Updated: August 2009
My first completed nixie tube clock……
(Click the pictures to embiggen)
Key design features…
- Automatic time set from radio signal. Also works in manual set mode.
- Accurate RTC, with long life battery backup.
- Time and Date display, HH:MM:SS and DDMMYYYY
- Remote display for extended output
- Various display modes and effects
The project was completed in stages:
- Receiving and decoding the UK MSF time signal. This is not as difficult as it sounds. Honest! – This was important, as I think a clock should be able to set itself.
- Implementing a RTC. Takes a burden off the microcontroller, and has a handy battery backup. – Another critical aspect of my requirement. When the power fails, who wants to set their clock??
- Tube interface bus and communication. To keep it simple, every board, including the power supply is on a single 20 way IDC bus. I also wanted to use the SPI protocol. – It is a nice protocol, but you will need to RTFM!
- Finishing the software. Developed in Proton using PIC Basic Pro. A very nice program!
- PCB design. My priority for this first design was to get the clock made. Future designs will be more compact.
- Case. Finally bringing it all together. Don’t skimp on the case or underestimate how much work is involved.
I chose the 16 tube idea because I could use the posts in the tube as colons, and I wanted to display the time + date. The nixie tubes are long life GN4 made by ITT.
The clock is housed in a cast aluminium box, with a front acrylic shield. I over-engineered every aspect of the clock deliberately – it’ll probably outlive me!! .. so with that in mind, I wanted it to be safe to use and safe from prying fingers.
All the controls are on the back panel. There you will find…
- Time signal receiver interface
- Remote display interface
- AC power input – with voltage setting for World-wide use.
- Mode setting ( more on this later.. )
- Manual time setting controls.
The clock works with or without the time signal receiver module. Connecting it puts the clock into automatic mode. Without it the clock runs fine in manual mode. The remote display option was not part of the original design. I used a LCD extensively during the design and debugging phase, it was so useful I decided to keep it in.
Here are some pictures inside…
A Microchip PIC is at the heart of the clock. A Dallas RTC with backup battery keeping time.
The tube slave boards are microchip port expanders on a SPI bus. 5 lines of output from the PIC gives me individual control of the 16 tubes. Every segment is direct drive. No multiplexing. ( no multiplexing problems with driving large numbers of tubes and also the individual segment drive requirements.)
The power supply is a high voltage transformer. I decided against using a SMPS, as it would interfere with the time receiver, and it is not as reliable as a transformer.
The time signal receiver and ferrite I took from a very cheap radio controlled clock. A small circuit was needed to interface the receiver to my PIC. I also added a LED to “show” the time signal. Note the receiver board simply demodulates the time signal. Further processing is required to decode it!
Remote display, waiting for its housing.
I am in the process of archiving and digitising my design file. Sections of the file will be made available here in the near future.
Fantastic Job!
Very, very professional quality build and design!
Thank you! I am very pleased with it.
With such design it will definitely catch attention for decades. Beautiful work, especially “sandwich” PCBs.
Great clock, any plans to sell it as a kit? Are there any plans or possibilities to connect a 1 PPS input to use a 10 MHz Rubidium frequency standard?
I think if you made a housing with red perspex filter, it would emphasise the numbers allowing tube drive power reduction and hide any tube poisoning. Not to mention a more Retro look.
cheers,
Christopher
The sandwich pcb layering worked well. I was lucky with the case!
No plans to sell it as a kit, sorry. I might consider making one up as a special.
An external timebase is certainly an option I can put in.
The red filter idea is a good one, and yes it would give the clock a more retro look. I’ll get a piece cut down and try it out. Check back soon.
This is the type of nixie clock I am planning to build, the type with date as well as time. There are absolutely no instructions anywhere on how to build this type, do you have the time to publish instructions, both for the beginner but also with schematics? A list of componants and where to purchase perhaps? I have to say well done, this is the best nixie clock I’ve ever seen.
How did you drive the posts as colons? Looks like IN-12a / B5991 type tubes, and the posts connect to the anode screen. Is it a matter of reversing the anode/cathode connections, then adjusting the drive voltage so the screen doesn’t glow?
John, I’ll be putting some information online in the next couple of weeks. I havent had much time recently.
Mike, the post illumination is accidental. For some reason they appear isolated from the anode grid. Driving the posts is as simple as driving them as a numeral but at a much lower current.
I have an update pending. I managed to locate some red perspex with a tint I liked. The clock looks great in clear or red. Pictures soon.
What a great clock – the best of the type I have seen. The earlier post by John O’Loughlin is correct that there are not any decent instructions anywhere on the web for date and time type, especially with radio controlled automatic update.
Is there any chance of putting up more details if the clock’s inner workings and parts list? I would love to build one!
Cheers
Mark
Very impressed. Share please
New job, young baby and some demanding personal development have put the squeeze on my hobby time. Sorry folks, I have a real lack of time at the moment.
The date and time extraction is pretty simple. In Europe or North America you should be able to pick up a “local” amplitude modulated time signal. I recommend purchasing a cheap “radio controlled” clock and removing the demodulator ( ferrite and small circuit board). The cheaper the clock, usually the higher the probability the demodulator ferrite and PCB is separate inside.
The easy bit is to simply feed that demodulated signal into an A/D port of a microcontroller. Then it is fairly trivial to write the software to decode ( you will need to know the format of your time signal, usually it is BCD encoded and takes 60 seconds to transmit the date, time and set a few other special flags).
If you don’t want to use a microcontroller then prepare to use a great deal of logic. It is not impossible, but it will take you a considerable amount of time to complete, and would certainly be far out of the realms of a beginner.
Never used a microcontroller? Take a look at the Arduino.. jump in!