2 pin flasher

I have to asked the obvious, did you read the PDF?

Or are you trying to include a 'blown' bulb fault detection into it? If so that is quite bit more involved...

 

CocaCola, yes I think he wants to drive two bulbs in parallel and detect when one goes open circuit, and flash the other one twice as fast as a warning that a lamp has failed.

I assume this is for the front and back indicator of a vehicle?

I think that old-style electromechanical flashers have this characteristic, although I think it's more an accidental feature due to the way they work internally.

To the OP. Please give a LOT more details if you hope to get any useful suggestions. Specifically, what are the voltage and current ratings of the bulb(s), what exactly are you trying to do, and any other information that might help us come up with a solution that will be useful to you.

 

OK. The normal way to detect the load current is by inserting a resistor in the current path and measuring the voltage across it. In this case I would insert this resistor between the source of the main switching MOSFET and the negative rail.

At your nominal current of 3.8A I would aim for about 0.5V loss across the resistor, so it needs to be about 0.13 ohms. While the MOSFET is ON, the voltage on its souce will be about 0.5V if both bulbs are present, or about 0.25V if one bulb has failed.

Circuit changes to make use of this information and change the flash rate are somewhat involved. In fact I think my preference would be to use a microcontroller. A discrete solution is possible, but quite complicated I think.

I will have a go at designing a discrete solution but I may give up on it. Do you have experience with microcontrollers, or a friend who does?

Edit: I looked into possibilities for a discrete solution but I don't like it. I'm suggesting a small microcontroller such as the ATTiny5 or a PIC10F220. If anyone here has ideas for a discrete design, I'd be keen to see them.

 

I suppose a JFET + resistor across R3 with the gate driven by a comparator on the current sense.

Bob

 

I have had a think and now I am thunk.

I thought that two oscillators could be used and switched from one to the other. A gate in a 4093 can make a very simple oscillator and if unequal on/off times are required this can be done with another resistor and diode.

A current sensing resistor is required but this can be made somewhat lower if the base of the sense transistor is held up to about 0.5V in the off condition. If the 10k feedback resistor is replaced by a pot, then the sensitivity could br set to suit the lamps.

This needs to be simulated and optimised if it is thought suitable.

 

That's an interesting idea duke37.

Maybe I'm over-engineering here, but I have a bit of a list of requirements for the design. It's possible to satisfy them with a discrete solution, but it would be quite difficult.

1. The delay between activation of the indicator and the start of the first flash should be short, for best road safety. But it needs to be long enough to ensure that the reservoir capacitor is fully charged. I was thinking that a micro with two ADC inputs could use one ADC input to monitor the reservoir capacitor voltage. It could also use that voltage to adjust its current detection threshold according to the automotive supply voltage, though that's an unnecessary feature.

2. The load current detection needs to take into account the initial current surge at the start of each flash when the bulbs are cold. This could easily be twice the hot current. With a microcontroller solution, the firmware would sample the MOSFET's source voltage using an ADC channel, shortly before the end of the flash. Alternatively, some kind of smoothing delay would be needed, to ignore the burst of current at the start of each flash.

3. Since you're only measuring the bulb current while the MOSFET is ON, you need some kind of delay to "remember" the slow/fast decision during the OFF period.

I see an electrolytic in your design, from base to ground on the sensing transistor, which probably provides for item 3 above, but I'm not sure how well that design would meet my other two design goals.

Do you think these goals are reasonable? If so, would you like to see whether they can be met with modifications to your design?

 

Likely when a $2 over the counter turn signal flasher unit will produce the functionality discussed...

 

CocaCola's solution seems to be the easiest.

As for Kris's comments
1. A capacitor to supply could be added to turn on the fet as the voltage rises so there would be no turn on delay.
2 and 3. The base capacitor should be able to deal with the initial surge and average during the pulse duration. A diode across the sense resistor would limit the feedback voltage to 0.6V to reduce the effect of the surge.

I am not skilled in running simulation programs, particularly with Schmitt triggers.

I used a 4093 since I wanted one and went to a radio rally and passed over my pennies and, instead of one, I was given a tube containing a couple of dozen.

 

If an off-the-shelf flasher will do everything the OP wants, what's the reason for this thread?!

 

Some people like to reinvent the wheel, or? It was you that first brought up the fact that over the counter flashers do this early on in the thread, don't know why that route wasn't explored further, it's the easiest solution, straight forward and cheap... Flash rate can even be roughly tailored by using additional or less resistance in the circuit...

 

Oh right! I didn't know you could rely on that behaviour. It was just something I'd noticed.