A lot has changed since my last Nixie Clock post in 2010. Using Eagle I had a prototype board made through 4PCB, unaware of their Student $33 Each deal; If you are a student use this! After soldering in all the components and testing the circuit it turned out that multiplexing was not a good decision. I was unable to switch at a high enough frequency while maximizing ignition of the tubes. In other words, I would either see the numbers scrolling in a bright neon light or I would see all at once in a dim neon light. Unfortunately there was no software fix. Discouraged and busy the project was dead until about 6 months ago.
I decided to start over from scratch. Once again creating an equivalent circuit on the protoboard. This time around there were a few new requirements:
1. Drive each tube individually.
1a. This requires (4x6=)24 pins. Use minimal amount of pins possible on the microcontroller.
2. Integrate with a RTC (Real Time Clock).
3. Keep the time when the clock is unplugged with a rechargeable battery.
4. Alarm buzzer.
Now lets address these points and examine why I chose this:
1. With each tube driven individually the brightness can be maximized. Multiple levels of brightness can be achieved by having all the anodes hooked to one transistor. With this method I can oscillate at a higher frequency than before, achieving better brightness and control over the tubes.
1a. It would be a waste to dedicate that many pins on a microcontroller for simple H/L outputs. The cost of the microcontroller would go up, as would the area and the wiring. Instead, three pins can be turned into 24 using multiple serial in, parrallel out shift registers (SIPO). The 74HC595 was chosen because it also has a serial output which allows you to chain the output of the first register into the second, the second into the third, etc.
2. While I am using a PIC microcontroller and could calculate one second using software, a real time clock would simply things quite a bit. It would also allow me to write an I2C driver and learn about the protocol. There is the option to sync with a GPS on the clock but this method is less expensive for the user and also very accurate. I chose the PCF8563. I will be writing a more in depth tutorial on how exactly this is done.
3. A simple 5V rechargeable battery was used. When the circuit is plugged in the battery charges, and when it is not the battery powers only the RTC. With a max supply current of 550nA on the RTC and a battery capacity of 45mAh, this will keep time for around 9 years without being plugged in.
4. I find myself using an alarm more often these days. I spent a while choosing the right frequency for the buzzer. This task was harder than I expected. I ended up choosing a Kobitone buzzer, model #254-EMB125-RO. It outputs 85dBA min at 2.3kHz. It's quite annoying: perfect.
This time around I chose KiCad to create the schematic. I was tired of the limitations of Eagle. This software is open source and outputs Gerber files. It takes a little getting used to, but there is a flow to the software once you get the hang of it. Below is the product I ordered from 4PCB. Thank you very much to Ashley at 4PCB. I had a major accident in my first order and she put a free expedite on my second order. I have learned to never trust library files. The drill size for some of the through hole components were too small for the components!
Below is the circuit board.
Here it is with all the components installed in all of its beauty. I still need to determine what to do with a case. I have been designing one in SolidWorks, but the cost of fabrication is more than I expected. I'll figure something out!
More to come soon!