[voxmagna]

I've bought a few aftermarket led bulb replacements and run tests on them. The Chinese sell using brightness as their USP even trying to make theirs look brighter than other sellers in photos. I've dissected quite a few and whilst they may use led chips with a genuine power spec., they use lower value series resistors to increase current and do nothing about removing the heat. I've yet to find a replacement led equivalent lamp where they've considered thermal management. Their other design issue is they often design on the bench for 12 volt when all car lamps should be designed for 14.5 volts. Leds bought from genuine companies like Ring or Philips should be good, but they all find it hard to remove heat for the standard 2 wire push in base. A brass capped lamp offers more opportunity to use the the cap as a heat sink. You can sell a led replacement as 1W, 3W or 5W. They use the same leds and push more current through them with most buying the biggest number.

I've left some of these Chinese auto leds running on a bench supply for hours. Most go pop after 2-3 days on continuous test, some only lasted 3 hours. O.K that may translate to a reasonable mileage, but you get better life from a tungsten bulb. If I buy any of them I hack their resistors increasing their value to reduce current, whilst accepting a 10% reduction in brightness.

The best auto leds are those designed with constant current chips running at sensible currents with a metal body for the heat sink, they will be for larger led clusters like DRLs. Leds should not fail easily and I can't see how flashing a led causes it to fail, unless they are running too much current through the chips and peak voltage causes breakdown. My EOS like others has led clusters for the turn signals which flash often. The failure of individual leds in those is due to their highish running current to meet brightness regulations for the older version led chips, or water ingress.

You get what you pay for with Chinese led bulbs, if you buy any, connect one across your battery and leave it running to see how long the led lasts - if it starts feeling hot and hotter you might get 3 days if you are lucky!

The lamp failure test at key on flashes a led bulb because it needs far less current to see it flash than a tungsten lamp. But if it can't survive that it isn't going to last long turning the lights on and off either.

 

[voxmagna]

I think I would be confident with those. I bought a pair of Philips DRLs and was V. impressed with the solid finned ali heatsinks, although their black paint finish didn't weather well and I repainted them. The problem with bad thermal management of led designs on cars is worst when they are sitting in traffic in Summer with no airflow. Slightly off topic, I just got some cheap Chinese MR16 led replacements. Running 5W from 12V with a zillion led chips on a 50mm diameter ali pcb and no space between, they got so hot the solder connecting the supply wire to the board melted and the wire lifted off! Some of this stuff really is cr*p! I'll fix these lamps by increasing the current sense chip resistor value and dropping the output from 5W to 3 W.

 

[voxmagna]

What are you doing about the other lamps? I fitted 6K HIDs on a bike but the tungsten parking lamp looked sick against it. I had to led that too to get the right look for all lamps. I ran my HIDs on low beam all the time and I'd expect to do the same with leds. Some argue different, but a 6K low beam is very visible and IMHO safer in daylight. Do a simple sum of running hours for your annual mileage at an average of 50mph and you find it takes years to burn even 6000 hours for a HID lamp. Running a led lamp continuously as a low beam is why I would fit one and if it won't do the 10-20k hours life they claim, it can't be that good?

I have a later 2012 VAG group car which to me has strange in built drls. They modulate the tungsten low beams reducing their power, but they put out an extemely yellow light. I haven't even thought about doing anything to those lights as I'd have to deal with the modulator and bulb failure monitoring.

 

[voxmagna]

Its good to see somebody trying these things out and posting, even if they don't work. I've played with a couple of housed led spotlights and some mains led floods, worked out the led constant current drivers and modded their circuits. Apart from the Chinese over running them, most have the ali conducting plate of some kind with 'x' surface mounted leds. When the Chinese sell something saying it's 50W, accept and run it at 25W and it will be reliable. I'm going to experiment with a 35W H7 and try out some cooling ideas, even if it means chopping off part of the rear end. The V.W interface will come next.

 

[voxmagna]

If I'm playing around with bulbs and have some spare time, I'll investigate how a tungsten bulb is seen by the system and whether it can be easily emulated electronically? Each bulb would need an external 'dongle' of some kind. I won't go down the road of padding out with high wattage resistors since it defeats the object of using leds and for the trouble you might as well stay with tungsten. Which bulb and holder is the easiest to play with? I don't fancy messing with fogs in the front bumper, been there and not so easy to remove as they say. The problem with emulation is I might fool the system to think it see a normal bulb, but it won't then tell you if the led replacement has failed, is that something you can put up with?

 

[voxmagna]

It can get more complicated. Many of the regular bulbs are electronically driven from PWM or d.c controllers. Bulb failure sensing might only need to catch the first few cycles of PWM at key on, or is continuously monitoring peak pulsed current rather than d.c. You would need to 'scope each bulb you were considering replacing to detect PWM or pulsed lamp testing. Once you know how they do it, you can design the electronic 'dongle'. For d.c driven lamps they could generate a short duration lamp test pulse which enables lamps to be tested when they aren't lit. That's how you could get the bulb failure warning at key on during system initialization checks.

I was experimenting with small 1W led puddle lights which would normally be 5 watt tungsten. As supplied they always put up a diagnostics error but no bulb failure. On those I could have added a fixed resistor to emulate the full 5W normal load, but it would run as hot as a lamp and would need sufficient air cooling around it. By experiment I could choose a higher resistor value which kept the led lamp current just above the diagnostics fault trip level, but you have to do this for a 9-10v system voltage simulating a low battery charge or cranking voltage drop. Otherwise, you will still trip a bulb failure diagnostics error cranking on a low battery.

 

[voxmagna]

I can't be sure until I measure what they are doing but I was thinking along the lines of a low RDSon power mosfet into a low ohm resistor when lamps were off, then turn off the fet when the lamps are powered? The resistor can be low wattage as long as the fet is turned off fast enough. If they they sample the lamp load current once it's running, that could be simulated by pulsing the same fet aka pwm but very narrow power efficient pulses. If they are detecting d.c, it's more difficult. If their startup lamp test is done over tens of milliseconds, a capacitor on the fet gate will handle that.Tungsten lamps may be say 50 Watts, but the equivalent resistance at this wattage is for the lamp running. When the filament is cold, if they are testing for continuity, the filament resistance will be only tens of ohms. Whatever you do there will have to be one dongle per lamp. I'd try to avoid winding lots of 'L's. Changing coding is the smartest solution but I don't think there are options. Do Xenon kits give no problem because V.W use them? I suspect the Chinese kit builders will have reverse engineered from the OE parts? Some led bulb suppliers talk about using capacitors to fool the bulb monitoring, but until more is known as to how it's done I'm not sure.

A bulb that is dead (zero current draw) or pulling a current below the required amount will trigger the bulb out warning. At least That is what I am seeing in my experimenting.

That's what I saw on the puddle lights as I kept increasing the shunt resistance value. But I hadn't got a 'scope across it so I don't know if it was d.c fed or pwm. Thinking about it, the door puddle lights are on the cabin light system which is dimmed by pwm. If lamps are driven by pwm and you know the frequency, you can calculate capacitor values that will take the a.c current without dissipating power like a resistor.

 

[voxmagna]

OE Xenons could be using a different system for bulb failure or similar done in a different way. When a halogen bulb lights there is an initial high inrush surge current that doesn't last very long before the filament burns at the normal lamp running current. If the lamp is turned on and off within a short space of time e.g main-low beam-main, the filament doesn't have time to go cold and the inrush surge current will be lower. The lamp test/failure monitoring would have to allow for this as well as low battery reducing both the inrush and running current

OE Xenons have a totally different start up behaviour: Before the arc strikes there's a huge current surge for mSecs for the high voltage igniter, then the arc fires and the ballast circuit regulates (by PWM) the running current of the lamp. The load seen by the battery is a combination of very high short period current surge and the pulsed current for the ballast. When the battery voltage falls e.g low battery or cranking, the current into the ballast increases - which is the opposite behaviour to a tungsten bulb.

I can only look into the behaviour of tungsten bulb monitoring because that's what I have. If I'm right about OE Xenon I would expect different versions of the headlamp controllers for tungsten and Xenon, or the same modules with different software inside, or an adaptation option to set different bulb monitoring behaviour. Somebody with a parts list or who has fitted OE xenon could confirm? The addition of capacitors and resistors are the hacks I've seen used to sell the Xenon kits, but I wouldn't rely on them working for all car brands and hidden from you could still be the nuisance of having DTC fault codes triggering all the time whilst not putting on the bulb fail.

 

[voxmagna]

Turning off the cold bulb monitoring with diagnostics solves half the problems of bulb failure warning (key on check). Simulating the load with a fixed resistor can work, when the lamp is running but it's not a neat solution. If the hot and cold bulb monitoring are similar they are sampling the lamp running current every 300-400 milliseconds. I've got some ideas but no practical solution yet to syncing a load, concurrent with the sample time, but I haven't given up.

 

[voxmagna]

As aku says, cold start lamp out warnings for ALL lamps is easily disabled but hot failure where each lamp current is monitored is rather different. There's a YouTube vid explaining how a Skoda (cheap V.W!) was modified using the V.W VAS workshop tool with various screenshots showing a list of lamps that could be selectively removed from hot lamp checks. There was a supplementary script file required for the actions to be applied. I looked with my VAS but couldn't even find the same screens as were posted and went no further, but perhaps the Skoda VID needs a bit more investigation? Or the controllers in the EOS don't fit the Skoda template and options for selectively disabling the lamp test are not available?

 

[voxmagna]

VCDS (VAGCOM) doesn't offer recoding options for disabling hot lamps and if a special script was required with V.W VAS to do this on a Skoda, I doubt any of the others will do it? There seems more to it than just plugging in a diagnostics tool and searching for an option to selectively disable the lamp warning. My impression of the Skoda hack was somebody had really gotten inside the VAS tool software to understand what it was doing and then write a custom script for it?