Hardware finished

Front view.


Side view.


Back view.

Microprocessor problem

The uP ATmega32 was dead when I try to program it many times, then, it raised out another problem, how can I remove it from the PCB? We can use a machine called SMD Rework to detach it from the board. Luckily, my friend Mok has this kind of machine :D


The ATmega32 removed.

UART buffer overwritten

Refer to the application note "AVR306: Using the AVR UART in C" from Atmel, the rubbish texts problem solved, this happens due to the gps data error processing in the buffer. Since I just copied the buffer when it received, but did not write any protection to ensure if there a buffer overwritten. The correct path should test the maximum buffer size first, to prevent the buffer overwritten occur.

Problem when the GPS module parsing data

The screen capture shown some data output, but it contains with some rubbish texts between the good data packets when it receiving...

Final test & assembly

This small unit is a GPS receiving module, I have made an inch of acrylic board supporter for it, this is my first time to bend a board with 90° degrees using the heat from a solder.


Fixing a GPS receiving module to the back cover.


Drilling some holes at the corner, it's for the screws use.


Final test before assembly, also, making some wires for a GPS receiving module and the power jack connection.

Case design

The acrylic boards layout was designed with the CorelDRAW software, cutting and drilling those acrylic boards performed by the laser engraving machine.


Using the Chloroform (CHCl3) to glue the acrylic boards.

Switch debouncing

I've designed the clock with three micro switches for system control, but the mainly problem will let the system uncontrollable is the switch bouncing effect, I tried many different ways to combat this effect, such as "delay loop", "while loop".... finally, a small function called "loop_until_bit_is_set" can do this, it come with the AVR Libc library.

RTC with battery backup

A close-up view of the tubes.


When the input power failure, a CR2032 lithium battery keeps the time running.

Cathode Poisoning

Most nixie tubes have more than one cathode, if a cathode is illuminated, material is sputtered from this active cathode to the inactive cathodes...


If a cathode is not used for a long time while others are lit, the coating on the inactive electrode(s) gets thicker. This coating is highly resistive, so if it gets thick enough, it prevents the inactive cathode from glowing...


If the inactive cathode gets switched on, parts of it will be dimmer or totally dark.

Soldering components (part 2)

All components soldered to the PCB, above shown the front view after its assembled. These six tubes represented the time "HHMMSS", date "DDMMYY" and GPS coordinate.


Above shown the rear look. Three micro switches control the mode change and date/time settings.


The tube with fine legs, it may broken if not careful to install.

Soldering components (part 1)

Using some 0.51mm diameter solder wires for component soldering.


Using a torch to check the soldered line connectivity.


Some 1206 and 0805 SMD components.

PCB from the manufacturer

I received the PCBs today, they all produced by a Shenzhen's PCB manufacturer, I found this manufacturer from Taobao.


Making these PCBs are not expensive, because our country is a manufacturing kingdom. For this standard FR-4 PCB with the size of 100mm x 80mm, total 10 pcs samples only cost RMB150 (~USD$23).

Generating the Gerber files to PCB manufacturer

After the schematic design and PCB layout work, I used a CAM process (it included with the EagleCAD) to generate some Gerber files. The Gerber File is a file format used by PCB manufacturing machines to lay out electrical connections such as traces, vias, and pads.

The gerbv 2.4.0 used for Gerber files viewing.

3D PCB design

Eagle3D generated a beautiful 3D image. It is a very useful tool before the actual PCB fabrication, I used this "virtual layout" to check the component orientation and the space is enough or not.

Schematic & Board

EagleCAD 5.9.0 used for the schematic drawing and PCB layout.


This screen capture shown many air wires (yellow lines), they need to route perfectly (red & blue lines) on PCB.

A lot of Nixies

A lot of B-5853S Nixies from my friend Mok, thank you very much :D

Real-time clock added

A DS1307 RTC chip added in the circuit prototype, it is a low-power serial real-time clock (RTC) IC, the data are transferred serially through an I2C interface. The final design will replace by DS3232, its built-in a TCXO with very accurate timekeeping, temperature-compensated, integrated RTC.


The above video shown the seconds counting...


A drafted circuit of my Nixie Clock design.


Different voltages measured in the circuit.

Multiplexing two Nixies

A test for multiplexing two Nixie tubes. In this circuit, all the corresponding cathodes of two Nixie tubes are connected together in a bus structure. The anodes of the Nixie tubes are switched on one by one and the selected cathode is activated for every digit. This test is fast enough (anode turn on time is 10ms for each tube), so we get the illusion that two digits are on simultaneously.

Multiplexing method can reduce the pins usage in the microcontroller.

0-9 Counter

This is a 0-9 counter test for the IC KM155ID1 and a Nixie tube B-5853S, its counted per second. The KM155ID1 is a industrial grade high voltage driver IC, it converts BCD and direct connect to each cathodes of the tube.

Testing for the load currents

In order to prevent a large current pass through to the Nixie tubes, and heat rising in the switching transistor, I need to make some calculation for the load current limitation.


I used a FET IRF740A for the switching transistor, it was very hot when I set a 93% duty cycle in the AVR's PWM output. However, an interesting thing was happen, I turned down the duty cycle to 30%, the output voltage dropped to 137V and all the tubes keep operation. The IRF740A was not hot and could be detach the heat sink.

Light up!

Two Nixie tubes were displaying the "Answer to the Ultimate Question of Life, the Universe, and Everything."


Working voltage for the B-5853S is at least 140V, and the anode current around 2mA.


I used a microcontroller to generate a PWM waveform, it is 93% duty cycle to turn the FET switch on and off, then charge the inductor.

Boost Converter simulation

Nixie tube needs 170-250V to operation, but how to change a DC source from 9-12V to high voltage level? I used a Boost Converter to do that, it is a type of SMPS (switching mode power supply). The screen capture above shown a software simulation, it will be use in this project.

The Proteus 7.4 used for the software simulation.

Tubes from the Lithuania

I bought the Nixie tubes from a Lithuanian, those tubes manufactured in Russia.


Nixie tubes IN-12 (top) and IN-8-2 (bottom).


VFD (vacuum fluorescent display) tubes are IV-21, and some neon bulbs.


The industrial grade high voltage driver IC - KM155ID1. It is a BCD-to-decimal decoder designed to drive the gas-filled Nixie tubes, compatible with well known 74141.