Light fixtures - 24v LED bulbs

[TomC]

Unless you are using your bus now, I would wait to by any LED fixture for lighting in the bus.  LED's are being heavily worked on since it is the next best lighting system.  In the past LED's are doubling their brightness every year.  If you just want exterior lighting, for stop/turn/running lights, plenty are available.  But for interior lighting, I would wait another year at least.  Good Luck, TomC

[Tim Strommen]

Holy Krapp, you've invented the FLUX Capacitor 

LOL! Only my improved version doesn't require 1.21Gigawatts, nor does it need to be driven to 88MPH to light up (the side effect of course is no time travel ).

...When you buy the MR16 LED bulbs listed on the websites cited in the previous post, they will already have the resisters and diodes that are needed built into them so they are pretty much plug and play...

...At least that is what I found, they are already set up to just ad the correct voltage or power source, a switch, and light them up, therefore all you have to do is find or build an appropriate fixture for them...

I believe that I brought this up with a previous thread about a dash board LED, but in this case it's even more important:  A series resistor with high-power LEDs is a very inefficient way of doing things.  When you place a resistor in-line with an LED to consume the extra voltage from the supply, it burns the power not used for lighting the LED as heat – this gets worse the higher your supply voltage relative to the operating voltage of the LED...


For example:  Let's say you have a Luxeon K2 TFFC cool-white LED (3.65volts @ 1Amp – for about 160 lumens).


If you were to drive that LED from a 5 volt supply, you would need a 1.35Ohm resistor rated at 2Watts or higher to consume the power above 3.65 volts.  That would mean that you are burning 5Watts of power (3.65Watts for the LED + 1.35Watts for the resistor = 5Watts). This is about 73% efficient.


If you were to drive that same LED from a 12 volt supply, you would need a 8.35Ohm resistor rated at 10Watts or higher to consume the power above 3.65 volts.  That would mean that you are burning 12Watts of power (3.65Watts for the LED + 8.35Watts for the resistor = 12Watts).  This is about 30% efficient.


If you were to drive that same LED from a 24 volt supply, you would need a 20.35Ohm resistor rated at 25Watts or higher to consume the power above 3.65 volts.  That would mean that you are burning 24Watts of power (3.65Watts for the LED + 20.35Watts for the resistor = 24Watts).  This is about 15% efficient.



Granted – you can offset some of the in-efficiency by putting more LEDs in the series string, for instance putting in 6 LEDs in a 24Volt system.  This would raise the diode voltage loss to 21.9 volts, so a resistor would need to be 2.1Ohms and only 5watts or more.  This would also raise the efficiency of the string to about 91% (21.9Watts for the LEDs + 2.1Watts for the resistor = 24 watts, much less being burned by the resistor).  This does however come with a catch – if your voltage drops below 24 volts, your LEDs will dim or go out (since the resistor value does not change with the supply voltage).


        This is why the most common way to regulate "Power" LEDs from a higher voltage supply is something called a "buck" regulator.  This is a solid-state device that can regulate the current of an output (using a very low ohm value "sensing" resistor), and will switch on and off at very high speeds a MOSFET or Transistor to charge something like an inductor (think of that like the coil in a relay – only without the contacts...).  The inductor resists having voltage come in at a very high speed, so when the driver turns on the "switch" the voltage/current at the output rises slowly (this is a relative term – since this happens in microseconds or less).  When the voltage at the output is high enough, the driver turns off the switch and the voltage/current begins to fall very slowly (again relative term).  When the voltage/current gets low enough, or a time-out occurs – the driver again turns on the switch and the cycle repeats.  Because the driver is either on or off, the resistance of the switch is very low so only a little heat is generated.  Also, because only the power required to sustain the output is drawn – the efficiency is very high.  This type of regulator is know as a "switched supply".

        This has some benefits as well – most LED regulators have a dim input (either analog or PWM) so dimming is very easy (relative to finding a 5Watt 200Ohm rheostat...), they are also very small (since they don't put out a lot of heat – they can be smaller – making the whole design much smaller/cooler), and they allow for other advanced features, such as temperature compensated current regulation (where the driver will reduce the current output as the LED's temperature rises – so that the LED won't burn out prematurely).  These regulators also adapt to the supply they are given – so if you only need a few LEDs, you can drive them from a supply with a wide range of voltages (instead of designing it for just one voltage...).

        This is the basis behind my design (shown on the first page of this thread) – it allows a wide operating range so I don't need to replace it if I change its supply from 12volts to 24 (a change I had planned to do), and it offers temperature compensation so that if the LEDs get hot, they will not be driven as hard and will cool down (a self stabilizing loop).  If you have a whole rig power management scheme (like the one I'm drawing up), you can de-rate (turn down) lights so they don't draw as much power, but still put out light as your battery wears down (so they don't have to just go out when the load gets dumped).  Using the LED driver method just opens up a whole lot of possibilities that a resistor does not allow for.

Cheers!

-Tim

[Chaz]

Wow, talk about over my head!   
I think I will just use the 24v. LED bulbs I can find with my 24v. system. I did get the two I ordered, along with the sockets, in today. I think I can make things happen. 

Thanx for all the info guys!
   Chaz

[pvcces]

Very nice reply, Tim.

I've never used these buck regulators. I can see that it's something I will have to get used to.

Thanks.

Tom Caffrey

[JohnEd]

Tim,

Is a buck resistor a speed control with a coil in line?  I thought the speed control/dimmer was the way to go.  Is there any reason they can't be used and what is the advantage of the buck resistor over them.

The temp and voltage sensing features are just so interesting to me.  Very much want to see your design schematic.

You have to go a ways to impress PVCCES....congrads!

John

[DrivingMissLazy]

I used a 6 volt rated zener diode, low wattage, to drop the 12 volts to six volts on my restored 46 Ford. I converted the Ford to 12 volts and wanted to still use the 6 volt instruments. It worked great and never got hot at all.

I really see no reason why a person could not do the same to drop 24 volts to 12 volts. Cheap and easy by using a 12 volt zener.

Richard

[TomC]

The only problem with a MOSFET switching voltage regulator, or dimmer switch, is I can see the light flicker and it bugs me.  At least this is on a halogen light with dimmer switch when on the lower lighting setting.  I assume that it would in fact be worse with an LED since LED's react so quickly to on or off (no lingering light after it is turned off).  I would think the best regulation for your output is just to get as many LED's necessary to run at that voltage without regulation (like series wired Christmas lights).  But-then I'm not an expert on LED's either.  Good Luck, TomC

[Tim Strommen]

Below is a simplified schematic of the described switch supply.

Is a buck resistor a speed control with a coil in line? I thought the speed control/dimmer was the way to go.

Hi John,

That's the basic theory behind it - but it's a little bit different (and please allow me a slight terminology correction - it's buck regulator).  A PWM speed controller is an device that simply cuts-off and turns on the power in rapid sequence - based on the user's set-point, not based on any circuit conditions.  Since the setting of the dimmer is external to the system, and does not get affected by the conditions of the system - it is defined as an "open loop system".

The "coil" (an inductor), is not a "consumer" of energy but rather a "resistor of energy chage" - much like a capacitor.  Where a capacitor acts like a "short" for high freqencies (i.e. it allows changes at a high rate) and blocks lower frequncies or DC (resists change by dumping its stored energy, or storing surplus energy), an inductor acts like a "short" for low frequencies (i.e. it allows changes at a slow rate) and blocks higher frequencies (resists change by dumping its stored energy, or storing surplus energy).  With the buck regulator, there is also current-sense resistor in the circuit (this IS a "consumer").  This resistor works on the same principal as every other resistor - only the circuit is designed to cash in on the fact that for a fixed resistance, if either the voltage or current changes - the other will change in same direction.

E
---
IxR

Where:

E = voltage
I = current (in amps)
R = resistance

If you use a fixed resistance - like say 5-Ohms, now multiply a current value like 700mA (0.7Amps) to get the voltage across that resistor (5 x 0.7 = 3.5volts).  If the current changes to say 900mA (0.9Amps), the voltage will change to (5 x 0.9 = 4.5volts).

In order to measure across a resistor, the driver IC needs two points.  Typically with high-side switch drivers, this current resistor is between the load and ground, so if the driver IC has a ground, that's one point.  A current sense pin that hooks up to the side of the resistor between the resistor and the load is the other point.  In the schematic the second pin is marked as "CS" on the driver part.

This lets the driver IC "watch" the current of the load by watching the voltage across the current sense resistor (if the current goes up, the detected voltage goes up, and if the current goes down, the detected voltage goes down).  This is compared to an internal voltage reference (since this is a very low current reference a very accurate linear regulator can be used without generating a lot of heat, or burning a lot of power).  When the voltage is above the reference the switch turned off, and when the voltage falls below the reference it turns the output back on.  This makes this type of control, a "closed loop system" where the system reacts to conditions within the same system.

Based on the size of the coil (measured in "Henrys", and marked by "h"), the speed of the switching action can be very fast or very slow (but with faster frequencies, the overall efficiency goes down).  A common frequency range is between 30kHz an 100kHz (30 to 100 thousand switch cycles per second).  A low-Henry inductor causes the switching frequency to go up, where a high-Henry inductor causes the switching frequency to go down.  Higher Henry inductors for a given current rating require more wire to be wrapped - so as a result they end up being physically larger.  There are some small inductor designs built for very compact system (at the expense of a few percent of efficiency), which have very small Henry inductors - and they can have switching speed in the megahertz (MHz) range (I've seen a few designs at 3MHz and higher... that's 3 million on-off switch cycles per second).

The only problem with a MOSFET switching voltage regulator, or dimmer switch, is I can see the light flicker and it bugs me...
...I assume that it would in fact be worse with an LED since LED's react so quickly to on or off (no lingering light after it is turned off)...

Hi TomC,

With PWM dimming, this can be an issue depending on the implementation - and based on the dimmer topology used, the switching frequency CAN be very low.  I too have noticed that with low frequency PWM (like that which is used on newer cars for their brake lights), I can detect a flicker as the car passes into my peripheral vision.  This is because the PWM is not a true high frequency PWM (above 30kHz), but instead is a modified "pseudo PWM" at a very much lower frequency (sub-100Hz, so it uses differing "chunks" of on-off time - this is probably what you are seeing).  I've discussed this with some people in the industry and it seems to be a trade-off in dedicated hardware vs. the use software on slightly more expensive in-vehicle networks.  Basically, more and more systems in a car are being pulled into a computer as it helps with diagnostics and management (something the car makers will kill for).  There just aren't enough processor cycles to run PWM at high frequencies (the change of the current sense line would be too fast for the software to react to), and there isn't enough incentive to use dedicate hardware drivers in every fixture (yet - prices are coming down, and this will soon be a "past" problem).

I have never met anyone who could see the flicker in a true PWM dimmer on LEDs at a frequency over 1kHz (and I'm in the video industry - where everyone's eyes are about as good as it gets ).

One other point of interest - because the current sense voltage does not have to fall all the way to 0 volts (which means 0 current across the resistor - and through the load), the brigtness doesn't change as aggressively as with PWM dimming.  It may fluxuate about 5-10%, and you will not be able to see that (people need at least a 30% change over a long period of time to notice a difference).

I would think the best regulation for your output is just to get as many LED's necessary to run at that voltage without regulation (like series wired Christmas lights).

I did mention this before as an option to help with efficiency - but I mentioned the primary draw-back of this approach:  It's designed to work at or slightly above the design voltage.  If the voltage changes, one could over saturate the LEDs (causing them to over-heat and have a very short life), or they could just turn off from under saturation (no electrons can "jump" the gap in the LED die, so no photons are generated).  This is how the lower cost "single voltage" LED fixtures are designed.  As anyone in the LED industry will tell you, "The #1 killer of LEDs is HEAT".  Over-heating the LEDs will cost you longevity and reliability.

I used a 6 volt rated zener diode, low wattage, to drop the 12 volts to six volts on my restored 46 Ford. I converted the Ford to 12 volts and wanted to still use the 6 volt instruments. It worked great and never got hot at all.  I really see no reason why a person could not do the same to drop 24 volts to 12 volts. Cheap and easy by using a 12 volt zener...

Hi Richard,

First, I don't imagine that the instruments draw a lot of power.  This is a big help with zener regulation.  With these LEDs, we are frequently talking in the 500mA and above range (the highest power single-die LED I've seen draws 1.5Amps at 3.7volts - that's 5.5 Watts!). To run three of those from a 12 volt supply (the max you could do, because of the voltage drop of each LED [11.1 volts]), you'd need an 11 volt zener at 2Watts or more (this is achievable).  This would also cost you about 40-50 cents.  But, if your supply voltage dropped below 11 volts (a lead-acid battery is considered "dead" below 12.01v) due to some high transient load like a starter or something else, the LEDs would shut off.

Now if you throw in an alternator to that, where the charge voltage is now 13.65 volts or up to 14.4 volts - the zener solution is a bit more risky.  You can't just add another LED to raise the regulation point up to 14.8 because the LEDs would never get enough voltage to turn on - also, because the voltage is now 2.65 volts above the regulation point of the zener, it now has to dissipate 4Watts of power to keep the LEDs safe.  This means your efficiency has gone down from ~92.5% (with the 12volt supply) to only ~80.4% with the 13.65 charging supply.  That's a big penalty, and a lot more heat that has to be dealt with.  If your supply does actually go up to 14.4volts, your efficiency takes another dive to ~76.4% [almost a quarter of the power consumed is lost to heat!] and you need a Zener that can absorb more than 5Watts (these power ratings start to get hard to come by in small quantities...).

Buck regulators are more constant with their efficiencies, with the National Semiconductor part I mentioned staying above 94% no matter what voltage you give it - up to 60 volts!  One other problem is that power LEDs tend to change their equivalent series ressitance (ESR) in the negative as they get warmer.  This means that they can draw more power as they get warmer, shortening their life even further.  A current regulation supply like a buck which watches a voltage based on the load - has a better chance of regulating a dynamic load like a power LED.  More advanced LED driver ICs even integrate a negative temperature coefficient resistor that changes with temperature variations - to create another reference voltage to compare to (and if that voltage changes - it can again be used to either add or subtract drive current based on the circuit design).


Again to each their own, and there are many "owns" that will work.  One person's solution may not be another's - or it might be.

Cheers!

-Tim

[Tim Strommen]

TomC,

  I just re-read your post - and I think I discovered the problem.  It sounds like the PWM device you are using is a fixed on-time, variable off time PWM modulator (aka a variable frequency PWM).  This is not a good device to use on lights, rather a better method is with a fixed frequency, variable DUTY CYCLE.  This does not change the start time interval of the "on" signal, but shortens the length of "on" portion of the total period.

As illustrated in the below picture, I've shown a 50% wave-form (what the voltage would look like on a scope) for both the fixed frequency and variable frequency PWM generators.  Note that they both have four "on" (rising) events from left to right.  Now look at what happens when you only want 25% of the period to be on - with the variable frequency PWM, it adds time to the "off" portion of the signal, whereas with the fixed frequency variable duty cycle PWM - it subtracts the same ammount of time from the "on" portion as it adds to the "off" portion.  This retains the four time periods for the same frequency (for both 50% and 25% outputs), but the variable frequency PWM has dropped to only 3 time periods - this causes the load to have more "coast down" time - and for a halogen bulb this would let it cool off and put out less light until the next "on-time".

With a fixed frequency PWM, there is a consistent period until the next on-time, which will not in all likelyhood allow the load too much coast down time to cool off - so there should not be flicker as the duty cycle is dropped off lower and lower (the output should dim smoothly).  If we were to assume that the below wave-forms were looking at a second in time, this would show that the 50% wave-forms had a 4Hz frequency.  For the 25% wave-forms, the constant frequncy PWM kept the 4Hz frequency, while the variable frequency PWM dropped to 3Hz.

Normally, a fixed frequency PWM would be above 100Hz (more like 28kHz or higher), to keep loads from "whining" (a motor will emit a tone at the frequency of the PWM - and if it's in the human audible hearing range, it will sound like the motor is squeeling... because you can actually hear the current "banging" through the windings).  TV systems are designed with a 60Hz or 50Hz refresh rate - while movies (Film) have a 24Hz refresh rate.  With these low rates often a flicker or "judder" is noticable in the picture.  At PC rates over 85Hz, most of these issues go away.


At any rate (punn) - if you have PWM that does variable duty cycles, your result will be better.

Cheers!

-Tim