure? yes | no

Smartwatch 10 Years Ultra-Long Battery Life

Description

Details

Project Logs

Collapse

Smartwatch PCB Alpha R5

Inflection Point :: Polyfurcation

LCD Display Power Circuit

Energy Harvester

Battery Consumption Experiment Real World

Day 4 :: Hike

Day 3 / Hike and Bike

Flex ribbon programmer/debugger connection OK

Smartwatch: Automata :: Third Assembly

Smartwatch: Automata :: Second Assembly

Discussions

Similar Projects

ESP8266 Plant Watering

High autonomy GPS Tracker

Dumbwatch V2

Franken UPS

Smartwatch 10 Years Ultra-Long Battery Life

Become a Hackaday.io member

Just one more thing

Description

Details

Project Logs Collapse

Enjoy this project?

Discussions

Become a Hackaday.io Member

Similar Projects

Does this project spark your interest?

Report project as inappropriate

Send message

Remove Member

Project Logs

Collapse