Need help with LED flicker.

[ryanj]

Hi, I must start off by saying I am not an EE I have a very basic understanding of electronics so please forgive me if I sound like a complete idiot, but I am out of ideas on how to solve my problem.

I am trying to retrofit a car with LED turn signal assemblies into a car that originally had incandescent bulbs. The car has a bulb monitoring system to detected when a bulb is out. This system sends out a very short burst of power (maybe about 10 -15 ms?) about every 30 seconds when the turn signal is not in use. This short burst of power causes the LED's to light for a split second when not in use.

The LED's have a resistor in series with them that stops any bulb out warnings from occurring, but this flash is still present. I have been messing around with capacitors in series and parallel to try and solve this, but they only made the problem more apparent. Looking for any suggestions on what to try.

 

[alec_t]

Welcome to ETO!
Is the turn signal switch in the +12V supply wire to the LEDs/bulbs, or in the chassis connection to them? More likely the former, but we would need to know.

 

[ryanj]

Ground is constant and +12v is switched

 

[MikeMl]

The obvious answer is to parallel the LED assembly with a power resistor that emulates the incandescent lamp. You will have to determine the max resistance that still satisfies the "burn-out" detector. Then you will have to determine a wattage rating for the resistor that will dissipate the wasted power...

 

[AnalogKid]

If this is the current connection:

12 V source > resistor > LED > GND

Then put a capacitor to GND between the resistor and the LED. This will act as a lowpass filter. With the right cap and R sizes, the pulse will not charge up the cap far enough to forward bias the LED. What is the resistor value?

ak

 

[ryanj]

I will find out my resistor value and get back to you. Also just realized the resistor is in parallel not series as I stated earlier.

 

[MikeMl]

If this is the current connection: 12 V source > resistor > LED > GND...

Likely, the current through the LED assembly is much less (20%?) of that drawn by a lamp. Also likely, the LED assembly has several LEDs in series, with only ~2V dropped across the current-ballasting resistor (if it is even a resistor; might be an active circuit).

AK's idea might work, but you would have to add the resistor/capacitor combination external to the LED assembly.

 

[dr pepper]

For bikes you can get special indicator relays for led's, you can probably get them for cars also.
If your indicators have a can bus interface with bulb monitoring then you'll have to use a load resistor.

 

[Diver300]

I think that the OP has got "series" and "parallel" mixed up. The LED lamps will have a series resistor or some other current control device, or they would fail when 12 V is applied.

There is another resistor in parallel to take current that is approximately equivalent to that of an incandescent bulb. That current will fool the car's systems into thinking that the indicators are working.

The problem is that the car is attempting to detect the incandescent bulbs with a very short duration pulse, that is too short to cause a bulb filament to get red hot, let alone white hot as it would be when lit. The LED lamp responds instantly and lights briefly.

What is needed is something to slow down the turn-on of the lamps by a few milliseconds, so that the LEDs don't come on. I suggest a relay. For an ordinary automotive relay, connect terminal 30 to ground, terminals 87 and 85 to the indicator feed wire, and terminal 86 to the LED indicator. Leave the load resistor connected to the indicator feed wire, not the LED indicator. You need to do this for each side of the vehicle.

When the indicators come on, the relay will operate and light the LED. However, there should be a sufficient delay that the very short bulb testing pulse will fail to operate the relay, leaving the LED off.

You should be aware that the resistor that fools the car's system into thinking that the indicators are working means that you won't get any warning if the LED indicators fail, and that a working bulb failure system for indicators has been a legal requirement for decades in most countries.

 

[Diver300]

For bikes you can get special indicator relays for led's, you can probably get them for cars also.
If your indicators have a can bus interface with bulb monitoring then you'll have to use a load resistor.

The CAN bus interface is a total red herring. What makes a difference here is that whatever circuit drives the indicators checks the current that they take. How that circuit communicated the fail status to the driver is totally separate. It just happens that the electronics that allowed bulbs to be monitored with low currents and/or short pulses got applied to cars about the same time as CAN bus.

 

[MikeMl]

I'm guessing that the original lamp is ~5W, so it would draw I=P/E = 5W/14V = 0.36A. That is a hot resistance of R=E/I = 14V/0.36A = 39Ω. The cold resistance of an incandescent lamp is usually much lower, about 1/3, so ~10Ω.

Likely, the LED assembly operates on less than 100mA, so has an "effective" resistance of 14V/0.1A = 140Ω. You will have to fool the "burn-out" detector into thinking that the LED assembly has a cold resistance of ~10Ω instead of 140Ω. To brute force this with just an external shunt resistor would take a power resistor of 10Ω 10W, or a 4W 14V lamp

Some power saving could be had with the following circuit:



Can you buy a 3300uF 20V electrolytic cap cheaper than a 10Ω 10W resistor?

 

[alec_t]

What is needed is something to slow down the turn-on of the lamps by a few milliseconds

Agreed. Perhaps use a PFET in the +ve rail, with an RC circuit to delay turn-on, and at the same time control a NFET to disconnect a shunt resistor emulating the incandescent bulb.

 

[Les Jones]

I suspect the lamp test is done by applying the testing pulse via a resistor and checing that no significant pulse is seen on the lamp end of the resistor. The OP says that the test does not see the lamps as open circuits. I think MikeMI's bottom schematic in post #11 should work. I also suspect that R5 could be increased in value and the value of the capacitor reduced to save space. The value of the 1K resistor R6 could be increased as the capacitor has about 30 seconds to discharge.

Les.

 

[alec_t]

While Mike's circuit nicely fools the bulb-test, how does it stop the LEDs flashing briefly, which is the OP's concern?

 

[dr pepper]

My last car had a black box at the front behind the rad, it sensed current to all the lamps at the front, if one failed it reported to the dcu that there was a failure.
Al cars are diffrent though, my mrs's berlingo doesnt have a flasher relay its electronic and is part of the dcu, my car which is 4 years newer actually has a 3 pin flasher relay not controlled by any black boxes.

 

[AnalogKid]

While Mike's circuit nicely fools the bulb-test, how does it stop the LEDs flashing briefly, which is the OP's concern?

Aha! That's why I suggested putting the capacitor after the 18 ohm current limit resistor. For a few milliseconds it keeps the voltage available to the LEDs below (Vf x 4). I guess that to do this the "right" way, split the 18 ohm resistor into two 10 ohm resistors in series and put the capacitor between them to GND. Now the pulse sees 10 ohms per Mike's calc, but the LEDs see are fed through 20 ohms for some power savings.

ak

 

[MikeMl]

While Mike's circuit nicely fools the bulb-test, how does it stop the LEDs flashing briefly, which is the OP's concern?

I forgot about that. Perhaps the TS can do something with this:

 

[MikeMl]

Aha! That's why I suggested putting the capacitor after the 18 ohm current limit resistor. ...

Doubtful that the TS can get hold of the node downstream of the current limiter in a sealed LED assembly. Likely, it has just two wires accessible without cutting into it... The circuit of post #17 has only added parts external to the LED assembly.

 

[AnalogKid]

You're no fun anymore.

 

[Diver300]

My last car had a black box at the front behind the rad, it sensed current to all the lamps at the front, if one failed it reported to the dcu that there was a failure.
Al cars are diffrent though, my mrs's berlingo doesnt have a flasher relay its electronic and is part of the dcu, my car which is 4 years newer actually has a 3 pin flasher relay not controlled by any black boxes.

Somewhat off-topic, but the adaption of electronic innovation in cars is very sporadic. I had a 1985 Montego that had electric windows controlled by electronics, giving one-touch operation, obstruction detection. By 2004, when my daughter's current Nissan Micra was made, it was fitted with electronic injection control, OBDII port, ABS, keyless entry etc, but the electric window circuit was just switches and motors, no electronics at all.