Smartwatch 10 Years Ultra-Long Battery Life

Some random notes.

Component selection is key to success. Not reading documentation is for quiche eaters.

Did you know you could find and SMD an accelerometer with 500nA (yes, nano-Amperes) consumption and 50 nA power-down mode? Putting things in perspective, on a 1Ah coin cell battery this equates to over 200 continuous measurement years run-time, or if you get creative you could exceed 1000 years...

Put that in your pipe and smoke it.

While waiting on Revision 2 of the PCB, I started working on Revision 3.

In parallel I'm also adding the alarms features in the SW/firmware, going is slow as I have a full-time job, obviously.

For HW Revision 3, I will include:

1. Accelerometer --- steps, body activity, what else? Taps? Since we have no buttons, and no touch screen, the watch enthusiast could use taps and shakes to control certain functions of the watch that would allow for standalone/independent watch operations from the smartphone.

2. Magnetometer --- digital compass feature.

3. Depth sensor --- since the watch has no buttons, it could withstand extreme depths. I really wonder how salient/relevant such a feature could be though, might add it as an optional? Could that watch be used as a dive watch?

4. Altimeter --- again, I'm not really sure how relevant an altimeter would be. An altimeter goes with a GPS, and we don't have a GPS (or not yet, huh), so what gives? Might add it as an optional?

5. Heart/oxygen sensor --- this is a hard no. The body sensors are notorious power hogs and would drop the battery life down to months from years. Case in point, the lowest power blood sensor is about 100uA, which under the best circumstances, if I get really creative with the power management, would drop the life to 1 year.

6. Full power-down battery disconnect --- that's a must, but how (???), since the watch has no buttons? I have a crazy idea, but it needs to be tested because this would involve some really creative work.

7. Screen LCD light with an LED --- that's a hard no. Massive power hog. Yet, again, I have a really crazy idea, but it needs to be tested because this would involve some really creative work and perhaps some legal research.

8. Alarm piezo --- about to test the validity and power consumption.

9. Vibro alarm --- as bad as it sounds from power management standpoint, it might be an option, considering the massive advantage of having a vibro inside the watch.

10. --- ??? Anything else? GPS is a hard no, that's a massive power hog.

11. Atomic clock sync up --- probably also a hard no, unless I find an easy way to do it.

Lots of things to do.

Cheers!

Making progress, will be on my desk within a few days!

This weekend I've been trying to implement alarms.

My idea is to have pre-defined alarm types the watch enthusiast can enable/disable, and also set, for example, for Time Zone 1, set a daily alarm to buzz at 7am every morning, or for Time Zone 3 set a weekly alarm to buzz every Sunday at 5pm.

I'll post my progress here, perhaps pictures. The real question is how the buzzer is going to work, I'm not clear on the concept there.

Since the power draw is so low, I wanted to check feasibility to power it with a small solar cell.

Went to the local thrift shop and bought two solar calculators on sale for 50c and gutted them for the solar cells. Then de-soldered the solar cells and made my own solar power module, and took it out for a spin.

It worked! Unlimited power! Well well... even shading one of the solar cells still kept the watch ticking. Awesome. FYI one cell delivers 3.3V @ 2.5mA, so at 100uA burst and 10uA average running the BLE watch on solar power is whole-sale feasible.

That's very encouraging, as I was considering a re-chargeable battery with a solar element, which would bump the life to infinity.

One step at a time.

Cheers!

While waiting for the R2 of the alpha version of the PCB, I've been working on a watch face to get the UX a lot more pleasant. 

My idea is, since we have so much memory, to have a number of watch faces the watch enthusiast chooses, based on use cases and work they do. You can choose the watch face from the smartphone app, as the watch itself will have no buttons.

I left one setup in a box for 10 days, with BLE advertising, this is the latest.

The 200mA battery voltage is a robust 3.21 volts. I'm very happy to report that my original estimates are holding up. I feel confident that after I optimize the hardware and the code I can reach 10uA average to achieve 2 years on a small coin battery and 10 years on a large one.

Also, the time is keeping pretty well, we are less than one second off my citizen GPS-adjusted watch, which also could be due to me fat-fingering the epoch time, so the 5ppm TCXO I selected is actually working well with the NRF52. I was really worried, because building a real-time clock is somewhat of a black art, and usually everyone goes for an external RTC, while I went full native and designed my own circuit with hand-picked components and wrote my own code around the (sorry Nordic!) crappy real-time counter, yeah, you heard me right, NRF52 doesn't have a real-time clock but a real-time COUNTER, lol. So I ended up using the counter to literally write my own time real clock to step around all that, what a doozy.

The upside is, we go back to power consumption, we don't need an external RTC! Less components, more power! Well, less power but more saved power, that is.

I mean, seriously, why do people need external RTC components, that's just lazy. You got all that processing power, use it. I'm waxing poetically here but to put its computing power into perspective, the Apollo 11 AGC (board computer) barely could perform around 40,000 instructions per second, and had 32,768 bits of RAM and 72KB ROM, and landed people on the Moon, whole 55 years ago. In comparison, we have 64MHz MCU with 512k ROM and 64kb RAM and only have to keep time. Note to self, don't be a quiche eater, and get to work!

Prepared a second revision of the PCB based on lessons learned in the first round.

Should be done by end of this week.

Things are moving fast.

In parallel, I'm trying to find out why the antenna on the BGA design didn't work.

Contacted the Nordic dev help, and had a few extremely nice and knowledgeable engineers give me very constructive feedback.

In short, Nordic reached out to me and offered to optimize my antenna design in-house in their lab in Oslo. The only thing I need to do is mail to Norway the designs with some extra work to allow them to access the MCU. I was extremely impressed by the level of customer service. Such attention to details is rare these days. I'll take them up the offer once I finalize my case design.

Hats of to the Nordic engineers, thank you, and looking forward to learning from you.

Fully assembled, and ready to move on to the next stage of productizing the alpha design.

Complete sealed stainless steel body, mineral crystal, silicone strap.

Expected life of this watch >10 years as is with BLE.

Is it too early to celebrate?

One full steel case and two polymer.