Turn Signal Blinker using MOSFET

[irobot]

I am working with the simple attached circuit diagram used as a vehicle turn signal flasher using LED indicators.
It works, but the LED's do not turn off completely. They switch between full and about half brightness (at a rate of about 1 Hz).
I can't figure out why they are not turning off completely.

I thought MOSFET's had an "off" resistance on the order of 100 or more megohms.

Any ideas why?

[BradC]

The current through the rest of the circuit is enough to light a LED.

[Kim Christensen]

Yup... You've effectively got 1.53K (R2 & R3) in parallel with the FET... Let alone the rest of the circuit.

[Zero999]

Add a resistor in parallel with the LEDs. Try 1k5, which should limit the voltage to just over 6V. If that doesn't work, try a lower value.

[Peabody]

Perhaps the bottoms of C1, C2, R3 and R4 should just be connected to ground.

[Zero999]

Perhaps the bottoms of C1, C2, R3 and R4 should just be connected to ground.

No, that won't work. The circuit is designed so an N-channel MOSFET can be used to switch the high side.

Another option is to use a CMOS oscillator, which draws much less power, making the LEDs very dim, probably unnoticeable under most lighting conditions. R1 & R2 can be increased to 10M and C2 reduced to 47nF for even lower power consumption, but I wouldn't recommend pushing it much further. I've added a zener diode to protect against voltage spikes.

[Shay]

Is there a specific reason for using N channel mosfet as a crude p channel? I'd personally just utilize a P channel/PNP or a npn with a resistor from the collector to VCC, with the output from the collector.

[Zero999]

Is there a specific reason for using N channel mosfet as a crude p channel?

How is it using the N-channel MOSFET as a P-channel? It uses a technique called bootstrapping, which allows an N-channel MOSFET to switch the high side, which is very common for motor drivers. When the MOSFET is off, C1 charges via D1 and the LEDs. When it turns on, D1 prevents C1 from discharging through the MOSFET and powers the oscillator circuit, whilst it's on. The MOSFET's source sits close to +12V. When Q1 turns off, C1 charges again.

I'd personally just utilize a P channel/PNP or a npn with a resistor from the collector to VCC, with the output from the collector.

Please post a schematic. Note the high side needs to be switched, rather than the low side. P-channel/PNP will definitely work. If bootstrapping isn't used, NPN would need to be an emitter follower, which will drop more voltage, but will probably work.

[Kim Christensen]

Probably why when you browse LED turn signal bulb replacements, many also include resistors which you have to wire in parallel with the bulbs. All to keep the stock car flasher and "bulb-out" circuitry functioning properly. Kind of defeats the point of buying aftermarket LEDs since regular turn signal bulbs last a long time anyway.

[irobot]

This is a cheap Chinese flasher I recently bought. It was simple to reverse engineer to draw the schematic.

I forgot to notate that the LED's have built-in current limiting resistors. Not sure what the R value is (for nominal LED current), but they are designed to work with 12V across them for full brightness.

[Zero999]

I plugged the circuit into LTSpice and had a play. By increasing the resistor values and reducing the capacitor values, I managed to keep the flash frequency at 1Hz and reduce the off current to a tiny fraction of the original circuit. If it's not practical to add a parallel resistor to the LEDs, this might be a better option. You should test it using an oscilloscope or frequency counter to ensure the flash rate meets your local regulations. In the UK it's 1Hz to 2Hz and the simulation gives a 1Hz. You might need to tweak the component values to give the correct frequency.


I would still be tempted to use a different circuit, based on a CMOS oscillator.

[james_s]

You need a resistor *across* the LED, not just in series with it. You have to shunt enough current around the LED to power the flasher circuit without making the LED glow.

[irobot]

I plugged the circuit into LTSpice and had a play. By increasing the resistor values and reducing the capacitor values, I managed to keep the flash frequency at 1Hz and reduce the off current to a tiny fraction of the original circuit.

I would still be tempted to use a different circuit, based on a CMOS oscillator.

Thanks for the effort. I'll try it out and provide an update!

UPDATE: I placed 470 ohms across the load LED and bingo! Turns off completely.
I may try a 4011 oscillator and continue to experiment. Thanks

[broly]

isn't this a complicated way of doing it?

i'm looking to do something similar for my HANDA saber

specifically i want to use a single LED array as both a turn signal and the parking light

i will make a new thread because this one seems like it made it harder than it had to be.

[Zero999]

Since the other thread has gone off on a tangent, so I'll post this here.

It's a Schmitt trigger oscillator which demonstrates bootstrapping. Hopefully it's easier to understand than the other circuits. I haven't used a model of a real Schmitt trigger IC, but an ideal one made with SPICE elements, voltage controlled switches and resistors.

The top plots show the voltages relative to the 0V or ground nodes (triangle symbols).

The bottom plots show the voltage relative to the output node, which is also the Vss pin of the Schmitt trigger.