Silly idea: RGB LED with two red LEDs in series

[nitro2k01]

Here's a simple and silly idea I came to think of. If you have a RGB LED powered to a 5V supply, using resistors for current limiting you will have a different corresponding voltage drop over the various resistors, due to the various voltage drops of the individual colors. So I thought, what if you put TWO red LEDs in series in the RGB package. For a ~2 V red LED, you now have a ~4 V LED drop and a ~1 V. According to my calculations, for P=UI of the resistor, you could get a 1/3 reduction in the U term because the resistor voltage drop decreases from 3 V to 1 V and 1/2 reduction of the I term because you could run the now two individual LEDs at approximately half the current for a combined equivalent luminous intensity.

Some back of the envelope calculations for a typical RGB LED specified for 20 mA per individual LED.

Resistive losses for single red LED: .060 A * 5 V -.020 A *(2 V+3.2 V+3.2 V)=0.132 W
Resistive losses for dual series red LED running at 10 mA: .050 A * 5 V - .010 A * 4 V -.020 A*(3.2 V + 3.2 V)=0.082 W

0.082/0.132=62%

Or about a 38% decrease in (specifically) resistive losses in the current limiting resistors. Or a 1/6=17% decrease in over-all power due to the decrease from 60 mA to 50 mA. Similar calculations could be done for other supply voltages and optimized configurations.

This wouldn't do much difference for a single LED, sure, but could be a pretty significant improvement for something like a display made with neopixel type LEDs, or for a gaming keyboard with individually backlit keys, where you have loads of LEDs. Both in terms of heat and total power consumption.

The idea seems so deceptively simple, so I'm wondering, are there any RGB LED manufacturers that do put two red LED in series. Or is this impractical? (Don't care because the RGB LEDs are so inefficient anyway? Maybe if you need increased efficiemcy, you would use a more efficient drive chip anyway?) Or is perhaps this idea in patent limbo?

Thoughts?

[Zero999]

Well there are RGB LED modules available with two wires for each LED so two could be connected in series like that.

If it's for a high power, high efficiency application, I prefer to use a current mode switched mode power supply so power isn't wasted in series resistors.

[mikeselectricstuff]

Red is sometimes the weakest colour - pretty sure I've seen 2x red, but probably discrete parts rather than die in a package.
 

[Siwastaja]

Why not.

Still, the voltage matching to 5V supply is still quite poor - the green has now most resistor losses, and the Vf of green is typically annoyingly close that you can't put two in series for 5V supply, like you did with red, but you almost could.

You need some voltage drop (20-30% maybe) over the resistor, even with well regulated voltage supply to provide stability against thermal runaway and chip variation, but, depending on green Vf, 50-60% is annoyingly lot.

Voltage-supply-series-resistor driven LEDs are typically used as indicators. If you are using it for illumination, you may want to have better efficiency than 50%, and hence use current-feedback SMPS.

But hey, you fixed the red, which had the most losses, and 2xVf(red) is quite close to the optimum you can get with voltage-based drive...

[mikeselectricstuff]

IME green and blue Vfs are usually very similar

[Zero999]

Red is sometimes the weakest colour - pretty sure I've seen 2x red, but probably discrete parts rather than die in a package. 

Yes, red is generally dimmer because the V_F is lower, there's less power for the same current.

IME green and blue Vfs are usually very similar

Yes, unless you have an old dim LED.

[mark03]

Still, the voltage matching to 5V supply is still quite poor - the green has now most resistor losses, and the Vf of green is typically annoyingly close that you can't put two in series for 5V supply, like you did with red, but you almost could.

But 3.x V is the new 5 V---what good is a 5V rail in the typical design these days?  Unless you're a beginner playing with arduinos, 5V is just plain inconvenient---the supply rail required by that one "problem" part, ruining an otherwise simple power-supply architecture    So that would be my answer to why not two reds in series:  4.4V is too high.

[nitro2k01]

But 3.x V is the new 5 V---what good is a 5V rail in the typical design these days?  Unless you're a beginner playing with arduinos, 5V is just plain inconvenient---the supply rail required by that one "problem" part, ruining an otherwise simple power-supply architecture    So that would be my answer to why not two reds in series:  4.4V is too high.

Blue LEDs are already on the high side for 3.3 V. They might light up dimly as low as ~2.8V but lighting at full rated brightness often requires 3.4V or slightly higher. This leaves little to no margin for current control and allowing for voltage drop in wires.

The benefit of the idea wouldn't be for Joe Duino blinking his first LED anyway, but for applications with a whole bunch of LEDs, where a few % reduction in heat dissipation could actually be valuable. Examples: RGB LED display. RGB backlit keyboard. (In this case, you have a handy 5V supply from USB.) If not using USB, you could be free to design your own power distribution. And for example, "intelligent" RGB LEDs like Neopixel/WS2812B are essentially 5V parts. They're specified down to 3.5 V supply voltage, again owing to the 3.4 V drop that's common for green and blue GaAn LEDs at full intensity, so nevertheless slightly above the comfort zone for a 3.3V supply if you want consistent light intensity.

Another application 12V or higher LED strips and other RGB lighting. In this case, just doubling up the red LEDs would lead to 6 LED in series instead of the usual 3, which would again be too high, but you could design a lighting panel such that you have 4 or 5 red LEDs in series.

[Siwastaja]

But 3.x V is the new 5 V

This vomit of characters doesn't make any sense.

---what good is a 5V rail in the typical design these days?

What the **** is a "typical design"?

I design in the power rails I need. 3.3V, 5.0V, or 6.7V, or 12.34V, doesn't matter, the design process is the same! The latest thing I designed, which is rather complex, has a 5V rail, and it's by far the biggest rail, 3V3 is one tenth of the current rating of the 5V. Why? Because it's designed to drive loads requiring 5V .

So where the 5V rail could be used? Maybe power up things that either require 5V, or when 5V is convenient for circuits you design? For example, for many analog circuits, or for charging a li-ion cell, 5V is handy with either a linear or buck charge controller. Also, driving MOSFETs. Or, some LCD displays require contrast voltages at 5V.  .... or... tadah... driving blue LEDs!

Have you though about why every stupid demo board / dev board has an eye-blinding blue POWER ON led, but all the IO/status LEDs are green or red? Yep, it's because they can't use the blue leds with the 3.3V IO, at least they would be dim and inconsistent, or very dependent on the exact LED part number and maybe even batch, and they want to blind you, and do it simply and cheap! So POWER ON LED for the "blue blinding experience" it is, they can use their separate 5V rail, or unregulated input supply, which today usually is externally regulated 5V, to drive a cheap blue LED at 20mA, it doesn't matter whether the Vf is 3.2V or 3.6V.

Of course, if your point is that MCUs (bare chips) requiring 5V supplies are obsolete, as well as 5V as digital IO voltage, you are quite right. But what does this have to do with anything discussed here?

[mark03]

Sorry to annoy you.  My comment was somewhat intended to be tongue-in-cheeck.

But really, what are these "loads requiring 5V" of which you speak?  AFAIK, the whole rationale for 5V traces back to TTL logic.  Nothing in physics, no inherent sweet spot, would suggest a rail near 5V, only tradition.  The 3-4V range on the other hand, is very natural for Li-ion battery chemistry, which in turn probably powers 80% of the electronic devices on the planet (wild-a** guess).  For non-battery-powered designs, of course, no voltage is sacred---you can choose whichever voltage you like.  In my experience, though, I'm just as likely (if not more so) to need 12V as I am 5V.  All of this to say, while I am sure there are applications where 5V is "just right", it seems an archaic standard by and large, and not a good one on which to base the development of future components.

The point about 3.3V being a bit low for blue is well taken.  But again, 3.3 mainly exists nowadays because of Li-ion chemistry, so you either have 3.6V available, or you are unconstrained and can set whichever rail voltage you like.

[Siwastaja]

5V loads that simply can not and could not work at 3.3V?

A HUGE number of analog chips such as certain amplifiers, comparators, etc. Many are even specified to surprisingly tight range of voltages around 5V, clearly only characterized at 5V.

Many transceivers, PHYs, etc.

Li-ion charger chips, because they need to go to 4.2V output, and there's always some dropout, even in switch mode design, bootstrap topology can't go to 100% duty cycle without being more expensive - and the FETs need 5Vish for gate drive anyway. 5V input happens to be fairly close to the best possible sweet spot for both linear and buck chargers.

Many MOSFETs need gate voltage of at least 4.5V.

Most blue and white LEDs need more than 3.3V from the supply.

Yes, all of this comes from physics. Some of it can be changed by changing the designs, i.e., what physical principles are used and in which way. But it's not trivial.

As a consequence, most "iDevices", board computers, etc. today take 5V input - and they are one more example of potential 5V loads you might find a need to supply power for.

Yes, digital logic and modern MCUs do not tend to work at 5V anymore. But there is much more to electronics than just digital logic.