Dimmable LED Instrument Lights, Using the headlight switch dimmer Options

[Larry.Hubby]

I’ve seen some threads on changing the instrument bulbs to LEDs, and noticed that several of the respondents have wanted a scheme, not only for dimming the LEDs, but one that also uses the existing rheostat built into the headlight switch to do this. I built a system like this for my car, so I thought I’d post a how-to covering what I did.

I apologize in advance that my description is so long and detailed. I did it this way because I’ve noticed several people here claim to know little or nothing about things electrical. Those of you familiar with electronics can skip ahead to the circuit and substitute your own practices for what I did in many places.

The positive features of my scheme are:

1. No need to disassemble the gauges to install the LEDs.
2. Uses the existing dimming control in the headlight switch.
3. Easily provides 2-3 times the light output of the original instrument bulbs, while drawing half the power. If desired, even more light is available by re-adjusting circuit parameters.

And, as with any scheme, there are always prices to be paid:

1. The LEDs require some work to mount.
2. Some re-wiring and construction of external circuitry is necessary, although the circuitry involved is very simple and quite inexpensive.
3. The LEDs are rather expensive (~$5/ea), but only 6 are required (for a standard 3 gauge 914 instrument panel; 8 are needed for the 911-style 5 gauge panels that some of us, myself included, have).

My basic approach is to use high power LEDs designed for general lighting applications, one to replace each existing instrument bulb, instead of a large number of much lower-powered LEDs. Instead of having a maximum current rating of 30ma like most small LEDs, these can be driven with up to 1 amp and produce correspondingly more light. This higher current draw is what makes it feasible to use the existing dimmer rheostat, although some additional circuitry is still required to get both the desired brightness and a wide dimming range simultaneously.

The LEDs I’m talking about are the Lumileds Luxeon Rebel series, which look like this:



They’re small, just 3 x 4.5 mm, and available in colors from red to blue, and three versions of white. The data sheet for the general purpose white ones is available on-line at: http://www.philipslumileds.com/pdfs/DS64.pdf,
and small quantities can be purchased on-line at: http://www.philipslumileds.com/products/luxeon-rebel-white .
First click on the color you’re interested in under “Products”, then find a part that has a small shopping cart icon on the left in the list that opens up. If the icon is that of a telephone, rather than a shopping cart, that particular part is not available in small quantities. A click on this icon will take you to the Future Electronics website, an online dealer for Lumileds parts. For the case of warm white parts, which will produce light closest to the stock appearance, the part number I recommend is LXML-PWW1-0050. These were available in small quantities as of the date of this writing at a price of $4.73 ea.

The package size of the Rebel parts is small enough that we can mount them in the same locations as the existing bulbs, although this isn’t quite as straightforward as it may appear, and will be described in detail later.

We can get all the design info we need for this application from the datasheet and the following set of measurements. Below is a graph of measured data for light output vs current, comparing one of these warm-white LED to a stock 2 watt instrument bulb:



The upper curve is the LED and the lower one is the stock bulb. As you can see, in addition to producing over twice the light at the same current as the full-on bulb, the LED’s output is nearly linear vs current, while the bulb still draws 2/3 of the full-on current when dimmed nearly out. Consider also that the LED is running on about 3.2 volts at 150 ma while the bulb needs nearly 13 volts at the same current. In the stock arrangement in the car, there are six of these illumination bulbs running in parallel, two in each instrument, so the total current is about 900ma. We can hook up the LEDs in series/parallel combinations to make up part of the voltage miss-match between the 12V car electrical system and the LED requirements, and to share the same current among more than one LED.

It might be tempting to use four LEDs in series, which would add up to a voltage requirement of 12.8 volts at ~150ma, which is almost exactly actual battery voltage. I have tried using two such series strings of four LEDs in parallel for the 5 gauge dash in my car, and it does work, however, it is not a very good solution for several reasons. Since the combined current draw of the two strings is only 300ma instead of the 900ma that the rheostat is designed for, and since the LEDs dim more linearly with current than the bulbs, the dimming range is reduced to about 2:1 instead of about 10:1. Also, the LEDs respond instantly to changes in current, which can be caused by system voltage fluctuations due to the action of the voltage regulator or large current load changes, while the stock bulbs tend to average these changes out a bit because of their thermal “inertia”. So, if we hook things up this way, the instrument illumination will often flicker with changes in engine speed, etc. The voltage requirements of LEDs also tend to change a bit over time and “age’ with use, and, although the change is small, it doesn’t take much change in terms of voltage to greatly change the current draw of the LEDs. If the aging increases the voltage requirement, the brightness can drop permanently and substantially, leaving you wondering what happened to your LEDs that are supposed to last forever. Another potential problem is protection from damage due to reverse-voltage spikes that can occur in a car’s electrical system due to inductive components like relay coils, electric motors and alternators. LEDs are diodes, but, due to their structure, they have very little ability to withstand reverse voltage. Hooking them up backwards to a source designed to run them usually won’t hurt, but inductive spikes can often reach over a hundred volts, and even a brief reverse spike of that magnitude can burn them out. Some LEDs intended for automotive use have reverse voltage protection diodes built right into their package, but, alas, these do not. We could simply add a diode with a high PRV (Peak Reverse Voltage) rating in series with our strings of LEDs, but this protection diode will drop an additional 0.6 – 1.0 volts, which will reduce the current through the strings of four to under 100ma, and practically eliminate the dimming range of the rheostat. Additionally, a series string of four lamps is a very unhandy number for the majority of dashboards, which require six lamps total.

OK, that’s a long-winded explanation of what won’t work very well, now let’s get to what will work. We can eliminate most of the problems mentioned above if we use fewer LEDs in each series string and run the strings in series with an active current regulator that continually adjusts its series impedance to keep the current through the LEDs constant at a pre-set value. The regulator will then adjust for LED aging and system voltage variations, and there will be enough voltage available to include a reverse voltage protection diode. Fortunately, there is a readily-available, inexpensive IC that can do just this. Even more fortunately, the constant current control circuit for this IC doesn’t require a high resistance potentiometer, and instead works fine with the 15ohm range of the headlight switch rheostat and with the fact that it’s a two-terminal rather than a three-terminal device. Here’s the circuit:



The way the regulator IC works is that its internal circuit tries to maintain the voltage of its “out” terminal 1.25 volts above that of its “adj” terminal by adjusting the impedance between the “in” and “out” terminals. The current that it will regulate to in the above circuit is then given by I = 1.25/(R1+R2). Variable resistor R1, the headlight switch rheostat, varies from 0-15ohms and then jumps to infinite ohms (open circuit) when the knob is rotated fully counter-clockwise. Resistor R2 is in the circuit to set the maximum current when R1 is at 0 ohms. When R1 is at infinite ohms, no current can flow from the “out” terminal to the LEDs, so they turn off, as one would expect. Actually, a small amount of current (~10ma) still flows from the “adj” terminal to the LEDs, lighting them very weakly. This doesn’t hurt anything, and normal operation is restored as soon as R1 is turned back to a finite resistance. At intermediate settings of R1, we can calculate the current from the above formula. From the current and the known voltage drop of the LEDs, we can calculate the power dissipated in each component. This lets us see what the maximum values are, and then to know what sort of power rating and/or heat sinking may be required. Shown below are some spreadsheet calculations like this for the cases of three and four strings of two LEDs each.



In each case, I’ve chosen R2 to give a maximum current through each LED of about 175ma. From the initial graph, you can see that this will produce about three times the light output of the standard 2 watt bulbs. As you rotate the headlight switch knob counter-clockwise, the intensity will dim down by either about seven or about nine times before going out completely at full CCW. The dimming range is equal to the ratio of the maximum to the non-zero minimum current, which is equal to (R1max+R2)/(R1min+R2). Since R1max = 15 and R1min = 0, this is (15+R2)/R2. Thus, the smaller R2, the larger the dimming range. The range for both these calculated cases is reasonably close to the stock range, hence probably acceptable for most of us.
The calculated power dissipations for the regulator IC, R1, R2 and D1 are also shown. As you might expect, the maximum power dissipation for R2, D1 and the regulator occur at full brightness, while, since a zero ohm resistor can dissipate no power, that for R1 occurs at a somewhat lower brightness setting. A standard 1 watt resistor will work fine for R2, and the less than one-quarter watt requirement for R1 is substantially less than the power it was designed to handle originally. 400PRV fast recovery diodes rated at 1A are readily available, and will be more than adequate for D1. The roughly 3 watt dissipation in the regulator IC is no problem, but is enough to require a heat sink, and even with a heat sink the device will heat up to 130-140 deg F when used in normal warm ambient temperatures. In addition, there is a mounting problem with the regulator, since the metal tab that is the thermal path for heat removal is in electrical contact with the “in” terminal, i.e. the switched 12 volts, so it can’t directly touch anything that’s grounded to the body of the car. Probably the best solution is to mount the IC, R2, and D1 in a small metal box that can be clamped to the body somewhere. The box will then protect the circuitry against shorts and mechanical abuse, and will form a suitable heat-sink for the IC. There’s nothing special about the box I specified, other than that it’s the smallest standard size I could find. The regulator’s TO-220 package can be attached to the box with a mounting kit that provides a thermally conductive but electrically insulating pad to go between the IC’s tab and the mounting surface, as well as a screw/nut/flat washer/lock washer and insulating shoulder washer combination that allow clamping everything together without creating any shorts. The table below lists the items required to build the regulator circuit as described, together with their part numbers at Mouser Electronics (www.mouser.com), one of several electronic component suppliers from which such parts can be ordered on-line with a credit card and shipped to your door. The shipping charges for all the parts on this list will likely exceed the cost of the parts themselves.



To mount the completed box as described herein, you will also need 2ea. #6 captive nuts such as PEM #KFS2-632, and 2ea. 6-32 x ½” screws, plus some scrap aluminum sheet, 1/16” or so thick Mouser doesn’t stock these hardware items, but local hardware stores are likely to have them. If you can’t find them elsewhere, McMaster-Carr (www.mcmaster.com) definitely has them.

We use series strings of only two LEDs here because two is the number of lamps in each gauge, and two adapts well to the cases of both three and five gauge installations. You could use strings of three LEDs instead, if, for instance, you have a three gauge dash but want to also dim the three gauges in your center console, which would require nine LEDs (assuming each gauge in the console has only a single lamp). You would then hook them up as three strings of three LEDs each, and use the 2.4ohm value for R2, as in the three string/two LED case above. The regulator will work just as well with no other circuit changes. It will just drop less voltage across its internal pass transistor. You could also change the color and/or type of LED (as long as the LEDs used can run at the same current) without problem. The main requirement is that all parallel-connected serial strings have the same number (and similar type) of LEDs. If you violate this rule, the string with the smallest number of LEDs will get almost all the current, will light up enormously brighter than the others, and may heat up so much they burn out. A side benefit to using three LEDs /string is that the power dissipation in the regulator IC will drop to about a watt, which would make it feasible to use a clip-on anodized heat sink if you really wanted to avoid mounting the circuitry in a metal box.

The ceramic package for these LEDs incorporates a metal pad that’s intended to provide a good thermal path to an external heat sink for the large amount of heat generated when these devices are operated at full rated current. We won’t need to run them nearly that hard to produce more light than the stock instrument bulbs, so the heat sinking isn’t necessary.

The conventional lamps we want to replace are mounted in cylindrical plug-in sockets about 3/8” in diameter and an inch long. Since the entire LED package will easily fit within a 3/8” diameter circle, a simple way to make a package equivalent to the lamp sockets is to glue them to the ends of posts that are made the same general shape. The posts can be made of almost any material as long as the LEDs can be glued to it. The posts I made for this purpose were of aluminum, and are described by the drawing below:



The larger diameter flange at one end is just to prevent them from being pushed too far into the sleeves that they fit in on the instrument cases and to provide a grip so they can be pulled out, and the hole down the center is to provide room for the wires to the LED contacts that need to be brought out to the back. I designed these parts to be the same size and shape as the conventional bulb holders they replace, however, the only important considerations are that the LEDs be held in close to the same position as the small light bulbs they’re replacing, that wires to both electrical contacts of the LEDs are brought out where they’re accessible, and that both contacts are kept electrically isolated from the instrument case and everything else that’s grounded.

Once you have the posts, you can assemble the LED lamps as follows.

1. Solder wires to the LED contacts.

This is the most difficult part of this project. The LEDs are designed to be surface-mounted on printed circuit boards, not hand-soldered, so their contact pads are quite small (~.025”x.050” with a .015” gap between them). Making good solder joints to these small pads requires a small soldering iron with a fine tip. The one I used looks like this:



This iron has replaceable tips, and this tip measures .05” wide at the end, which was fine for soldering to these small pads. If you’re not familiar with electronic soldering, two important things to keep in mind when making connections of this kind are: use high (60% or more) tin (and not acid core) solder, and keep the tip of the iron clean by frequently wiping it against a damp sponge.

A hole, drilled with a #37 drill in a small block of plastic (or something soft enough to not scratch the transparent plastic lens of the LED), can serve as an jig to hold the LED while soldering. Place the lens of the LED facing down into it, which will prevent the part from moving and will put the contact pads on the opposite side of the ceramic substrate from the lens on top where you can get at them.



The finished lamps will have to be wired in series/parallel combinations later, so allow 12”-18” of wire at this stage for each contact. Small, flexible 28 gauge stranded wires are fine. Use different colors for the + and – terminals. Red for + and black or brown for – are good choices. Strip only 1/16” or so of insulation off of each wire on the end to be soldered. Tin (coat with solder by heating with the iron and melting on a small amount of solder) both the contact pads on the LED and the stripped ends of the wires before making the final connections.



Lay the wires over the LED so that the tinned ends are over the contact pads with the wires themselves extending away from the edge nearest the pads. After tinning, the insulation of the wires may have shrunk back, exposing more conductor. If so, position the wire so that the insulation covers the larger thermal contact pad at the opposite end of the substrate. You can trim the excess conductor back even with the substrate edge afterward.

(To be continued on the next post)