Converting a 12v LED flasher circuit to work on 6v

[canadaboy25]

I have a vintage motorcycle that I am working on converting to LED lighting.  The bike runs a 6V electrical system.  The original incandescent bulbs pull a large amount of current and it is just too much for the 50+ year old wiring and switches.  It is a fairly rare bike, so no 12V conversion kits are available.  The bulbs are standard so I was able to purchase 6V LED replacements.  The LED taillight/brake light works well.  The problem is with the signals.

The original 6V signal flasher is just a bi-metallic strip that heats up due to the current flowing to the lamps, which breaks the circuit, cools off, and reconnects the circuit.  This obviously does not work with LED lamps.  I have tried buying 6V "LED" flashers, but they either end up being for 6V incandescents, or for 12V LED bulbs.

I bought a pack of 12V LED flashers and reversed the schematic from them.  They work for 12V LED bulbs, but do not work for my 6V LEDs.  I would like to try and modify one to work on 6V.

I've attached a schematic of both the wiring on the bike, and the circuit inside the 12V LED flasher to be converted.

You can see on the bike, the flasher is connected to +6V at all times, and grounds through either the left or right signal depending on which way the turn signal switch is pressed.  The bulb marked "indicator" is inside the speedometer and originally worked since it was a much lower wattage bulb than the actual signals, and grounded through the bulbs which were not active, causing the indicator to light, but not the actual signal.  I am aware that this indicator will not work with the LED replacements.

The flasher has two pins labelled "B" and "L".  Q1 is a UMW 2300A, and Q2 is a 2SA812 (Datasheets attached).  VR1 is a pot on the bottom of the flasher that can be used to adjust the flashing frequency.

I've drawn the circuit in LTSpice and it works as expected at 12V.  However, dropping the source voltage down to 6V causes it to stop working completely.  I've fiddled around with the component values but nothing I've done has allowed it to work at 6V.

Can anyone give me some ideas on what to change to get this circuit working on 6V?

[ArdWar]

Looks like a relaxation oscillator that waits for Q2 emitter (C1) to charge until it gets one V_BE above R2-R4 voltage divider.

The problem is that with 6V the maximum settings for VR1 only allows for 6V * 2k / (2k + 10k) = 1V, just barely above V_BE, and you know supply isn't going to be 6V since LEDs do drop some voltage even at very low forward current.

So, try either reduce R4 and/or increase R2.

[aliarifat794]

I think you can add a buck converter to the circuit. That will step down from 12V to 6V.

[Zero999]

I think you can add a buck converter to the circuit. That will step down from 12V to 6V.

I thing you misunderstood. The idea is to use a 12V flasher, with a 6V system.

I have a vintage motorcycle that I am working on converting to LED lighting.  The bike runs a 6V electrical system.  The original incandescent bulbs pull a large amount of current and it is just too much for the 50+ year old wiring and switches.  It is a fairly rare bike, so no 12V conversion kits are available.  The bulbs are standard so I was able to purchase 6V LED replacements.  The LED taillight/brake light works well.  The problem is with the signals.

The original 6V signal flasher is just a bi-metallic strip that heats up due to the current flowing to the lamps, which breaks the circuit, cools off, and reconnects the circuit.  This obviously does not work with LED lamps.  I have tried buying 6V "LED" flashers, but they either end up being for 6V incandescents, or for 12V LED bulbs.

I bought a pack of 12V LED flashers and reversed the schematic from them.  They work for 12V LED bulbs, but do not work for my 6V LEDs.  I would like to try and modify one to work on 6V.

I've attached a schematic of both the wiring on the bike, and the circuit inside the 12V LED flasher to be converted.

You can see on the bike, the flasher is connected to +6V at all times, and grounds through either the left or right signal depending on which way the turn signal switch is pressed.  The bulb marked "indicator" is inside the speedometer and originally worked since it was a much lower wattage bulb than the actual signals, and grounded through the bulbs which were not active, causing the indicator to light, but not the actual signal.  I am aware that this indicator will not work with the LED replacements.

The flasher has two pins labelled "B" and "L".  Q1 is a UMW 2300A, and Q2 is a 2SA812 (Datasheets attached).  VR1 is a pot on the bottom of the flasher that can be used to adjust the flashing frequency.

I've drawn the circuit in LTSpice and it works as expected at 12V.  However, dropping the source voltage down to 6V causes it to stop working completely.  I've fiddled around with the component values but nothing I've done has allowed it to work at 6V.

Can anyone give me some ideas on what to change to get this circuit working on 6V?

It appears to be marginal design for 12V. Q1 only has a maximum V_GS of 12V, which would be exceeded on a 12V automotive system.

I've put this circuit into LTSpice.

The LED will need a resistor in parallel to work, otherwise the forward voltage drop will be too much.

I had to increase the capacior values in order to get the frequency down to something acceptable. I don't know about Canada, but in the UK, the indicator signal lamps need to flash between 1Hz and 2Hz. The duty cycle is also quite high, which I don't know is a problem or not.

I would go for a 555 (the CMOS TS555, or 7555 would be better) timer oscillator, with the same bootstrapping configuration, which will be more reliable.

[sy]

Not sure if this is a thing but I wonder if the capacitor C1 is not fully discharging after the mosfet is switched on?

So it kinda just locks itself in a loop and there is no charging and discharging as intended?

But if this is the case, not sure why it works with 12V 

[canadaboy25]

Thanks to everyone for all the helpful responses.  I've tried out your suggestions and have some results below.

Looks like a relaxation oscillator that waits for Q2 emitter (C1) to charge until it gets one V_BE above R2-R4 voltage divider.

The problem is that with 6V the maximum settings for VR1 only allows for 6V * 2k / (2k + 10k) = 1V, just barely above V_BE, and you know supply isn't going to be 6V since LEDs do drop some voltage even at very low forward current.

So, try either reduce R4 and/or increase R2.

That analysis makes sense and has helped me understand how this is working a bit better.

With everything connected, I am only getting 2.6V across the flasher.  The emitter of Q2 is at 2.1V, and the base is at 1.7V.  The collector is at 0V, so the MOSFET is not being turned on.  The LED is glowing very dimly whenever it is connected.  The Q2 base voltage divider must be passing enough current to dimly light the LED.  Reducing R4 causes the LED to glow brighter, but still not flash.

The LED will need a resistor in parallel to work, otherwise the forward voltage drop will be too much.

I had to increase the capacior values in order to get the frequency down to something acceptable. I don't know about Canada, but in the UK, the indicator signal lamps need to flash between 1Hz and 2Hz. The duty cycle is also quite high, which I don't know is a problem or not.

The parallel resistor seems to be the ticket.  Placing a 100 ohm resistor across the LED causes the circuit to start working at 6V.  I need to have the pot adjusted very close to the point that it is not in circuit, but modifying the values of the resistors ArdWar mentioned would likely fix this.

The duty cycle is indeed quite skewed from 50%.  The LED is off for most of the time with just short pulses of being on.  If I could get the circuit working in LTSpice, I could likely find a way to adjust that.

Not sure if this is a thing but I wonder if the capacitor C1 is not fully discharging after the mosfet is switched on?

So it kinda just locks itself in a loop and there is no charging and discharging as intended?

But if this is the case, not sure why it works with 12V 

This kind of seems to be the behavior I am seeing in LTSpice at 6V.  The simulation does one flash, and then gets stuck in a state where the FET is partially on, sinking a bit of current through the LED and not flashing.  I have played a bunch of component values and nothing seems to get it working in the simulation.  Perhaps it is because I do not have models for the actual transistors I am using, and that is throwing the simulation off.


I will try some higher values for the parallel resistor and see at what point the circuit stops working.  I would like to not have to go as low as 100 ohms.

Does anyone have suggestions on how to get my Spice simulation working at 6V to more closely reflect the actual circuit?

[Zero999]

I've simulated it with LTSpice. I changed some of the capacitor values to reduce the frequency to something sensible. R4 is the potentiometer and fixed resistor in series. Here are the files.

[canadaboy25]

Thank you to everyone for the helpful tips.  I was able to get my spice simulation working.  Turns out I had modified some component values and forgot to set them back.     Reducing the value of R4 got the circuit running.  However, it still does not replicate the physical circuit very accurately.  I guess this is not surprising since I don't have accurate models for the LEDs or transistors.  It was good enough to get an idea of what was going on though.

I changed some of the capacitor values to reduce the frequency to something sensible.

I never had an issue with the frequency being off in either the physical or simulated circuit.  The duty cycle was very low but the frequency was around 2Hz.  It looks like in your circuit the values for the emitter and collector resistors are swapped with the values for the base resistors.  That's probably why the frequency was off.


I was able to increase the value of the parallel resistance across the LEDs to 4.7K which makes the circuit work.  My understanding is this allows enough current to flow to make the circuit function since the LEDs are basically an open circuit when turned off.

I then had to decrease the value of R4 as suggested as the pot was at the bottom limit of its range.  Tacking another 10K across R4 gives me a good range with the pot.

The duty cycle was still very low.  Adding 22uF across C2 gives very close to 50% duty cycle.


I would like to get the indicator on the tachometer working if possible.  It will stay an incandescent bulb since there are no LED replacements in that size, and it needs to be non-polarized based on how the bike is wired.  I've attached a diagram of how I think I can make the indicator work.  My idea is to use a transistor to short out the unused signal when the other one is active to give the indicator a path to ground.  For example, when the left signal is active, the top MOSFET will be on, shorting the right signal to ground and allowing the indicator to light without letting the right signals light.
(I realize my usage of the words "indicator" and "signal" might be confusing to those from the UK.  When I say "signal" I'm referring to the actual directionals that other motorists see, and when I say "indicator" I'm referring to the little light on the tachometer that lets me know the directionals are active.  Sorry about that)

Does anyone see any issues with this solution?

[Zero999]

Thank you to everyone for the helpful tips.  I was able to get my spice simulation working.  Turns out I had modified some component values and forgot to set them back.     Reducing the value of R4 got the circuit running.  However, it still does not replicate the physical circuit very accurately.  I guess this is not surprising since I don't have accurate models for the LEDs or transistors.  It was good enough to get an idea of what was going on though.

I changed some of the capacitor values to reduce the frequency to something sensible.

I never had an issue with the frequency being off in either the physical or simulated circuit.  The duty cycle was very low but the frequency was around 2Hz.  It looks like in your circuit the values for the emitter and collector resistors are swapped with the values for the base resistors.  That's probably why the frequency was off.


I was able to increase the value of the parallel resistance across the LEDs to 4.7K which makes the circuit work.  My understanding is this allows enough current to flow to make the circuit function since the LEDs are basically an open circuit when turned off.

I then had to decrease the value of R4 as suggested as the pot was at the bottom limit of its range.  Tacking another 10K across R4 gives me a good range with the pot.

The duty cycle was still very low.  Adding 22uF across C2 gives very close to 50% duty cycle.

Oh yes. It was a drawing error. Increasing the value of C2 sorts the duty cycle out. SPICE isn't perfect. It's just performing calculations.

 Bootstrap astable.asc (1.92 kB - downloaded 13 times.)

I would like to get the indicator on the tachometer working if possible.  It will stay an incandescent bulb since there are no LED replacements in that size, and it needs to be non-polarized based on how the bike is wired.

Presumably it needs to be small. How about replacing the lamp holder with an LED and appropriate resistor?

I've attached a diagram of how I think I can make the indicator work.  My idea is to use a transistor to short out the unused signal when the other one is active to give the indicator a path to ground.  For example, when the left signal is active, the top MOSFET will be on, shorting the right signal to ground and allowing the indicator to light without letting the right signals light.
(I realize my usage of the words "indicator" and "signal" might be confusing to those from the UK.  When I say "signal" I'm referring to the actual directionals that other motorists see, and when I say "indicator" I'm referring to the little light on the tachometer that lets me know the directionals are active.  Sorry about that)

Does anyone see any issues with this solution?

Wouldn't it be easier to change the wiring so the indicator lamp is connected to both signals via diodes and 0V?

[canadaboy25]

Presumably it needs to be small. How about replacing the lamp holder with an LED and appropriate resistor?

Yes, a very small bulb.  I have considered this, but the lamp holder is molded into a rubber seal to keep moisture out of the tach.  I'd like to not modify the bulb holder on the bike.  I could likely smash the glass out of a bulb, and install some LEDs in the base of the old lamp.  Not trivial, but not impossible either.

Wouldn't it be easier to change the wiring so the indicator lamp is connected to both signals via diodes and 0V?

Yes, that would be easier and likely more reliable too 

I found a B16100 schottky rectifier in  my parts bin which is 2 schottky diodes in a TO220 package which is perfect.  Low voltage drop for the lamp and a nice rugged package.  I was able to solder wires to the diode, and crimp bullet connectors on.  This was then able to be plugged into the factory wiring harness with no modification.

The bike has a front and rear signal, so there are 2 LEDs active at once.  The extra load from the second LED was enough to make the flasher work without the parallel resistor across the load.  The extra load from the incandescent indicator threw the duty cycle off, so I had to reduce the 22uF cap I added to a 10uF.

The signal lights are now working perfectly with a 50% duty cycle, and adjustable frequency with the pot on the flasher.  Couldn't be happier with it.

Thanks for the help!

[Zero999]

Good. I'm glad you got it working. Thanks for posting an update.

EDIT:
It's probably better to keep the incandescent lamp for the indicator for now. If you need to replace it, then you can for for an LED, which can be potted in translucent silicone.