Light-Emitting Diodes (LEDs), especially white ones, are gaining ground very quickly due to their practicality and energy saving. After revolutionizing the signaling area, they begin to enter the branch of lighting.
The price of white LEDs has dropped steadily, allowing them to be used for decoration of environments and production of various consumer items such as lamps, lamps, lanterns and emergency lights. LED panels and strips can also be found, adaptable to a wide range of uses. And we are just at the beginning of this new technological era, as alternating current LEDs (Cree, Luxeon, Samsung, etc.) and OLED (Organic LED, from Verbatim) are now commercially available.
Being a recent technology, fueled by manufacturers’ eagerness to set their market share, we witness a flood of low-end products-to say the least. To reduce the price of equipment that uses white LEDs, its drive is simplified to the extreme.
Therefore, the cheaper LED equipments have serious problems as they tend to burn with little time of use. In this article, I outline how white LEDs are manufactured and what requirements are required to safely energize them, using as an example a table lamp.
The white LED, a diode
Every LED is a diode, with the added feature of emitting light. In figure 1, the diode and the LED are shown. As you can see, in symbology they are very similar. In behavior, they also have several similarities.
Internally, the diodes are formed by a PN silicon semiconductor junction. To know the semiconductor diodes in more detail, consult the University Institute of Lisbon  and the University of Brasilia .
When the junction is directly polarized, the diode leads. This occurs when the anode is connected to the positive and the cathode connected to the negative of the power source.
The junction allows the passage of the stream in only one direction, in the same way as a retaining valve in water installations. The semiconductor junction, when directly polarized, does not begin to drive at once. For this to occur, it is necessary to cross the potential barrier. This barrier is the minimum voltage for the device to start driving. In the case of LEDs, to make them emit light.
The potential barrier is also called a voltage drop, because that is precisely what it does. If we connect two diodes in series, for example, the voltage necessary for them to start driving will be the sum of the potential barriers of each.
The value of this voltage drop varies according to the type of semiconductor.Common silicon diodes, lead from 0.5V. Schottky diodes, used in PC sources, for example, have a voltage drop that can start at 0.13V, on the higher power components.
On the other hand, the LEDs have different potential barriers, starting at 1.5V for the infrared LED and up to 3.8V for the ultraviolet LED, according to OKsolar’s LED Color Chart . In Wikipedia , there is another table, which shows the tolerances of these values for each color.
More precisely, the Trainweb modeling site  highlights a characteristic common to all diodes and very important for the activation of LEDs: the voltage drop is dependent on the current flowing through the component. In the case of the white LED, the voltage drop starts at 2.5V (with current less than 1mA) and goes up to 3.7V, with 20mA of consumption.
Figure 2 shows the voltage/current curve of a hypothetical white LED, based on this information. On the horizontal axis is the voltage ( v ) on the diode terminals. On the vertical axis, the current ( i ) is passed through the component.
The LEDs light when working within the direct bias region, more specifically in the voltage range of 2.5 to 3.7V. It is the part of the curve that resembles a knee.
Because of the high voltage drop, white LEDs can not be evaluated as diodes by ordinary digital multimeters. On the continuity scale, which is usually also for diode, the Meter provides a small current for the measurement.
What the multimeter measures, on the diode scale, is the size of the potential barrier of the semiconductor junction of the component. That is, its voltage drop, relative to the current supplied by the instrument.
And the diode scale usually reaches up to 2V. As white LEDs show a voltage drop above 2.5V, the multimeter will indicate overflow (open circuit), even though it can dimly illuminate the LED.
Reverse polarization and disruption
The graph of Figure 2 is divided into three areas: direct polarization air (discussed above), the reverse polarization region, and the region of rupture. Each of these parts has particular characteristics and provides valuable information for understanding the diodes.
The reverse bias region is that formed when the semiconductor junction of the LED is inversely polarized (it is the wrong connected component, in which situation it does not work). The junction presents the same behavior of direct polarization, with the difference that the reverse potential barrier is larger.
In the case of white LEDs, the reverse potential barrier (or allowable reverse voltage) is generally between-5 and -6V.
Common diodes have this much higher potential barrier: the 1N4007 diode, for example, can receive up to 1000V reverse voltage without any damage. For him, the graph equivalent to the region of reverse polarization, with that scale, would not fit on the page…
As a curiosity, there are components that take advantage of the region of rupture, such as zener diodes and photodiodes. For details on these and other diodes, refer to references  and .
The leakage current
It is noted in figure 2 that within the reverse bias region there is a small leakage current, which increases slowly with increasing negative voltage. After this region has passed, disruption occurs, with the consequent avalanche current, which can burn the component.
With white LEDs, this is very easy to happen because the breakpoint is only-5 or -6V. If the LED does not have the ESD diode (discussed later), the reverse voltage will be a serious threat to it.
The reason is that, when inverted voltage is applied to the chip, the migration of the minority carriers begins to occur. When a diode is biased inversely, the P-layer is connected to the negative. It has the minority charge carriers, which are attracted to the positive pole (which in this case is connected to the N-layer). This results in a leakage current, which is increased by the gaps of the N layer, attracted to the negative side. Migration tends to increase leakage current and decreases LED performance and can render it unusable.
Manufacturers literally recommend that all necessary care is taken in the design of the drive circuits so that the LEDs never receive reverse voltage. All to prevent premature failure of these components and to let them away from the avalanche current.
Looking at the LED closely
One thing that strikes us when we look closely at white LEDs is that, internally, they are not connected like the red LEDs. There is an extra gold thread on the silicon wafer, which is attached to the negative. In figure 3, one can see an RGB LED, where the green color is on the left, the red in the center, and the blue on the right. One can notice the similarity of the connections of the green and blue LEDs, which have two wires each. The darkest tablet, the red one, has only one binding thread.
A white LED, very closely, appears in figure 4, while in figure 5 we have a red led.The difference between the connections is obvious.
A Matsushita patent, which shows how to make a blue LED emit other colors, with the application of a fluorescent layer, has the form of bond very similar to that currently used in the assembly of the white LEDs (figure 6).
In figure 7, there is a drawing of the silicon wafer model used to make the white LED, from the Korean blog Polytech Lab . The reason for the existence of two bonds on the chip is that the substrate (base layer) shown in the model is aluminum oxide (Al 2 O 2). Which is, in principle, an insulator. There are technologies already in use in the more sophisticated LEDs that make the connections underneath the insert.
The white LED manufacturing process
Did you know that white LEDs are actually blue? Yeah….
Currently, there are 3 ways to build a white LED, according to Rohm . In figure 8a, there is the first one: 3 pads ( chips ) are mounted in the same housing, each emitting a color. The colors follow the RGB (Red, Green and Blue, or red, green, and blue) standard of televisions and monitors. When the 3 chips are connected simultaneously, the resulting light is white.
For example, we have an RGB LED, with SMD package, model 5050, soldered on a flexible printed circuit board, whose transparent silicone layer has been removed, to give more clarity to the photos. In figure 9, it appears off, where you can see the 3chips , corresponding to the colors. Next, in the same figure, the LED appears completely on, forming white light.
Figure 10 shows the slightly excited exciter inserts one at a time. In this case, as each LED within the encapsulation can be triggered separately, the component has the ability to emit several colors. Just measure the current for each of the 3 inserts (colors). Commercially, there are circuits, called RGB controllers, that make this current dosage and allow the user to choose the desired color. Even some have remote control.
The second method of construction (Figure 8b) uses an LED that emits light in the near ultraviolet and excites the phosphor pigments deposited just above. Each pigment responds by one of three RGB colors. Could not get such an LED to show for example.
The third method (figure 8c) is the most used today (in 2013). A yellow fluorescent layer is applied, which when combined with the blue LED, emits white light. To improve color reproduction, other pigments that correct the color temperature are added.
In figure 11, there are 2 power LEDs, mounted on a star heatsink. The left one emits warm white light (3000 to 4000 K) and the other, cold white light (6000 to 7000 K).One can note the small difference in hue of the fluorescent layer.
The yellow fluorescence layer can be applied in two ways. One of them is shown in figure 12a: the pigments are mixed with the LED encapsulating resin. Referring to Figure 11, it can be seen that these LEDs were assembled using this method.
In figure 12b, the other way of applying the fluorescent layer appears: directly on the silicon wafer. Figure 13 shows the use of this manufacturing process in an Osram LED of the Golden Dragon Plus line .
The emission spectrum
Faced with so many methods to improve color reproduction and try to get close to the index achieved by incandescent, there has been a lot of progress in recent years.But we still have a lot of optical illusion, because what looks white is just a well-balanced mix of a few individual colors. There are several tones that simply are not emitted. As an example, Figure 14 shows an emission spectrogram of an RGB LED, from Photal . That mix of only 3 colors makes us see the white. But all other tones have weaker or even zero emissions.
This is what affects the color reproduction index (IRC or CRI = Color Reproduction Index ), which can reach 100% incandescent bulbs. In 2013, the cheapest white LEDs reach, at a great cost, an IRC of around 80. But there are already manufacturers selling white LEDs with IRCs over 90, but they are still expensive.
In order to better understand the electromagnetic spectrum, and within it, the part that fits the visible light, we extract from the page of the Center of Sciences of the Education (CED) of the UFSC , figure 15. In it, one can see the spectrum of Visible light, with the colors and respective wavelengths. The visible light corresponds, on average, to the range between 390 and 700 nanometers.
Nanometer (nm) is the unit of the wavelength of the frequency of light (the color).That is, the unit nm indicates the size of the wave of a certain color. A nanometer corresponds to the millionth part of the millimeter, that is, a millimeter divided into a million pieces.
In figure 16, an overlap of 8 spectral distributions of the Luxeon white LED line, Rebel ES, appears. High performance (125 lumens / W) LEDs, specific to lamps and luminaires, operate at color temperatures between 2700K and 5650K. Very similar, therefore, to incandescent bulbs. For more details, consult Philips Luxeon .
In figure 17, the emission spectrum of three lamps – fluorescent, incandescent and LED-of the same color temperature (3000K) is shown. This image was taken from an article on the science of colors, from the excellent magazine LEDs Magazine , which includes information from all the major manufacturers in the area and has a free version to consult on the internet.
Another image, very didactic (figure 18), comes from the page of Popular Mechanics , which shows a comparison between the sun’s emissions and 3 types of lamps.Seeing the graphs in perspective seems to make comparisons much easier.
Note that it is still necessary to improve three points. The main one is the range between 450 and 500 nanometers (nm), which corresponds to a stretch between light blue and cyan. Then it has the range above 650 nm, which pulls to bright red and the part below 470 nm, belonging to violet. Despite this, there are already commercial LEDs in the infrared and ultraviolet range, which are at the ends of the visible spectrum.
Sensitivity to electrostatic discharge
All manufacturers, without exception, comment in technical texts on the problem of electrostatic discharge (ESD = ElectroStatic Discharge). A good text, from Sharp , informs the proper way to handle the white LEDs. It is in booklet form and is suitable for printing and use as a guide.
It is important to pay attention to this problem because it can severely reduce the life of electronic components. Whoever deals with white LEDs – in fact, with electronics in general – needs to have protection against electrostatic discharge on the bench. It is composed of various anti-static equipment (grounding, blanket, wristband, tools, furniture, packaging, etc.).
The reason for this precaution is that current electronics are much more sensitive than those of past decades because of the miniaturization of silicon chips. Thus, small discharges, which formerly caused “tickling” in semiconductors, today can burn them.
Another text, by Osram , reports that industry losses from static electricity discharges are in the range of 8 to 33%. The company article explains that the generation of static electricity occurs mainly through the contact and separation of two materials, which is known as triboelectric loading.
The effect is caused by the transfer of electrons from one material to another. The material on one side, which receives electrons, becomes increasingly negative, while the other, which loses electrons, becomes more positive.
The name “static electricity” is used because the electric charges do not flow, they are stationed (static) on the surface. Depending on the material, the charge may last for more or less time.
For example, a person walking on a carpet (or carpet, or carpet). The shoes do just that contact and separation between the surfaces. On dry days, with up to 25% humidity, one can generate up to 35kV! If the humidity is high (greater than 65%), 1500V can still be generated. This is why many “take a hit” when touching a surface or catching the bus. Obviously the discharge voltage is high and the current too, but the human body can only store energy equivalent to a capacitor of 100 to 150pF, just enough to give an audible click and cause some discomfort. Of course, if there are flammable gases nearby, the situation can become very dangerous.
The so-called triboelectric series features a list of materials that may or may not attract electrical charges. The list is classified into materials that attract negative charges, which are neutral, or which attract positive charges – see references InfoEscola  and Science Fair . The human body is the most positively charged, whereas Teflon is the one most likely to get negative charge. Steel and cotton are considered neutral. Synthetic clothes, therefore, can help generate static electricity (and maybe change people’s moods …).
An electrostatic discharge event ( ESD event ) occurs when there is a body electrically charged and another that is close when suddenly receives this charge.The human body is a frequent cause of ESD events. Discharge usually occurs without physical contact. It is a small spark between the two nearby points.
Damage caused by these events is determined by the ability of the devices to dissipate the energy of the discharge or to support it. This characteristic of the components is known as ESD sensitivity (sensitivity to electrostatic discharge).
A highly illuminating image on the behavior of LEDs in relation to electrostatic discharge surges is shown in figure 19. It was extracted from the Acrosentec site . The curves shown refer to the region of direct polarization. The left most curve is normal, very similar to figure 2, shown at the beginning of the article. As the damage increases, the current required for the LED to turn on is increasing, as the “knee” of the curve goes up, to a stage where the LED does not start. The “knee” of the curve is precisely the operating region of the LED.
According to company page information, static electricity discharges are often associated with reverse bias. Blue and green LEDs are much more sensitive to reverse voltages than red ones.
To avoid this problem, manufacturers place an electrostatic discharge diode (ESD diode) inside the LED housing when space – or cost-benefit-allows. Figure 20 gives an idea of how the process in an RGB LED, reference 5050, of Wah Wang . One can compare this model with that other, which has the same reference, in figure 3, but does not have the protections. Incidentally, very cheap white LEDs usually have no ESD diode…
In Figure 21, some forms used by manufacturers to implement ESD protection are shown. The design of the SMD LED shown here can be compared to the Osram LED, in Figure 13, where you can clearly see the connection of the ESD diode.
But the ESD diode can not do everything, because it usually protects against surges up to 3kV, depending on the manufacturer. Further precautions against electrostatic discharges are required in the design of equipment that will use LEDs.
Continuing with the Osram article , there is a guideline to help protect against electrostatic discharge in the case of several LEDs connected in series: place a zener diode in parallel with all LEDs, as shown in figure 22. In the case of An ESD event occurs, the zener voltage is exceeded and it provides another way to drain the electric current. The zener voltage must be greater than the sum of the voltage drops of all LEDs.
In addition, Osram’s article addresses the equipment required for a static electricity-protected work area, the constructive characteristics of LEDs, and what happens on the microscopic level with silicon chips when subjected to electrostatic discharges.Lastly, there is a check list for ESD Control ( ESD Control Checklist ).
The current surge
Another problem is caused by current surges on the LEDs at the switching instants.Philips  has an article that addresses this issue, and clarifies that simple switching of a transistor can cause a surge of current so intense that it damages the LEDs.
LEDs are devices that are sensitive to current variations. For this reason, designers should be aware to avoid current spikes when implementing their drive circuits.
One can compare the behavior of these circuits with a garden hose. With the tap open, but the hose nozzle closed, the pressure on the hose will be maximum. At the moment of opening of the nozzle, the existing pressure causes the water to flow suddenly over the small outlet orifice. Soon after, the flow stabilizes. Everyone has probably seen the squirt of water come out when opening the hose nozzle. The time it takes for the flow to drop to a standard level is fast enough for many to overlook this initial squirt.
Similarly, in electrical circuits, the time required for the current to fall to a regular level is too rapid to be noticed. In fact, we usually think that the LED turns on immediately. However, semiconductors react millions of times faster than humans.A very short current pulse – lasting a few milliseconds – can destroy a semiconductor wafer.
To clearly demonstrate this effect, let us consider a simple circuit, consisting of a current source, a key and a set of LEDs, connected in series (figure 23). In this example, let’s assume that the maximum voltage drop (Vf) of the LED array is 36V.Source specifications include a direct current (If) from 350mA to 50V at maximum – similar to a typical 25W source for these devices.
The circuit limits the current to 350mA when the switch is turned on. But when the switch is opened, the current flow to and from the circuit is unable to self-regulate itself. In a short time, the voltage at the source output rises to 50V of the technical specifications. This is easy to check by measuring the voltage between the source terminals (the voltmeter of figure 23).
The output of the power supply, plus the wiring, act as a large capacitor, storing charge, in the same way as the household water pipes. When the switch is closed, this load flows rapidly through the circuit until the power supply begins to stabilize.But the total load on the wiring may be enough to destroy a semiconductor (such as an LED) in that short time.
Preventing destructive current surges is the job of the designer. For a layman, the difference between a current surge and a soft-start drive is best visualized on a chart. In Figure 24 the y-axis corresponds to the current (left side, in mA) and voltage (right side, V), versus time (x-axis, horizontal, in milliseconds). There are three areas of interest in these curves: before starting the device ( off ), during the time the source takes to regulate the current ( transient ), and after the current stabilizes ( regulated ).
Using Ohm’s law (voltage is equal to the current multiplied by the resistance, or V=RI ), we can calculate the current through the LED. It is easy to see in the graph of Figure 24 how the voltage transient injects a large current peak into the LED. In this case, three times greater than the projected value of 350mA. These current spikes can cause permanent damage to any semiconductor such as integrated circuits, microprocessors and, of course, LEDs.
By specifying a soft start, undesirable outbreaks can be easily avoided. A soft start circuit ensures that the wiring is at 0V when the switch is open (figure 25). With the switch closing, the current rises from zero to the desired level, without the peak current shown in the previous figure. This is analogous to leaving the hose with the nozzle unlocked, when the tap is opened. It takes a short time for the water to reach the hose, and when this happens, the flow is at the desired intensity.
Peculiarities of white LEDs
There are other interesting features about white LEDs that can help you use them and handle them better. The Nichia datasheet , for example, informs that white LEDs emit a stable color, should always work above 10% of the maximum currentcapacity.
The same text also warns that some LEDs do not tolerate ultrasonic cleaning , whereas the Wah Wang SMD 5050 LED datasheet  directs that power below 15 W is used and the cleaning bath is less than 1 liter. Osram  also does not recommend ultrasonic cleaning. And all manufacturers surveyed accept or recommend cleaning the LEDs with isopropyl alcohol.
The working current
Perhaps the most important feature of the LED is the working current. Although the LEDs show a voltage drop, they can not be considered as resistors because they do not offer any current limitation. If you connect a 4.5V power supply directly to a white LED, it will consume all the electric current the source can supply, because the resistance of the LED when driving is almost zero.
Since the current is I=V/R, and R in this case, tends to zero, the current will tend to an infinite value (V, of any value, divided by a much smaller value). In practice, there are other limiting factors involved, but the intention here is to give an idea of the behavior of the circuit.
If we look again at figure 2, the segment we are talking about is the vertical part of the curve, just after the “knee”, inside the region of direct polarization. This is when the current rises infinitely after exceeding 3.7V. Reminiscing: LEDs work (emit light) only on the front of this point, inside the “knee”. Above that, they burn, when the maximum current for which they were designed is exceeded.
Therefore, a direct connection of the LED to the power source, without current limitation, is discouraged. But there are many LED flashlights on the market that use this damn form of connection because they rely on the low current capacity of the batteries.
The idea is simple: if the current is too high, naturally the batteries will lower the voltage (decreasing current) and the circuit will come into balance. But the LEDs are losing the ability of illumination over time, which is largely to blame for the initial surge of current to which they are subjected.
This is because, as the LEDs are not identical (they have different voltage drops), when connected in parallel one of them will come into operation first, because of the smaller potential barrier. It will receive all the current from the batteries at a very high intensity. Hence, as the higher working current on the LED increases its voltage drop , the other LEDs will start to turn on. This happens in a fraction of a second.But the damage is already done (see figure 19). In the short term, the LEDs do not turn on anymore.
Limiting the current in the LED-the simplest mode
Some time ago, I got a small LED light, powered by 3 batteries (4.5V total) and also a USB cable, which could be connected to the computer (5V). Without knowing the product, I connected the luminaire to a 5V source of the bench and the LEDs lit up, but they started to erase one by one. They smelled strongly because they were burning.
I opened the luminaire and was surprised to find all 12 LEDs in parallel, connected directly to the power source. Figure 26 shows the wiring diagram of the original product. I went to the store, complained and changed to another lamp, new and equal. Then, before I turned it on, I made some modifications to avoid the problem created by the manufacturer.
The proposal is to implement a limiting resistor for each LED, which is the simplest way to limit current. The calculation, in this case, made for one LED, is valid for the others, because all are in parallel. Thus, there are 12 resistors for the 12 LEDs of the luminaire. The changed lighting scheme is shown in figure 27.
Few simple accounts
Ohm’s Law says that V= RI. That is, voltage (V) is equal to resistance (R) times current (I). Figure 28 brings Ohm’s Law graphically, easy to learn.
So the luminaire in question can be powered by voltages between 4.5 and 5V, coming from 3 batteries or from the computer source, respectively. Let’s also accept the rechargeable batteries, which has 1.2V. So we have a power range between 3.6V and 5V, in which the flashlight should work.
Many use, as a general rule, the current in the range of 10 to 25mA for common white LEDs. But they work from 1mA. To save energy, but still have a reasonable light level and also to increase the durability of the components, we determined a minimum current of 10mA and maximum of 20mA.
Let’s review the Trainweb page . From there, we extract the voltage x current graph for the LEDs (figure 29). This graph shows only the operating range of the LEDs (the “knee” of Figure 2). It was assembled by the author of the page, after making several measurements. Let’s consider only the right-most curve of the white LED.
For the minimum current of 1mA (0.001A), the graph of Figure 29 reports a voltage drop around 2.5V. The minimum supply voltage shall be 3.6V. Hence, current limitation should be based on 3.6V (batteries) minus 2.5V (LED), which gives 1.1V.The calculation is then R =V/I.
R=110 ohm. This will be the maximum value of R, ie we set the minimum current, for when the batteries are weak.
Now let’s calculate the resistor for the highest voltage value. The voltage for calculation will be 5V minus 3.6V, which is the voltage drop of the LED when powered by 20mA current. We have 1.4V.
R=70 ohm. This is the value of R for the maximum current when we power the circuit with the USB cable.
So we should choose a commercial resistor between 70 and 110 ohm. We chose 82 ohm, which is a closer commercial value (could also be 100 ohm).
We also want to calculate if the chosen resistor does not exceed or fall below the planned current. Then, I=V/R.
I=1.1/82 , which gives 13mA for the 3.6V supply.
I=1.4/82 , which gives 17mA when the power supply is 5V.
In addition, a 82 ohm resistor will activate the LEDs until the batteries are well spent with 2.6V. For this voltage will be reached the minimum current of 1mA per LED. In theory, the circuit will use the batteries up to the stalk…
Changing a luminaire
A little practice. Figure 30 shows the 82 ohm resistor (gray, red, black bands) used in the luminaire. For those who want to know more about the colors of the resistors and their respective values, consult the Science Fair . The steps for effecting modifications of the LED luminaire are shown in Figures 31 to 50.
In Figures 34, 35 and 36 is the sequence for interrupting the tracks of the printed circuit board, which houses the LEDs. The intention is to individualize the connections of one of the poles of each LED (no matter what, as long as all the interrupts are from the same pole). In our case, the interruptions were made in the anodes of the LEDs. The resistors were connected to these interrupts and then joined at one point, which connects to the positive (Figures 37 to 41).
With the set assembled, the total current of the LEDs was measured with a 5V source (figure 42). The consumption was 170mA, which, divided by 12, gives approximately 14mA per LED. Which shows that the calculations were reasonably correct. It can be considered that the inequality of the LEDs of the luminaire increased the error of the calculations, since the LEDs are of unknown origin and have not been tested individually.
As the modification was made hastily, there was a mishap. When reassembling the luminaire, it was found in the middle of the compartment that houses the LEDs, literally a small post… It seemed to be a mechanism of centralization between the upper and lower part, and since I did not find another reason for it to exist, it was excised (Figures 43 , 44, 45 and 46). If he had been less adventurous in assembling the resistors, he could have placed them looser, which would allow him to bypass the plastic obstacle and keep the lamp intact.
In figures 48 and 49 the luminaire is already installed, and in figure 50 the cable, used to connect it to a USB port.
Ways to power the LEDs
The Nichia datasheet  states that its LEDs can be connected in parallel when connected to a current source. When the LEDs are well-known, such as these from the Japanese company, the quality is evident and there may be a lot of similarity between the products in a line, which may make the connection in parallel acceptable. But I have my doubts, I still prefer to connect them with individual current sources. Or in series when there is enough supply voltage.
The same document informs that when the LEDs are connected to a voltage source, they must have a resistor in series-which, after all, transforms the fixed voltage supply into a fixed current supply… That is, always turn on the LEDs On current sources!
Current source , for those who do not know, is the one that keeps the current flowing steadily. That is, the value of the current is controlled, not the voltage. The simplest current source is the resistor. The electronic circuits that act as a current source mimic the behavior of the resistor. It is desirable for the current source to have a high impedance.
To carry out this task, higher input voltages are required than for fixed voltage sources. In a hypothetical example: it does not matter if the load is 1 ohm or 100 ohm, the source must supply the same current for both cases. The voltage on the load may vary, but the current will always be the same in any case.
The regulator integrated circuit LM317 – see reference , a common component in adjustable power supplies, can be used as a current source. Manufactured since the 1980s, it is very common still today in the electronics trade and facilitates the implementation of any current source up to 1.5A.
It supports up to 40V of voltage difference between the input and output and is sold in multiple encapsulations. The TO-220 housing dissipates up to 20W. Obviously, with higher currents there will be a need for a heatsink.
In Fig. 51 a current source is shown with LM317. The calculation to determine the value of the current that the circuit will provide is simple:
Where I is the desired current, in Ampere, Vr is the reference voltage (1.25V) and R is the resistor value. So,
The National datasheet for the LM317  directs the R value to be between 0.8 ohm and 120 ohm, which means that the minimum current it provides in this configuration is 10mA (1.25 / 120 = 0.0104A).
In addition, it is necessary to add to the drive voltages of the LEDs, the voltage drop caused by the LM317 – which is approximately 2.5V. The circuit can only be supplied with voltages higher than this sum.
The page of Francesco Sacco, Epic Electronics , has the translation of a technical note from ON Semiconductors about current sources specific to LEDs with the LM317.There are shown, for example, ways to connect LEDs in parallel, with individual drive.
Another source of current with the LM317 is in Talking Electronics , which shows how to make an LM317 provide from a few milliamperes up to 500mA in a continuously adjustable mode. The layout is poorly drawn, as the entrance is to the right, in the inverse position to the usual (figure 52). Apart from this, it is an interesting circuit to test, because his proposal is to pass most of the current on the transistor, allowing the use of common potentiometer. But it should be noted that the voltage drop on the circuit will increase even more.
More current sources
The 78xx regulator line (7805, 7808, 7809, 7812, 7815, etc.) can also be used as a current source up to 1A (for TO-220 encapsulation). The scheme shown in figure 53 uses the 7805, connected in the same way as the LM317. In addition to the additional capacitors, change the calculation, which must consider the working voltage of the regulator ( Vreg, which with the 7805 is 5V) and the quiescent current Iq, which is 4.2mA for this regulator-see reference Newpic  . The formula for the calculation of the output current Is is:
The same Newpic article shows you how to make a continuously adjustable current source with the 78xx. There is also a datasheet from Texas Instruments  which shows the current source settings for the LM340 or LM78xx. The LM340, through the data sheet, has a much lower quiescent current: 1.3mA.
Turning the old fashioned LED
The Troniquices page  has a simple source of constant current, with transistor, stabilized by zener diode (figure 54). The values of the components established therein maintain a current of 20mA for the LED, under supply voltages between 9 and 20V.
The author calls the circuit “magical resistance” because it functions as if it were a self-adjusting resistance, depending on the supply voltage. This is what characterizes, in fact, all the sources of current. The article explains in detail how to perform the simulation of the circuit in SPICE, but does not inform the calculation method of the circuit.
From Tritin , we have a current source scheme for 5A, under 5V, dedicated to electrolysis (figure 55). To function properly with LEDs, you must recalculate your components. The reason for this circuit to appear here is that it is quite similar to the previous one, but it handles much more power (25W).
This is made possible because of the Darlington coupling of the transistors, which tremendously increases their gain. In this type of coupling, the gain of one transistor is multiplied by the gain of the other. The two transistors, in this case, work as if they were one. The higher the gain of a transistor, the less base current will be required to handle the current between the emitter and the collector. For more details on this connection, consult Electronics-tutorials .
But these circuits have limitations, in relation to the supply voltage. If it changes too much, the resistor that polarizes the zener diode should be recalculated to avoid overloading. In addition, it is difficult to operate through other devices, such as microcontrollers.
The operation of this implementation is based on the zener diode , which stabilizes the voltage at a given value and derives part of the current that receives to the base of the transistor. In this way, the base current will always be equal, so that the current between collector and emitter, much larger, is also constant.
But, too much current goes only to the zener diode, especially when the power supply reaches the maximum value. Under 9V, the zener drains 2mA, which rise to 7mA when the voltage reaches 20V (figure 54). In times of energy saving, it is not acceptable to waste, in the form of heat, almost 30% of the energy to power an LED.
And the configuration of Figure 55 seems to have a basic design problem. Each transistor, when conducting, maintains between the base and emitter a voltage around 0.6V, due to the voltage drop of the semiconductor junction. In the case of a Darlington joint, the tension doubles: 1.2V. So, what’s the point of connecting a zener diode, much higher voltage (3.3 V) between the base and the emitter, if the direct voltage at the junction will not increase?
Of course, the implementation works well if there is an emitter resistor, as in figure 54. But in the diagram in figure 55, the current will not pass through the zener diode. That is, this component will not do anything, because all the current will be taken to the base of the first transistor.
Simple but efficient and functional circuits
There is an interesting current source from Elektor’s “311 Circuits” under the title “Simple LED Constant Current Source” . The tip (only the schematic) was published on the Internet by Silveira Electronics . The configuration maintains an approximate current of 20mA and can be powered by voltages between 5 and 24V (figure 56).
To perform the current calculation, only the law of ohm is used: R=V/I, where Ris the desired R1 value, V is the junction voltage of T2 (at which it starts its operation: 0.7V) and I the working current. The current can not exceed 20mA with those components.
Thus, we have R=0.7V/0.02A, which in this case results in 35 ohms. The closest value of a commercial resistor is 39 ohm, precisely what was used there.
Another circuit, slightly different, comes from the site Eletroalerta . It accepts a working voltage between 6 and 15V (figure 57). The difference between the previous circuit and this is that the components form a block, which can be connected as a resistor: in series with the LED. In addition to using a Darlington transistor (BD679 or TIP120), which extends the power of this implementation because it handles up to 100mA of current-according to the text.
The author used a reference voltage around 0.56V because the calculation is exactly the same as in the circuit of figure 56: according to the article, if the value of R2 is 56 ohm, the current in the device will be 10mA . If R2 is 5R6 ohm, the current will be maintained at 100mA.
Because of the serial connection with the component block, the LED can be mounted in another way: connected to the positive branch of the power supply. This would facilitate the implementation of an on / off command, since it would be sufficient to disconnect the R1 (10K) resistor from the positive and to couple it to the output of any digital circuit, such as a microcontroller or a logic gate. Figure 58 illustrates this possibility.
From the Bright Hub Engineering page  comes another circuit of the same type (Figure 59) and the same calculation formula. It calculates both the emitter resistor (R2) and the base resistor (R1), which is also very simple:
Where Vb is the voltage over R1 (the maximum potential it will receive when switched on),
Hfe=is the current gain of T1 (beta – available in respective datasheets ) and
I=current on the LED.
This circuit (Figure 59) already comes with the ease of external control. In the Schematics from Delabs site , we find another similar circuit (figure 60), also dedicated to LED power, usable up to 25V, with maximum 20mA.
The last current source, in figure 61, from the Electronis DIY page , operates with voltages between 2 and 18V. The author adopted a power MOSFET transistor (although he designed it incorrectly), which reduced the minimum voltage for the operation of the circuit and enables its use in portable devices. The Mad Scientist Hut site  also describes a current source with MOSFET transistor, with calculations.
This configuration facilitates the implementation of the soft start/stop feature because the MOSFET port only needs voltage to be triggered (the current it drains is negligible). For low working voltages, it is necessary to employ a logic level MOSFET transistor. These transistors usually switch from 2.5V. On the other hand, the common MOSFETs can only fully actuate the load with a door voltage around 10V.
All of the implementations of Figures 56 onwards work in the same way. Therefore, we will explain the operation only of the scheme of figure 61.
Transistor Q2 is a power MOSFET, N-channel, which functions as a variable resistor.R3 is the sensor resistor , connected between the Supridouro-S terminal and the earth, which exhibits a voltage between its terminals, proportional to the current of Q2.
The value of R3 is calculated to display 0.7V on the basis of Q1 as soon as the maximum current is reached. That is, transistor Q1 only comes into action-to reduce the voltage of Door (G)-, if the current exceeds the predicted value.Otherwise, the MOSFET Q4 transistor is not bothered and conducts fully, which indicates that the current will always be at its maximum value (if the source has enough power).
These circuits have the advantage of accepting the command by any external device, including PWM drive. They can handle currents from a few milliamps, up to 20A or more, provided that the appropriate components are chosen and the heat dissipation conditions are satisfied.
Which current source to choose?
The current sources shown in the previous topics have advantages and disadvantages, which may indicate which circuit is most suitable for each application.
Using voltage regulator integrated circuits, such as the LM340, LM317 or 7805, among others, we have a device ready to supply currents up to 1 or 1.5A, which is easy to implement. When you do not want to control the LED through an electronic circuit, it is an interesting option. The configuration with 7805 can be assembled with scrap components, but the LM317 is of more rare use in electronic equipment.
The current sources that use zener diode to stabilize require careful design so that they can be used properly, and are also not suitable for external commands. In addition, the calculations of the components are many and make difficult the suitability of the circuit for other cases, as can be seen in Wikipedia . Another difficulty is finding the exact value of the zener diode in scraps.
The most versatile circuit is the current source with resistor sensor (resistorshunt ) and MOSFET transistor, as it can handle a lot of power, facilitates the use in portable equipment and allows the configuration of the external control, if desired. In addition to using very common components in scraps, such as computer sources.
A practical circuit, to drive LEDs safely
Figure 62 shows a current source diagram with MOSFET transistor and some additional components. This circuit has been tested by me and works very well. It has soft start and shutdown, plus overcurrent protection and reverse polarity on the LED.
This circuit does not define how many LEDs will be supplied. Many of them may be used in series provided that the voltage drop of all of them is at least 2V below the supply voltage.
It should also be ensured that the transistor door receives no voltage higher than the drive voltage, somewhere between 5 and 10V-depending on the type of MOSFET.This can be done with another zener diode, with a slightly higher voltage (7.5 or 12V, respectively), connected between the door and the supr.
The overload protection, made with a zener diode, with a voltage slightly higher than the voltage drop of the LED, was placed in parallel with it. This zener diode (D1) should be changed if more LEDs are connected, since it must always have a voltage greater than the sum of all the voltage drops of the LEDs connected in series.
It could be used (for a LED) a zener diode of 3.9V. But in order to reduce the leakage current, as it would be only 0.2V difference in relation to the white LED voltage drop (3.7V), a zener diode of 4.3V was chosen, giving a gap of 0 , 5V. With more LEDs in series, the total clearance can be reduced. If the application is to trigger LEDs of other colors, the value of the zener diode must also be adjusted according to the resulting voltage drop.
To protect against reverse voltage, the D1 zener diode itself could be left because the datasheets report that the direct voltage drop of them is 1.2V. This value would still be below the voltage drop of any LED. Again, as a precaution, a reverse polarized diode (D3) is placed.
Here, it is most appropriate to use fast diodes, such as 1N4937, very common in fluorescent lamp reactors, or some other, from PC sources or the horizontal stage of televisions. The reason for using short response time diodes is that since MOSFETS has a fast diode reverse polarized between the Drain and Suprucker, some reverse voltage might be present on the LED in a short time in which it would be unprotected. If the LED used has internal ESD diode, D3 is unnecessary.
Soft start-up and shut-down. MOSFET transistors chaveiam when there is a sufficient voltage on the gate ( Gate ). The impedance of this G-terminal is so high that current consumption is not considered relevant. Thus, there is enough tension in the door to be able to actuate them.
In the case of MOSFETs that accept activation by logic levels, a 5V supply voltage is generally considered. The device turns on when the voltage rises beyond 2/3 of the power, and shuts off when it is 1/3 below 5V. Many of these transistors already switch from 2.5V. Other traditional MOSFET transistors connect the load fully when the door reaches about 10V.
So, in order to get smooth starting or stopping on MOSFET transistors, it is necessary to raise or lower the door voltage slowly. This behavior can be easily configured with a capacitor (C1), connected between the port and sup- port of Q2 (figure 62). Through a high-value resistor (R1), the capacitor charges slowly.
The resistor R2, placed in parallel with the capacitor, ensures that it will be discharged at the next drive. Its value determines the discharge velocity of C1 and must be at least twice that of R1, so as not to decrease the gate drive voltage too much. With low supply voltages, the reduced door voltage can cause erratic operation.
The free end of R1 can be connected to any circuit that provides the logic levels high(value close to the power supply) and low (value close to zero). Sort of as if we were to connect this resistor to the positive, to turn on the LEDs, or grounded, to turn them off.
Attention should be paid to the behavior of R1 when grounded (low logic level). It will, in this case, be in parallel with R2, making the discharge of C1 faster. If the technician does not want this behavior, a signal diode (D4) can be inserted in series with R1. As it is connected (figure 62), D4 can only inject voltage into the circuit, not withdraw it.
If a PWM drive is desired, it is desirable to eliminate C1 because it may influence the switching depending on the frequency used. Or its value must be reduced until the interference, seen through the oscilloscope, is insignificant. R2, in this case, will also be removed.
With this configuration, we have a simple, reliable, efficient and low cost drive system.
Common Darlington transistors can also use a capacitor for the delay function. But since their base current is much higher, the value of C1 will have to be higher. Not so much for Darlington transistors, which are a middle ground between bipolar and MOSFET, due to high Hfe.
Other practical circuits for LEDs
On the internet there is a lot about LEDs. I have chosen some circuits that I consider interesting and that help to give an overview of what it is possible to do in the area of illumination.
Belza is an excellent electronics page, from the Czech Republic, which has very interesting ideas. For example, a complete design of a 230V LED bulb  where the author shows what happened to the white LEDs after 7 months. Or a small night light, from those that are in the socket, also to connect in 230V . Some articles have dual language (English and Czech). Even so, the Czech language is not a problem, as you can ask Google to translate the page …
In Circuits Lab , there is a diagram of a cluster of red LEDs for brake light (brake light ). From the same site, comes a pen-type flashlight scheme , which uses 4 button cells. It’s based on an Elektor circuit.
Finally, a control project for RGB LEDs, coming from the Electronics-DIY page .It has printed circuit board design and can be modified to feed RGB LED strips. The circuit continuously modifies the emitted color, in the same way as some commercial products. It does not use microcontroller, only a CMOS 4029 and a 555, plus a few more components.
For all the circuits shown it is necessary to observe the recommendations on the maximum allowable voltage, reverse voltage, static electricity and overcurrent. By carefully designing LED power circuits, with the proper protections, respecting their limits, they can provide good services for a long time. Incidentally, this should be the rule: mount a circuit and then forget that it exists, because whenever you need it will be doing its part. Life changes too much for us to always do the same things.
It is noted in the text that the reference voltage for the semiconductor junction is not unanimous. Depending on the source consulted, we find values from 0.5V to 0.7V. It is not an appreciable difference, beyond which, in the assemblies, deviations of the projected values always occur. In electronic circuits, practice always improves theory, because depending on the origin of the component, more or less important changes in expected behavior may occur.
In this article, I did not think of powering the LEDs with pulses, only with direct current. PWM-type power supply for regulating current intensity is more complex to implement and is better suited to circuits with dedicated microcontrollers or integrated circuits such as the TL494 or SG3525.
Another issue is the estimated useful life of the LEDs. This article is intended, firstly, to demonstrate suitable ways of connecting them, so as to ensure their useful life. Another objective is to expose their characteristics so that they can be better known. Because the use without any care makes their durability much smaller than advertised. This has occurred on a large scale in the manufacturing industry, which seeks to reduce costs just where it should not: the LED drive circuit.
And a detail about energy expenditure. Despite the much-hyped economy, even LEDs are not a perfect example. They are only the reflection of the current stage of our technology. Because if incandescents turn light only 5 to 10% of the energy they receive, the LEDs are only slightly better. Of the total, only 30% turns light, the rest is heat. And the fluorescents are in the middle of these values, as they are worse than the LEDs. So we still have a lot to improve, for our future.
Connecting LEDs to 220v without using a power supply …
LED Tape connectors
Physical Computing With Scratch
how to connect LED light 250V AC without transformer
Blue light LED works without any resistor and doesn’t burn …
How to connect led strips without soldering
MacBook Repair Tip: Burned or Damaged LED connector port …
How to Install LE 12V Led Strip Guide ?
Powering lots of LEDS from Arduino – Erik Katerborg
How to parallel series connection the LED strip lights …
How to Connect Color Changing LED Strip Light to RGB …
LED Light System for RC Remote Radio Controlled Plane. How …
How to Build an LED Display, #1 Basic Wiring and Setup …
50W High Power LED is as bright as a 500W Halogen
Arduino Tutorials for Testers: build a Light Dimmer
How to connect and control WS2812 RGB leds with an Arduino …
How to Install a Dimmer for LED Strip Lighting
how to connect led lights to 220VAC
Light-Emitting Diodes (LEDs)