As someone who loves building custom projects that meet a need that the market can't or won't supply, I thoroughly enjoyed this article. I may be putting one of these together very soon.

I'm also out of unused Pi Zeros so may have to slightly modify it to use an ESP32 though. And it'll either use ESPHome or control over local MQTT since I don't need HomeKit. And I'm just now realizing I'm thinking out loud lol

I wholly sympathise with your comments re: Autodesk software.

If you’re wanting to do parametric modelling work in the future, or just any modelling work or CAD, I’d personally recommend Rhinoceros 3D from McNeel. Its a one-off licence, and comes with the node-based programming system, Grasshopper, by default.

We were taught AutoCAD at school, and it had its uses, and don’t even get me started about Revit…

What, no ambient light sensor?

Love this (I also love my Yamaha TSX-70 so won't be building one any time soon).

What, no ambient light sensor?

....damn, yeah, that would mean I could ditch the schedule-based bright/dim setup, too.

Time to start planning version 2.0!

Is this what you were looking for?

View attachment 139530

Nope—"auto time" clocks like this don't actually sync their time from any source. They're simply factory-set and have a backup battery, and auto-adjust themselves to DST. There's no external time source sync. (I fell way down this rabbit hole, believe me.)

Is this what you were looking for?

It does say its for seniors (as in technical editor?).

I have been thinking about a project like this but I want the clock to include seconds and tenths of seconds too.
The tenths of seconds should be shown with 10 LEDs in a circle.

Why?
Answer: Why not?

Nope—"auto time" clocks like this don't actually sync their time from any source. They're simply factory-set and have a backup battery, and auto-adjust themselves to DST. There's no external time source sync. (I fell way down this rabbit hole, believe me.)

Ah, good to know. I just assumed autoset was atomic.

Why the "sub second" description in the text?

And was the drive current to the display not accessible? A PWM running at 10 kHz or so should allow you to dim to nothing. It'll become nonlinear are you reach the turn-on/turn-off time of the elements.

This is a great article and the detail you put into it show passion. Thank you for sharing!

What would it look like with NIST's WWV integration for time accuracy?

Edit: It was already mentioned.

Why the "sub second" description in the text?

And was the drive current to the display not accessible? A PWM running at 10 kHz or so should allow you to dim to nothing. It'll become nonlinear are you reach the turn-on/turn-off time of the elements.

To the first, I wanted the displayed time to be exactly accurate, and I wanted the display to switch from 10:01 to 10:02 at exactly the same time as my watch, computer, phone, and other NTP-backed sources—not 1-2 seconds before or after.

To the second, the clock looks better with a smoked acrylic shield over the numbers anyway, so rather than get jiggy with the voltage, I leaned into NDF and plexi.

What would it look like with NIST's WWV integration for time accuracy?

I'm not 100% sure, but "more complicated," probably. I've never done anything in the RF world, and NTP is "free" with the host OS. I imagine syncing to NIST via radio would mean some more horizon-expanding.

edit - @Wickwick's got it, then. Sounds like it might not be that hard at all! (Though I think I'd still prefer the NTP solution—I'm a lot more familiar with NTP than anything RF-related.)

This is a great article and the detail you put into it show passion. Thank you for sharing!

What would it look like with NIST's WWV integration for time accuracy?

You'd start with something like this.

I dare say the amount of coding is then reduced to almost nothing.

Didn't see it mentioned in the article or in the comments yet, so I'll ask:

Did you consider a build using a GPS/GNSS module instead? What led you to decide to lean on local network instead?

Going off on a tangent here, but since author lives in the US he should be within range of those radio signals that are synchronised with atomic clocks (like the one from NIST, for example). I’ve seen some of these types of clocks with a similar red LED aesthetic sold in Asia. You could probably source the modules online somewhere. They’re pretty much the same as the one’s you’d find inside Casio G-Shocks that have the radio wave sync feature.

This was my first thought too, though with this being a self-admitted exercise in over-engineering, I think he was right to go with the less elegant solution. To go off on another tangent, I don't think enough technology producers make elegance a first-class citizen of their design process, and far too much of the enterprise and consumer IT estates are vastly over-engineered.

edit - @Wickwick's got it, then. Sounds like it might not be that hard at all! (Though I think I'd still prefer the NTP solution—I'm a lot more familiar with NTP than anything RF-related.)

There's no more RF-related challenge to using an element like that than there is using your WiFi router. The antenna is fully integrated into the component. It uses digital communication to the rest of the board and there's even Arduino code on that page.

That's just the first receiver I happened upon with a search. I'm sure there are others. I had a battery-powered LCD clock some 25 or 30 years ago with this feature built in. I knew it couldn't be expensive.

Being a craftsman and Maker, I can really appreciate the motivation behind this. "Why can't someone make a simple (insert something so esoteric and outlandish for the average person there's no chance in hell it would ever be profitable to sell)!?!?!"

Then I go off and make it myself in a huff of slightly OCD indignation.

I have a room full of that kind of thing around me. If I can't find it, and feel I need it, I make it.

But I have a clock that ticks a lot of the requirements here, other than the precision and Ethernet connection. It has red numbers, (in varying levels of intensity). It doesn't display seconds, but it does give day, date time and temperature. At my age, sometimes I need all of that.

It's usually AC powered, but has a self-charging battery backup for power outages (and they're changing the power poles in my area every few weeks now, so we get a lot of them lately), so when the power comes back, it's still set for the proper time.

It doesn't keep perfect time. It's off by about a minute per year or so. But I don't have a lifestyle or mindset that mandates microsecond precision, either.

A REALLY cool article, though. I love these DIY things, especially when shared by the people who have both the expertize to do it, and the communication skills to clearly relate exactly how and why.

A fun read.

A project to control a timed device at work was given to me because somebody saw an Arduino laying on my desk for another project and thought I might like to take this on.

Up to that point I'd only coded Arduinos for small, single task controllers. This was much bigger. I ended up scoping it out to have a 16x2 LCD on the front panel with a single button to turn the backlight on for 5 seconds and also connect to the LAN to fetch the time. How hard could it be?

Lots of examples of what I needed were available on the internet to make the basic functions work but what ended up being the hardest, by far, was dealing with DST. I had code to detect which Sundays the time change occurred but it took awhile to get the hour of the time change to work reliably in testing. Top tips of coding for time: always keep a master time in GMT and converting that to Unix time you end up with time that is now very easy to compare.

The project has ended up working out very well and was an awesome learning experience.

This was the point where the project stopped feeling fun and started feeling impossibly hard.

(looks at various projects started but never finished) I'm feeling targeted

Really cool project!

For the next version maybe consider incorporating a EverSet ES100 or CANADUINO V4 AM to receive atomic clock transmissions.

Lots of examples of what I needed were available on the internet to make the basic functions work but what ended up being the hardest, by far, was dealing with DST. I had code to detect which Sundays the time change occurred but it took awhile to get the hour of the time change to work reliably in testing. Top tips of coding for time: always keep a master time in GMT and converting that to Unix time you end up with time that is now very easy to compare.

The classic Tom Scott/Computerphile video on the challenge of time zones and DST convinced me long ago that I should always outsource anything involving timekeeping to an existing technology stack

I did something similar with the Evil Mad Scientist Alpha Clock Five (I used white LEDs, but it appears to no longer be available unfortunately).

Originally I found the kit when my girlfriend at the time wanted a clock with large white LED segments and this was the only one I could find, but as it was a kit there was some assembly required and it was my first soldering project. The microcontroller can be reprogrammed, so I added some personalizations to it (greeting her by name when turned on, displaying "HAPPY" "B-DAY" on her birthday, "MERRY" "X-MAS" on Christmas, etc.).

I later built one for myself. While the kit includes a Chronodot with a coin-cell battery to maintain the real time clock accuracy, I wanted a clock that I wouldn't have to set. Turns out it supports adding a GPS module, so now mine sets its time from GPS satellites.