Note: this is a revised version of an article that originally appeared in The Neon News, issue number 17, December 1995 (a wonderful but unfortunately out-of-print newsletter for the neon sign industry published by Ted Pirsig & Val Crawford. Many thanks to them.)

While originally intended to control the brightness of neon signs, the described circuit is equally applicable to any inductive load (ex: motors) that can be controlled by phase-controlled voltage. This revision adds only minor corrections and additions.

Click on any figure to get a large version.

Sometimes you have to go backwards before you can go forwards. Trying to figure out the secret of dimming neon was like that. Here's the full story:

I had always heard that regular everyday incandescent-type light dimmers won't work with neon. Being somewhat of a nonbeliever in conventional wisdom, however, I decided to try it to find out for sure. To make a long story short, the conventional wisdom is right. Regular dimmers don't work. Trust me.

It all started one day when the local hardware store had a sale on light dimmers, I think they were going for about three to four bucks [in 1995!]. These were the cheapie type with the round plastic knob which doubles as a push on/off switch.

To find out whether they would work as neon dimmers, I dragged the trusty 15kv/30 burn-in transformer up onto the bench, grabbed some miscellaneous tubes and jumper wires, and hooked the whole mess up. After turning the dimmer up to full brightness, I plugged the creature in.

Initial success! The tubes lit to full brilliance. Things were looking really good, except that I then decided to turn the knob on the dimmer. As I rotated the knob, the tubes started to dim slightly, and then suddenly the room was filled with weird flickering stroboscopic light effects, followed by darkness. All in a matter of seconds!

After that, no amount of wire-wiggling or knob-turning would restore the light. We're talking serious dead here. Giving up, I bypassed the dimmer with a jumper wire, which erased the darkness. Glowing tubes indicated the failure wasn't the transformer, which only left the dimmer. Connecting the dimmer to a 100-watt light bulb yielded more darkness. Whatever died inside the dimmer was apparently going to stay dead.

With nothing left to lose, I pried the cover off the dimmer to assess the damage. Reverse-engineering the circuit yielded Fig 1.

Original Dimmer

The circuit (before flameout) worked like this: device Q1 is a triac, which is a power-switching device. When triggered, it switches to a fully conducting state, and stays that way until the current passing through it goes to zero. As it is connected to alternating current, which changes direction 120 times a second, it will be turned off as often. The trick to making it "dim" lies in how and when you turn it on. That task is left to components R1, C1, and D1.

Assume the following: the alternating voltage applied to the upper black wire in fig. 1 has just (time-wise) crossed through zero and is increasing. Current starts to flow through pot R1 (the thing on the other end of the plastic knob) and starts charging up capacitor C1. The setting of R1 controls how fast C1 charges. As you turn the knob clockwise, you decrease the resistance of R1, causing C1 to charge faster. Turning counterclockwise increases the resistance, charging C1 slower.

C1 will charge up until it reaches the breakdown voltage of diode D1 (known as a diac). In my unit, this occurs around 20 volts. Once diode D1 reaches its breakdown point, it switches into a conducting mode, discharging C1 into triac Q1, turning it on. Once on, power is no longer available to R1, keeping it from recharging C1 for the duration of the cycle. This whole process repeats 120 times a second. Because the time delay caused by the charging of C1, the portion of the AC cycle over which Q1 conducts regulates the time period during which power is applied to the light bulb load, and thus its brightness.

So what, you ask. Why did the dimmer die, and what's this got to do with neon? The problem is this: with a light bulb as a load, once the triac is triggered on, voltage is applied to the bulb, which immediately conducts current. This current keeps the triac turned on until the current cycles to zero, some milliseconds later. The neon transformer, however, is a partially inductive load, not purely resistive like the light bulb. Applying voltage to an inductive load does not cause current to immediately flow - it takes a little time. The problem is, it takes too much time for C1, which uses up all its charge before the neon transformer gets going, and triac Q1 turns off again. This allows C1 to start charging again, during the same line voltage half-cycle! This process repeats, at breakneck speed, until the diac overheats and dies.

Herein lies the secret: What we need to do is somehow keep trigger current flowing into Q1 long enough for the neon transformer to get going, without overheating diode D1. I did this, as shown in Fig. 2.

Modified Dimmer

Fortunately, when I was buying dimmers, I somehow had the sense to buy two of them. With my first dimmer lying in smoldering ruin on the bench, there was not much left to do but scrap it out for parts, the most useful being the still good triac, Q1. Opening up the second dimmer, I mounted the salvaged triac, now known as Q2, to its frame.

I rewired the dimmer as shown. Added resistor R2 is a 3,500 ohm, 5 watt power resistor. I mounted it to the dimmer frame (as a heat sink) with a small bracket and some silicone grease. Wire the modified circuit EXACTLY as shown in Fig. 2. Reassemble the dimmer cover, using small screws in place of the original rivets or pins.

Note the colors of the new wires: they are critical, as there are now three of them. When connecting up the finished unit, make sure the white wire goes to neutral, as does one end of the neon transformer. The other end of the transformer goes to the red wire. The black wire goes to the hot AC supply. The switch MUST break the black wire!

The new circuit works like the original, with the following exceptions: (1) When D1 triggers, it turns on triac Q2. Because it has a resistive load (R2), it turns on immediately. Current though Q2, limited to a reasonable level by R2, continuously triggers triac Q1,giving the transformer enough time to start drawing power. Later, as the line voltage goes to zero, Q2 turns off, removing drive from Q1 which, because the transformer's inductive current time lag, turns off sometime after zero crossing.

Miscellaneous notes:

Application notes:

  1. DO NOT use this circuit with high power factor type transformers or solid state type transformers. It won't work.
  2. It sometimes helps to use one size larger transformer if you're going to operate neon down to very low brightness levels. Experiment to see what you need.
  3. Working with 120 VAC is dangerous - use all possible caution. To limit flame and damage during bench testing, temporarily connect a high-wattage light bulb in series with your AC supply wire. This limits the maximum current flow in case of a short. Use a suitably sized isolation transformer to power your test setup.
  4. A final note: never work on the high-voltage wiring to your neon tubes on a dimmer-controlled setup, unless you KNOW the power is off (and, preferably, locked out.) With a dimmer turned all the way down, your neon may not light up, but you will if you grab the wiring.

Well, there's the secret: buy two cheap dimmers, and make one neon dimmer. It's easy when you know how (and why).

UPDATE: if you play with most standard light dimmers, you will notice an anomaly in their operation: if you turn down a lit light bulb, it will dim to a very low value (or go out completely). On the other hand, if the bulb is extinguished, as you turn up the dimmer you will note that the bulb brightness jumps suddenly from 'off' to about 1/3 brightness, rather than brightening smoothly. This anomaly is caused by unused residual charge left on C1 from one line cycle to the next.

The circuit shown below fixes this problem by removing residual C1 charge any time the line voltage is less than that on C1. C1 is discharged via either D3 and R3, or D5 and R4, depending on the voltage polarity on C1. When the line voltage is not zero, diodes D2 and D4 keep resistors R3 and R4 from contributing charge current to C1 by clamping any possible charging voltage to near ground potential.

Modified Smooth Dimmer