CANBUS-friendly LEDs
CANBUS-friendly LEDs
Author
Discussion

Watchman

Original Poster:

6,391 posts

269 months

Wednesday 8th March 2017
quotequote all
This is diving me mad. I'm trying to "correct" Mercedes' odd decision to equip my 2011 GL with half and half LEDs and filament bulbs by replacing the bulbs with LED equivalents. But I'm having problems getting LEDs that don't throw up CANBUS errors.

Anyone have any UK-based recommendations? Ideally somewhere in the Midlands where I can actually go and visit and test-fit them. I'm bored of ordering them by post only to have to seek a refund when they clearly don't work.

Thanks,

rgv250ads

434 posts

138 months

Thursday 9th March 2017
quotequote all
100% want to follow the answers to this too, especially for T5 sidelight bulbs!

Watchman

Original Poster:

6,391 posts

269 months

Thursday 9th March 2017
quotequote all
From the little info I've found, you need a resistor load in parallel that is low enough to emulate the original bulb which then will then generate a lot of heat.

The festoon type reg plate LEDs I found have heatsinks but clearly the resistance is still too high because I'm still getting a dashboard error message.

The reverse light works without error.

The foglight (rear) won't even power the LED at all.

I can't get to the front sidelights to pull them out and replace them with the LEDs I have bought.

steve-V8s

2,924 posts

272 months

Friday 10th March 2017
quotequote all
This is an interesting conundrum. Plainly to prevent an error message the system has to be convinced that a functioning lamp is fitted. In simplistic terms fitting a resistor which provides the same load as a lamp may work in some situations but is far from a good solution mostly because it will generate a lot of heat.

One of the issues is the resistance of a filament lamp changes significantly between cold and hot so the system will probably look for different measurements when the light is on or off. On a different German brand I looked at the system briefly polled the lamps when the ignition was switched on but the lamps off. Then with the lamp on it measured the current and seemed to require fairly close result to the expected value.

A single LED typically has a forward voltage of around 1.2v so the plug in replacements tend to be constructed by connecting a number of LED in series along with a limiting resistor and then stacking the result in parallel to achieve the required light output. Different brands seem to use a different series parallel combination which is probably why some are less likely to result in an error message.

None of the ones I have seen are type approved so are not road legal, if you look at the packaging they are approved for “off road use only”

Just to be pedantic it is not a “Can Bus” error. Can Bus goes nowhere near the lamps, communication between the bit of electronics which is controlling the lamps and the electronics controlling the display may be Can Bus but could be another communication system.

Probably best to stick with the original filament lamps, they look rather nicer anyway.

Watchman

Original Poster:

6,391 posts

269 months

Friday 10th March 2017
quotequote all
I've performed similar analysis, and even wrote a similar system for a pre-modern car-ECU based on the characteristics of both filament bulbs and LEDs - it was part of my electronics degree about 25 years ago. It's this change in a filament's resistance throughout the heat-cycle that is tricky to replicate with an LED. If you're right, and the ECU can detect this, then I'm probably stuck.

Where I have *some* optimism is that this last point doesn't appear to be true in the case of my GL. In an old Cavalier I had, I know it did this short test of all bulbs because it reported failures even before I switched them on. In the GL, it only tells me of the failures after they are switched on.



I also know it's not a real CANBUS error - but that's the term everyone uses. CANBUS is a networking protocol for all the car's various ECUs to converse with each other. It's similar-ish to Ethernet.



I might have been tempted to leave the filament bulbs in place if they were all filament bulbs but the light clusters are a strange mixture of filament bulbs and LED panels on the rear, and filaments,LEDs and HIDs on the front. In particular the sidelight bulbs jar against the HIDs - the former being very yellow against the latter's white. And festoon reg plate lights are terribly dim.

As mine has this mixture from new, I can only assume the ECU is capable of measuring the failure of LEDs as well as bulbs. It's just whether it's hard coded measurements against particular clusters, or even measured with hardware instead of software - meaning no changes would be possible for me to influence.



I expect LEDs could work without ECU fiddling if:

1. You had a cluster of LEDs large enough to replicate the load of a bulb (still not sure about this initial detection-test), or

2. You use a resistor and a heatsink large enough to dissipate the heat safely. Trusting a solution like this to the cheapest Chinese supplier troubles me.



I'm still hopeful of finding a supplier who has an out of the box solution I can trust.