Recently I bought two new clocks for my home with the intent to replace the analog ones with digital variants.
Mainly because the batteries were drained quite fast on the analog design plus I got annoyed by the constant clicking sounds.
When looking for cheap ones to buy, I struggled with many aspects:
First off, I wanted the function to be more or less limited to time only. I don’t need a weather forecast or any of that fancy stuff.
But apparently simple designs with minimal displays are not that popular these days.
Another issue was that many digital designs worked with LEDs, causing a high energy consumption. So my preferred clock would use something like an LCD display only.
Of course price also played a role, as I don’t want to pay too much for something as simple as a clock.
As I searched for the perfect wall clock however, I thought about maybe developing my own clock specifically for my demand.
Immediately a lot of interesting ideas came to mind.
I first began to think about the energy efficiency: Not only do analog clocks need a power hungry stepping motor but also everytime you hear that clicking sound, this represents wasted energy of some sort, as it is the electronic energy converted to acoustic sound without any real usage. Things you don’t worry about, when using their digital counterparts. The only upside of this analog design lies in its low complexity, only consisting of one or two stepping motors, and a quartz and not much more.
Nonetheless, not a design I encourage.
Next up I thought about building a binary format clock with LED’s in which one row displays the hours and the next row displays the minutes in a binary format.
Something like this:
O O O O HOURS
1 2 4 8
O O O O O O MINUTES
1 2 4 8 16 32
I could accomplish a prototype for such a design relatively quickly, with an Arduino™ or similar using low power LED’s.
However, this design is still rather energy expensive.
Maybe very cool as a general project, but not really efficient.
Even though low power LEDs came quite some way, physics turned out to be a hard wall to hit.
Current low power LED’s run at roughly 2V and 2mA, which makes up for 4mW (2V*0.002A = 0.004 W(att) -> 4mW) per LED.
This is simply too much for a battery driven device. In our design there are normally more than one LED running (depending on the time to display), so you end up at an even higher power consumption.
Given that a normal battery has a capacity of 2000mAh, we end up with a runtime of a few hundred hours. In all fairness however I have to mention two strategies to bring battery consumption down by a bit in this design:
Multiplexing the LED’s by using a Multiplexer and so turn only one LED on for every millisecond (although this can also be achieved easily by software when using an Arduino™ ) and then switch to the next one or even turning them all off for longer periods of time, to reduce the time, where the LED is lit (which also gives a dimming effect on the other hand).
This alone could save us some energy but combined with afterglowing materials there is maybe room for some improvement.
But this project is also easily realisable without coding, by using a NE555 or similar, giving signals every minute. This signal goes into a CD4040, which is a binary counter, which is exactly what drives the LEDs for the minutes.
Everytime the minutes reach the configuration of 60 (= 32 + 16 + 8 + 4), the CD4040 needs a reset. This can be achieved by using an AND-Logic gate. Furthermore you need to send this pulse by the AND-Logic Gate to the CD4040 responsible for handling the hours, which then counts one hour higher.
Of course this CD4040 needs to be reset after reaching 13 (= 8 + 4 + 1).
Also add a possibility to manually trigger the CD4040, to set your time.
The display could be powered by the CD4040 directly, as they only require a resistor before them and the output pins of the CD4040 should be able to provide enough current for some low current LEDs.
My next thought experiment went into using actual Displays. As I only need two colours, most OLED’s are simply overkill and regular LCD’s are better suited for this job.
To my surprise monochrome LCDs are far more efficient than I thought. Old Nokia™ displays and similar only use currents in the range of uA, while taking 3.3 V (e.g. 250uA*3.3 V = 0.000825W -> 825uW).
Unfortunately, there is not much room for improvement with this solution, since they have to run all the time and can not be turned off in between the minutes, when there is no need to display any change.
So here I am bound to the model's power consumption, even though it’s better than the LED solution.
Taking things to the next level, I came across e-papers, the same sort of display used in kindles™ and various electronic (price) tags.
The strength of those epapers is their bistable nature, as they only require energy in the moment, when the display needs a refresh. Once the content of the display is set, it doesn’t require any power to maintain its content. This is the main reason they are favoured for solutions, where hardly any page refreshing is needed and power consumption is a concern.
When using an e paper we therefore have to limit the amounts of refreshes in order to cut down on energy consumption.
For a normal wall clock, the favoured option is, to get rid of displaying the seconds.
Taking this a step further, we could also just update the display every 2 minutes or even only display the hours and nothing else, which dramatically reduces the amount of refreshes.
Some e papers also allow for partial refreshing, only updating the affected area, leaving the hours unchanged for example, when only the minutes part changes. This holds potential for even more power savings.
Assuming our e paper has the following stats:
Power needed for page refresh: 26.4 mW
Refresh time: 1440 * 1s = 1440s (per day)
We end up with: 1440s * 26.4 mW = 38016mWs / 38.016 Ws per day
A normal AA battery is about 3 Wh
So: (3 Wh*3600s)/38.016Ws = 284 days of runtime, with just one battery. Not too bad for just one battery!
In general bistable solutions seem to be a great option for building a clock of this kind.
But we can even go further: Sharp™ developed “Memory-LCD”’s, which allow for a drastically reduced power consumption, as they also make a difference in energy consumption, when not changing the content of the display. Plus the energy consumption in the “active” and the “passive” mode are extremely low.
This design: https://www.sharpsma.com/products?sharpCategory=Memory%20LCD&p_p_parallel=0&sharpProductRecordId=1504542
only takes about 100uW to run in its “static” state and 600uW in its “Dynamic” state.
As we only need to update every minute, the combined power consumption would be in the range of 108uW.
This makes up for:
P:108 uW
Battery power: 3 Wh
Runtime: 3 Wh/0.000108 W = 27777.78 h -> 1157.40 days -> 3.17 years.
Now, this is quite a number, considering we are only using one battery so far!
With those two designs however, we are also able to use low energy harvesters such as long lived efficient solar panels, which could provide our clock with power viá charging supercapacitors.
This could allow for wall clocks without batteries in general, reducing the maintenance costs and general energy consumption by a large margin.
For this discussion I left out the microcontroller part, because first, they only need very very small amounts of power if handled correctly and second is it a topic on its own.
Oh, and I forgot to mention what wall clock I bought. I have chosen to just buy two Pearl™ clocks (https://amzn.to/3jDdeAs *) for about 12 bucks each as they also last a very long time and saved me a lot of time by not having to develop my own ones.
Maybe they aren’t that energy efficient, as the shown examples but they do a fairly decent job as well.
However I wanted to provide you with some honorable mentions:
Pocket sun clock*:
A “digital” sun clock:
https://www.youtube.com/watch?v=wrsje5It_UU&feature=emb_title
Really Large sand timers:
https://www.hourglasses.com/largeglasses.html
Water flow clock:
https://en.wikipedia.org/wiki/Bernard_Gitton
DIY Wristwatch:
*Amazon Affiliate Link
