My car (years before CANbus) has a very basic mechanism for detecting bulb failure that uses a heating element and a bi-metalic strip with electrical contacts, as per this circuit.

The input is +12V on R, output to the bulb is RG. WS is the feed to the bulb failure warning light through a NC pair of contacts, attached to a bi-metalic strip. When you switch the light on, the heat from the coil causes the bi-metalic strip to bend and break the feed to the warning light. When the bulb fails, the coil cools and the contacts reconnect. The advantage of this system is that for a few seconds, while the coil initially heats up, the bulb warning light will be lit to indicate that the warning light itself is OK. The downside is that it is prone to failure, cannot handle high current loads like a headlamp, and not very accurate when multiple bulbs and LEDs are connected to the same circuit.

There must be a way of sensing the current flow electronically to provide a more reliable system. What would I need to build a monitoring system for multiple bulbs, in high power circuits and also for low power LEDs, which includes the ability to perform the warning light check when turned on?

 

... There is a device called a reed switch, which can be closed when a magnetic field is activated.
The general idea is to wrap a number of turns of wire around the reed switch so that the current flow to an individual bulb creates a magnetic field that is sufficient to open or close the reed switch, as the case may be, and de-activate a light bulb out warning.
reed switch
... This will require some testing or modification, but may work for your application.

 

You could probably attach a suitably-rated Hall effect sensor to the cable feeding a bulb, to determine whether the expected current is flowing or not. Depending on the sensor, a ferromagnetic flux-concentrator might be needed to enhance the magnetic field the sensor is subject to.

 

I seem to recall my old XJS Jag had fibre optic conductors no larger than a small gauge wire ran to a small indicator panel.
Max.

 

I seem to recall my old XJS Jag...

Strangely enough, the car is question is an XJS, so I know it well. The fibre optic set up was for distributing illumination to dashboard switches when the side lights were on. It was not part of the bulb failure warning system.

 

+1 on the reed switch,, easily done..

 

Strangely enough, the car is question is an XJS, so I know it well. The fibre optic set up was for distributing illumination to dashboard switches when the side lights were on. It was not part of the bulb failure warning system.

It has been quite a while, so my memory has dimmed.
But the technique could still be used..
Max.

 

The general idea is to wrap a number of turns of wire around the reed switch so that the current flow to an individual bulb creates a magnetic field that is sufficient to open or close the reed switch

I see one problem with this approach - I would have to be careful about isolation of the reed switch so that it wasn't affected by power flowing in other parts of the wiring loom. To pick one example, with 4 headlights, including main beam, that is 6 circuits to monitor and each reed switch would have to be pretty much in the headlamp bowl, with shielding from the rest of the powered wires. Then I have to run each switched signal back to an indicator on the console, I was hoping for something that would be installable close to the fusebox in the cabin. It also doesn't give me the self test feature of checking the warning light.

 

Since you already have a monitoring wire at the lamp, why not wire a low value resistor in the return path that develops at least 0.7V with the lamp current flowing; then use the voltage to switch on an NPN transistor, signalling that the lamp is on.

 

Since you already have a monitoring wire at the lamp, why not wire a low value resistor in the return path that develops at least 0.7V with the lamp current flowing; then use the voltage to switch on an NPN transistor, signalling that the lamp is on.

The monitoring wire is not at the lamp, it is behind the fuse box, and there may be more than one lamp on a circuit. It gets kind of ugly, this is just the side lights:

Some further searching gives me this: https://www.electroschematics.com/auto-12v-bulb-failure-warning/ - detecting the voltage across a couple of diodes in series with the load. But I need to scale it up to multiple bulbs, and it doesn't give me the warning lamp failure test on start up. I'm thinking that this problem is not an easy one to crack in a purely analogue fashion, and I may have to use a micro controller approach.