[ Español ]

Nixie clock I − Vintage electronics


This is my first digital clock using nixies to display the time. Nixies were invented in the fifties but were soon replaced by LEDs, fluorescent displays or LCDs and nowadays are valuable antiques. A nixie consists of a sealed glass tube filled with gas, typically neon, a grid and ten electrodes each shaped like a number. When enough voltage is applied between the grid and one electrode, gas around that electrode ionizes and glows displaying the corresponding number.

This project is possible thanks to Estebitan, Miguel Gimenez and Ronald Dekker. Estebitan sent me an old board with three nixies and some 74xx TTL ICs I use in this clock, Miguel Gimenez gave me another nixie and Ronald Dekker provided a larger neon bulb for the tens of hours. Thank you very much!

Whenever possible I use original components and circuit design common in the seventies so this clock keeps that vintage electronics style.

The clock

This is my clock finished. Click on the picture to enlarge:

Click to enlarge

I have mounted it in a transparent box of chocolates where I have drilled holes to screw the board, nixies, push buttons and to let in air flow for cooling. The nixies are inside the box secured with spacers.

Update january 2017: I replaced the tens of hours nixie with a similar sized neon bulb as this was the original design idea, a clock with only three nixies.
Actually the color difference between the neon bulb and nixies is much less noticeable, the camera exaggerates it a lot. Click on the picture to enlarge:

Click to enlarge

Top view:

Top view


This is the clock before mounting in its box, the neon lamp represents tens of hours since I didn't have the larger neon bulb yet.

Clock outside its box

Original components used

Clock board

Some integrated circuits are original TTL chips from early seventies and are labelled with their manufacture date: three 9315 in the top row are dated 7327 (year 73 week 27, June 1973), two 7490 in the second row left and right are dated 7205 (February 1972) in ceramic package and a 7490 in the third row right is labelled 7236 (September 1972) in grey plastic package. They are made by Fairchild, draw around 20mA and get warm.

The 9315 is a Fairchild variant of the standard 7441 TTL decoder and nixie driver, pin compatible and differs only in its overrange outputs (10 to 15). More information at Nixie drivers reference (Brent Hilpert).

I have reused the current limiting resistors, nixie sockets, wires and a BC107A transistor with gold plated pins. From another old board I got a selenium rectifier AEG model B20C450 and a 1000µF 10V Bianchi capacitor to build the power supply.

Power supply

Circuit board

Circuit design

Like many vintage electronic circuits from that era, the entire circuit is live, uninsulated, directly connected to mains power. This is a simple, cheap and effective way to get the necessary voltage to drive nixie tubes but has the drawback that the circuit cannot be touched in any way while powered. This type of design is not meant for playing or experimenting with circuits, instead it is meant to be built and enclosed in an isolated case before power is applied for the first time.

Safety warnings!

Having said that, for improved safety the circuit has resistors limiting current from both mains poles that limit the impact of an accidental short or electrical shock. Also once unplugged it does not contain capacitors that could store a dangerous electrical charge.

The same safety procedures, precautions and hazards apply just like working on any other electrical live mains circuit. Mains has two dangers to be aware of: on one hand, a dangerous voltage exists between the two poles and on the other hand, a voltage also exists between each pole and earth ground. This is because the power facility connects to earth ground one of the poles or the neutral in a three-phase system, thus a dangerous current could shock you if you touch one pole and earth ground which could happen if you go barefoot, touch a wall, pipe or plugged appliance.

To probe the circuit you can disconnect diode D3 and resistor R9 to isolate the digital circuit from mains, then it will operate normally however the nixie tubes will not light up.

Circuit schematic

Before using this circuit be sure to have understood all the information provided in the previous section.

Nixie clock circuit

Circuit updated on August 2014. View original design (April 2003). Changelog:
- moved resistor R9 to the opposite mains pole to limit current from both poles, improving safety.
- added a low-pass filter (R3 and C6) to the 50Hz signal to filter out transients and improve time-keeping accuracy.

The entire circuit is live and under dangerous mains voltages, it must be isolated properly and never touch any part of the circuit while plugged.

As a time reference it uses 50Hz mains frequency. To use only three nixies this clock counts hours up to 12 and displays tens of hours using a neon bulb, resistor R5 has to be selected according to that bulb.

Power supply

Nixie supply voltage is obtained by rectifying mains AC with D3 diode and limiting current with resistors R5, R6, R7, R8 and R9. Rectification is necessary so that only the numerals glow and not the nixie grid or internal wires and current has to be limited because neon once ignited does not limit it by itself. Half-wave rectification reduces by half the time the nixie glows, extending its life with no noticeable flicker.

R6, R7, R8 (and R5 if a nixie is used) resistor values are calculated according to nixie specifications: ZM1020 (Dieter's data archive). 47k with common 12k ½W were the resistors on the original board that I reused, this equates to 95k per tube because the common resistor supports 4 times the current (12k * 4 + 47k). In ZM1020 datasheet page 7 we get that 95k with pulsating 330V supply gives a nominal current of 1mA per tube, the minimum for full glow coverage. Peak current is 2mA ((330 − 142) / 95k).

The maximum nixie voltage drop is 170V at turn on (page 2, ignition voltage), 142V while lit (page 4, maintaining voltage with 2mA peak) and 118V to turn off glow (page 2, extinguishing voltage).

The 9315/7441 is designed to drive nixie tubes directly, for this reason its outputs handle up to 70V (F9315PC at Datasheets360). In normal operation each 9315 has always an active (grounded) output, in my design nixies are never blanked. With one cathode grounded, the anode voltage will not exceed the nixie ignition voltage of 170V, so that the inactive digits will see no more than 100V (170 − 70), that is below the 118V extinguishing voltage ensuring inactive digits do not glow.

Let's see what happens in the inactive cathodes. From the current vs floating voltage curve in ZM1020 datasheet page 5 we know the nixie weakly pulls them up to the anode voltage with currents in the µA range. 9315 datasheet does not include a similar curve but states a cut-off leakage (Ioh) of 40µA at 55V, because the nixie inactive cathode current is higher it will pull the voltage up to some value between 60 and 70V with a current between 50 and 200µA. This sets the nixie operating point in the nominal operation area, the area to the right of curve N so it will work optimally.

In the event that no output is active (for example if 5V supply voltage is missing) the voltage at the 9315 outputs will reach 70V and leak some current resulting in a dim glow in the nixie. This current will not exceed 1mA, well below the supported 2mA at 70V as per datasheet, thus no risk of damage.

The tens of hours neon lamp is the only one that can be fully off, therefore I use a BF422 high voltage transistor with a voltage rating of 250V. In the off state the bulb anode voltage is 258V (330 minus 72 that drop at common resistor R9), this leaves 8V insufficient to ignite a neon bulb.

Due to the high current drawn by these old TTL ICs a transformerless power supply is not suitable. Following strict vintage minimalist style a resistor is suficient to set the 5V supply voltage given that current drawn by these TTL ICs is constant enough. R10 resistor value has to be selected according to circuit current draw and resulting supply voltage with that load. These TTL IC operate between 4.5V and 5.5V, I recommend to adjust slightly below 5V so there is margin against power surges, I also recommend to place a 5.6V zener diode in parallel with C1 cathode to positive and a 250mA fuse in series with the resistor to protect against overvoltage.

Some 100nF filter capacitors are recommended from 5V supply to ground distributed across the circuit to filter noise and improve stability.

Circuit logic

Nixies are controlled by three 9315 (7441) ICs designed to drive nixie tubes, they contain a decoder to be connected directly to binary counters. In this circuit, minute and hours units are connected to 7490 decade counters and tens of minutes to a 7492 that counts to 5.

Time-keeping is performed by a set of concatenated counters that divide the 50Hz mains frequency down to one-minute pulses. R3 resistor connected to U10 pin 14 brings 50 pulses per second from transformer output to the first counter clock input, R3 together with C6 form a low-pass filter to eliminate transients that could make the clock run faster. U9 and U10 divide by 5 and 10 respectively counting to 50 and generating a pulse per second on U9 pin 11, and U7 and U8 divide by 6 and 10 respectively counting to 60 so that on U7 pin 8 we have a pulse per minute to drive the minutes counter U6.

For tens of hour I used the free flip-flop in U5, output from this flip-flop pin 12 controls the tens of hour neon lamp by switching transistor Q2. Flip-flop state flips when it receives a pulse on pin 14 and this happens when the hours units change from 9 to 0 and from 2 to 1 (9 to 10 and 12 to 1 hours), in other words, when hour units are 0 or 1. This is because the way the nixie is wired to the decoder, note it is shifted two units.

Hour units nixie is wired so the relevant values 0 and 1 correspond to binary values 8 and 9 in the 7490 counter, these values activate U4 pin 11 (the most significant bit) and signal the flip-flop to change state through transistor Q1 acting as an inverter, since the flip-flop reacts on the falling edge. Now the tens of hour lamp switches on and off properly and all is left is to have hours count from 12 to 1.

To reset hours from 12 to 1, MS1 and MS2 inputs are activated to reset the counter to value 9 (1 displayed). The tens of hour flip-flop status is input on U4 pin 6 and diodes D1 and D2 implement a logical AND that sets high U4 pin 7 when the counter reaches value 1, so at 13 hours both inputs activate and reset hour to 1. D2 stops the counter to be set at 11 hours, counter value 9.

Finally push button debounce is achieved with capacitors C3, C4 and C5 and a 100k resistor paralleled with MIN+ button. These capacitors block DC voltage and only let through a short pulse when signals change state, effectively mixing them. The 100k resistor keeps capacitors charged so that there is no voltage difference on push buttons terminals and no increment when pushed down, the counters increment cleanly only from pulses from U9 pin 9.

The values of these capacitors were selected experimentally. Excess capacitance (for example 4.7nF) made the counter react on both the rising and falling edge, incrementing twice each cycle. With 470pF it no longer happened and with 4.7pF it still was fine, therefore 47pF was taken as a middle value. Note that when a rising edge occurs, the capacitor is charged and raises the input voltage above the supply voltage, for this reason to minimize adverse effects it is not advisable to use a larger capacitor.

It is possible to modify this circuit to run at 60Hz mains frequency changing U9 to a 7492 IC and connecting it like U7, C4 capacitor to pin 11 and wire to U8 to pin 8.

Possible faults

At turn on there are two numbers lit, a diffuse glow inside a nixie or it counts from 60 to 69 minutes
This clock has no reset circuit for the counters and when plugged they contain a random value that may be out of range and cause the 7441 decoders to turn on two or more digits. This is normal and is fixed advancing hours and minutes.
A diffuse glow is visible around numbers
This happens when the 7441 does not activate any output and can be caused by missing its 5V supply or due to the reason stated previously.
Front grid or connecting wires glow
The rectifier diode is reversed or its maximum reverse voltage is too low. You must use a diode with at least 350V reverse voltage, like 1N4004 or BY127.
Pressing push buttons to set the time, the clock does not advance
The clock is slow or does not increase minutes
Try to double C3, C4 and C5 capacitor values.
The clock experiences a small deviation of some tens of seconds but in long-term keeps time correctly
Small deviations in short-term are normal, but long-term precision should be good. This clock uses mains AC frequency as a time base, depending on the power drawn by nearby industry and buildings mains frequency may experience small deviations that the power utility counteracts when power demands drop by synchronizing with a precision clock.
The clock advances one hour, one minute or both when some appliance is switched or plugged
Time-setting push button wires are somewhat sensitive and can pick up strong mains transients. To minimize this effect keep them short, twisted and apart from other wires.

Project completed on October 2007 by Jeroni Paul.
Copyright © 2007 Jeroni Paul.

Back to index