(Re)-Lighting: [n] Sourcing Refreshingly Cool Illumination

June 2003 - Ongoing (Revised Mid-2006)

Compiled by Craig Dewick

The re-lighting project has been made possible thanks to the help of Robert Bennett from Ultraleds in the UK!

Stage 1 - Genesis of the Project

The rating of the small 'festoon'-style filament lamp was 10 watts. Using Ohm's Law to do some simple calculations, that means the lamp presented a load equivalent to 14.4 ohms resulting in a current of 0.8333 amps out of the nominally 12 volt car battery... That's a lot of current for a very small lamp!

Over the course of a night (say around 8 to 10 hours), drawing almost 1 amp constantly from the car battery was enough to reduce the battery's level of charge to the point where it could not supply enough current to maintain the terminal voltage at a level above about 12.5 volts which was needed to crank the engine via the starter motor when I tried to start the engine early the next day.

Once I further realised (via a tip-off from the person who'd owned the car before me! Later confirmed via wiring diagrams) that the light in the boot is not actually connected to any switch that's set up to be actuated by the closing opening or closing of the boot lid (similar to the way the door switches operate the interior light behind the rear-view mirror and on the left hand side at the top of the B-pillar), the genesis of trying to find a solution was created!

Having a background in electronics from when I was in my mid-teens, I knew that LED's (Light-Emitting Diodes) were an excellent low-current, high-efficiency light source, and since about 2000 (through my work as a train driver with a natural interest in railway signalling systems) I also knew that replacements for traditional filament-based lamps built from arrays of LED's were being produced commercial for use in railway signalling applications.

Some searching on the Internet and in other places turned up Ultraleds in the UK as the only place (as of the middle of 2003) which was actively involved in marketting LED-based filament lamp replacements suitable for use in automotive applications. Once I had a look at what was available, the decision to commit to purchasing one, then later a couple more, batches of LED-based lamps to try in my car, was made.

Examples of LED-based Lamps Sourced from Ultraleds
Selection of LED-based lamps	Lower LED board in red stop/tail and amber indicator lamps	5 mm white LED 'wedge' lamp useful in switches, etc.

Stage 2 - Initial Trials

At first I thought it might be something wrong with the LED lamps but via a little bit of googling discovered that the rear earth points at the very back of the central chassis member that supported the false floor in the boot along with the earth connections to both of the rear light clusters were failing simply due to age and being prone to becoming moist and corroded if any water got into the space under the false floor and then evaporated inside that space resulting in condensation on everything in the space, including the rear earth points. That started off another small project which actually has very long term benefits...

The standard 21 watt filament lamps used for four of the five lamps in the rear light clusters and in the front light clusters as well produce a large amount of localised heat and over an extended period this can cause the plastic mouldings of the light clusters to soften and warp. In addition, stressses on the pins in electrical connectors (especially in the rear light clusters which can be passing up to 10 amps at times through small contacts that rarely get re-seated or checked for good contact) cause the metal contact fingers for the single common earch connection to each rear light cluster to become a heating point themselves. This comes about due to excessive contact resistance that continues to increase over time until the contacts are no longer able to safely conduct the flow of electricity, and something happens to either break the circuit continuity completely, or partially, which can cause extreme heat generation and perhaps arcing, melting contacts and surrounding plastic, or at worst, start a fire!

Once the earthing problem for the rear light clusters was solved, I did more tests of the LED lamps and they were a success. They worked as expected, and immediately showed off one of the chief benefits of LED-based lamps - a big reduction in heat resulting from the far lower current consumption that also had links to the correction of the rear-earth problem.

Stage 3 - Full Service Trials

With each indicator circuit comprising two 21 watt filaments lamps (one in each front and rear light cluster) and a small (5 watt?) filament lamp in the instrument panel, that's a total current drawn by each indicator circuit when the relay is connecting power to the lamps of almost 4 amps. The frequency is normally about 0.5 Hz (ie. one 'on-off' cycle every 2 seconds or thereabout), but with LED-based lamps replacing the 21 watt filament lamps in the front and rear light clusters, the total current draw is dropped to well under half of the 4 amp original total, and this results in the flasher relay being unable to cycle the contacts on and off fast enough. The relay is effectively forced to try and work outside it's design limits and either sounds like the engine in a model plane on a bad day trying to cycle the contacts at a rate much faster than the limits of mechanical design allow, or it doesn't cycle the contacts at all and the relay 'sticks' with the contacts either closed all the time or open all the time. Not good.

The solution, and it's not the best one, is to insert 'ballast' or load resistors into each indicator circuit to 'simulate' the load that the filament lamps put on the 12 volt DC supply into the flasher relay. A much better solution would be to replace the flasher relay with one which does not depend on the current flow in a circuit to set the flash rate and uses a small electronic timer instead. That's an idea that sits in the back of my mind waiting for the inspiration to either create a solution myself or find a ready-made replacement flasher relay with a built-in electronic timer.

The images below show the fitting of LED-based lamps to the left-side rear light cluster in 'the Beast':

Full Service Trial in Left-Side Rear Light Cluster
LED-based lamps fitted ready for full testing	Close-up showing LED-based lamps and location of new earth connection bolt	The lovely glow of success with the ignition switch in the centre position

Note that the bottom right lamp doesn't light up during the test as illustrated in the third picture above. During the work I did as part of the wiring research for the foglighting project I discovered that the inner-most red lamp on each rear light cluster is set up as a rear fog light and that the wiring to fit a special switch actually exists complete with a switch connector sitting inside the wiring in the dash. I confirmed this by fitting a switch to the connector and seeing that each inner-most red lamp did indeed light up when the switch was pushed 'on'!

Unfortunately there are not enough spare switch mounting holes in the dash to have one switch each for the front and rear foglights so I have left the rear foglight switch connector in the dash wiring without a switch fitted since the seperate switch to operate the mean-looking yellow front foglights used up the one spare switch mounting hole. sigh.

Stage 4 - Solving Fast/Hyper/Ultra/Mega/Supa (!) Speed Turn-Signal Indicator Flashing!

When you replace your car's standard 12 volt indicator lamps (bi-pin 21 watt at the front, uni-pin 21 watt at the back in Saab classic 900's) with LED lamps the first thing you will notice is that the indicators will probably flash extremely fast, or not at all! In all but the most modern and expensive cars, flasher relays for turn signal indicators 'set' the flash rate using the current drawn by the lamps.

A 21 watt, 12 volt lamp will nominally draw around 1.75 amps per lamp when lit, so at any one time there will be approximately 3.5 amps flowing in each indicator circuit when the lamps are lit (though it will vary with the temperature of the filaments).

As we know, LED lamps draw drastically less current than their filament cousins, so the result is a flasher relay that tries to flash the lamps at a rate which is either much faster, or so fast that the relay contacts can't be opened and closed fast enough! That is actually VERY BAD because it will result in the relay coil overheating and/or possibly burning the relay contacts if they chatter at a very high rate.

Two simple solutions are:

1. - to wire industrial wire-wound metal-cased load resistors in parallel to the indicator lamp circuits to make the current draw similar to the load drawn by the old filament lamps,
   
2. - or to replace the flasher relay with one that is designed with a fixed flash rate from the outset that doesn't depend on the load drawn by the old filament lamps.

The first solution is one I have tried and it certainly works but you still end up with the same or similar load on the indicator circuits and because the same amount of energy isn't dissipated in the LED lamps as is by the filament lamps they're replacing, the additional energy balancing the load to that of the old filament lamps has to be dissipated by the load resistors. They DO get hot if the indicators are flashing for any significant amount of time - which is why they have to be wire-wound metal cased industrial resistors and you need to think carefully about where you locate them.

Ultraleds sell special resistor kits especially to use for this solution (see the pic to the right), and other places probably have them too but all you need is the right value and wattage rated resistors and a way to connect them in parallel with the indicator lamps.

The second solution is better because you don't need load resistors, so you still get to take proper advantage of the low current draw and therefore much lower heat output of LED lamps without having to waste energy as heat in load resistors. There are a lot of different replacement flasher relays around ranging from simple two-wire ones up to complete replacement relays for standard ones found commonly in European and other types of car.

Saab classic 900's use a 3-pin flasher relay which is located in the relay cluster located under the left-hand side of the dash, alongside the wiper relay (and others, depending on the build year, option level and target country). I will shortly add a picture of this, along with a circuit diagram. Note that the wiring for the indicators and hazard controls may be different in your car.

A number of places sell replacement flasher relays (some are pictured below), and Ultraleds sells one too.

Saab Classic 900 Indicator Relay Location, etc.
Relay panel under LHS of dash in 1983 8VT C900	One generic make of electronic 3-pin flasher relay	Another make of electronic 3-pin flasher relay

In the left-most pic above (taken in the 1983 8V turbo car), the flasher relay is the three-pin relay on the left, and the wiper control relay is on the right. The blue hose goes to the turbo over-pressure switch from the intake manifold, and the black hose goes from the over-pressure switch to the boost pressure (aka intake manifold pressure) gauge in the dash.

There are two-types of 3-pin flasher relay - the ISO-type and the JSO-type (Japanese standard). My 1983 8-valve turbo C900 had the wiring to the flasher relay socket set up in the the JSO configuration (which surprised me!). The chart below explains the different types.

The pin numbers and colour coding for the flasher relay socket are 49 (red) for +12 volts, 49a (green) to the load, and 31 (black) for ground. I have tried one of the black-cased relays shown above and initially it didn't work because of the incorrect (for the new relay) wiring configuration. Swapping around the red and black terminals (ie. reversing the location of pin's 49 and 31) in the socket to the ISO configuration got it to work so the car now has indicators that flash at the correct rate regardless of what sort of lamps are fitted and no load resistors are required in the turn-signal indicator circuits!

Comparison of Saab (Hella) and 3rd-Party Flasher Relays
Top of original Hella relay	Circuit board inside original Hella relay	Comparison with circuit inside new flasher relay
Back view of circuit boards in new and old relays	Close-up of changed socket configuration for new relay	This space left intentionally blank 8-)
		This space left intentionally blank 8-)

I don't know if Saab 9000's use the same type of flasher relay setup - they may have something totally different so be careful to make sure of what you're doing before pulling and changing relays and wiring configurations!

Fire Alert! - What Happens When LED Lamps Mysteriously Fail!

The failures have ranged from one of the lamps giving up completely all the way to to a fire inside one of the rear number plate light housings which I only discovered a couple of days later (I think it happened on the drive to or from a SCCA meeting in 2006) when I opened the boot and noticed one of the housings was very blackened compared to the other one! The two pics below show the results of the fire - look how the LED lamp has completely self-destructed and it's badly burnt and melted the clear housing as well:

The actual lamp holder part (removed so the camera could see inside) was melted just a little bit. That destroyed LED lamp is melted into the burnt plastic mess that formed the bottom face of the housing!

There have also been a couple that, for no apparent reason, have started flashing the array of LED's on and off at an annoying (!) rate even though the power supply is a nice clean nominal 12 volts. I've suspected that 'ripple' in the DC voltage has caused something in the circuit inside each 'festoon' LED lamp that did this to turn into some sort of a slow duty cycle oscillator but I haven't been able to prove it conclusively.

That lamp which caused the small fire was probably an extreme case of the circuit starting to oscillate on and off which ended up over-heating and burning.

I've replaced all of the voltage regulators fitted to the alternators in my three running C900's with brand new ones (they only cost about US$15 from one of the places selling them on Ebay!) and haven't noticed and flashing failures since, but again that might just be co-incidence.