My partner was not impressed by the stock headlights on her XJ Cherokee Jeep and she requested some form of headlight upgrade. This is not a step by step guide, just some notes concerning my choices and findings.

The engine layout may look a bit weird to some people as this vehicle is a 1998 (i.e. post update), righthand drive (RHD), 2.5l diesel (VM brand), XJ model, Jeep Cherokee. The discussion begins with a review of H4 Halogen bulb specifications and the performance of our stock XJ Cherokee's headlights. Then provides an overview of the upgrade with some installation hints. Finally moving on to the results of the upgrade with positive and negative aspects. Also included a components list with indicative 2011 pricing (~$500 total cost) and some further notes concerning components and their selection.

I'm not an automotive or electrical engineer therefore you are just getting my best guesses and you cannot rely on the accuracy or validity of any information in this document. Basically if you try to use or implement any of this and it ends in tears or there are performance or reliability issues don't come crying to me. You have been warned. I'm not consistent e.g. interchange terms like bulb with globe.


Diesel XJ Jeep Cherokee Engine Bay

High-Intensity Discharge (HID) lights are not an option as they are currently illegal to retrofit to the XJ headlights in all States and Territories. The same restriction probably applies in Europe, USA, etc. Although HIDs in driving lights maybe legal in this State.

Higher power (wattage) H4 style globes (e.g. 80/100W Osram Off-road Only series, or the H4 style 130/170W bulbs advertised on eBay) exist but are not legal on public roads. These very high power globes can try to draw approximately twice the normal current through the headlight wiring. A side effect of higher current through the circuits is hotter wiring and connectors. If the electrical system is in perfect condition then this higher current shouldn't cause an electrical fire. But if the resistance of wires, connectors or contacts have increased due to corrosion, contamination or the conductors in the wiring are starting to fracture, then you have the pre-conditions that can lead to burnt out contacts, or wiring/connector failures, with the worst case an electrical fire. The XJ Jeep Cherokee headlight circuits were never designed for these higher wattage H4 style globes.

There is more detailed discussion below about stock wiring and its impact on performance of higher wattage bulbs in "Not The Brightest Bulb In The Pack" section.

Obvious answer is a headlight wiring upgrade to improve the existing headlights. There are various headlight wiring upgrade projects published on the intertubes, e.g. Go Jeep's XJ Cherokee Headlight Loom Upgrade project info or Toyota 4Runner. Google for other articles. I've included more background information as I've made different choices to other upgrade projects, plus I don't agree with some of the current comments concerning Halogen bulb life in the Toyota 4Runner project.

It would be possible to make significant cost savings by simplifying this headlight upgrade. The idea is to delivery the brightest headlights using standard H4 globes without compromising reliability of the headlight wiring or switching. Choices have been made that deliver very minor voltage increases but it is these many small increases that contribute to the final result. This isn't the ultimate headlight upgrade project as there are still some minor improvements to be had.

Given the difficulty of my project (e.g. box stuffing and complexity) it should only be attempted by somebody experienced enough to not need any of my project information. As such a person wouldn't require circuit diagrams or wiring schematics these haven't been included.

H4 Halogen Bulbs

Nitty, gritty details

United Nations Economic Commission for Europe (UNECE) Regulation 37 specifies the design, construction and performance of standard H4 bulbs. It specifies the values to be marked on the bulb of "Rated Voltage" (12 Volts) plus the driving-beam (high) and passing beam (low or dipped) filament "Rated Power" (65 Watts / 55 Watts). This standard explains why you see so many competing brands of seemingly interchangeable H4 bulbs with identical power ratings.

Rated power of H4 bulbs for 24V trucks and buses is higher at 75W/70W and are specified about 20% brighter than the 12V globes. So if you thought truck and bus headlights were brighter than the average car there maybe some science backing up your observation.

A small complication

Light output varies with applied voltage and UNECE Reg 37 specifies a 13.2V test voltage. Interesting twist here is there are two different voltages in this regulation for the H4 bulb. Rated Voltage of 12V is used to mark or label the bulbs and you might assume the 13.2V test voltage is just something used for a short testing period during the manufacturing process. The 13.2V test voltage is in fact the designed operating voltage of the globe. The standard states the Test Voltage is:

Even though the H4 bulbs are marked 12V they are designed to operate at 13.2V. Interesting to note the test voltages for other types of 12V marked bulbs (i.e. not H4 type) can be 12.8V. 13.2V, 13.5V or even 14V. This can lead to confusion as the same bulb can have multiple ratings of power and light output at different voltages. Just for completeness here, the test voltage for Osram 24V globes is 28V. When the actual voltage applied to a bulb's filament is increased the power (wattage) increases so the maximum power at 13.2V is specified as 75/68W for a bulb that is marked 12V 65/55W.

Now moving onto the light output of the globe. For background information refer to the Wikipedia article on Lumens. At 13.2V for Luminous Flux there are Objective Values. The standard states that Objective Values are:

Reference light ouput at 12V is 750 lumens (lm) while at 13.2V it increases to 1,000lm. Therefore a voltage change of +10% (1.2V) gives +33% (250lm) change in light output. Unfortunately, due to the non-linear characteristics of normal eyes the world isn't suddenly perceived to be 30% brighter.

The complication has a small complication

Due to the characteristics of H4 bulbs the higher light output at 13.2V results in a shorter life span when compared to the same bulb running at 12.0V. To quote Wikipedia's article on Halogen Lamps:

Tungsten halogen lamps behave in a similar manner to other incandescent lamps when run on a different voltage. However the light output is reported as proportional to V3, and the efficacy proportional to V1.3. The normal relationship regarding the lifetime is that it is proportional to V-14. For example, a bulb operated at 5% higher than its design voltage would produce about 15% more light, and the efficacy would be about 6.5% higher, but would be expected to have only half the rated life.

Halogen lamps are manufactured with enough halogen to match the rate of tungsten evaporation at their design voltage. Increasing the applied voltage increases the rate of evaporation, so at some point there may be insufficient halogen and the lamp goes black. Over-voltage operation is not generally recommended. With a reduced voltage the evaporation is lower and there may be too much halogen, which can lead to abnormal failure. At much lower voltages, the bulb temperature may be too low to support the halogen cycle, but by this time the evaporation rate is too low for the bulb to blacken significantly.

Plugging these formulae into a spreadsheet gives the following results:


The formulae used to generate this table may not apply over the full voltage range. I haven't read the sources used to create the Wikipedia article so there could be other factors that invalidate these results as you move further away from 13.2V. I don't know if these other factors actually exist so I'll continue to use this table and formulae in this document. Take this as a warning that the following discussions, the results for this headlight upgrade and my conclusions could be suspect. This is best guess territory.

Start rant. Ordinary incandescant light bulbs made in the dim, dark past are capable of burning for over a 110 years (refer Livermore's Centennial Light site) and can still be going strong. You won't see any H4 bulbs that match that longitivity as higher light output is traded off against a shorter bulb life.

Estimated life for a particular make and model of H4 bulb may not be published and could be 80, 700, 1400 hours or some other number. The bulb manufacturer's design and material choices determine this performance. Maybe they use more or higher grade materials (i.e. higher production costs) for longer life, or achieve extra life by trading off some brightness (UNECE Reg 37 specifies a tolerance of ±15%). It is trivial for a manufacturer to say a particular bulb lasts "twice as long" but twice as long as what? Is that twice as long as 80 hours, 700 hours, or some other number? I spent some time searching some H4 bulb manufacturers websites and could only find technical data for Osram products. Seemed Osram was the only one of these particular companies willing to publish its technical data in a readily accessible form for the general public. I really wonder if those other companies are ashamed or if they have problems with the quality of their products. Next time I have to buy bulbs I'll be looking for Osrams as I know what I'll be buying. End rant.

Osram does deserve this special mention as they do publish their H4 technical data, the data on estimated life is given in a table in their Index section. Due to their construction high and low beam filaments have different estimated life values. Just to complicated things even further, Osram specifies two different values for each filament.

Of course the average punter probably prefers to know the 'average' life (50% failure or B50) of a bulb. It should be possible to derive a formula to calculate or from graphs determine the B50 failure rate from the Osram data based on the Weibull distribution. There are references to Weibull distribution on the web with worked examples using Microsoft Excel spreadsheets. Unfortunately, or fortunately depending on your point of view, my PC doesn't have Excel. My OpenOffice version doesn't have the same range of high level functions as Excel and I'm too lazy to go back to first priciples to calculate the B50 data.

Alternator and Battery

Jeep Cherokee battery charging circuits have evolved over the years and this discussion is based on the 1998 diesel XJ Cherokee. In this model of Cherokee which uses a Nippon Denso 90HS series alternator the output voltage of the alternator is controlled by the Powertrain Control Module which monitors the battery voltage and battery temperature. Earlier Cherokees used a different battery charging system. The starting point for the headlight circuit is the combination of the alternator, battery and what may come as a surprise, the Powertrain Control Module. There are many references to alternator theory on the intertubes (e.g. Joe Guilbeau's Alternator Theory Page) and even more sites on the theory of batteries (e.g., battery university, Wikipedia, etc., etc.) so I will skip these general topics. My experience/knowledge of automotive electrical systems is limted so this is definitely my best guess territory.

Bosch Alternator graph I couldn't find a graph for a Nippon Denso 90HS alternator used in the 1998 XJ Cherokee so I'll use a graph of a 90Amp alternator from a Bosch catalogue to illustrate the issues. The graph shows the maximum output current in Amperes (Amps) that the alternator can supply into a load. The RPM scale is the alternator's RPM and that isn't the engine RPM. Due to the alternator pulley being a smaller diameter than the crankshaft pulley the serpentine belt will rotate the alternator faster than the engine. At engine idle speed (875±25RPM) the alternator will be 'geared up'. I need to measure the pully diameters to determine the ratio. At idle the alternator will be capable of supplying something like 40 Amps. Catch is I don't have the 90HS alternator graph and 40A is a reasonable guesstimate. This is enough current for the headlights, power requirements of the engine, charge the battery, etc.

If the idle speed were to drop to 600 RPM then the alternator could struggle to support the maximum electrical load, the battery may have to supply current to the vehicle electrical system and start to discharge. At the other extreme the engine will be capable of driving the alternator past 6,000 RPM and the alternator will then be capable of supplying more than 90Amps.

At engine idle speed with the headlights switched on the alternator increased the battery terminal voltage to 13.86V and there was no appreciable change in battery voltage when increasing the engine above 2,000 rpm. Although the battery charging circuit is trying to maintain a constant battery terminal voltage this can still vary depending on the type and age of the battery, state of charge of the battery, temperature, current being drawn by the headlights and other electrical components, plus the voltage losses in the wiring.

Remember this is a diesel so engine revs tend to be lower than the typical petrol engine. Battery terminal voltage is not a constant and will vary over time as well as between different vehicles. Due to the resistance of the cable and connection of the battery's negative terminal to the bodywork, when current is flowing through the battery leads the battery terminal voltage is different from the voltage measured between the battery's positive terminal and chassis/bodywork. The voltage drop/loss will be measured in millivolts and is only significant as another source of error in the calculations and it is one of the headlight circuit losses not addressed by this upgrade.

Problem To Be Solved

When current flows any resistance in the headlight circuit decreases the bulb's filament voltage. The losses in the circuit will include the resistance of the wires, relays, fuses and connectors. The higher the voltage losses in the circuit the dimmer the headlights but on a more positive note the bulbs should last longer. The higher the power of the filament the higher the current drawn by the filament and the greater the voltage loss.

Some of the headlight circuit shares wiring with other circuits. E.g. the battery or alternator leads. The voltage lost on a shared path is determined by the total current being carried on that shared path so the voltage drop/loss will be higher than just the headlight current alone. Except during engine starting, using an electric winch or recharging the battery, the headlight current should be largest component of any current through shared wiring. Even then there are exceptions like an extremely high power audio system or RF linear (QRO).

With a fully charged battery and the engine running just above idle the alternator provides all the current to the headlights. With the headlights switched on at engine idle speed the alternator increased the battery terminal voltage to 13.86V and there was no appreciable change in battery voltage when increasing the engine above 2,000 rpm. In this model of Cherokee voltage regulation is controlled by the Powertrain Control Module which monitors the battery voltage and battery temperature. Earlier Cherokees will use a different battery charging system. Although the battery charging circuit is trying to maintain a constant battery terminal voltage this can still vary depending on the type and age of the battery, state of charge of the battery, temperature, current being drawn by the headlights and other electrical components, plus the voltage losses in the wiring. There are many references to alternator theory on the intertubes (e.g. Joe Guilbeau's Alternator Theory Page) so I will skip this topic.

Remember this is a diesel so engine revs tend to be lower than the typical petrol engine. Battery terminal voltage is not a constant and will vary over time as well as between different vehicles. Due to the resistance of the cable and connection of the battery's negative terminal to the bodywork, when current is flowing through the battery leads the battery terminal voltage is different from the voltage measured between the battery's positive terminal and chassis/bodywork. The voltage drop/loss will be measured in millivolts and is only significant as another source of error in the calculations and it is one of the headlight circuit losses not addressed by this upgrade.

I measured the following headlamp voltages on the stock wiring loom with standard wattage H4 bulbs. Each of these measurements were taken with the engine running but were not taken on the same engine run, i.e. there will be differences in battery state of charge, engine/alternator temperature, etc., and these changes could cause measurement errors. Results were:

HeadlampLow BeamHigh Beam

Only a small number of measurements were taken during the course of the upgrade and when checked the battery's positive terminal voltage varied between 13.84V and 13.87V referenced to the bodywork. It is possible the terminal voltage varied outside this range during the upgrade. Unfortunately this writeup was started after the upgrade work was completed and I didn't take enough measurements during the upgrade. Some filament voltages were recorded without the corresponding battery voltage therefore it appears as though the righthand bulb's low beam filament voltage is higher than the lefthand filament. On a RHD diesel given the longer length of cable to the righthand bulb (battery and relays on lefthand side of the engine bay) I would have expect the righthand headlight's filament voltage to be lower for both low and high beam.

A petrol Jeep has the battery and relays on the righthand side of the engine bay. I would expect for a stock, petrol driven, XJ Cherokee that the righthand filament voltages would be higher than the lefthand side due to the longer cable runs.

Given the strange low beam filament voltage measurements there maybe a minimum error of 0.2V at 11.76V for the righthand bulb. This represents a miniscule 1.70% error at 11.76V but with righthand delta V (low beam/high beam) only being 0.41V a possible 0.2V error may represent a massive 48.8% error. Without knowing the battery voltages at the time of both these low beam measurements the results of the calculations below are suspect. Due to the alternator charging the battery it is possible the battery voltage increased over the time of the upgrade and the righthand bulb's low beam filament voltage is valid and there is no error. Due to manufacturing tolerances, aging, etc., there could be significant differences between the two H4 bulbs and the filament voltage measurements could be perfectly valid and within the measurement error of the digital multimeter used. Now I wish I had taken more measurements during the upgrade

Using 13.2V as the reference voltage these formulae estimate the stock XJ Jeep Cherokee headlights generate about 70% of the bulb's rated light output on low beam and about 64% on high beam.

HeadlampLow BeamHigh Beam

So there is room for improvement. As vehicle design is a compromise between price and quality (plus superconductors are not an option) the XJ Cherokee's headlight circuit does have some losses that can be reduced by throwing money at the problem. Eliminate the voltage losses and there will be an immediate improvement in headlight performance.

Not the brightest bulb in the pack

Using Ohm's Law it is possible to get some ballpark figures for higher wattage H4 bulbs by making some dubious (to put it mildly) assumptions like bulb resistance is a constant and no shared current paths with other circuits. The average headlight bulb voltages are Low beam VLo55=11.66V and High beam VHi65=11.375V, Alternator/battery voltage VAlt=13.86V and assume these H4 bulbs were identical and exactly 12V 55/65W. Just to clarify, these calculations are not based on measurements of the H4 bulbs actually fitted to the Jeep being upgraded.

From R=E2/P then filament resistances for the standard 55/65W bulbs at 12V are RLo55=2.618 Ohms and RHi65=2.215 Ohms. In reality filament resistance is not a constant and these values will not apply at any other voltage. I'm using H4 Standard's 12V specifications as these are closer than the 13.2V specifications to the filament voltages measurements for the stock wiring.

From circuit resistance RLoCct=RLo55 * (VAlt - VLo) / VLo then estimated circuit resistances are RLoCct= 0.494 Ohms while by the same logic RHiCct= 0.485 Ohms. The cable gauge and average length of wiring produces roughly the same estimated average resistance values. More accurate calculations would provide separate results for the lefthand and righthand circuits.

Substituting Osram 80W/100W H4 style offroad/rally bulbs (note: power is specified at 13.2V not 12V) gives RLo80=2.178 Ohms and RHi100=1.7424 Ohms. Filament voltage VLo80=VAlt * RLo80/(RLo80 + RLoCct) therefore VLo80=11.30V while VHi100=10.84V, theses voltage estimates maybe on the low side given the assumptions that were used. Filament resistance is non-linear and it should increase with increased filament temperature (i.e. increased filament voltage). Shared current paths should have less impact on voltage loses.

Assuming the test voltage for these globes is still 13.2V and the formulae are still reasonable so far from the test voltage (another very dubious assumption) then the spreadsheet says these bulbs will only be putting out 63% of their rated light on low beam and 55% on High beam. Even I don't believe the higher power bulbs would perform this badly, just because it is in a spreadsheet it doesn't make it real, but without taking measurements I'm in the dark (sorry, couldn't resist that). Although in reality these 80/100W bulbs would be brighter than standard bulbs and probably brighter than these estimates, no way will they match the bulb manufacturer's perfomance claims in the stock XJ Jeep Cherokee headlight wiring, but hey, they may last longer.

All this does is confirm a headlight wiring upgrade is the way to a brighter future.

Pick a number, any number

Due to UNECE Regulation 37 standard H4 globes are labelled with their 12V characteristics. So a 12V 60/55W globe is at 13.2V a 75/68W globe. I.e. for your buck you are getting a bulb that is really rated at 1.25 times the labelled power at 13.2V. If a manufacturer produces non-standard H4 style globes they don't have to conform to this labelling regime. Osram's technical specifications show that their Offroad 100/80W H4 style globes are basically using a 13.2V power rating. If I was trying to compare power ratings I'd be dividing the labelled power by 1.25 to roughly get a 12V power rating, in this case 80/64W.

Assuming that 80/64W is only going to be a little better than a standard 60/55W bulb is jumping to the wrong conclusion. Looking at the light output data shows low beam output has jumped from 1,000lm to 1,500lm while high beam output has jumped from 1,650lm to 2,900lm. So the power rating is not really telling you that much, certainly doesn't tell you what the manufacturer has done to optimise light output nor anything about beam shape or spread.

If a manufacturer doesn't provide technical data then you are up the proverbial creek in a barbed wire canoe when trying to make a purchasing decision in a shop based on a cursory examination of all the competing claims on the packaging of the different products sitting on the shelf.

Plan Of Attack

Solution is to reduce the voltage losses in the headlight circuit by installation of heavier gauge wiring (thicker with more copper), minimise wiring shared with other circuits, more direct (shorter) wiring, etc. Driving lights, winch and a second battery are going to be installed at a later date so suitable cabling, a fuse, plus driving light wiring will be incorporated into this project. The object is not to modify any of the existing wiring by picking up the Low and High beam control lines from the existing headlight connectors. If something with this upgrade goes really pear-shaped on the road I want the option to restore the stock wiring and isolate the modified circuitry from the battery.

To minimise my costs there will be a preference to use items from my collection of bits and pieces that has been assembled over the years. Catch is I don't need to compromise on cost while most people would have to purchase most if not all components at considerable cost. Consider Go Jeep's XJ Cherokee project info for a cheaper fix to the same problem.


Diesel XJ Jeep Cherokee Engine Bay

First issue is to decide where under the bonnet to locate the new fuses, relays, etc. On a righthand drive, diesel, XJ Jeep Cherokee there is an open space on the lefthand guard between the relay/fuse box that sits immediately behind the battery, and forward of the airconditioner's receiver/dryer which sits in front of the firewall.

The lefthand guard is curved metal and has a couple of vacant holes that will accept 8mm bolts. Construction would be simpler and neater using a flat sheet spaced above the Jeep's guard. My bits and pieces collection yielded some 80mm wide, 3mm thick aluminium bar and some 1.5mm galvanised sheet. The galvanised sheet seemed flimsy but may have been suitable if I had been desperate. The aluminium was more rigid, simpler to cut, plus fatigue and corrosion would not be an issue in this application.

Mounted the aluminium bar by using standard 8mm nuts as spacers. Also has the advantage of holding the bolts in place when it comes time to mount the flat sheet and control box. Putting the 8mm bolt heads on the wheel side of the metal guard meant there was no chance of a long bolt digging into a tyre when cornering, etc. Bolt length is critical, obviously too short doesn't work and too long can make it difficult to mount the flat sheet or it could foul the lid of the box.

All the mounting hardware was stainless steel with nyloc nuts so that corrosion or nuts working loose wouldn't be an issue. These days I avoid zinc plated hardware like the plague. In a severe environment it tends to rust, plus usually they are made from mild steel that sometimes doesn't survive my ham fisted attempts at tightening nuts and bolts. Maybe zinc plated hardware is OK when kept dry and in low humidity.


Diesel XJ Jeep Cherokee Engine Bay Front Left Guard Close-up

Component Selection

Main Fuse

If the upgrade was only going to be handling the headlights then a suitable Maxi-blade or 5AG fuse would be appropriate in the battery lead. As a second battery will be installed in the rear cargo area plus a winch at the front there needs to be a higher capacity fuse and fuse holder. My bits and pieces collection provided the choice between a decorative ANL and a more utilitarian looking MEGA fuse holder.

To change these ANL fuses you have to unscew one end of the fuse holder then slide the blown fuse, terminal posts and one lead through the transparent body before you can unbolt the blown fuse. This requires careful handling to keep the live battery lead away from the metal bodywork. The gold plated ANL fuse holder may look pretty under the bonnet but it may become a pain to maintain. The ANL fuses from local supplier Jaycar range from 80 to 250Amps while the MEGA fuses range from 150 to 500Amps. The selected winch is rated at over 300A so the MEGA holder was selected.

As well as a high current, battery fuse my personal preference is to have appropriate fuses in the individual headlight and driving light circuits. Normal gauge headlight wiring is likely to catch fire or vaporise before the main MEGA fuse will blow therefore it is important to incorporate lower rated fuses to protect the headlight and spotlight circuits. There is further discussion on fuse selection below.

Control Box

There were multiple options when digging into the bits and pieces collection. Mounting the MEGA fuse holder, a distribution block, a Blade fuse holder block (for the lower rated, individual circuit fuses) plus relay holders was not practical on the available aluminium plate. Drilling holes into the guard to mount components would create a wiring rats nest that would solve the space problem. But the Blade fuse holder and horizontal mounted relays were not waterproof or even water resistant. Another of the possible Jeep upgrades is a snorkel so there is the potential for the fuse holder and other components to be drowned. Solution is to put these items into a sealed box.

Raided the bits and pieces collection to find an ABS plastic box large enough for all components but it was too flimsy and the lid didn't wasn't a waterproof seal. In this case my bits and pieces collection came up empty and I had to buy a rigid, IP65 rated, sealed enclosure. The IP65 rating means it is dustproof and will survive immersion. It was more rigid than the Jiffy box so the MEGA fuse could be mounted on the lid but there was not enough room for the relay sockets or the Blade fuse box.


Fuses for the individual circuits was mandatory. In the bits and pieces collection were 12V automotive style relays that incorporated a Blade fuse. Solved the problem of finding a place for a Blade fuse block and this provided enough space to use five relays.

Three is the minimum number of relays required for low beam, high-beam and driving light switching. With just three relays there are more shared current paths that will increase voltage loses. Although unlikely, if a fuse blows you instantly lose both left and right headlights. Solution is to use five relays each with their own fuse so blowing a fuse only takes out one headlight. 10 Amps at 12V is 120W and there is always a chance that a 130W or higher power bulb will be fitted so 15 Amp fuses could be installed in the headlight relays while a 30A fuse fitted to the Driving Light relay.

Using three relays means the current for the left and right headlights is flowing through a shared relay, fuse and wiring. This doubles the current on this shared path and any voltage loss would also be doubled. Installing five relays is a trivial change for a theoretical improvement. I have not performed any tests to determine what if any improvement to voltage loss is made by installing five relays.

If a H4 filament fails it could blow a fuse hidden inside the box. It would be convenient to ascertain relay and fuse states without removing headlamps or pulling the control box apart. As there will be five fuses plus the relays buried in the box it would be nice to monitor all the inputs and outputs. Bought some red and green coloured, 12V, Light Emitting Diodes (LEDs) mounted in chrome bezels. Used three red LEDs wired across the Low Beam, High beam and Driving Light relay coils and five green LEDs for the relay outputs (in circuit after the fuses) plus one green LED after the MEGA fuse and before the 12V distribution block in the box. Hopefully it will be posible to disgnose any fault conditions just by looking at LEDs, headlights and as a last resort measure all box input and output voltages on the terminal strips.

Switching off the current through the relay coils can generate a Back EMF (Electro-Motive Force) that could destroy the LEDs. Simplest solution is to wire a flywheel diode across the three inputs. Bits and pieces yielded 5x 1Amp diodes so it was trivial to make each relay circuit identical with a supression diode across each coil. Any additional switching delay due to a diode is irrelevant for this application.

Battery cable

High current cable used for vehicles can be labelled in a few different ways.

  1. Simplest and I'm guessing least reliablelabelling is a current capacity or current carrying specification. Looking at catalogues from a couple of local surpliers that both have 8 Gauge cable constructed with the size conductors (same cross-sectional area), one is rated at 56A and the other at 61A. One has 2 Gauge sized cable rated at 160A and the other supplier labels a thinner (smaller cross-sectional area conductor), 3 Gauge cable as 200A.

    Just to complicate the issue, current carrying capacity of a cable is highly dependant on the temperature rating of the insulation, common ratings are 65°C, 75°C, 90°C, 105°C while the high temperature silicone rubber insulated wire I used is rated at 250°C. So the difference in carrying capacity maybe down to the rating of the insulator. Catch is the temperature rating for the insulator is not given in either case so all you can do is place your trust in the manufacturer and seller to have it right.

    In the dim, dark past I inspected various purchases for a large company. Randomly I would be sent to sample production of a manufacturer who shall remain nameless, check their production records, and have my samples tested by staff in the manufacturer's registered laboratory. Most of the batches I observed being tested would pass inspection but a percentage would be rejected. But I never saw in the manufacturer's own laboratory testing records any rejected batches. Guess it was just bad luck that the only time this certified, quality assurance scheme manufacturer using their own traceable standards laboratory failed to meet there own quality standards was when testing was independently supervised. Other manufacturers would have their own issues with production quality. At the time there were notable exceptions like Aegis and Hewlett-Packard where I can't remember them ever presenting a defective batch of equipment for inspection. Doesn't mean they didn't have equipment fails in the field or there weren't design issues, just that they passed inspection tests. Now back to cable insulation; higher temperature rated insulation is more expensive to produce so the manufacturer has an incentive to cut corners. When there are no quality standards specified it is all down to trust, and even when there are standards it can still be down to trust.

    Cable labelling of 'High Temperature' is another rubbery area (pun intended) where marketing seems to rule. I've seen different cables with ratings between 90°C and 400°C labelled 'High Temperature'. A 'high temperature' label without a temperature specifications is pretty useless. Rule of thumb is the higher the insulation temperature rating, the higher the cable's current rating for the same cross-sectional area of the same type (e.g. copper) of conductor. Neither supplier published a temperature rating for insulation on these 2 and 3 Gauge cables so you cannot know if one cable's insulation has a superior temperature rating, or whether the labelling difference is just a marketing exercise or the result of different testing standards. Is one supplier under specifying the current rating so that you buy more of their heavier therefore more expensive cable to avoiding current carrying issues with your project? Ignorance is not bliss.

  2. Next up is an American Wire Gauge (AWG) specification. In theory this is a measure of the cross-sectional area of the conductor. In practice different suppliers can label the same cable with different AWG sizes. Eg. the same size cable could be labelled 2 Gauge or 4 Gauge, one is probably based on the cross-sectional area of the conductor plus insulation, while the other supplier is just using the cross-sectional area of the conductor without insulation. If I don't trust the the suppliers size labels then measuring the weight of larger diameter cables may tell you which has the most copper but it doesn't tell you about the insulation quality. Although heavier is generally better.

  3. Next up is less user friendly but provides a better picture of what you are buying. The supplier specifies the conductor's total cross-sectional area. Alternatively, the number of strands with the individual strand's cross-sectional area, with this information you can estimate the total cross-sectional area. If you're told the temperature rating of the insulation then you are in the best position to compare cables.

With smaller gauge wires there can be surprising differences in the thickness of the insulation and the amount of copper for wire labelled as the same gauge. If cable/wire is protected by conduit or tubing, will not be rubbing against sharp edges or the frame, and will not be tightly bundled, then more copper and less insulation is the way to go. If you are running around the vehicle without protection my preference is to select cable with thicker insulation and sacrifice some current carrying capacity.

Control Box Stuffing


Control Box Internal Heatshrink On

Bits and pieces yielded a power distribution block with a clear arcylic cover for insulation. It was designed for 1x 4 Gauge cable in and 4x 8 Gauge cables out. With five relays the optimal configuration was 3x 8 Gauge cables to the High beam and Driving Light relays as they would draw the highest current therefore have the greatest voltage drop over 12 Gauge wiring. The one remaining 8 Gauge position received 2x 12 Gauge cables for wiring the two Low beam relays in parallel. As they draw the lowest current they would receive the least benefit from parallel 8 Gauge wiring plus have the lowest voltage drop over the smaller diameter 12 Gauge wires. There was enough clearance to run the 8 Gauge cable over the top of the relays but you have to be careful not to place excessive strain on the relay connectors. Also the 8 Gauge cable's insulation could creep due to any tight bends, therefore I used heatshrink in all appropriate locations throughout this project.

The control lines from the input connector were run in a five wire automotive cable that is typically used on trailers. All other low current wires in the box were run in high temperature, silicone rubber insulated cable. To maintain some semblance of water resistance IP68 waterproof cable glands were used on all cables entering the control box.

Relay Installation

When securing the relays I drilled another mounting hole through the middle of the distribution block's acrylic base. Also enlarged the mounting hole of one relay so that it is secured by the 8mm bolt. This hole is through very thin sheetmetal so great care is required or you will damage the relay mount, relay or yourself. The 8mm bolt also secured a couple of round crimp connectors to provide earth connections for the LEDS and relay coils. An earth lead was also extended from the round crimp connectors to the input connector strip just so I can test the state of the box's earth connection and use it with a multimeter to check circuit voltages without having to attach an alligator clip lead to the bodywork. Do you really need to be told I'm lazy?

Light Emitting Diodes (LEDs)

All the LEDs plus the Input connector strip are mounted on the control box body so that the lid can be removed without disturbing the more delicate wiring. It would have been cheaper to use resistors with various coloured low voltage LEDs mounted with plastic bezels that clip into place, unfortunately these bezels won't be waterproof. The 12V bezels only have a limited range of colours but do use a metal body with a threaded section and nut that pulls the bezel against the box to seal the mounting hole. Therefore 12V LEDs were the obvious choice.

Low and high beam control signals are taken from the lefthand headlight connector so the Cherokee's headlight switch gear will continue to operate the headlights as origionally designed by Jeep. The driving light's control line is terminated on the input control strip and by extending this with the highbeam control line to a switch on the dashboard the driving lights can be optionally switched with high beam.

Control Box LEDs with Main Fuse LED ON

Only Red and Green coloured 12V LEDs in bezels were available so some compromises were made with layout and choice of colours and that could confuse. Green was used for fuse outputs and Red for inputs to the control box.

The bottom row is for dipped/low beam, middle row is high beam while the top row is driving lights plus the main fuse indicator. For low beam, high beam and driving lights the Red LEDs are monitoring the relay coil voltages and these LEDs will only be lit when 12V control signals are applied to the box.

Only the main fuse's LED (Green in top row, righthand column) should be continuously lit. Lose this one and obviously there will be total loss of all headlights and the driving lights will just be decorative baubles. You had better be caring spare MEGA fuses as they will not be available over the counter at your friendly local servo.

Control Box LEDs with High Beam LEDs ON

If high beam is selected the middle row's Red Led indicates a 12V control signal has been applied to the control box, and the two Green LEDs indicate 12V power is being supplied to the headlight filaments.

If all these high beam LEDs were lit with one or both headlights not shining I expect headlight filament failure and replacement of the affected H4 bulb/s would be required.

If the Red LED is lit but one of the Green LEDs is not then I would expect bulb filament failure has blown a fuse. But if everything appears to function normally after the replacement of a blown fuse I would suspect an intermittent wiring problem between the control box and the headlight connector, or the fuse rating may have been marginal. If the fuse was not blown I would suspect a relay failure or the quality of my wiring within the control box. I bet at this point I will be muttering darkly about the quality of cheap relays, the company that sold them and ancestory of the management team that produced them.

If the Red LED is not lit it is likely to be an open circuit in the headlight control wiring from the lefthand headlight connector or a failure further back in the Jeep's original headlight circuitry. Looking at a blown fuse, broken wire, connector or Jeep relay failure, etc., therefore you are back to an XJ workshop/service manual. Don't get me started on the availability, let alone quality or lack thereof, of all the aftermarket workshop manuals for a diesel, XJ Cherokee.

Control Box Mounting

Control Box mounting to aluminium plate

The lid holds the MEGA fuse holder and the output connector strip. The box is held onto the aluminium bar by some of the distribution block/relay mounting bolts plus one of the 8mm bolts, so securing components and the control box to aluminium plate was done after the component wiring was basically finished and tested. I called this section "Box Stuffing" as it was a relatively delicate operation with much care so that there was no chance of any physical damage or compromising any of the headlight circuitry or components.

Control Box lid seal A piece of small diameter tubing is used as a gasket for sealing the lid and it must be trimmed to size.

Control Box external


Final Assembly

Once the box is mounted in the vehicle the next stage was connecting the new headlight circuitry to the 12V supply. Although the alternator will normally be supplying the headlight current with a small voltage loss in the alternator's connection to the battery, I didn't want to disturb or break into the alternator's circuit as it could compromise reliability. Remember the aim is not to disturb or modify the original Jeep wiring. The logical place to connect a lead is on the positive battery terminal.

Battery Lead with Anderson Type Connector

As I don't have an appropriate crimping tool for 2 or 4 Gauge crimp connectors I can only solder them with a Weller electric plumbers iron, or BernzOmatic gas or welding torches (and boy do you pay through the nose for the BernOmatic oxygen cylinders when you can buy them). In this location the new cable can be cable tied back to the existing battery lead so there is no strain or mechanical movement on a soldered connector. A better alternative is to approach an electrician or installer of high powered, automotive audio equipment to produce the crimped cables.

To change a MEGA fuse both cables have to be unbolted from the fuse holder. Having a live battery cable waving in the breeze is just asking for trouble so a single pole, high current, Anderson type connector was included in the circuit to provide a battery disconnect for the MEGA fuse holder. The length of 2 Gauge cable from the battery to the Anderson connector was determined by the length of an off-cut. Upto this point the 2 Gauge cable is supported and properly soldered connections are acceptable.

All bets are off if you create a faulty solder joint. For example, if you can't get enough heat into the connector and cable, or you overheat the solder, or the cable joint moves around while the solder is solidifying there can be problems with the joint. There are too many ways to screw this up if you don't really know what you are doing. I only listed crimp connectors and mention this soft-solder jointing method as it provides a last resort when no better assembly techniques are available.

From the Anderson style connector to the MEGA fuse the cable is not supported and the cable connector should be crimped. The local cheap auto parts shop has prepackage cables of various length using crimped connectors. Although the packaging is clearly labeled 2 Gauge they only had the same size conductor as my smaller diameter 4 Gauge cable. Remember the larger the guage number the smaller the diameter of the cable. So not all 2 guage cables are not created equal.

Once the 12V supply was connected a final test of the box confirmed the relays and LEDs were functional and the next stage of the installation could start.

Lefthand Loom Tube Feeding Part 1

Two Wiring Loom Tubes with 12 Gauge cables were installed in stages from the control box to the headlights. The tube to the righthand headlight also contained the Driving Light cable. The tube to the lefthand headlight also contained the Low and High beam control cables. These control cables were silicone rubber insulated and are rated for 250°C. Only used the high temperature cable because it is very flexible, temperature under the bonnet was not going to be an issue, plus no cost as a reel was in my bits and pieces collection. There is no justification for actually purchasing this cable just for a headlight upgrade project.

It is not possible to push all the cables through the full length of the tube. It maybe possible to use a draw wire but the tight bends will probably cause too much grief. Installing the cables into the loom tube prior to feeding the tube into its final position might be considered an option but this will make the loom tube less flexible and probably prevent installation.

A solution is to partially insert the empty tube, next insert the cables then finally feed the combination through to the next stage. The lefthand installation is relatively straightforward but getting this tube through the hole between the battery and body then feeding through to the back of the headlight shell is a pain. The right hand tube had to be installed prior to the lefthand and it is just visible in these lefthand headlight photographs.


Lefthand Loom Tube Feeding Part 2


Righthand Loom Tube Feeding

The righthand headlight routing is much more complex with multiple steps required to just get past the lefthand headlight area.


Headlight connectors

Male connectors compatible with the H4 headlight sockets were bought cheap on eBay. Looked like somebody had a production stuffup by running the common connector and High beam in Red wire and the Low beam connector in Black wire. I can live with the colour combination although if it had been up to me there would have been three different colours on the connector. To keep crud out of the unused RH headlight connector all the wiring was remove from one male connector and it was inserted into the unused headlight connector, covered in insulation tape as well as cable tied together, then placed out of harms way. Obviously the existing lefthand headlight connector allows you to pickup Low and High beam control voltages without cutting or modifying any of the stock wiring. All you have to do is remove the male connectors to To revert to the stock wiring and this will disable the low and high beam control box relays.

The new H4 female sockets couldn't accomodate cables larger than 12 Gauge. I crimped then soldered each connection to ensure a reliable connection. The last step is to secure the loom tubes with cable ties taking care to avoid the Overhead Console's temperature sensor in front of the radiator.

Results:- The Good, The Bad And The Ugly

Next time I change both H4 bulbs I really need to take another set of voltage measurements to minimise any errors and confirm these tentative results.

The good

To stress test the new wiring I replaced the old bulbs with some new Navara 55/100W H4 bulbs then measured filament voltages. Low beam has the same power rating so these measurements show directly any improvements while the High beam measurement will highlight the capability of the new wiring to support higher wattage bulbs when setup for off-road.

HeadlampLow (Old)Low (New)High (Old)High (New)

Obviously there is a significant improvement and as would be expected High beam gets the greatest improvement. It is interesting to note the old 65Watt High beam voltage was approximately 0.285V less than the old 55Watt Low beam voltage. The new 100Watt High beam voltage is about the same as the 55Watt Low Beam. The voltage difference is down in the measurement error (±0.09V) of the multimeter.

Using the spreadsheet to calculate the new light output compared to the stock wiring gives:

HeadlampLow BeamHigh Beam

A great improvement in High beam light output was expected and delivered. This magnitude of change in brightness should be noticable if you sat a Jeep with unmodified wiring next to an upgraded jeep. Our area has good street lighting and the driving experience around the suburbs with the lights dipped on low beam is not so spectacular, yes there is more light thrown on the road but the street lights are providing most of the distant illumination as the headlight lenses are determing the beam shape. Higher power low beam globe is not the answer if you come across a long section of road or freeway without street lighting as you will want to go for High beam if travelling at speed. On country roads high beam is great. When the driving lights are added it will be as good as it gets.

The bad

As the upgrade has delivered more than 13.2V at the filaments the low beam light output will be approximately 112% of the manufacturer's rating and the bulb life will be shortened to 59%. No such thing as a free lunch. Osram publish their technical data and the impact is:

CodeDescription13.2V High/Low Beam(hours)13.7V High/Low Beam(hours)
64193Original line500/900295/531
64193CBCool Blue150/35088.5/206.5
64193NBPNight Breaker Plus160/40094.4/236
64193ULTUltra Life500/3000295/1,770
64194100/80W Off-road only75/15044.25/88.5

Remember these times are for 63.2% of the bulbs to have failed. One way of looking at this table is you have to be luckier than the average punter for your bulbs to last this long, and much luckier than the average punter for them to last significantly longer than these times.

To select the appropriate bulb as a replacement then life plus light output is of interest. Osram give the light output data of the globes at 13.2V as a value in lumens with a ±tolerance. This data doesn't show any directional information so the following table is a bit limited. It is amazing how much more light the Off-road bulb generates but the tradeoff is an extremely short life. Although if it was crippled by stock wiring the bulb life maybe acceptable. The Cool Blue bulb appears to have traded reliability for appearance. The Silverstar has much less variation between the minimum and maximum values so the maximum might not be brighter but the minimum is guaranteed to be significantly better than the poorest of the other standard wattage bulbs. Unfortunately it is like the Cool Blue bulb, not reliable enough for my purposes. Really comes down to a choice between the Original Line and the Ultra Life bulbs. If bulb life ever becomes an issue I'll be hunting up the Osram Ultra Life bulbs.

CodeDescriptionLow minLow maxHi minHi max
64193Original line850lm1,150lm1,452lm1,848lm
64193CBCool Blue850lm1,150lm1,403lm1,898lm
64193NBPNight Breaker Plus850lm1,150lm1,403lm1,898lm
64193ULTUltra Life850lm1,150lm1,403lm1,898lm
64194100/80W Off-road only1,275lm1,725lm2,465lm3,335lm

The ugly

This isn't a list of the items I purchased. This list is my best guess of the parts I used. For example, I don't know the lengths of cable used so I've guessed lengths for this table. There is a good chance I've guessed wrong so don't use this as a shopping list. Some parts were drawn from my bits and pieces collection so I've had to guess details for some parts. Just to repeat... this is not a shopping list. Items can be packaged in multiples but only some of the parts might be used in this project. Therefore packet or minimum order quantities (i.e. 1m of cable) is used for pricing. The unused portions will become spares and added to the bits and pieces collection. The price list is incomplete but it provides a starting point.

SupplierPart NumberQtyDescriptionPrice (2012)Cost (2012)
JaycarHB-6134 1 IP65 Sealed ABS Enclosure, 240x160x90mm$37.95$37.95
JaycarZR-1004 5 1N4004 Diodes (Qty 4 per pack)$0.50$1.00
Supercheap Auto15024 1 2 Gauge battery lead 36" Lug/Lug$22.98$22.98
JaycarSL-2644 3 12V Mini Chrome Bezel LEDs, Red$3.65$10.95
JaycarSL-2645 6 12V Mini Chrome Bezel LEDs, Green$3.65$21.90
Autobarn48911BL 2 Narva 12V 55/100W P43t, H4 globes (for testing purposes)$19.35$38.70
JaycarSY-4076   15A Fused Automotive Relays (see discussion below)$8.95 
JaycarSY-4077 5 30A Fused Automotive Relays (see discussion below)$9.95$49.75
JaycarWH-3070 1m 2 Gauge Cable, Red$14.50$14.50
JaycarWH-3064 1m 4 Gauge Cable, Red$8.90$8.90
JaycarWH-3060 1m 8 Gauge Cable, Red$3.60$3.60
JaycarWH-3080 16m 25 Amp (12 Gauge) Auto & Marine Cable, Red$2.20$35.20
JaycarWH-3082 8m 25 Amp (12 Gauge) Auto & Marine Cable, Black$2.20$17.60
Supercheap Auto120312 1 Crimp connector to suit 2 Gauge cable (Qty. 2 per pack)$11.48$11.48
Supercheap Auto120311 1 Crimp connector to suit 4 Gauge cable$10.98$10.98
RS Components359-683 1m Silicone rubber insulated cable, Red (Qty. 25m roll) $20.10$20.50
RS Components359-712 1m Silicone rubber insulated cable, Green (Qty. 25m roll) $20.10$20.10
JaycarWH-5520 1 Comprehensive Heatshrink Pack$14.50$14.50
eBay  2 H4 style connectors, male$0.00
eBay  2 H4 style connectors, female$0.00
   1m 5-way automotive cable$0.00
JaycarHC-4020 1 4G - 4x8G Distribution block$12.95$12.95
JaycarHP-0744 2 IP68 3-6.5mm Plated Brass Gromets$7.45$14.90
JaycarHP-0745 2 IP68 4-8mm Plated Brass Gromets$7.95$14.90
JaycarHP-0746 1 IP68 6-12mm Plated Brass Gromets$8.95$8.95
JaycarHP-1225 10m 10mm Loom Tube$15.95$15.95
JaycarSF-1980 1 High Current Bolt Down Fuse Holder (MEGA Fuse)$19.95$19.95
JaycaySF-1982 1 125A Bolt Down Fuse (MEGA Fuse)$9.95$9.95
JaycarSF-2136 4 15A Blade fuses$0.85$3.40
JaycarHM-3198 1 Terminal Strip, 30A$2.85$2.85
RS Components290-6196 2 Anderson Type Connectors, 120Amp Single Pole, Red$6.70$6.70
Bunnings    1x 8mmx25mm stainless steel bolt$0.00
Bunnings    1x 8mmx35mm stainless steel bolt$0.00
     2x 8mm stainless steel washer$0.00
     2x 8mm Stainless steel nuts (used as spacers)$0.00
     2x 8mm Nyloc nuts$0.00
Bunnings    2x 6mm Stainless steel bolts (mounting MEGA fuse holder)$0.00
     4x 6mm Stainless steel washer$0.00
     2x 6mm Stainless steel Nyloc nuts$0.00
Bunnings    10x 4mm Stainless steel bolts$0.00
     14x 4mm Stainless steel washers$0.00
     10x 4mm Stainless steel Nyloc nuts.$0.00
     250x80x3mm Aluminium bar/sheet$0.00
AltronicsH 1825 2 Ring type, 8mm stud Crimp connectors. Yellow$5.70$5.70
JaycarPT-4725 12 6.8mm Crimp connectors, female, Yellow (Qty. 8 per pack)$3.75$7.50
JaycarPT-4625 7 6.8mm Crimp connectors, female, Blue (Qty. 8 per pack)$2.95$2.95
JaycarPT-4525 1 6.8mm Crimp connectors, female, Red (Qty 8 per pack)$2.75$2.75
Total        $469.99

It didn't cost me anything like the list total. For example there is over $40.00 allocated for two reels of coloured, 240V rated, high temperature, silicone insulated wire that had been bought to repair tumble dryers, but for this upgrade some cheap, standard automotive wire would have been equally acceptable. I used the silicone insulated wire so it is included to inflate costs. The following could be deleted: LEDs, enclosure, Mega Fuse, negative leads to headlights (use bodywork for return path), connectors, distribution block, stainless steel hardware, etc., etc. Look at Go Jeep's XJ Cherokee Headlight Loom Upgrade project for a more reasonable upgrade. Although when I lost a headlight filament the LEDs made it very easy to diagnose the problem. For my project the power distribution block was recycled from a powered speaker box. Crimp connectors are just consumables here. Don't know what past project needed the aluminium bar, etc., etc... Although it will be interesting to finish populating the component prices to work out how high I can push the total. Hopefully that figure should frighten off anybody contemplating this upgrade project. In the introduction I noted ~$500 cost and I wonder how many will suffer sticker shock and not get as far as this paragraph.

SY-4077 Relays: There are two similar fused relays sold by Jaycar, 15A (SY-4076) and 30A (SY-4077) with a significant difference in price, 15A at $8.95 and $9.95 for 30A. Although they are different prices I had expected the relay's mechanical components to be identical (i.e both 30A relays) just with different fuses installed to differentiate the products. Scratching around in Google I found a similar looking relay with supplied specifications.

Contact Arrangement1A (1H)(SPSTNO)
Contact MaterialAgSnO 2
Contact Rating (resistive)15A , 30 A /14VDC
Max. Switching Power420W
Max. Switching Voltage75VDC Max. Switching Current: 30A
Contact Resistance or Voltage drop≤ 50 m  Ω Item 3.12 of IEC 255-7
Operation Life Electrical100,000 Item 3.30 of IEC255-7
Mechanical10,000,000 Item 3.31 of IEC 255-7

This specification gives the same value of "Contact Rating (resistive)" for both 15A and 30A versions. My asumption is the 15A and 30A relays have the same relay contacts and the only change is the fuse, part number and price. Guess somebody would have to destroy a couple of relays to confirm this assumption.

Calculated current (refer Ohm's Law) at 12.0V for 55W (H4 low beam) is 4.58A and for 65W (H4 high beam) is 5.42A. When using off-road H4 bulbs the current to be switched could increase significantly i.e. 130W at 12.0V is 10.8A while 170W at 12.0V is 14.17A. Needless to say it is not that simple. A high power globe rated at 170W may have a manufacturer's tolerance ±15% so theoretically the maximum power could be 195.5W at +15%. However unlikely this would seem it could be specified at 12.0V. So this power could be equivalent to 16.3A at 12.0 Volts.

A bulb's resistance is non-linear which means when the bulb is cold the resistance is lower and the current drawn at switch on will be higher than the normal operating current. This is termed the inrush current, refer to Freescale Semiconductor's Lamp Inrush Current Computational Tool Application Note for further details. A simple estimation of the inrush current can be calculated using Ohm's law if you measure the resistance of the filamant when it is cold and the voltage that will be applied.

As the applied voltage increases the bulb will run hotter, its resistance will increase and the current drawn will increase but not as much as a simple Ohm's Law calculation would predict. I'm going to ignore all these complications and just go with simple Ohm's Law calculations to give indicative currents.

Description Power (W)
at 12.0V
Power (W)
at +15%
Resistance (Ω)
at 12.0V
Current (A)
at 12.0V
Current (A)
at 13.8V
Low Beam, Standard H4 Bulb5563.252.27675.276.06
High Beam, Standard H4 Bulb60692.08705.756.61
High Beam, Non-standard H4 Style Bulb1001151.25229.5811.02
High Beam, Non-standard H4 Style Bulb130149.50.963212.4614.33
High Beam, Non-standard H4 Style Bulb170195.50.736616.2918.74

At a worst case of 18.74A for a 12V, 170W rated bulb at 13.8V (without considering non-linearity or the inrush current) the obvious choice is the 30A relay. Given reliability is paramount a 30A relay is still appropriate with a worst case 14.33A for a 12V, 130W rated bulb at 13.2 Volts. For bulbs rated at 100W or lower 15A relays should be acceptable. 30A fused relays were selected for this installation so that contact rating will never be an issue. 8A fuses would probably be the minimum rating that could be used while I would select 10A or 15A fuses due to bulb 'inrush' current for cold filaments (resistance is lower and current drawn will be higher until filament heats).

If using a minimal 3 relay version (single relays for low beam, high beam and spotlights) the 15A relays could become seriously underated for high power globes and not suitable for the purpose. While for the 5 relay version the 30A relays are overkill for 65/55W globes.

Further Improvements

There are some obvious changes to further reduce voltage losses . Increase the cross-sectional area of the copper wire (e.g. use heavier gauge wire from the control relays to the headlights), decrease complexity (e.g. delete Mega Fuse), improve existing wiring (e.g. upgrade battery cable from negative terminal to chassis), shorten leads (e.g. the cable from battery's positive terminal to the control relays), etc., etc... And this is without considering more radical solutions like placing control relays directly behind each headlight and running heavy cables from the battery to these relays.