Power indicator LED in High Voltage Circuit

[MikeMl]

Here is how I did it. Note the use of two resistors. Each dissipates ~0.52W, so 1W rated resistors are ok. Note that the voltage at the ends of the LED are near ground potential (in the US). This is what I had in mind when I posted this:

 

[MrAl]

Steven, Mike,

Mike:
That circuit needs a series diode as well which will halve the power dissipation without losing any light output. With only one anti diode the resistors have to work twice as hard for the same light output. Adding another diode in series corrects this and so the power in the series resistor drops by half from about 240^2/R to about 240^2/(2*R).

Steven:
I'm sorry i reported the wrong value for the power of the LED circuit i was using. It's actually half what i wrote before because it is half wave rectified with one diode in series and one in anti parallel.

So for 120vac operation with one diode in series with the LED and one in anti parallel to the LED the power is:
PTotal=120^2/(2*R)

so for R=82k we get 87.8 mw total, and for R=100k we get 72mw, both of which are under 100mw.

If you wanted the same current in the LED at 240 as i have at 120 and with 100k, then you'd have to use a 200k resistor and this would lead to a power of 240^2/(200000*2)=144mw total.
To get the equivalent current as with 82k and 120v you'd need a resistor value of 164k which would lead to a total power of 176mw.

If we use full wave rectification we get more light output but then we use more power too.

Also keep in mind that the resistor should be oversized in power by a factor of 4 to keep it very cool, but 3 might work too. I use a 1/2 watt resistor with less than 100mw actual dissipation.

 

[MikeMl]

Mike:
That circuit needs a series diode as well which will halve the power dissipation without losing any light output. With only one anti diode the resistors have to work twice as hard for the same light output. Adding another diode in series corrects this and so the power in the series resistor drops by half from about 240^2/R to about 240^2/(2*R)...

I have lots of 1W and 2W resistors; I would have had to buy some 1000PIV rectifiers...

 

[MrAl]

Hi Mike,

Yes, and they are about 7 cents each

 

[MikeMl]

Hi Mike,

Yes, and they are about 7 cents each

Oh, but the aggravation of having to get on the internet, or phone, and pay shipping... I get a perverse pleasure of doing a project out of what I have in the junk box...

 

[MrAl]

Hi Mike,

No problem there Mike, i just wanted to point out that you can keep your resistor cooler with the extra diode. If you use one it's not hard to use two. It sounds like you are telling me that you only own one single diode and that's hard to believe

Also, i was playing around with the Neon Bulb formula and found that we can get a 'decent' estimate by making the turn on voltage VH equal to the turn off voltage VL. This means instead of using say VH=90 and VL=65, we just make VH=VL and then make VL=65 volts for example. The new formula comes out simpler to:

PTotal=-(Vpk^2*asin(VL/Vpk))/(2*pi*R)+(Vpk^2*acos(VL/Vpk))/(2*pi*R)-(VL*sqrt(Vpk^2-VL^2))/(pi*R)+Vpk^2/(4*R)

This should be accurate to about +0 percent -10 percent for most common voltage levels. It's not exact though and assumes Vpk is quite a bit higher than VH, so for Vpk=170 and VH=90 and VL=65 it comes out pretty close but we only have to use Vpk and VL then. So in other words the time spent between VL and VH just before turn on does not add too much to the power unless the peak voltage is comparable to the turn on voltage. This is never going to happen for most applications. This also means the new formula will always yield a power dissipation that is a little less than actual but never more.

 

[Paul_L]

There are lower voltages that are accessible in vacuum tube amplifiers. I worked in the AF we had an old radar (1942) the source of several power indicators was from the tube filament voltage right off the secondary of the power transformer (AC volts). They were incandescent bulbs (GE47) I enjoyed this particular system it was decommissioned in 1993.

 

[MrAl]

Hi,

I thought the filament voltages were either 6.3vac or 12.6vac in the old tube stuff?
I used to have a couple filament voltage transformers, usually made to put out fairly large currents like 10amps or 20amp to run all the filaments in the product. It's amazing how things have changed where we never need that anymore, at least not in most stuff.

 

[Paul_L]

We had over 700 vacuum tubes in this old radar. Most were 6.3 and 12.3 volts.

 

[steven7890]

So if I put neon bulb NE2 with a resistor R, then the total power consumption is (see #66 above) PTotal=-(Vpk^2*asin(VL/Vpk))/(2*pi*R)+(Vpk^2*acos(VL/Vpk))/(2*pi*R)-(VL*sqrt(Vpk^2-VL^2))/(pi*R)+Vpk^2/(4*R), setting VL at 65V
For 240V AC and R = 100K, PTotal = 0.435W or 435mW
For 240V AC and R = 220K, PTotal = 0.198W or 198mW

If I use one diode in series with the LED and one in anti parallel, plus one current limiting resistor R, the LED circuit power is PTotal=Vrms^2/(2*R), see #62 above
For 240V AC and R = 82K, PTotal = 0.351W or 351mW
For 240V AC and R = 100K, PTotal = 0.288W or 288mW

Thanks to MrAl

 

[MrAl]

Hi,

You're welcome steven and i am glad you can use this information. I am not sure what kind of brightness you are looking for, but i found the lower current levels we are talking about here good enough for indicator lamps which dont have to be super bright like flashlights. Usually if we look at the indicator and we can see that it is lit up in normal room lighting then that is considered good enough. I have no problem seeing these indicators when used with high brightness LEDs, probably the ones rated 5000mcd or above.

I also used a multicolor red/blue/green LED for an indicator on a hot glue gun. BLUE for low, GREEN for high. Very nice looking and surely bright enough.

 

[Scotophor]

Then we have the zero diodes version. This requires two LEDs however, connected in anti parallel. The series resistor always conducts, but when it conducts it always produces light in one of the LEDs.

Sometimes capacitance is included to help filter the DC and so supply a smoother current to the LED which reduces motion flicker.

Also, sometimes a series capacitor is used to help reduce power consumption and get higher brightness. This is a bit more complicated and is usually not needed in an indicator if the LED is a high brightness type.

There was another type (actually a subset of your "zero diodes version"), but the LED type required don't seem to be available anymore. I assume you're familiar with 2-lead bicolor component LEDs (usually red/green, but other combinations are available), which are actually two LED dice in anti parallel in a single package -- switch the polarity of applied current to change the displayed color, or apply AC to view a flickery mix of both colors. Some years ago there were available, LEDs made exactly the same way, but having two dice of the SAME color in anti parallel. One of those would also require only a series resistor to operate from mains AC power -- if you could find one.

Another method that hasn't been mentioned yet in this topic, is the use of a series capacitor instead of a series resistor, to limit current when powering an LED from AC with much reduced wastage as heat.

EDIT 13 December 2014: I went on yet another hunt for the bipolar, unicolor LEDs -- lo and behold, I found some! As it turns out, such construction is actually becoming common with surface-mount devices, but Kingbright still makes them in through-hole packaging. Their part numbers L-57IID (GaAsP/GaP high efficiency red), L-57GGD (GaP green), L-57YYD (GaAsP/GaP yellow), and L-57SRSRD (GaAlAs super bright red) are obsolete non-RoHS parts but the popular colors high efficiency red, and green, may be available from remaining stocks from some suppliers, at least according to the UK website of **broken link removed** where I first found them. Replace the "L-" in each part number with "WP" for the current-production, Pb-free version, and you might be able to find some more suppliers of them, such as **broken link removed**, who list 6 part numbers (in both 5mm, T-1 3/4 and 3mm, T-1 sizes) but only stock two of them (red and green 5mm). Kingbright's numbers for the 3mm size are WP-937IID (high efficiency red), WP-937GGD (green) and WP-937YYD (yellow).

 

[MrAl]

Hi there Scotophor,

It was good of you to bring up these two interesting ideas. Yes, the ready made bicolor LED is an option. We havent talked about the series capacitor yet because i assumed that the readers here so far did not want to go through that trouble. It's a good idea however.

First, the bicolor LED with two LEDs in antiparallel.
I have one of these somewhere from long ago, but i can tell you that with this thing you would not want to use it as an indicator. It's red and green, and when the colors are both displayed (AC) it looks like, well, a dead fish
Crummy amber color, nothing like the nice bright yellow we can get today.
Also, it is not high brightness meaning we would have to use a lower series resistor or move to the series capacitor idea you mentioned. It's doable i think but i dont like the color of this type. Perhaps a more modern version if they make one with better colors.
For a better example of nice colors see the tri color LEDs. These things have beautiful color, at least for the three pure colors. Some colors mixes look nice, some dont. They would be harder to drive though because they have common anode or cathode to all three LEDs. So no back to back connection (anti parallel again) possible there, too bad.

The series capacitor idea is a good one if you dont mind going through the extra trouble of using the right kind of cap and building in some circuit protection against the cap shorting out on failure. A small bridge rectifier and it should work nice.
I actually built a couple of these just to see how well it would work. I can get the full 20ma that way. Right now i dont need that much brightness so i dont use them, but i could see how someone would want to use them for various things.
I was sure to add a surge protection resistor and zener and fuse too, things that make it safer to operate. I used to have a full blown schematic on my web site until AOL decided to kill AOL Home.

 

[Scotophor]

See my edit above - bipolar, unicolor LEDs are in fact still available.

 

[MrAl]

Hi,

Oh so you want the same color for both forward and backward LEDs. That's cool too

 

[Scotophor]

I ordered a small supply from Digi-Key, mainly for replacing neons as they fail but a few of them might find their way into some of my new projects too.

 

[MrAl]

Hello,

The dual color two lead LEDs i have used in the past where the red and green. The mix of colors didnt work out well though when used on AC because the color was a dim 'amber' color which was not that great. I prefer the yellow LEDs as they can have a nice bright yellow appearance rather than a dim looking mud yellow

Some colors seem harder to get too, like orange, when mixing colors of more than one LED like with the tri color LEDs. Aqua isnt too hard to get but does take some balancing of colors by adjusting each LED (red, green, blue).
They should include yellow (red, green, blue, yellow) for four colors.

It's also not hard to wire two in anti parallel. That way you can use any color and brightness you need. I like to use a very high value for the resistance so that the power dissipation is very low (little heating of the resistor) so i need high brightness so the low current lights it up enough to see without any problem.

 

[schmitt trigger]

Fond memories....#47 indicator lamps.

 

[schmitt trigger]

Fond memories....#47 indicator lamps.