PWM replacement for car lighting dimmer

[PlaneCrazy]

Greetings, all

I drive a 1980s Mercedes-Benz, with a series-connected rheostat dimmer for the instrument cluster and console lighting. These rheostats are notoriously fragile, and when they fail, the car is legally and practically unsafe to drive at night. Also, I would like to upgrade the backlighting to LEDs due to their greater brightness and lower heat. So I have been toying with the idea of a drop-in PWM replacement for the finicky, fragile and NLA rheostat.

I have access to two terminals between the battery (via some fuses and switches, of course) and the load. I don't have access to chassis ground. And most OTS dimmers have 4 terminals, not two; furthermore, I would like to use a potentiometer with a through-hole shaft, so I can have a shaft custom-machined to fit the existing knob.

Before I even begin looking at packaging, I have designed this CMOS 555-based circuit that I would like you all to review for me, please.

The yellow circles indicate the terminals where the rheostat unit connects to the rest of the system. The maximum power consumption, based on standard equipment bulbs and max optional equipment, is ±20W. I am aiming for a frequency high enough to prevent flickering for LEDs while being robust enough for incandescent bulbs and quiet enough to prevent artificial tinnitus. I have excluded the fuses and main switch upstream of the battery-side terminal for brevity.

I understand that I need to use a CMOS 555 and a P-MOSFET, since I have no choice but to control the high side of the circuit.

I am open to suggestions, as well as red flags.

Thanks

 

[Diver300]

My immediate red flag is that changing to PWM and to LEDs will mean that the dashboard lights will be flashing at the PWM frequency. You are then relying on the slowness of response of your eyes rather to smooth out the flashing.

In situations where the eye is moving across a flashing light, with high contrast, flashing lights become far more obvious and intrusive. As the dashboard is high contrast, and you hopefully only look at it briefly when driving, both the conditions for PWM being visible are met.

People's sensitivity to flashing lights varies enormously, and I am particularly sensitive.

I would use a linear regulator or make sure that the PWM is smoothed sufficiently before reaching the LED panel lights.

The other red flag is that LED replacements for bulbs may behave in unexpected ways at low duty cycles and at reduced currents or voltages.

 

[PlaneCrazy]

Good observations, but I have been using LED backlighting in this car for a while. The LEDs cause no problems for backlighting, except that it cannot be dimmed with a simple potentiometer. The dimmer does not work on the warning lights, which should be obvious but I will mention it all the same.

I will experiment with the frequency, but I'm guessing 150Hz should be fine for either incandescent, LED or a mixture of both. I don't know how a higher frequency will affect the incandescent bulbs. I do not have epillepsy or any other affliction, and am not bothered by some flickering. But, perhaps an enhancement could be to limit the minimum dimming? Also, how to smooth the signal for LED?

 

[Galgso]

When you say that you haven't got access to chassis ground, do you mean you can't access chassis ground at all or just the return path for the lights? I'm not sure why you've shorted VCC and GND for the 555 but you'll need a separate ground line for any type of electronic PWM dimmer they can't just go in series.

 

[PlaneCrazy]

When you say that you haven't got access to chassis ground, do you mean you can't access chassis ground at all or just the return path for the lights?

Where the rheostat unit connects to the instrument cluster, it interrupts the +12VDC line in series with the bulbs. This is the only way I can implement a drop-in replacement. You can see on this unit that the rheostat has been shorted, so the lights are at full brightness all the time; as mentioned, a common "fix" on these cars.

 

[PlaneCrazy]

View attachment 150406
I'm not sure why you've shorted VCC and GND for the 555

Good catch! This is why I need help. Thanks.

 

[Galgso]

I understand that the PWM dimmer needs to go in series with the lights which have to be high-side driven, but the control circuitry generating PWM will need a chassis ground connection. If there's no convenient place behind the dash to tap chassis ground, then I doubt you'll be able to get a PWM setup working because you won't be able to power the 555 or whatever IC you're using.

 

[Galgso]

Also I'm not sure how your 555 setup is supposed to work, but here's a 555 PWM circuit that I've used many times with good results:

Where there would normally just be a resistor between discharge and trigger + threshold, there's a potentiometer and each side of the pot is connected in series with a diode. Because the diodes are going in different directions, if you move the pot one way it increases the resistance in series with one diode and decreases the resistance in series with the other. This keeps the frequency constant but changes the charging/discharging times giving you a variable duty cycle.

 

[PlaneCrazy]

I understand that the PWM dimmer needs to go in series with the lights which have to be high-side driven, but the control circuitry generating PWM will need a chassis ground connection. If there's no convenient place behind the dash to tap chassis ground, then I doubt you'll be able to get a PWM setup working because you won't be able to power the 555 or whatever IC you're using.

There is a grounding point behind the cluster. I could certainly use that.

 

[danadak]

Dont forget to protect circuit from load dump :

https://www.ti.com/lit/snva681

https://www.ti.com/lit/pdf/snoaaa1

https://diotec.com/en/application/application-notes/application-note-about-automotive-load-dump-protection.html?file=files/diotec/files/pdf/service/applications/AN_automotive_load_dump_protection.pdf

Regards, Dana.

 

[Diver300]

Also, how to smooth the signal for LED?

That depends on the LED lamps.

The LEDs themselves can be dimmed by reducing the current. The LED lamps will have some sort of driver to limit the current, and how those will behave with a reduced current or voltage is anyone's guess. Also they may work OK with PWM, but some designs just won't dim until the PWM duty cycle falls below about 10%

 

[PlaneCrazy]

I will test the LEDs with my Arduino.

 

[PlaneCrazy]

I cannot simulate the load dump circuits, but I will look into it.

@Galgso, your circuit seems to work well, even if I have to add a ground wire.

 

[Galgso]

With the resistor values and capacitance I used the frequency will be around 230Hz. I doubt the PWM frequency would be an issue on a dashboard but if you want to be safe I'd run it at 1kHz. Here's a nice video showing the effects of PWM frequency on visible flicker:

Also, one of the limitations of this circuit is that the duty cycle cannot go down to 0% meaning that you will not be able to get full brightness. At 10% duty cycle, the PMOS will be on for 90% of the time and that will be the brightness limit. This is due to the 1k discharge resistor. The lower this resistor value is relative to the potentiometer resistance the closer the duty cycle will be able to get to 0%. Also, for a higher current circuit it would be necessary to artificially limit the duty cycle to between 10% and 90% to ensure that the switching MOSFET gate had enough time to charge and discharge otherwise there is the risk of excessive power dissipation in the MOSFET but this will probably not be necessary in your case.

About load dump and transients that can occur on a car's low voltage system, you can get TVS diodes designed to suppress such transients and absorb the excess energy.

https://diotec.com/en/application/application-notes/application-note-about-automotive-load-dump-protection.html?file=files/diotec/files/pdf/service/applications/AN_automotive_load_dump_protection.pdf

Danadak's link here shows an app note with some devices designed to do this. If you are connecting your circuit to a car's 12v system it may not last long without some kind of overvoltage protection.

Also, it is advisable to add a reverse polarity protection diode (this can be any rectifier diode capable of handling the current of your lights) in series with the positive rail of the circuit to prevent it from being fried if the polarity is reversed accidentally during testing.

 

[ZipZapOuch]

People who claim that your eye will not notice a flicker of PWM above 70Hz or so, say that for a static situation (you staring at the LED and your persistence of vision helps bridge the off-times in a helpful way). In a dynamic situation, however, you can see the results of flicker at much higher frequencies (again because of persistence of vision but in an unhelpful way).

A dynamic situation would be like the first Cadillacs to have LED tail lights/brake lights. They used LED brake lights on at full power but used the same LEDs as the dimmer tail lights with a 30% to 50% duty cycle PWM. Unfortunately, GM used a 256Hz PWM which, to the engineers who never used PWM in a dynamic situation seemed like very fast rate. The unfortunate part is that people driving behind the Cadillacs SCAN their eyes across the roads (left to right) can create a series of light images across the trailing driver's field of view.

 

[PlaneCrazy]

I am aiming for 25kHz. It is beyond the audible frequency range, it hardly seems to affect incandescent bulbs, and seems well-suited to LED's.

 

[Diver300]

I will test the LEDs with my Arduino.

Make sure you test them in similar conditions to where they will be used. Looking directly at lights will not detect flicker nearly as well as in other conditions.

(Edit: just noticed this is a repeat of what ZipZapOuch said)

 

[Diver300]

People who claim that your eye will not notice a flicker of PWM above 70Hz or so, say that for a static situation (you staring at the LED and your persistence of vision helps bridge the off-times in a helpful way). In a dynamic situation, however, you can see the results of flicker at much higher frequencies (again because of persistence of vision but in an unhelpful way).

A dynamic situation would be like the first Cadillacs to have LED tail lights/brake lights. They used LED brake lights on at full power but used the same LEDs as the dimmer tail lights with a 30% to 50% duty cycle PWM. Unfortunately, GM used a 256Hz PWM which, to the engineers who never used PWM in a dynamic situation seemed like very fast rate. The unfortunate part is that people driving behind the Cadillacs SCAN their eyes across the roads (left to right) can create a series of light images across the trailing driver's field of view.

VW used 100 Hz, with 10% duty cycle on their tail lights. That gave gaps of 9 ms. That was considerably worse for following drivers than 30% at 256 Hz, which would have been gaps of 2.7 ms.