which method should I use for C7 DIY LED window candles?

2) Once you've decided on your current , add a polyswitch per path. To make a bit more failsafe add some heat insulation around each polyswitch and reduce the thickness of insulation if they falsely cutout. If you don't want to go overboard with current limiting circuitry

For 4 leds in parallel you want to *increase* the current. Assuming each led has an equal voltage drop, the current is divided among the leds, so at 10ma total, each led is getting 2.5ma. I'd go with a series configuration so you don't need a matched voltage. In a series configuration each led gets the same current with the voltage divided among them.

Since you have leds to spare, try increasing the current for a single led until it fails, this gives you an approximate upper bound on the current you can use.

** IME leds do not fail in any simple way when driven with excess current.

Years ago, I repaired Mesa Boogie guitar amp that was almost new. The owne r had purchased it in the USA so it ran here from a 120V step down transfor mer. After a few months, it developed multiple strange faults: the volume l evel would suddenly drop and then return, tone settings varied all over the place in a similar way and the problem got worse the longer the amp was ru nning.

Another tech decided it must be due to bad output valves an changed the lot - for no good reason. The fault remained and then it came to me.

After verifying the owners weird story, I removed the chassis and eventuall y spotted something odd. The changing behaviour of the amp corresponded wit h varying DC voltages on the leds that were part of the many Vactec opto-co uplers in this model. The Vactecs controlled tone settings and channel swit ching.

The 5VDC supply to the leds was quite steady so it must be faulty leds - bu t all 20 of them! How ?

Due to a screen printing error on the PCB, resistor positions that should h ave been 470ohm were installed with 47ohm parts increasing the led current to 50mA or more. The leds tolerated this for a couple of hundred hours then began to fail by dropping light output intermittently. Some would go dark and then come back minutes later, when dark the full 5V was across the led terminals.

New leds and resistors fixed it completely.

Naturally, diagnosis and repair took a fair bit of time (though leds and re sistors are cheap) the owner got a substantial bill which she was quite gru mpy about.

Because the amp was purchased in the USA, there was no warranty cover here in Australia. The tech who unnecessarily replaced the output valves worked for the warranty agents here in Sydney and she had paid a big bill for that work too. Her bargain priced Boogie turned out to be no bargain at all and she had to carry around a 6kg step-down tranny too.

.... Phil

The brightness of each bulb is proportional to the current thru that bulb no matter what the configuration. Your numbers represent a 4:1 (or 16:1 if that's 10mA total for all 4) difference in current thru each bulb with a 2:1 difference in the number of bulbs. So, the answer is "no" but the relationships aren't entirely linear.

The hardest part of any project is setting the objectives/specs.

My favorite system architect, Beyonce, said it best: "If you want it, put a ring on it." You need to tie down the objective. "Longest Possible" is not quantifiable and a useless constraint.

Sounds like the real objective is, make the wife happy.

Since we know zero about the LEDs, I'd assume that 17mA is a good number for reasonable life. 40mA is probably too much.

First, decide if you have 12V. If it's a regulated wall wart, you will. If it's one salvaged from a 20 year old screwdriver, it won't. Adjust the resistor to whatever it takes to get the current right.

See if there's anything you can do with tinfoil and LED positioning to aim the light where the wife wants it.

Beyond that, start adding lights at 20 mA until the wife is happy or you run out of space. With 12V, you can put up to three LEDs in series with appropriate dropping resistor to set the 20mA current. At four, I'd put two sets of two, with a separate dropping resistor for each series string. For maximum life, you probably don't want to connect LEDs directly in parallel.

Depending on the size of the lights you bought, you might find that the 27 LED's from a Harbor Freight free flashlight are smaller and can be packed in more tightly. Or you can buy unpackaged LED's dirt cheap. Just because you have Xmas lights doesn't mean that it's an optimum means to achieve your objective. Xmas lights are designed for low cost. They run for a few hundred hours and nobody worries much about whether they run at all next year. If you want high quality candles, it's better to start with high quality components.

Remember that, for most electronic stuff, reducing the temperature by 10-degrees C doubles the life. And that's the CHIP temperature, not the ROOM temperature. Chip temperature is proportional to current. If you pack stuff too tight and raise the temperature, you're defeating your purpose. While you can certainly buy single LED's that are bright enough, they're also hotter and will likely require some heat sinking. And the resistors add heat too. There is no free lunch.

If that's not enough brightness, you have to make a tradeoff. Shortening the life by increasing the current increases the wife's happiness. Anybody who has the slightest indecision about which way to go has never been married.

The lights probably came with sockets. Use the sockets and just replace them when they burn out. Spin it like a politician. Every time the light burns out, you're a hero for fixing it. Maybe you want them to fail every

28 days...be sure to get the phase right. ;-)

Bottom line is that it's not about the technology. Your tradeoffs are about the benefits of a happy wife.

Thanks for the tip! A "polyswitch" wasn't something I was familiar with until I looked it up. It looks like a fuse that resets itself once a fault is corrected, is that right?

The configuration I've gone with is four LED's for each C7 bulb. The LED's are driven in series by a 12 V power supply with current set at 10 mA.

I have a current resistor set for each four LED candle since each one drew a different current amount with the same resistor. All set for 10 mA.

The question is, where do I place the polyswitch and how do I determine what amperage/ voltage I need? For two rooms, I have a 12 V supply powering two candles, with each candle set as above (3V @ 10mA) and in another room, I have four candles powered by a 12 V supply. Ideally, I'd like to place the polyswitch at the power supply output before it branches off to either the two or four candle arrangement.

Any tips here would be helpful.

Thanks, Sam

The reason for that was mentioned earlier. 12V is not enough voltage to reliably run 4 in series. Two sets of two in series with two resistors will be far more stable over temperature and with variations in led characteristics. The cost is twice the energy consumption.

You might find that 3 in series at 15mA gives you more light and only

50% energy penalty. It's always something...

On the surface, it seems silly to replace incandescents with LED, then use an incandescent as a current limit. BUT Incandescent lamps make excellent positive-temperature-coefficient resistors. You just use a 12V lamp. Pick the lamp current so that, when you short the output, the lamp doesn't overload the power supply. Can get them at any auto store and they last forever in this mode.

The polyswitch has a hockey-stick resistance curve and is clearly a superior device for this application. But you might already have a bulb in a drawer somewhere. I just measured a 1458 lamp. It drops 0.44V at 40mA so you'll need slightly smaller series resistors...or just not worry about it. Probably won't work well with 4 LEDs in series. Higher current incandescent will have lower voltage drop at 40mA

You will still want to use two sets of two series LEDs/candle. There will be some interaction between candles if you plug/unplug one.

And you get a "short indicator" for free.

Life is a tradeoff...or two...

I wasn't clear on your reason. So far, so good. The 12 V input divides down to 3V per bulb in series with each holding at the 10 mA level.

I like the lamp idea and actually forgot about it. I used to use such a set up with a high voltage circuit. Been so long ago that I forgot about it. I'll have to see how much total current the circuits use and then try out a bulb. I don't see any reason I couldn't place the bulb at the power supply point before it branches off to the 2-4 candles. Would certainly be a lot easier than trying to wire one up per candle. Thanks for the idea here!

So you're saying I would have been better off with two sets of two LED's in series per bulb. A series-parallel combination, but that would have required more resistors to set the current correctly per set.

I'm saying that if you start with 12.0 volts and drop .44V across the bulb and need 3V per LED...you don't have it. The purpose of the resistor is to normalize the current given variances among the LEDs. With 4 in series, the resistor gets so small that it doesn't normalize very well.

Sure, the candles may light, but you'll get a different intensity per candle and the resistance of the bulb will create interaction between the candles.

I'm suggesting that if you put two LEDs in series with a resistor. put two of those series combinations in parallel in each candle put the candles in parallel put one bulb at the power source You'll be at the sweet spot of low cost, long life, fault tolerant, happy wife and just works.

In reality, you can probably tweak it to work with 4 in series.

What I'm trying to teach you, and people reading this next year, is the importance of system design where the performance is largely independent of the parameter variations of the devices. And single point failures will not set anything on fire. And if you built a thousand of them, they'd all perform the same. And if you substituted a red LED, it'd still work the same. And if you put it outside in below freezing temperatures, it'd still work the same. And if your power supply was 11V or 13V, it'd still work the same. And the margins are so wide that you'd never have to test any of that. It will just work. The obvious tradeoff is that it takes twice the energy to run it.

We haven't discussed your wall wart. If it's electronically regulated, it's likely current limited to a level that protects itself and won't supply enough current to set your house on fire. If so, forget the bulb/polyswitch and let the wart take care of the current limit.

If you wanna discuss more "clever" designs, we can do that. If you want a bullet-proof design that anybody could build using random LEDs without any test equipment or exotic devices, this is my recommendation.

Are we having fun yet?

For me, this was getting too complicated for the short time I have off from work, and time available to complete this project. Tried the bulb in series first, but the one I found-- 12 V @ 1.4 W-- pulled down the current by 1/3 so it dimmed the bulbs too much. Other bulbs to try are too expensive to keep purchasing, so I just went with 100 mA fuses. Problem solved.

Thanks for the suggestions though, may come in handy for similar projects in the future.

I would put the leds in series. Don't worry about the lifetime, it far exc eeds a few hundred holiday seasons. The leds should be good for 50,000 hou rs if you don't run them close to their max rated current. 10 hours per da y x 30 days x many many years to get close to 50,000 hours lifetime.

t.

ner had purchased it in the USA so it ran here from a 120V step down transf ormer. After a few months, it developed multiple strange faults: the volume level would suddenly drop and then return, tone settings varied all over t he place in a similar way and the problem got worse the longer the amp was running.

ot - for no good reason. The fault remained and then it came to me.

lly spotted something odd. The changing behaviour of the amp corresponded w ith varying DC voltages on the leds that were part of the many Vactec opto- couplers in this model. The Vactecs controlled tone settings and channel sw itching.

but all 20 of them! How ?

have been 470ohm were installed with 47ohm parts increasing the led curren t to 50mA or more. The leds tolerated this for a couple of hundred hours th en began to fail by dropping light output intermittently. Some would go dar k and then come back minutes later, when dark the full 5V was across the le d terminals.

resistors are cheap) the owner got a substantial bill which she was quite g rumpy about.

e in Australia. The tech who unnecessarily replaced the output valves worke d for the warranty agents here in Sydney and she had paid a big bill for th at work too.

Can you blame people for wanting to just buy new?

Update as of May 26, 2015: Well, using 4 of the warm white C7 LED bulbs in series @ 10 mA current should have done the trick, but my resulting lights have started burning out one by one. Lost two so far and another is flickering. I can't figure out why, current is low and they didn't seem to generate much heat to speak of. Also, when I was just using a single bulb @10mA, none of the candles burned out for over 3 years and even then when I decided to brighten them by using 4 in series, none of the original singles had burned out. Any idea why the four in series are slowly failing with same current and voltage as the single lamps were? I'm disappointed in this for sure.

Sam

Part of the cooling comes from the mounting wires, and if the leds are close together, they also get hotter. That is bad. Also a real current source instead of a (small??) resistor might increase survival, because voltage drop for each led is rather dependent on temp/current.