Keeping things simple, firmware is identical in all four of the PIC16F1509s. When power is first applied to the system, controllers wake up, "sniff" the slots (specified using pull-up/pull-down resistors on two of their pins), then set their I2C addresses accordingly.
Of particular interest is the way in which the Microchip engineers used tools like Excel to determine the X-Y coordinates for the Nixie Tube lead holes, and for the orientations and locations of the surface-mount drive transistors (one for each cathode). The PCB was created using CadSoft EAGLE PCB Design Software and fabricated through ValueProto for $25 each, which is a really good deal for a small hobbyist prototyping run.
The Nixie Tube clock circuit board.
The video below shows the clock displaying time, followed by a random sequence, then temperature.
Day and Cappy also will discuss future enhancements, including the
addition of a pressure transducer, a humidity sensor and Wi-Fi
capability. They alos will introduce a variety of Nixie Tube resources,
including alternative Nixie Tube Clock implementations.
1 whole MCU plus 10 transistors to drive a Nixie? There's a 40-year old chip called a 74141 that does that (and you can still get them....)
Yeah, yeah, they draw more current, and you can't easily blank them..... but were I doing this I'd use them, and just remove power to the whole display chain if it was not required. But then I always have been a bit of a luddite.....
One of the aspects of this presentation that I really enjoyed was the walk through of their thought process when making design decisions. i.e. what MCU and how many, how to program, and things of that sort.
All in all they did an excellent job with this.
I don't see what is so revolutionary about this clock. I first put one uC per tube into my Microchip based B7971 nixie tube 'Smartsockets', back in 2006. I've since made several versionjs for different display types. I was also the first person to put coloured LED's behind nixie tubes. There's a lot about this clock which is familiar to me....I keep my project up to date and I've recently finished VFD display based smartsockets.
Nice application and at the first look I agree to John Pote: One uC per Nixie seems really a luxury. - The Nixies could also be controlled by some latched shift registers. So you use just one uC port pin for shifting and a second one for latching. But then you will need a uC that has some more PWM outputs for the independent RGB LEDs.
So all in all the 4 uC solution is really a 21st century solution. uC are that cheap and breaking up a big problem in parts resulting in more, but smaller problems (thinking of hardware and software) is a very good way to solve engineering problems - IMHO the century doesn't matter in this case... :-)
WOW Nixie tube nostalgia! First desk calculator I saw, early 70s' in school, used them. LEDs and VF displays that followed just didn't convey the same quality. Curious how the circuit design is a 21st century solution to a 20th century problem, 1 MPU per Nixie. Luxury! Back then 1 MPU/CPU was all we were allowed but we could have a few of octal latches and lots of driver code to multiplex them. How well I remember filling a 2k byte eprom with intricately interwoven monolithic assembler. I still wind up colleagues by threatening to put a 'goto' in my 'C'. They fall for it every time.
Awesome project, very niffty application, even at $25 each board that is a great value, as youo pointed it out Max.
I have to confess to been also a fan of Eagle, for their powerful and user programmable scripts. I was able to place 25 items (SOP6 LEDs)with perfect spacing in a square array 5x5.
So Excel is a handy tool but Eagle scripts could have been as well suited for that job.
Thanks for sharing, wish I could make it, but work and other obligations would make it hard for me. Max, hope you can give us a photo gallery of the event.