This project updates the Neon lights in my boat's shorepower panel with LED versions. The lights are used for indicating NORMAL and REVERSE wiring conditions. The boat is a few years old, and over the years, the Neon lights have lost some of their brilliance, so much so that they are barely visible. So I decided to replace them with more modern LED indicators.
There are potential lethal voltages and fire hazards connected with this project. This project is not UL tested or certified. Anyone considering this project must acknowledge that www.boat-project.com or its authors are not liable for any action that you may take in connection with this project. This information should be considered amateur in nature, and should be used for entertainment purposes only.
BlueSea (P/Ns: 8034, 8066, 8169) and PanelTronics (P/Ns: 048-066, 048-016, 048-017, 048-021, 048-022, 048-023) make 120VAC LED panel lights, and they could be used for this project. However, their mounting hole dimensions are 5/32 to 1/4" dia", which is smaller than the 3/8" hole the Neon lights mount in. So they just won't work.
After I searched the electronic supply houses (Mouser, Allied, etc) I did find some 120VAC LED assemblies, however, they are quite costly - up to $25 per unit. As I need 4 of them, that is a hefty price for indicator lights. So, I made the decision to make my own, as I can buy bezels and LEDs in the size that will fit into the holes. I am estimating $2 to $5 for each LED assembly, which is a lot better than the alternative.
First a little LED theory:
A Light Emmitting Diode is similar to a generic silicon diode in that it is basically a PN junction, and can provide a rectifier function. In reality, the only major differences are its forward-bias voltage (1.7V to 2V), and low Peak Inverse Voltage (PIV), of anywhere between 5V and 30V. PIV is rarely a concern in DC electronic circuits, but becomes a factor in 120VAC circuits during the reverse biased half of the AC cycle. Applying 120V across the diode when it is reverse biased typically exceeds the LED's rating, and consequently will damage/destroy the LED.
Since a LED is a current device, all that is really needed is to provide a means to limit the current, which is easily enough done with a resistor as shown here:
In this example, when the LED is forward biased (top drawing), about 3mA of current flows through the resistor and LED, thereby illuminating it. A current of 3mA is sufficient for most T-1 3/4 sized LEDs. Since the LED is a diode, it's self-rectifying. A half-watt resistor is required, since 3mA @ 120V is around 300mW.
But notice what happens on the reverse bias cycle (bottom drawing). The LED does not conduct, so no current flows in the opposite direction (well, other than a small amount of leakage current). Consequently, the resistance of the LED is much higher than the resistor, so almost all of the 120V supply voltage will be across the LED. And since they have a reverse bias (PIV) rating of as little as 5V, you can expect the LED will soon be damaged.
The solution then is to prevent the LED from having 120V applied across it during the half-cycle it is reverse biased. One method of doing this is to use a standard silicone diode clamped across the LED, with the anode and cathode in the opposite direction. I am using a 1N4004 diode here, which has a PIV of 400V, but that is really not needed as the LED will complement the 1N4004 by restricting the voltage to about 2V when the 1N4004 is reverse biased.
Since the silicone diode has a characteristic forward biased voltage of about 0.7V, the LED will be limited to this voltage in it's reverse biased state. Notice here that current now flows in both directions. Also, the current flowing through the 1N4004 when the LED is reverse biased is wasted current, but its only 3mA of wasted current.
The observant will notice that the current flowing through the circuit will be slightly different in the two half-cycles, as the LED drops 2V when it conducts, and the 1N4004 drops 0.7V when it conducts. But this variation is minimal, and well within the current handling capability of the resistor.
Since the LED is conducting for half-cycles only, its effective frequency is 30Hz, which if you look closely enough, a slight flicker can be observed. This flicker is not normally noticed however. But if it is an issue, you can always use a bridge rectifier in place of the clamping diode. This has an additional advantage that there is no wasted current as the LED will be on for both halves of the cycle. This concept is shown below:
The conducting path for each half-cycle is shown by the Red and Green arrows. Note that irregardless of which direction the current flows into the bridge, the current always flows thorugh the LED in the forward biased direction - which illuminates it.
A popular kit manufacturer, Velleman, makes a product that would be satisfactory; although you would have to change the LED to whatever color you wish, as it comes with a blue LED. It uses the full-wave bridge rectifier approach, but adds a capacitor in series to help flicker and to reduce voltage (which also limits current) to the LED. Since Velleman is an European company, this kit will work on 50/60Hz, 120/240VAC. At $5 bucks per kit, the price is right. If you are interested in this, look for Velleman "AC Power Voltage LED" kit MK181.
However, I am going to stick with the more simple half-wave approach as was originally discussed. To test this circuit, I am going to build a little test mock up. I want to verfiy the resistor does not overheat, and I want to ensure that 3mA is sufficient to illuminate the LED that I have selected. Here is the final circuit:
The fuse was added on the test circuit so that I would not inadvertantly burning my house down. Again, the resistor is a 1/2 watt resistor, and I used a 1N4004 diode which has a 1A current capability and a 400V PIV (both of which are much more than required). But the cost for the diode is literally pennies.
Building the Test Circuit. This is a very simple circuit, and the components are simply soldered together. The only real issue is to ensure you get the Cathode on the LED and 1N4004 wired in opposite directions. After that, it matters not which side the resistor is connected to, or which side of the AC power line the wires are connected.
To remain somewhat safe, I mounted the circuit in a NEMA 5-15P plug, one that has enough interior space to house the few components.
Successful test. I am pleased that even at 3mA, the light is sufficiently bright. And since the lights will be in the cabin of the boat, they will not be in direct sunlight. And they are certainly brighter than the current Neon lights. I also was able to confirm that after hours of operation, I could feel only the slightest warmth in the resistor.
I'd say the test was a success. The next step; order the parts for the modification.
For the production system, I am going to create two modules; with each module supporting two LEDs. The modules will contain all of the parts except for the LEDs themselves. Since I am going to all the trouble of creating modules, I decided to go with full bridge version. I figured, for 20-cents in additional diodes, I can take advantage of the reverse-biased cycle and have a brigher LED with less flicker. Its not uncommon for a project to take a left-turn such as this.
|For the prototype board, I used a small Radio Shack board with holes on 0.100 centers. The terminal posts are Keystone 4902, and I had to enlarge the holes just a bit for the posts to fit. I will be placing the proto board into a Bud PB-1558-BF potting box. This box is 1.5"x2" so you have to work small.|
On the underside, I soldered all of the components (8 1N4004 diodes, 2 22K Ohm 1W resistors, and two 0.125A fuses). I went with 1 Watt resistors (even though they only need to dissapate 1/2Watt) as they will be encapsulated in potting compound, and I wanted to compensate for any thermal increase in the box.
This wiring scheme is known as point-to-point, and was actually the method of choice up until the late 1960s, when transistors and circuit boards became more popular. Until then, your grandad's tube TV set was probably wired in this fashion.
|Back to the top side of the proto board. The two green components on the top side are the fuses. They will be exposed after potting so that they can be replaced if necessary. Also notice that I offest the fuses to one side slightly. This was so that I could tell which side of the module is the input side (120VAC) and which side was the output (to the LEDs). Otherwise, after potting, you'll have no idea which terminal is what. Another technique would be to mark one side of the case with a dot of paint to identify the terminal orientation.|
|Now the potting compound pour. Potting compound is essentially epoxy, but with low conductive/low static properties. The potting compound has a 24 hour cure time, so you want to mix it up, let it sit for 20 minutes, then stir it again - which removes the bubbles. When pouring, it comes out pretty slow, so make sure you allow time for the potting compound to level out. Otherwise, you could end up overfilling the potting box. You don't want to cover the terminals or fuses.|
|And here is the finished module. Two sets of 120VAC circuits connect to the 4 terminals across the top, and the two LEDs connect to the 4 terminals across the bottom. Using spade terminals allows the connections to the module to be relatively protected, although it will be mounted inside of the boat's electrical panel. Notice that the solder-tail fuses are exposed, so they can be replaced if necessary.|