No announcement yet.

IMPORTANT! New Member LED info

This topic is closed.
This is a sticky topic.
  • Filter
  • Time
  • Show
Clear All
new posts

  • IMPORTANT! New Member LED info

    this is for all of those new members interested in led modifications that are drowning in the sea of information that is this led section .
    my goal is to keep the primary things you need to know right in this one location for easy reference and hopefully a better understanding of what you're getting yourself into!

    this thread will remain closed to keep it as uncluttered as possible, below is a link to the discussion thread for anyone interested in adding anything to this guide. i'll start it off with what i can think of right away, but i'd love to get input from everyone else too.

    1. led selection
    2. series/parallel information
    3. resistor selection
    4. voltage/current regulation
    5. pulse width modulation(pwm)/dimming
    6. circuit board design
    7. soldering methods/techniques
    8. high power led's
    Last edited by soundman98; October 9th, 2013, 11:10 PM.
    The time you enjoy wasting, is not wasted time

  • #2
    1. LED Selection

    by now, everyone knows of some source for led's. there's thousands, if not millions of retailers for every led ever made. so how do you pick the 'right' ones?

    a good start is to go over the defining qualities.


    just like trying different food items, brand can be very important. there are some brands that just leave a bad taste in your mouth, where other brands taste great.

    led's are no different. luckily, this forum has been around for many years, so all the hard work has already been done, and the recommended brand list has become quite short.

    recommended led brands are as follows:
    Luminous Devices
    Seol Semiconductor

    these are the main brands, there are some more, but require careful research and understanding of what and how they're meant to be used.

    but i found some led's on 'site such-and-such' for a fraction of the price!!! why can't i just use those?!?!? they'd be so much cheaper, and would save me a ton of money!

    when this led section was first started, many of the users did exactly that. i mean why not--when the product appears the same, what's the use in paying more then you need to?

    and they found that no-name led's had a multitude of issues, all stemming from very poor quality control.
    many will have poor binning.

    because of the way led's are made, each one is slightly different. they will each have a different color value, required voltage, and brightness. binning is the act of establishing specific boxes, or bins, for led's of similar qualities. in really simple terms, no-name led's use much wider binning tolerances, where name-brand led's will be binned under closer tolerances, so the final batch of product you receive is as consistent as possible. in some cases, name-brand tolerances are so tight, it is nearly impossible to distinguish between different bins without specialized measurement.

    many no-name led's will also have poor/inconsistent life spans. many older users had issues with some led's failing within 3-6 months while other led's continued working for years. this is the point where the cost-effectiveness of name-brand led's takes over. why constantly re-do sections of a project, when you can pay a little more for a higher quality product, and never need to touch it again? it's one thing to desire to do it again, but it's quite another to be forced to be constantly be re-doing portions of a project.


    there are many types of led's available, each one has a different use and requirement.

    for the most part, the most common type discussed and recommended is the 'superflux' type. it is exactly like the one pictured at the top of this post. it is a 7.62mm square, with one corner cut off, which usually indicates the 'cathode' or negative side of the led(this is typical, but is not a requirement!! always check the datasheet for the led's you use to verify things like this!!).

    why not 5mm leds? because they don't get rid of heat as easily for starters. a major misconception about led's is that they don't get hot. led's still create heat, although it is substantially less then other light sources, and heat is really what kills them over time, so the superflux design is better then normal 5mm led's because there are 2 extra legs to conduct heat away from the middle of the led. and with 4 mounting points, they are easier to mount flat, and are also more rugged when dealing with vibration, like the vibration of driving down the road.


    what color is best for automotive brake lights?

    while personal preference can play somewhat of a role in determining that, red-orange is the unofficial industry standard for brake lamp illumination color. this is because the color appears brighter to the human eye without actually being any brighter then a typical red led. how can that be?

    the easiest way to see that is to look at a human eye sensitivity chart:

    stolen from here:

    as you can see, we are naturally most sensitive to green, and our response drops off sharply to either side of that.

    a side benefit of using red-orange led's is that the color will also better match any filament bulbs that are used for the same function.

    if your head isn't spinning yet, it's about to be.

    first, a link to a datasheet--this is where everything comes from that needs to be known about any electrical device. when designing any electrical circuit around a component like a led, consider the datasheet the bible. there is a reason for every bit of information in it.

    what the heck am i looking at?!?

    let's go through it one page at at time:

    the first page is pretty boring-- typical marketing write-up. these are always nice to read, as they will always hit on the key features of the component. so if you're looking for a very specific feature, it can help so you don't need to waste any time going past that first page.

    but then this is at the bottom:

    that lists exactly what model numbers apply when looking at this datasheet. depending on the manufacturer, some products will have a different datasheet for every model number instead of combining a bunch of different model numbers to one datasheet.

    next, page 2 is probably one of the more useful pages.

    as you can see, this is the breakdown of exactly what each of those part numbers on the first page really mean. for the most part, the HPWT-MH, and HPWT-DH are the most commonly used versions around here. which one really depends on the needs of the tail lamp. for the most part, i personally prefer the DH version, as i believe the 60 degree lens offers a nice balance between crazy-bright, and width, without being too tight of a beam to make it difficult to see the signal off to the side of the vehicle.

    next is the mechanical drawings page.
    this is very helpful for anyone who decides to make their own circuit board in any auto-cad like program, where they need to make their own solder pad layout. i believe dedicated circuit board creation programs like eagle already have a superflux design installed.

    next is page 4, and now we're really getting into the meat of it!

    this is probably the most-overused page of the entire document. please note the title says MAXIMUM RATINGS. they don't give any recommended ratings, they just say that you can't go above the ratings listed here.

    50mA and above will not have any visible difference in brightness, so i don't recommend running them at 70mA because i don't see a need for them to consume any more power when there is not any difference in output.

    the 'forward current' is just the amount of power that is applied to the led. led's are known as 'current devices' instead of voltage devices. meaning that they are more effected by current changes then voltage differences.

    current is the pressure of power flowing through a device, where voltage is the 'carrier' for that pressure.

    'power dissipation' i've never used this, but it can be used to better calculate how many watts are being consumed by the final design.

    'reverse voltage' the way led's work, if they are connected backwards, this is the voltage level that it takes before any voltage pass through the led. this is bad, don't plan on doing this

    the next important part is 'soldering temperature'. notice how it specifies about 235 deg celcius soldering temp, and for no more then 3 seconds. while this is a really good spot to shoot for, realistically, this is very hard to do in the beginning, but don't keep the iron on one led leg for any more then 5-10 seconds at the absolute most..

    moving on to the next chart, it's mostly gibberish to me for the most part, it is not needed-- nice to know, but i haven't had a use for the flux measurements.. but if you decide to try different led's, the most important part of that chart is the wavelength part.

    osram is particularly bad about this. their 'amber' led's are the same wavelengths as everyone else's red-orange led's.

    the next page is possibly the most important chart out of the entire document.

    this is the page that specifies the forward voltage of each model number of led.

    the forward voltage is the level of voltage required to get the led to light up.

    for the most part, you will always want to use the 'TYPICAL' measurement on this page, as that will work with that model number of led's. for the most part, it is not worthwhile to attempt to test each led to check it's forward voltage beyond that.

    after that, comes 2 pages of output graphs. for the most part, there's really nothing to do with these, as most of the time, you'll be using the raw led. these are really only helpful if you need to custom design some sort of secondary reflector to better channel the light output.

    and then comes the binning data:

    and the next page.

    these two pages are essentially useless, as this next document delves fully into the binning of each lumiled.

    when ordering lumiled's, you can typically specify a 5-digit option code(g4000 for instance) but cannot specify a category, or bin code(typically a 3-digit code, b14 for instance). the option code is a fine-er-tuned category, though is still quite broad.

    in car terms:
    'chevy' is the same as 'lumiled'-- quite a broad subject, but narrowed down to manufacturer.
    'chevy caprice' is similar to the option code. it's further narrowed down, but still quite broad-- it doesn't specify the color, or engine type yet.
    'chevy caprice in blue with 3.8 liter v6' is like the full bin code-- we've narrowed it down to a very specific model with very specific features.
    Last edited by soundman98; March 19th, 2013, 08:50 PM.
    The time you enjoy wasting, is not wasted time


    • #3
      2. LED Series/Parallel Information

      i was going to jump straight into selecting a resistor for led projects, but i think it's important to go over this first, because it plays a major role in determining the requirements of a resistor.

      Series/Parallel wiring

      series/parallel circuits are always confusing, partly because there are so many options in creating things with them.

      when wiring a circuit in series, it means that power needs to flow through every single component.

      if any of the components are removed from the circuit, everything will stop working.

      when wiring a circuit in parallel, it means that there is more then one path for power.

      if any component is removed from the circuit, the rest of the components will continue to keep working.

      most led projects use a combination of these 2 techniques. referred to as a 'series/parallel' circuit

      if led #2 is removed, led's 1+3 will stop working, but led's 4 through 9 will continue working like nothing happened.

      this is done for a few reasons, most notably for power efficiency, and reliability. all of which will be talked about more later on in their respective sections.
      The time you enjoy wasting, is not wasted time


      • #4
        3. Resistor Selection

        led's normally require a low voltage to work, and in many cases, they are going to be connected to a higher voltage then what they are rated for. the only way to get them to work for more then a few seconds is to add a resistor.

        what does a resistor do?
        in some places, you will see the term 'current limiting resistor'. that is really describing what the resistor is doing. it's limiting how much current(amperage) the led(s) are allowed to take from whatever the input is.

        how do i figure out when i need to use a resistor?
        a resistor needs to be used whenever the voltage going into the circuit is expected to be higher then the voltage that is required by the led.

        what resistor do i need for my circuit?
        the size of resistor is determined by a couple things.

        before you can start to figure that out, a few questions need answers.
        1. what is the input voltage?(if using car voltage with no regulation, enter 14.5v, but voltage regulation is recommended-- usually 12v)
        2. what is the typical forward voltage of the led's?(if using HPWT-DH-G4000, enter 2.5v)
        3. what is the current level of the led's?(if using HPWT-DH-G40000, max current is 70mA, i normally run 50mA)

        after that, there's 2 ways to figure out what size resistor you need. one method uses one of the calculators listed below, the other method requires a pencil and paper. forewarning-- this site rounds all answers to the nearest E-series resistor, and will not show the actual calculated values, so multiple mA output settings can have the same recommended resistor values. this site appears to only be able to calculate for up to 32 led's, so is not normally used, but is nicer because it will give you a recommended common value, but also state what the actual calculated value is. i like that this one shows both the calculated resistance value, as well as the nearest common resistor value, but it lacks series/parallel calculations. in order to use this one, you can only calculate a single series string-- after adding up the forward voltage of the number of led's used.

        all recommend a resistor wattage-- just like a light bulb, wattage for resistors is how much power the resistor can take. if in doubt, it never hurts to step up to the next higher wattage resistor.

        this is a off-line downloadable calculator:

        i like this one because it gives the raw calculation values, but leaves the rest of the decision up to me on which resistors to actually use(when in doubt, always step up to the next highest resistor!)

        for those of you who still like to do things on paper, then here's the method for that:

        just like any electrical calculation, we need to start with all the formula's-- and they're all listed in one simple image:

        using the same specs listed in parentheses, we end up using a formula like this:

        (stolen from

        Es=input voltage
        Eled= led forward voltage
        Iled= led current

        but remember-- this calculation is for a single led. to calculate the resistor for 3 led's in series, the forward voltage needs to be added up. so instead of using the 2.5v figure from the spec sheet, it would be 7.5v. but because the led's are in series, the current level wouldn't change-- they all consume the same amount of power.

        but we're not done! we also need to calculate the wattage size of the resistor!
        to do do that, we would use the Isqrd x R calculation in the top right corner of the wheel. all that's left is to convert the answer from a decimal to a fraction. if the answer is 0.28, that would be more then a 1/4 watt resistor, so you would step up to a 1/2 watt resistor.
        Last edited by soundman98; October 9th, 2013, 08:50 PM.
        The time you enjoy wasting, is not wasted time


        • #5
          4. Voltage/Current regulation
          to restate my oversimplified definition of those terms:
          current is the pressure of power flowing through a device, where voltage is the 'carrier' for that pressure.
          googling "difference between current and voltage" turns up millions of results, many going into much finer detail then i intend to, if anyone's interested in learning more about what each of those terms really are.

          this topic is a little messy, as there are conflicting points, so it's bound to get a little confusing...

          for the most part, this guide will focus on voltage regulation, as it's the most common method used to control led's.
          current regulation is usually handled by devices called 'led drivers', and are usually referred to as 'constant current'. meaning they have a fixed current output level, but allwo the voltage to fluctuate based on the needs of the device. in order to try to avoid as much confusion as possible, that's as far as i'll get into constant current drivers, unless a need arises for it later.

          first- remember this from the led section:
          led's are known as 'current devices' instead of voltage devices. meaning that they are more effected by current changes then voltage differences.
          so while technically a current-regulated design is going to be better, most of the builds here regulate the voltage instead. i know-- it's a little confusing..

          for the most part, the led style that is most dealt with-- superflux led's-- can get rid of heat faster then they can create it. because of this voltage regulation works correctly. but if the led's were constantly put in a situation where they created more heat then they could get rid of, then a voltage regulated design would be the weakest link, and would be the cause of failures.

          a general rule of thumb is that led's that require 70mA or less are considered 'low power' led's, which means that typically, voltage regulated power supply designs are perfectly fine and will work with any led below that threshold.

          but any led that requires more then 70mA should use a current regulated power supply to help maintain a reliable working condition for the led's.

          as far as i know, there is no set standard for this (the above statements are primarily directed to brake light lamp design--where the lights will be on for extended periods of time), and are based on my own experience looking at a lot of different led's. in many cases, the 'on' time of the led's play a huge role in determining which type of power supply is required. if a higher output led is only used for a few seconds at a time, then a voltage regulated design would work in that circumstance.

          voltage regulation

          other then the fact that the most common led's used here can take it, the other main reason that this is the most common method is because it is the easiest to implement.

          voltage regulators are exactly what the name implies-- they control the amount of voltage going into a circuit.

          the most commonly used version here is the Sharp PQ**RD21. the * denotes the output voltage the 2 common one's are the 12v model, and 9v model, listed as '09' in that part number. there is also a 5v(05) model as well that isn't typically used.

          either are available from digikey.

          another model that is also very popular in electronics are regulators like the LM317, which is an adjustable-output regulator.
          and here's a list of a bunch of voltage regulators yellow_cake was kind enough to put together (KurumaOtaku suggested it be added to the list here)

          Originally posted by yellow_cake View Post
          I posted this in another thread, figure it would come in handy for someone searching later on...

          They are all LDO through-hole regs w/ on/off pin.

          Shows the mouser/digikey prices for 1 piece and 10 pieces. The prices may be different now as it was several months ago.

          PQ12RD21 Sharp (The Classic)
          2A Max Current
          On/Off switch
          Discontinued, still available through Digikey.
          Digikey price: 1 - $1.43, 10 - $1.14


          KA78R12 Offered from a few brands (Samsung, National Semiconductor, Fairchild...)
          1A Max Current
          On/Off switch
          Available through various suppliers such as Digikey, Mouser.
          Digikey price: 1 - $0.93, 10 - $0.82
          Mouser price: 1 - $0.92, 10 - $0.714


          KA278R12 (National Semiconductor, Fairchild...)
          2A Max Current
          On/Off switch
          Available through various suppliers such as Digikey, Mouser.
          Digikey price: 1 - $1.01, 10 - $0.895
          Mouser price: 1 - $0.98, 10 - $0.762


          KA378R12 (National Semiconductor, Fairchild...)
          3A Max Current
          On/Off switch
          Available through various suppliers such as Digikey, Mouser.
          Digikey price: 1 - $1.33, 10 - $1.18
          Mouser price: 1 - $1.33, 10 - $1.02


          KIA278R12PI (KEC Korea Electronics)
          2A Max Current
          On/Off switch
          Korean brand, available through Chinese suppliers (ebay, aliexpress)
          Price, around $0.70/piece for larger quantities (15-25), shipping included.

          voltage regulators come in all sorts of types, but any of the ones that look similar to the image above are normally called 'linear' or 'shunt' type regulators. very simply put, any amount of voltage over their set output level is sent directly to the negative wire coming out of the regulator.

          the linear type of regulator is the most common just about everywhere (in car terms, they are every electronics mfg's 4-door car-- everyone's got at least one model), as it's the simplest to use, because all models have a very low required external-parts count, and are very cheap to implement.

          the disadvantage to linear regulators is that they are not extremely efficient, because they will send whatever extra voltage over the output limit to ground, and by doing that, will create a lot of heat.. sidebar: i always recommend adding a heatsink to any linear vreg, at less then $5, it is cheap insurance to extend the life of the device. many datasheets will specify a limit of how much current the regulator can output before absolutely requiring a heatsink, but i feel it is best to include it regardless of that.

          linear regulators also have what is called a 'dropout voltage'. the dropout voltage is the voltage required by the vreg above the output voltage. look at this as a self-imposed 'tax' that the vreg requires to do it's job. so if the dropout voltage is 3 volts, that means that if you want the vreg to send 12 volts to the device, it requires a absolute minimum of 15 volts to do that. at this point, maybe you're seeing an issue with using a 12v regulator in a car hold on, there's more to this!

          from there, vregs can be divided into 2 categories:
          typically 'normal' or un-specified vreg's can have a dropout voltage of about 3 volts.
          but 'Low Drop Out' or 'LDO' vreg's will have a dropout voltage of less then 1 volt, usually closer to 0.5v

          the PQ**RD21 that was mentioned before is a LDO vreg, with a dropout voltage of 0.5v. which is how we can get away with using it in-car. typically, a running car will have a voltage of about 14.4v, while a car that is turned off will have a voltage of about 12.0v, depending on the state/condition of the battery. with only a 0.5v 'tax', that means that when the car is running, even if the voltage dropped down to 13.0v, it would still be 0.5v above what the vreg requires to work.

          but what happens when the voltage falls below the required dropout voltage?
          (edited by vegas f6)
          the vreg still needs to take the self-imposed 'tax', or dropout voltage, and then passes whatever voltage is left onto the device being powered.

          So in other words, if the input voltage is 12.6V and the required dropout is 2V (dependent on current and temps) then the output will be ~10.6V. But as the input voltage rises to 13V or drops to 12.2V then the output voltage varies the same amount. This is unregulated in that the output voltage is not fixed. But it will never exceed 12V.

          the difference in output voltage

          sall brought up that it would be a good idea to do a small section on the differences between the 12v and 9v versions of vregs.

          my first response to such a question is that it depends on the conditions surrounding the install.

          normally, i like to recommend the 12v regulator for any installation that is planned to be operated while the car is running(third brake light), and use the 9v version for any installation that is planned to be used while the car is turned off(dome light).

          selecting between either model has equal benefits and drawbacks.

          selecting the 12v model for anything means that running the device with the car off can almost always mean that it will be getting unregulated voltage. but when the car is running, this model will create less heat as it sends the extra voltage to ground, because there is less of a difference in the input and output voltages.

          selecting the 9v model for anything means running the device with the car off would still be getting regulated voltage. but when the car is running, this model will create more heat(which means it needs more of a heatsink), as there is a larger difference in the amount of input voltage to what the output voltage needs to be.

          there are formula's to determine how much power is 'wasted' as heat in linear regulators-- it's all in the ohm's law wheel
          as an example, we'll use a 14.4 v input, with a load that is 1A.
          for this, we'd use E x I = P on the ohms law wheel. or put a different way-- Volts x Amps = Watts

          12v regulator:
          14.4-12v= 2.4v. 2.4V x 1.0A = 2.4W

          9v regulator:
          14.4-9v= 5.4v. 5.4V x 1.0A = 5.4w

          according to the datasheet, the regulator can handle a load of up to 1.4 watts without a heatsink. in the case of this example, the 12v regulator could probably get away with a simple piece of aluminum bolted to it to work correctly. but the 9v regulator would require a much larger dedicated heatsink that is approved for getting rid of 5.4 watts of power.

          other types of regulators
          there are also 'boost', 'buck', and 'buck/boost' type regulators. all 3 are much less commonly built because they require a lot more parts, and calculations to design to correctly work. but they are typically much more efficient than linear regulators, as they all use the principle of turning on and off to control the voltage going out to the device, which also tends to create a lot less heat.

          buck vreg:
          contrary to the name, these still step the voltage down just like a linear regulator, but do so much more efficiently-- in many cases with much less heat output. it does this by using a capacitor and a inductor, and will switch the power on and off very fast. because of the properties of combining a capacitor and inductor, this creates the conversion process. more info can be found here:

          boost vreg:
          this design actually follows the name, and will boost the voltage to a higher level then what is input. it also switches the power on and off really fast to accomplish that, and similar components are used to accomplish that.

          it not only steps everything up to a certain voltage, but can also automatically step everything down-- which means that things like 'dropout' voltage aren't an issue anymore, and the design remains regulated all the time. these can be more difficult to design, and have their own set of rules that go along with using them.

          just as with led's, the datasheet should tell you everything you need to know about any model vreg, so ALWAYS check the datasheet.

          main things to look for in a vreg:
          with so many different models, how do you pick the right one?

          for starters, 2 things need to be determined:
          1. what is the planned input voltage and conditions the vreg will be used in?
          2. how much current is it going to need to supply?

          1. when using a vreg in a car, you'll need to decide if the circuit is going to be used only while the car is running, or if it will be used a lot when the car is off.
          in most cases, things like tail lights will be designed for use while the car is running, while things like dome lights are designed for use while the car is off.

          2. this is best determined by calculating the current using one of the resistor calculators from the "resistor selection" section above this one.

          after those things are determined, all that is needed is to install the vreg in the completed circuit.
          vIn always refers to the positive supply-- connected to the car-side of everything.
          vOut always refers to the positive output-- which will always connect to your device that requires regulation.
          gnd always refers to the negative connection-- which always needs to be connected to the negative input wire from the car, as well as the negative from the device that requires regulation.
          Last edited by soundman98; November 3rd, 2015, 08:55 PM. Reason: add vreg list
          The time you enjoy wasting, is not wasted time


          • #6
            5. Dimming led's

            there's really 2 ways to dim an led. pulse width modulation(pwm), and current limiting.

            pwm is the preferred method by many oem car makers because it provides the most consistent brightness control with led's.

            azdave did a excellent job explaining it in the past, so this is just copy-pasted from that explanation

            Originally posted by azdave View Post
            Update: Additional reading info added at the bottom of this first post. azdave 11-07

            I‚€™ve been trying to think of an additional way to explain to people on the forum how a pulse width modulator actually manages to ‚€œdim‚€? LEDs. I put the word ‚€œdim‚€? in quotes as the PWM does not dim them at all. It is an illusion for our eyes to enjoy.

            I‚€™ve always looked at wave form graphs showing duty-cycles and frequencies along with wordy PWM explanations (yes, some of my own too) and thought to myself "That is really not very helpful to people who are just learning about these things". I‚€™m a visual person and I‚€™ll understand things quicker if my eyes are involved in the learning process (other then reading).

            After building a PWM for a tech article, I decided to get the camera out at let it reveal the facts in visual form.

            Here are a series of photos that I think will help everyone understand a little better how these circuits work. In all of the photos shown the camera‚€™s settings remain the same. It was in full manual mode and the aperature was stopped down to F/8. Shutter speed was set at 1 second. I did not worry about a perfect focus as it does not matter for what I will demonstrate here. In the four photos where I was intentionally creating movement, the rate of motion was kept very similar so as not to skew the results.

            In this photo, the LEDs are not in motion. The PWM is adjusted to about 5% duty-cycle. To my eyes the LEDs looked quite dim. Exposure was locked at 1 second so it doesn't seem all that dim to my camera as it had one full second to keep gathering light.

            In this photo, the LEDs are still not in motion. The PWM is adjusted to about 96%. To my eyes the LEDs are extremely bright. Exposure was 1 second again.

            Now I lower the PWM duty-cycle back to 5% but this time I am moving the LEDs very quickly past the lens as the exposure records a single pass of the LEDs. The camera reveals pulses of light as the LEDs ‚€œfly‚€? past (if these LEDs were dimmed by using a big current-limiting resistor we would see a long arc of very, very dim green light). The LEDs here are off most of the time. When they are on, they are on at full current levels. This is the first clue of what is happening. With no movement, as in the first picture, those pulses of light were allowed to "pile up" on top of each other and appear brighter than they really were. Now with movement you can see them alone for time.

            Now I‚€™m up to about 25% duty-cycle. The arc streaks are getting a little longer but still the LEDs are off more than they are on. See how the LEDs streaks are the same brightness as the previous picture. They are on for a longer period of time but they are not any brighter.

            Now I‚€™m up to about 75% duty-cycle. The arc streaks are longer and now the LEDs are on more than they are off. The streak brightness is still the same.

            Now I‚€™m at 96% duty-cycle. The arc streaks are nearly continuous as the LEDs are barely turning off before they turn back on. Again the streaks are no brighter than the little dots of light we saw in the first photo showning movement.

            Pretty cool huh? The main thing to take note of is that in the last four photos, the brightness of the arc streaks stay the same. The streaks of light simply got shorter and longer as I adjusted the duty-cycle. When those streaks are allowed to "pile up" upon each other, our eyes see it as a brightness change when in fact that is not really what occurs.

            So now maybe you can understand a little more clearly. A PWM does not adjust the brightness levels of the LEDs. It only adjusts the time periods when the LEDs are on and off and our brains sort of "averages " what it sees.

            Now more than ever, the term Pulse Width Modulator should be a little easier to grasp. You've seen the pulse and it's width plus you've seen what happens when we change it.

            Our eyes and brain will fall for this illusion every time (unless there is movement to reveal the magician‚€™s secrets).

            More PWM explanation added Nov. 07.
            Pulse Width Modulators Explained
            (as used in retro-fitted LED automotive lighting applications).

            I‚€™ll try to explain, in simpler terms, the functions of pulse width modulators as used in this forum.

            First and foremost! A Pulse Width Modulator (PWM) is a device used to manipulate power going to your LEDs.
            It is NOT a voltage regulator or voltage level adjusting device. You can‚€™t put 14 volts in and set it for 9-12 volts out (You might think so if you try to measure it with a VOM but you are being fooled by the VOM trying to average the on-off pulses).
            Note: You must limit the maximum current to your LEDs the same way you would if you were not using a PWM in your design.

            Basic description: A pulse width modulator is a circuit that gives you the ability to turn power on and off very quickly to your LED arrays. Depending on how you manipulate the on and off pulses, a PWM can fool your eyes into believing that the LEDs are dimming when in fact the LEDs are simply turning on and off faster than your eyes can detect it. This is useful because you can adjust the brightness levels much more accurately and uniformly than when simply reducing current levels by using additional current-limiting resistors. Please note: While a PWM is a very cool way to dim LEDs for use when an LED must function at more than one brightness level, most LED retrofit projects do not require a PWM circuit. Using a PWM is an extra feature that allows you some flexibility in the design and also gives your project professional OEM looks.

            When should I use a PWM? In large arrays where it is undesirable to run LEDs at extremely low current levels. Why can‚€™t I run LEDs at extremely low current levels? Some LEDs, especially ones that operate at higher current levels like Lumiled‚€™s SuperFlux series, may not dim uniformly when running them at minimal current levels. In a small array you might not notice the difference. In large arrays driven at only 2-5mA, you can see the effect quite noticeably.

            The reason a PWM can be used to ‚€œdim‚€? LEDs is that the human visual system will retain an impression for about 1/30th of a second. The exact time depends on the brightness of the image. This image retention is known as ‚€œpersistence of vision‚€?. This is the reason a PWM can make you think you have dimmed the brightness levels on an LED array when in fact you are only turning them on and off at full brightness faster than your eye is capable of processing the information. This whole ‚€œpersistence of vision‚€? phenomenon can be debated for hours with technical people but for us, we know it works, however you want to explain it.

            Generally, turning the LEDs on and off at any rate above 60 times per second is fast enough for most people to not be bothered by the flickering. As long as your PWM frequency is above 60Hz you probably won‚€™t be irritated while viewing your lights. It‚€™s an effect you tend to notice while not looking directly at the LEDs. You‚€™ll notice the flicker in your peripheral vision more often than looking straight at them.

            More info:
            A PWM breaks up the power going to your LEDs into tiny little segments. The number of segments that can fit into a one second time period is the frequency of the PWM. In the examples below I‚€™ll use a PWM with a frequency of 100Hz or in other words, the power to the LED will be cut up into 100 segments per second.

            Now if you only divide the power into little bits it doesn‚€™t really do you any good unless you can also manipulate each of those individual tiny segments of power. This is where the fun begins.

            Not only does a PWM have the ability to divide power up into little segments, it also has the ability to turn the power on and off within those individual segments. This feature allows the PWM to perform its magic for us. The amount of time the power is ‚€œon‚€? as compared to ‚€œoff‚€? in any given length of time is referred to as the duty-cycle.

            A 100% duty-cycle would mean the power is always on. The PWM is sending power to the LEDs with no interruptions at the full current level you have designed into your circuits. This is effectively regular old DC current as if the PWM was not even in the circuit. A PWM can not normally be set fully to 100% but 99% is enough to look like full brightness.
            At this setting it would not matter whether your PWM had a 1000Hz frequency or a 10Hz frequency because the power within all the individual segments would always be on without interruption.

            Now let‚€™s assume for instance a 20% duty-cycle. Again I am using a PWM with a 100Hz frequency for the example: At 20% duty-cycle, the power in any measured time interval will be on for only 20% of the time and off for 80% of the time. Note that during the time the power is on, it is still on at full current levels. The LEDs will actually only be on for 20% of the time (100 times each second). Your vision will ‚€œblend‚€? all the quick on and off pulses into what appears to be much dimmer LEDs. What‚€™s cool is that 80% of the time that you are looking directly at those LEDs they are completely shut off but you can‚€™t see it. This of course is very closely related to why florescent lights in your home appear to be on all the time even though they also are quickly turning on and off too.

            Some PWMs have potentiometers so you can easily vary the duty-cycle during operation but many PWMs operate at a fixed duty-cycle and you must substitute resistors and/or capacitors to change the duty-cycle. I prefer a PWM that I can adjust with a potentiometer since I never know what duty-cycle I want until the LED arrays are built and out in the real world for testing. Most adjustable PWMs will give you about a 5% to 98% duty-cycle adjustment range. When I show off my LED taillights at a car show at night time, I like to have them on at very low levels so I custom built a PWM to have a 0.5% duty-cycle.

            as far as how to implement pwm dimming, there is more than one way to do it.

            this is just a brief sampling of some of the available pwm units:
            this is the 'quickar pwm'. the design has completely changed since when azdave used to suggest it. it now uses a programmable IC instead of a 555 timer chip, and is a lot smaller then it used to be

            these 2 are just a few of the huge amount of units off ebay. when i first started looking at pwm dimmers, there was really only 3-4 models on ebay sold by many sellers, now the number of slightly different units is well into the 6-8 variations and quickly expanding..

            this one is really just a very basic, no frills pwm unit. no on/off or special features.

            this one is more of a 'deluxe' model-- it has start/stop buttons, as well as a selectable pwm frequency. the display is also pretty nice to have in some situations:

            all three of these units can be simply connected between the input power, and the device that needs to be dimmed. in the case of cars where a brake and dimmer tail circuit are needed, the pwm would be connected to the tail circuit via a diode, and the brake circuit would bypass the pwm via a diode. the 2 diodes are required to prevent the brake and tail circuits from 'seeing' each other and causing issues(hitting the brakes and lighting up the front parking lights for instance).

            here are just a few methods to setting up a pwm system with some of the above units. these layouts assume that the pwm unit uses a common ground connection. always check your pwm to make sure it is setup this way.

            layout #1

            for this circuit, it puts the vregs ahead of the pwm module. the diodes are used to 'hide' the vregs and the inputs from each other.

            when you have dual vregs in a circuit like this, and both are powered up, one will always output slightly less voltage(due to mfg tolerances), so if they were connected without diodes, the one that outputs less will always be trying to get rid of the extra voltage it sees from the other vreg, and the vreg that outputs more will be constantly trying to add voltage because the other one is constantly taking it away.. the battle finally ends when one of them burns up from the continued fighting.

            layout #2

            this uses only one vreg, but requires the same amount of diodes, and a relay(which could be substituted for a transistor, but i'm just not good with setting them up predictably)

            layout's #1 & #2 can be universal in the sense that they will work with any vreg.

            layout #3 relies on the vreg having a positive 'enabled' pin(in other words, connect it to ground, and the vreg turns off)

            this is most suitable for the sharp vregs recommended around here, but does require an extra diode over the other designs.

            the 1k resistor is connected to do what they call 'drain' the enable pin. simply stated, voltage can hang around on the enable pin, which can cause the vreg to stay on at the wrong times. connecting this resistor gets rid of that voltage hanging around, and keeps everything working as it should.

            current dimming

            current dimming is different than pwm dimming in the fact that it limits the amount of power going to the led's, so it will make the led's visibly dimmer.

            the main issue with the current dimming method with respect to the projects here(meaning most projects will typically have more than a single series string of led's) is that manufacturing tolerances of led's mean that led's in the same lineup will naturally require different power requirements when stepping down the current. this can lead to some led's/series-strings looking dimmer or brighter than the other led's/strings in the array when driven at lower than optimal levels.

            because of the potential difference in brightness, that is the main reason that oems prefer to use pwm dimming---it gives a more consistent end-product.

            to figure out the appropriate resistor, i typically recommend getting a variety pack of resistors and simply trying different sizes(higher resistance=dimmer). while there is a DOT standard for brightness of lighting devices, there is no set value that i can simply say that would match the DOT standards. it depends a lot on the led's used, how many, and how bright as a collective group they are.

            i was bullheaded(my family would be shocked ) about building the front turn signals on my car, and insisted on using the current dimming method for the parking light brightness. and it shows. at first, i thought that it was all rumors and wrong information because everything looked ok at first, but it didn't take long before a couple of strings started appearing a little dimmer than the others. it's not a huge difference, nothing that makes itself easily known just walking by the car, but easy to see if you look for it..
            Last edited by soundman98; October 9th, 2013, 08:51 PM.
            The time you enjoy wasting, is not wasted time


            • #7
              6. circuit board design

              for starters, i'll go over the types of circuit boards.
              all circuit boards are composed of 2 basic parts. a substrate, or 'base', and a very thin layer of copper, also known as 'copper clad' in it's raw unfinished form, or 'traces' in completed board form.

              Board substrates are composed of a few different materials, depending on the needs of the final component. FR-2, and FR-4 are the 2 most common versions, though there are many others.
              • in either case, the "FR" designation indicates 'Flame Resistant'.
              • FR-2 substrate is made of resin bonded paper, and is less mechanically durable, and is less electrically stable then it's FR-4 brother, most typically it is used in extremely cheap devices because it has significant cost savings over FR-4 substrate. of note, it is not recommended for automotive use, as sustained vibrations can cause cracks to form which will destroy any circuit traces above the crack, leading to a board failure.
              • FR-4 substrate is made of layers of woven fiberglass cloth, and is considered to be resistant to vibration and mechanical flexing forces. this is the type of substrate most commonly used in most devices.

              outside of FR-4 substrate, it can sometimes be difficult to locate boards precoated with copper that have a different substrate makeup, but they do exist. usually they are sold in very large quantities, so the general amount required for most projects this forum is based on would be considered to be a very small amount, and as such, make it difficult to acquire, but not impossible for the persistent.. there are usually flexible copper-clad boards available on ebay, which have the copper applied to a thin layer of plastic-- usually polyimide type, which can allow for very flexible circuits, with properties very similar to a standard sheet of paper, except conductive. though i'm not aware of a consistent source for such boards.

              board selection: there are a few different types of boards available. the correct one for you depends on a couple of factors, most notably, tools available, as well as the level of quality expected from the end product.
              • protoboard is usually considered the best option to use when circuit board design software and workshop space/tools are limited, not allowing for extremely custom designs, or where a custom design is not absolutely required. it is generally the fastest circuit board to build, because the design phase involves placing the components onto the board just before they are soldered together. generally, protoboard works best for square or rectangular designs, as the pre-stamped holes maintain a grid with 90 degree corners. the negatives to protoboard are the appearance if the board is visible, and the pre-stamped holes generally do not allow more custom designs, most noticeable with rounded or elliptical shapes, where a slight contour is required for only one side of the design.
              • there are 3 types of protoboard available.
                • ​non-plated
                • plated
                • lined

              non-plated circuit boards look like this

              non-plated circuit boards are not really recommended for general use, as they lack any method to solder components to them. they have their uses, but i don't believe they are useful under the context of this forum, as the most commonly-used component here is superflux led's, and they do not have long enough leads to be able to be fastened down to the opposite side of non-plated boards.

              plated circuit boards look like this

              plated circuit boards are one of the more commonly used boards when protoboards are used-- they allow for components to be soldered onto them, and can be cut with more basic tools.

              lined circuit boards look like this

              lined circuit boards are kind of like an 'advanced version of plated circuit boards. the benefit to these is that it allows for solid connection points. to get between points on plated boards, jumper wires must be used. with lined boards, the lines can be utilized for a portion of each circuit trace, allowing for a little cleaner looking board.

              next are full copper clad boards. there are 2 main versions-- one sided, and 2 sided. as the description implies, 1 sided boards have copper on only one side, where the other side is uncoated. 2 sided boards have 2 thin layers of copper, one on each side. more interesting things can be done with 2 layer boards, but the complexity of getting both sides perfectly lined up is significantly more difficult.
              this is a one sided copper clad board

              circuit board design
              there are nearly as many different ways to design circuit boards as there are circuit boards.. but the general advantage to designing a circuit board over using protoboard is that it allows for more custom shapes, and is generally more professional looking. but it is more time consuming, both in the design aspect as well as the board finishing aspect-- as the board must be cut to size, and all the holes need to be manually drilled into the board.

              usually most designs start off with some sort of layout, most typically with some sort of computer program. though some circuits can be made free hand with a marker and a ruler on the piece of copper clad.

              computer programs used here in the past: (to be edited later, i know i'm missing 2 or 50...)
              each program has it's positives and negatives, and i won't get into that, as i believe much of it is based on personal preference. for example, i only use progecad, because eagle/diptrace/expresspcb just don't make sense to me. that is not to say that they are inferior programs, because they aren't, in fact they are very powerful programs in the right hands, they just don't work for me personally.

              but the general idea is the same regardless of the the program used. the design is created within the program, then the circuit layout is sent to a printer, where it is transferred to the copper clad board, and readied for creating the circuit paths.

              methods for creating circuits

              after the circuit is drawn up, there is really a few different methods i've come across to make the transfer from digital image to completed circuit a little easier.

              the quik-n-fast method/ the dremel method:

              i like this method because it gives the benefit of custom layouts, but is relatively quick, and doesn't include any of the nasty chemicals. the down side is that it doesn't look the prettiest.

              just print out the design on a standard sheet of paper-- tape the design to a new piece of copper clad, and using a dremel cutoff wheel, cut apart the traces to create the circuit layout.
              (bewarned-- in this example, i attempted to drag a dremel drill bit along the board to create the traces-- using a cutoff wheel results in a lot cleaner looking circuit)

              the toner transfer method
              this method is more involved because it requires a laser printer, as well as glossy magazine paper, or some press-n-peel blue toner transfer paper(link to seller). but either method follows the guide already posted on this forum:

              and the last normal method is photo etching, which is essentially the same as the above, except it requires a few different things-- like requiring the circuit layout to be printed on a transparent sheet, a UV light source, and a presensitized pcb. one of the positives touted by UV-method supporters is that the design can be re-used. personally i don't buy into that because i can just as easily reprint a new copy as it is to re-use the UV method...

              this guide covers how to etch using the UV method

              The time you enjoy wasting, is not wasted time