Bridge rectifier + capacitor for brighter non flashy LED Christmas lights?

[Red Squirrel]

Has anyone done this? Did you need to reduce the voltage (peak AC voltage is actually higher than 120) or add a resistor or anything or were you able to just plug it straight in?

I'm thinking of retiring the 7w C9's this year simply because the bulbs just don't last, and they're expensive to buy replacements for. So thinking of going LED. I just don't like how they're so dim and how they flash at 60hz and are so dim so I'd want to fix that.

I'd probably make my own rectifier out of some diodes from an old UPS, or maybe buy an actual bridge rectifier so I know what the rating actually is by looking at the data sheet, while if I build my own it's kinda guess work.

The premade ones seem surprisingly cheap and can handle quite a lot, ex:

http://www.digikey.ca/product-detail/en/GBJ2010-F/GBJ2010-FDI-ND/775736

Unless I'm not reading that info right, I can draw 20 amps DC off that thing? I imagine I need to put a heat sink on it though, but it would be inside a metal wiring box and I'd probably just use the actual wiring box as the heat sink, and considering this would be for Christmas lights it would also be sitting outside in -40's so cooling is not too big of a concern.

I'd then find a decent capacitor with like 200v rating and throw it on. Obviously the higher the capacitance the better as it will not ripple as much at higher currents. I'm not expecting to get a super smooth line here, just the full wave rectification alone will help a lot I think and capacitor will help a bit more.

Thoughts? Anyone done this before, anything I might be missing?

 

[MrDudeMan]

I need to see the lights you plan to run using this rectifier before I can tell you if it will work.

With that said, cooling is still a concern. Look at the derating curve on page 2. It can handle 4 A without a heatsink up to 100C and then it linearly decreases to 0 A at 150 C. That means the maximum power you can draw is approximately 500 W. That sounds like a lot, but it all depends how many lights you plan to string together. Also, relying on ambient air, even at -40 C, would be a possibly dangerous and definitely stupid mistake. The thermal coefficient of air is extremely low compared to the ability of that device to generate and output heat. Cold air will certainly help, but it's not a silver bullet.

I can tell you how big of a capacitor you need to make the flicker imperceptible when you give me more information about the lights and how many you plan to use. FYI: 120 VAC is the RMS value, which means peak is higher by a factor of SQRT(2), or 170 VAC. A 200 VAC rating on a capacitor would be insufficient for long term viability and you would probably fail at that rating once you took into consideration temperature, ripple currents, and material. I would shoot for 1.5x in this application at the very least. Keep in mind that capacitance can drastically change depending on applied voltage and material, which means you may need to actually look at the datasheet.

Don't build your own recitifier for this project. Use the one you linked or another similar model.

 

[Red Squirrel]

I don't actually have a specific set I want to use at this point, it depends whatever the store has. Probably something like this:

http://www.canadiantire.ca/en/pdp/n...ack-christmas-lights-multi-colo-0512316p.html

My fear is that they may split them in a way where they don't all take one side of the wave, if that's the case I would possibly need to do some rewiring.

I want to make this a rather generic plug where I can plug any set into it.

I could have sworn they had some where it was basically C9 LED bulbs that work on 120 directly and the bulb itself had some kind of step down circuitry (probably just a resistor). Those would probably be the best bet as I don't have to worry so much about how they work in series.

It is possible to buy retrofit LED bulbs for C9 sets, so perhaps that's my best bet.

http://www.amazon.ca/Faceted-Transpa...eywords=c9+led


As for cooling, do you think I actually need active cooling (fan) for the rectifier even if it will be at -40? I figured just connecting it to the wiring box so the whole box acts as a heat sink would be good enough. Idealy I'll want my box setup to be able to handle a decent amount of current so I don't have to worry too much about what I plug on the end of the light set. For outside I do have some garlands that will use incandescent bulbs. Though to be safe I can always plug those into a separate extension cord.

 

[MrDudeMan]

You don't need an actively cooled solution, but you need to realize the rectifier will get very hot if you try to use it to power a lot of lights without a heatsink. It's not going to be -40 C all the time so you can't rely on that. The junction box could work as a heatsink, but I'm not sure that's a good idea. Google to see if there's a specific issue with which to be concerned. My initial inclination is to tell you to buy a separate heatsink for various isolation and safety concerns. The tab should be electrically isolated, but it's worth reading to make sure there aren't additional issues to note and I don't have any relevant experience for this type of application.

The second set of bulbs are 120 VAC @ 0.45 W, which means each bulb is drawing approximately 3.75 mA. Red LEDs typically have a junction voltage of around 1.7 V, so this circuit is going to be massively dominated by the resistor in terms of voltage drop. Bumping the voltage to 170 VDC will cause 5.3 mA of current to flow, which is a fairly small change. Diodes aren't linear, so it's hard to tell what the brightness increase will be without knowing the LED specifications.

There's an additional issue to note when changing the voltage. Let's say the circuit is simply a resistor and an LED. The LED is dropping 1.7 V and the resistor is dropping 118.3 V, which makes the value of the resistor 31.5k. The power dissipated by the resistor is 118.3 V ^ 2 / 31.5 k, or 0.444 W. If you increase the voltage by 1.42x (170 / 120), the power dissipated by the resistor will increase by the same ratio, which means 0.63 W (0.444 W * 1.42) of power will be dissipated by each resistor. That's actually a lot more and uncomfortably close to a 1 W rating. I'm assuming each resistor is rated at 1 W, but it may be 0.5 W or even 2x @ 0.25 W. I would probably worry about this until I found hard evidence that each element/bulb/resistor/whatever could handle the extra voltage. Take one apart and figure out the values and ratings of the constituent parts if you're serious about doing a DIY project. You could wing it and probably manage just fine, but is it worth the risk?

 

[Red Squirrel]

Makes sense, now that you mention it I think the voltage increase could potentially (no pun intended) be an issue, even more so for a parallel strand.

I'm thinking to make this as universal as possible I need to see how easy it would be to bring it down to 120VDC. I imagine a regular light dimmer would not really cut it right? I could look into a regulator circuit but probably getting into something more complex at this point such as a SMPS. Could do linear but that's super inefficient.

I suppose another option is to adapt each light set individually to the device vs adapt the device to the light sets. So let the device dish out 170vdc and modify each set accordingly.

I do have a few indoor sets here I can experiment with. I have one here, it has 35 bulbs all in series.

 

[drebo]

Ceramic C7 lights are the best. The colors of the LED ones are aweful.

Ceramic or nothing, imo.

 

[Red Squirrel]

Ceramic C7 lights are the best. The colors of the LED ones are aweful.

Ceramic or nothing, imo.

They ARE nice, but those bulbs burn out all the damn time, and it seems nobody sells them in bulk. I have some C9's and they're great but I have like 10 burnt bulbs, going to cost more to replace them than to buy several LED sets. :awe:

I wish they would just have built in rectifier/caps and use higher quality/brighter leds (like in phone flashes, though maybe not THAT bright), but then nobody wants to pay extra and wants everything so cheap.

 

[MrDudeMan]

You don't need to do anything complicated to handle voltage translation. Converting 120 VAC to a higher or lower voltage can easily be done with a variable autotransformer (often called a variac). You basically turn a knob to adjust the output voltage while maintaining grid impedance and characteristics with respect to the load, or close enough that it doesn't matter.

Regulators, filters, and complex power supplies would work here, but why? A transformer can do what you need with no issues. Using a 3:2 transformer, you would get 120 VAC peak at the output instead of 120 VAC RMS, which would give you 120 VDC however crude it may be using a single filter capacitor. A variac would give you a range of voltages instead of a fixed value, but the concept would be the same. Note: some variacs can go above and below the input voltage (step down and/or step up).

A light switch dimmer may work, but the connectors will be harder to use. A variac usually has a standard NEMA 5-15/20 connector, so you can plug your lights into it directly.

 

[Red Squirrel]

Yeah was thinking a transformer, could go that route. Variacs are quite expensive but could get a small transformer. Though that will bring the rating down quite a bit, but probably not an issue if I strictly use it for LED lights.

I got brave, I just threw in a 1ohm resistor (I actually wanted to use the 1k and picked the wrong one LOL) in series and connected it directly to see what happens and it worked. No more flicker. Well there is one but I can't really see it unless I really look for it as now it's at 120hz instead of 60.

So just having the bridge rectifier (one I made) helps a lot. I think I will just skip the capacitor. Will this work as is with any random configuration of lights, or did I just get lucky? All I really did is inverse the second part of the wave, so it's still getting AC, sorta. The string is zero ohms though on it's own, which is really strange, I would have figured it would at least have some resistance. I guess this is where impedance comes in, that's something I'm not that familiar with. If I do turn it to 100% DC I'm guessing this is where I need to consider resistance. I may even be able to remove my 1 ohm resistor.... But think I wont push my luck here. :biggrin:

I'll also need to test to see what happens with a GFCI. I assume if I ground the negative the GFCI should still work, but I'll have to test. Right now it's just floating DC so it would isolate it from the GFCI. That's minor details I can deal with when/if I actually do this though.

 

[MrDudeMan]

Yeah was thinking a transformer, could go that route. Variacs are quite expensive but could get a small transformer. Though that will bring the rating down quite a bit, but probably not an issue if I strictly use it for LED lights.

Good quality variacs can be had on ebay for less than 30 dollars.

I got brave, I just threw in a 1ohm resistor (I actually wanted to use the 1k and picked the wrong one LOL) in series and connected it directly to see what happens and it worked. No more flicker. Well there is one but I can't really see it unless I really look for it as now it's at 120hz instead of 60.

Say the circuit is drawing 5 A. The resistor would be dropping 5 V, which wouldn't change the current draw at all. It can't be drawing anywhere near 5 A, though, because that would be 25 W of power dissipated across the 1 ohm resistor and I'm pretty sure you don't have an array of sand block resistors in stock. Feel free to take the resistor out as it's doing nothing.

So just having the bridge rectifier (one I made) helps a lot. I think I will just skip the capacitor. Will this work as is with any random configuration of lights, or did I just get lucky?

It should work the same with most light strands. You'll only run into problems due to power dissipation if I had to guess.

All I really did is inverse the second part of the wave, so it's still getting AC, sorta.

No, you have DC now, but it has no filter or regulator and, therefore, has a lot of ripple. The current always flows in one direction, which means it's no longer AC.

The string is zero ohms though on it's own, which is really strange, I would have figured it would at least have some resistance. I guess this is where impedance comes in, that's something I'm not that familiar with.

There's probably a reverse protection diode or some other circuit element that presents a low impedance path at DC. It's hard to say without seeing exactly what you did.

If I do turn it to 100% DC I'm guessing this is where I need to consider resistance. I may even be able to remove my 1 ohm resistor.... But think I wont push my luck here. :biggrin:

See above.

I'll also need to test to see what happens with a GFCI. I assume if I ground the negative the GFCI should still work, but I'll have to test. Right now it's just floating DC so it would isolate it from the GFCI. That's minor details I can deal with when/if I actually do this though.

A GFCI won't care about what you just did unless you're forcing current to go through an earth ground instead of the neutral wire. GFCIs trigger when a differential current condition is present because the hot and neutral wires pass through a toroidal inductor that produces a voltage when the currents are different in magnitude. As long as you used neutral instead of earth ground, a GFCI won't care unless there's an actual problem somewhere else.

Your circuit isn't floating. It has to be returning current to the grid through the neutral wire or earth ground (it's almost certainly not the earth ground, btw).

 

[rockyct]

I had a large, computer controlled light display for a few years (30,000 lights). Some thoughts:

Odds are you will have to modify the longer LED strings a bit. Most of the 70/100 count LED strings you buy at a store drive the half the string on the positive side and the other half on the negative side. You'll have to slice the string in the middle to get them matched up. I agree though, the cap isn't needed. You're doubling the time they are on and that's good enough to remove the flicker.

Just an FYI, but you can buy professional grade sealed LED strings with full wave rectifiers built in. You actually want sealed LEDs because replaceable LED strings often just rust out. Big box store incandescent strings are usually fine and the Christmas light nuts still buy them but they really stay away from store bought LEDs.

 

[Red Squirrel]

I had a large, computer controlled light display for a few years (30,000 lights). Some thoughts:

Odds are you will have to modify the longer LED strings a bit. Most of the 70/100 count LED strings you buy at a store drive the half the string on the positive side and the other half on the negative side. You'll have to slice the string in the middle to get them matched up. I agree though, the cap isn't needed. You're doubling the time they are on and that's good enough to remove the flicker.

Just an FYI, but you can buy professional grade sealed LED strings with full wave rectifiers built in. You actually want sealed LEDs because replaceable LED strings often just rust out. Big box store incandescent strings are usually fine and the Christmas light nuts still buy them but they really stay away from store bought LEDs.

Yeah that's a concern I have too and it's hard to know how they're wired until you take them out of the box, I don't really want to be splicing wire that may be going into a wet location. I've done it inside a computer for low voltage stuff but I don't really want to do it for something higher voltage that will get more movement. Though if I can find a place that sells heat shrink tubing I could use that instead of hot glue, electrical tape and tie wraps.

 

[rockyct]

Yeah that's a concern I have too and it's hard to know how they're wired until you take them out of the box, I don't really want to be splicing wire that may be going into a wet location. I've done it inside a computer for low voltage stuff but I don't really want to do it for something higher voltage that will get more movement. Though if I can find a place that sells heat shrink tubing I could use that instead of hot glue, electrical tape and tie wraps.

Heat shrink will probably give you a better seal than the seal around the base of an incandescent bulb so it should be good enough. There's always wire nuts for outside but in general I avoid using wire nuts and would never use them on my Christmas lights. I get my heat shrink at Harbor Freight. It's probably not the best, but it's dirt cheap: http://www.harborfreight.com/42-piece-marine-heat-shrink-tubing-67598.html or http://www.harborfreight.com/127-piece-heat-shrink-tubing-set-67524.html