My daughter asked for help creating a flickering candle effect for some night lighting in her new home. Battery operation is required, so LEDs are the only practical choice. There are various DIY approaches you can find with a quick search, but none suited me so I've been experimenting a bit and thought I would share my findings before the holidays arrive.
Many of the projects you find online require a microprocessor, and I didn't want to bother with that learning curve, especially when the candle flickering "program" is already available. Instead I chose to use the guts from one of those plastic, LED tea-light candle simulators.
They're less than 50¢ and use 3 button cells to power a single yellow LED. They have a chip (glob on PCB) that controls the LED and it's flickering resembles candle light. I gather there are similar lights where the circuit is built into the LED itself, but in my case the LED is separate from the controller. The LED and the controller both connect to the high side of the power supply. The controller also connects to ground and a third wire connects to the low side of the LED.
So the goal of my experiments was to use that pre-made controller, meant for one LED, to control a larger array of LEDs.
After a bit of playing around, I found that the glob draws only about 100µA when not loaded at the output, and that it needs a minimum of ~3V. That means you can power it from anything more than say 6V by using a simple resistor divider. The resistors can be in the 10K range, since 100µA will drop only 1V across a 10K. At times I powered the glob up to 7-8V and it survived, but I don't know what its limits are.
The simplest approach to scaling up the output signal is to route the signal directly to the gate of a MOSFET. A logic-level FET would be better at low voltages, but this works well as shown.
This inverts the flashing sequence but it's hard to see much difference from the original, positive signal. The most pronounced difference is that the original signal has a brief flash, that appears to be full on, once in the sequence. The inverse signal goes dark for that moment and that feels a bit unrealistic.
So, I added a small transistor to restore the original signal.
This works very nicely in recreating the original flickering effect, but with as many LEDs as you might want. With a bit of playing around with the resistors, I found I could accentuate the flickering by reducing the average voltage at the gate of the MOSFET, so that it is off more often. This seemed a bit more realistic than the original effect of the tea-light, more like a real candle.
To increase the realism of simulating a candle or gaslight in a lantern, I considered using two glob controllers so they could flicker out-of-sync. (I've since changed my mind in favor of a single control). Before I realized how simple the transistor circuit above could be - even when doubled, I built a circuit using a dual op-amp, using one amp for each controller. This accomplished the inversion of the signal to maintain the logic, but also gave a way to boost a low voltage control signal up to a higher voltage at the MOSFET gate. Only later did I find this unnecessary in achieving the effect.
Many of the projects you find online require a microprocessor, and I didn't want to bother with that learning curve, especially when the candle flickering "program" is already available. Instead I chose to use the guts from one of those plastic, LED tea-light candle simulators.
They're less than 50¢ and use 3 button cells to power a single yellow LED. They have a chip (glob on PCB) that controls the LED and it's flickering resembles candle light. I gather there are similar lights where the circuit is built into the LED itself, but in my case the LED is separate from the controller. The LED and the controller both connect to the high side of the power supply. The controller also connects to ground and a third wire connects to the low side of the LED.
So the goal of my experiments was to use that pre-made controller, meant for one LED, to control a larger array of LEDs.
After a bit of playing around, I found that the glob draws only about 100µA when not loaded at the output, and that it needs a minimum of ~3V. That means you can power it from anything more than say 6V by using a simple resistor divider. The resistors can be in the 10K range, since 100µA will drop only 1V across a 10K. At times I powered the glob up to 7-8V and it survived, but I don't know what its limits are.
The simplest approach to scaling up the output signal is to route the signal directly to the gate of a MOSFET. A logic-level FET would be better at low voltages, but this works well as shown.
This inverts the flashing sequence but it's hard to see much difference from the original, positive signal. The most pronounced difference is that the original signal has a brief flash, that appears to be full on, once in the sequence. The inverse signal goes dark for that moment and that feels a bit unrealistic.
So, I added a small transistor to restore the original signal.
This works very nicely in recreating the original flickering effect, but with as many LEDs as you might want. With a bit of playing around with the resistors, I found I could accentuate the flickering by reducing the average voltage at the gate of the MOSFET, so that it is off more often. This seemed a bit more realistic than the original effect of the tea-light, more like a real candle.
To increase the realism of simulating a candle or gaslight in a lantern, I considered using two glob controllers so they could flicker out-of-sync. (I've since changed my mind in favor of a single control). Before I realized how simple the transistor circuit above could be - even when doubled, I built a circuit using a dual op-amp, using one amp for each controller. This accomplished the inversion of the signal to maintain the logic, but also gave a way to boost a low voltage control signal up to a higher voltage at the MOSFET gate. Only later did I find this unnecessary in achieving the effect.
Attachments
-
14.3 KB Views: 1,135
-
26.7 KB Views: 1,117
-
44.4 KB Views: 1,123