This week marks the official opening of the Vivid Sydney outdoor light show for 2015. Sadly, the weather is absolutely horrible today, with scattered heavy showers all over Sydney meaning it’s another “stay at home” day.
This got me thinking – we don’t see many neon bulbs nowadays, with their warm and soft glow. The NE-2 used to be extremely common, used as an indicator lamp, night-light and even in circuitry. Sadly, they’re not as easily available as they once were.
While I was testing the Keysight U1461A Insulation Resistance Testing Multimeter, I used an old fluorescent starter as a substitute for a gas discharge tube in testing the step-voltage test mode of the meter. Maybe I could turn some fluorescent starters into a “glow lamp” – it won’t be bright, but it will have the soft gaseous glow and it would be a way for me to get vivid with my ideas!
Fluorescent Tube Starters
Fluorescent tubes running on old fashioned magnetic ballasts used to be a mainstay of commercial lighting, and even made it into many homes, although mostly relegated to garages. These were robust and simple, using essentially large wound inductor ballasts to keep current under control. They had a reputation of having a blinky slow start, and twice-line-rate flicker.
These are slowly being phased out, with electronic ballasts which offer much more reliable starting and longer tube life, and finally, by LED retrofit devices which offer higher lighting efficiency.
So why would I choose fluorescent tube starters for my little project? The main reason was that they are ubiquitous at the present moment and relatively low cost at under AU$1 a piece. They’re also well standardized by IEC 60155 standard and connectors are easily available which is a bonus.
That being said, the starter’s primary job is not to act as a glow tube. Its job is to:
- Let through a limited amount of current at power on to warm up the lamp’s electrodes for a pre-set amount of time.
- Short circuit (i.e. close the contacts) to suddenly put a high current through the ballast for a very short time.
- Open circuit (i.e. open the contacts) quickly, so that the inductive kick from the ballast causes voltage to build-up over the fluorescent tube, causing the gas to ionize, and thus allowing current to flow through the tube and produce light.
- To repeat the above processes in case the discharge through the lamp fails.
This is necessary, as most tubes have a striking voltage higher than line voltage, and thus cannot start without an inductive kick. The earliest fluorescent tube systems had a “manual” start switch position where a human could “toggle” the switch and act as a starter. Most units since then, rely on a starter to automate this procedure. Further to that, the starter needs to avoid failing “welded closed” or it could cause a fire risk.
An easy way to automate this procedure is to use a gas filled tube with bimetallic strip contacts, which comprises the majority of starters. The gas has a breakdown voltage less than the line voltage, so applying the mains will start the “glow” discharge. This glow allows a limited amount of current to flow through and “warm” the bimetallic strip electrodes which begins the contact closing procedure. Once the contact closes, the arc extinguishes as the voltage across the starter falls to zero, and without the arc to warm the bimetallic strip, the contact then separates abruptly. Voila!
That being said, the design and the gas mixture used in the starters vary as a function of manufacturer. Further to this, because of the switching involved, this can cause radio frequency “bursts”, like from a spark-gap transmitter, to be emitted during starting, thus most starters also include an RF suppression mylar film capacitor.
Another thing to note is that the traditional starter described above is not the only type of starter available. Some are purely electronic, involving no gas tubes, and are more expensive but feature more advanced starting procedures, such as the Arlen Pulsetarter.
A Teardown and Test of Starters
Fluorescent starters come in a standardized “can” shape enclosure, which opens easily by using a slotted screwdriver and prying at the base. The gas tube is clearly visible inside a glass envelope, and is normally connected internally in parallel with a mylar film RF suppression capacitor, which I have clipped off the right two. The base is normally either paper-card PCB substrate, bakelite or even plastic, with crimped studs.
Three different models of starters are shown above, from left to right, we have the DETA 6480 “generic” style starter featuring a bent U-shaped bimetallic strip contact, the Clipsal FE-U Cat 700 universal starter with twin “parallel” hooked bimetallic contacts, and the Philips S10-P Performance starter with asymmetric twin bimetallic strips, one with a hooked contact with a clear blue plastic can.
In order to see what happens when high voltage is applied across the contacts, I used my old Tenma Insulation Resistance Tester to subject each starter to DC. When DC is applied, only the negative terminal glows, thus we need to reverse the polarity to see the other electrode glow.
We can see the DETA branded starter has a mixture which glows a light blue. The curved bimetallic strip electrode with a “tip” seems to be a standard construction for lower-cost generic starters as it reduces the material to one hooked bimetallic strip.
The next starter off the list is the Clipsal unit, and it seems to glow a darker, deeper blue in comparison. It’s not a very bright glow, which is a little disappointing.
Now, for the surprise that is the Philips Performance Starter.
The Philips Performance starter has a different asymmetric strip layout, and you can see that it glows a nice pinky orange, sort of vaguely like neon, but not quite. In the right-most sample, it was being driven by an AC source, and you can see both electrodes glowing. Because of the heat from the glow, you can see the contacts are almost touching, as opposed to when cold on the left-most image.
As a result, if you want a nice glow, you should opt for the Philips S10-P Performance Starters. As an added bonus, the clear can will allow the glow to be admired from the outside without putting people at risk of touching high voltages, which is doubly cool. I managed to order a box of 25 from overseas (which are genuine, as checked with go4real.com) … not that I really needed that many, but it was cheaper that way.
The Rest of the Bits
Of course, to make these things light up, you really need a high voltage AC source of some sort. The discharge doesn’t start happening until about 185v (worst case) for these starters according to the U1461A, so we’d need at least that much.
Now, I could make an exercise for myself to build some sort of primitive inverter to drive it, but then I saw one of these online for AU$4.04, and I just said “screw it”.
This is a CCFL (cold-cathode fluorescent lamp) ballast, designed for 12v DC operation. This is normally used to drive “ring” lights around headlamps, and linear 300mm tubes used in computer case-mods. CCFLs are generally driven quite simply with a limited current and high voltage AC, pretty much what we need.
The unit claims a pair of 900v outputs, with a current of 5-8mA, at 50Hz. All within the ballpark we want, although it pays to remember that it is current limited, so it is possible to run a lower output voltage provided it is not so low as to fault the inverter itself.
The unit can be secured down with screws. The cover at the rear is clipped in, so lets take a peek inside …
As with most high-voltage devices, this one is almost completely covered in potting compound to keep out moisture and ensure reliability. That seems like a good idea, and a better result than I would have expected for my AU$4.04. Just be a bit careful around the outputs – you can get a nasty zap from them.
Prototype One: Go Big or Go Home!
Disclaimer: I don’t take any responsibility for whatever you may do, and whatever loss or damage which may be incurred directly or indirectly through following these instructions, inability to follow these instructions, omissions on my behalf, or negligence.
Lets start off with the adage – “go big, or go home!” That’s exactly what I decided to do, scavenging an old weather-proof box as a mounting base for everything, and scavenging resistors to keep the unit happy. The schematic looks as follows – pretty simple.
As the starters come in standard bases, I ordered some cheap starter holders online, which turned out to be AAG Stucchi units with push-fit wire connections from Italy. Since I didn’t know how to use the push-fit connectors, I got the wires in the wrong holes (oops!) but it still works. I decided to use some terminal connector blocks to make the internal connections, so I don’t have to solder (which is pretty rare for me), and I even got out the cordless drill and screws to make it all look neat.
Then I discovered, when I powered it on, absolutely nothing happened. As it turns out, it seems that the RF suppression capacitors were shorting out or causing the current limits to be exceeded on the CCFL inverter, so each of the starters had to have its can pried, and the capacitor removed using a pair of side cutters.
Initially, I did not have any resistors in series with the starters, so they started doing their dance.
While this is beautiful, it also throws out a crapload of radio interference, in essence, operating as a spark gap transmitter and will cause you and your neighbours headaches. As a result, you need to dial back the current to the glow so it’s not warm enough for the bimetallic strips to make contact. By experimentation, I chose 330k ohms which is incidentally about the same value I would use if I were to hook up a starter can across the mains to achieve a similar effect.
Nicely, I also discovered the CCFL inverter is short circuit protected, and will operate down to about 3v, although with severely diminished output voltage. Once fully constructed, this is what I get …
… pretty cool, although it’s probably something only a nerd can appreciate. It draws about 0.4A at 12v, and is understandably, not so bright. It would be even dimmer using blue-emission starters.
Just make sure not to remove any of the starters in operation – you will take out the whole series string, but you will also get sparking/arcing due to the high open-circuit voltage developed by the CCFL inverter.
Prototype Two: Smaller is Definitely Cuter!
Then, I had a thought – 12v and a big box is hardly convenient for a little glow tube action. Why not make a smaller one? As a result, realizing the inverter will actually run off 5v, and will provide sufficient voltage to strike a starter, I decided to use a CCFL inverter with just one starter on each channel. When supplied with 5v, the current is low enough not to even need a limiting resistor, which is convenient. I decided to cannibalize an old USB cable for the project.
Then came the issue of an enclosure – and then, I remembered I had a spare Raspberry Pi enclosure from the original Raspberry Pi series which I had no use for. By removing the potted module from the plastic enclosure, I could stuff it into the Raspberry Pi enclosure and hot-glue everything together. Hence this Frankenstein creation …
… isn’t it cute? It requires just 220mA at 5v, suitable for even the wimpiest USB ports, and gives you a pair of glows that should last a very long time. Just make sure you don’t remove any of the starters (same with the above) otherwise expect arcing/sparking from the connectors.
Turn off the lights, and enjoy the night-light ambiance …
I suppose this is an innovative way a geek could end up using some “off the shelf” components for things it was not intended for. It definitely gives me that warm fuzzy neon feeling, with its soft glow. It’s pretty unique as well, as I haven’t seen anyone else with a batch of glowing starters on their desk …
… call it a slice of Vivid on my desk … albeit a monotone one. It’s pretty easy to build your own too.
Just thought I’d throw in a note here at the end of the article – I’ve been extremely busy as of late with research for my PhD, so despite having a pile of stuff I really want to get online, I haven’t quite got the time. I’ll still try my best to use my “spare” time (or make some) to put up the occasional article.
Traffic to the site has been sporadically heavy, and in those periods, visitors may experience erroneous 404 messages with “Page not found” errors. These errors stem from the site being exhausted of resources, and often, the link is correct. If you do experience these errors, please be patient, and try reloading after 20 seconds or so. At this stage, this only seems to affect < 1 in 600 visitors (sometimes myself), so a resource upgrade isn’t yet justified. I will continue to monitor the situation.