Magic colour changing mosaic lamp

Introduction 

This is a project that I worked out together with my wife Mieke, who loves working with glass mosaic.

I love working with LED lighting projects, so I thought: let's combine the 2 hobbies and try to make a kind of art-project combining electronics and glass mosaic.

Combining RGB-LEDs with transparent glass mosaic gives very nice lighting effects.

Working with glass is a challenge, because it is not easy to make holes in glass. So I needed to find a way to power the electronics inside the glass work without making a hole for a power supply connector or to feed through a power cable. Inside the glass work is a microcontroller that is used to slowly change the color using an RGB-LED. The color changing is very slow, so you don't see it changing when only watching a few seconds. The microcontroller and LED don't need much power, so I didn't need a very powerful wireless power transmitter. Furthermore, the power did not have to be transferred over a great distance, but only over a few centimeters = thickness of the base cover + the thickness of the glass work.

Besides that, I didn't want to make a hole in the base for a power switch, to switch the wireless power transmitter on or off. So I decided to use a capacitive touch/proximity sensor for this purpose, because this type of sensor works through glass and when the capacitive sensor area is made big enough, it can work over distances of multiple centimeters. The sensor has to be temperature stable, very sensitive and using a small amount of power, since it is powered continuously.

The result is a nice looking color changing lamp using wireless power transfer and that is being switched on/off using a capacitive proximity/touch-sensitive toggle switch.

Here are some pictures of the finished project :

System diagram

Circuits

Capacitive touch/proximity sensor with toggle switch (temperature stable, low power and very sensitive)

The capacitive sensor circuit was published in Elektor Nr. 537/538 July/Aug. 2008.

It is a very stable and sensitive capacitive sensor, that can sense capacitance differences down to picofarads.

The circuit has virtually no problem with temperature changes, because the capacitance is measured using a differential approach, meaning that 2 signals are compared with each other. Both signals have the same temperature response, so this effect is cancelled out.

The relaxation oscillator build around U2B has a 50% duty cycle square wave output with a frequency of about 100 kHz. The output signal is fed to two RC networks, that both integrate the square wave signal. One RC network is fixed and formed by R2 and C1. The other RC network is formed by R4 and the capacitive sensor, that is made out of a copper clad PCB. The time constant of this last RC network will change with the capacitance between the copper clad PCB and earth. When capacitance is added, f.e. by moving your hand close to the copper clad PCB, the charge time of the RC network will increase, causing a delay, compared with charge time of the fixed RC network (R2/C1). The difference in charge times (thus delay) will be proportional to the extra amount of capacitance "felt" by the copper clad PCB. The compare operation, to measure this delay, can be done with an XOR. By feeding both signals to the 2 inputs of the XOR, the output of the XOR will show pulses with a width corresponding to the delay between the two RC networks. But the XOR will output pulses when the time constant of the capacitive sensor RC network is higher, but also when it is lower than the time constant of the fixed RC network.

When we use a D-flip flop (U3A) to measure the delay, it will only output pulses when the clock input signal (CLK) is delayed, compared with the data input signal (D). When the D signal is delayed towards the CLK signal, the output will be 0.

R3 and C2 form an integrator with a very high time constant compared with the pulse width of the output pulses of U3A. So the voltage over C2 will be proportional with the pulse width of the output pulses. This means that the voltage will be proportional to the capacitance difference or, in other words, the added capacitance.

U3B will toggle each time the voltage over C2 reaches the threshold voltage of the CLK input of U3B, which is about 1.5V. MOSFET Q1 is used as a switch that will toggle each time when the capacitive sensor sees enough extra capacitance. The MOSFET switches the +15V to the next stage, which is the buck converter for the wireless power transmitter.

R2 has to be adjusted so the U2A does not output pulses when the capacitive sensor does not "feel" any extra capacitance.

The capacitive sensor is very sensitive and will react to any change in capacitance. So when you connect the ground-clip of your scope-probe to the circuit, the total capacitance "felt" by the sensor is changed. The same thing when you connect the circuit to a lab power supply that is earthed. This will also change the total capacitance "felt" by the sensor.

So it is important to adjust R2 in the situation in which the capacitive sensor will be used without connection to a scope. Best is to put a LED at the output of U3A, so you have a visible indication of the capacitance.

Once R2 is adjusted, and you move the sensor over to another table made out of other material or that is higher, lower, thicker or closer to a wall, the capacitance will change, and you have to re-adjust R2. Also, when you re-route the power cable coming from the adapter to power the lamp, this might influence the capacitance, especially when routing it close to a wall or metallic objects.

Oscilloscope screen of the output of the oscillator U2B at pin4 :

Click here to download the LTSpice simulation file for the capacitive sensor

Buck regulator to power the wireless power transfer transmitter

The buck converter is used to bring the power supply down to about 5V for the wireless power transmitter. With 5V, the transmitter delivered enough power for my purpose.

The microcontroller that controls the RGB-LED and the RGB-LED itself do not need a lot of power. Furthermore, the distance between transmitter and receiver is maximum a few centimeters, so I did not need to burn a lot of power just to overcome the distance. This also keeps the temperature of the transmitter components cool, because the whole thing is assembled in a completely closed enclosure without any air vents.

The buck converter is nothing special. See application note for the MC34063. I added an extra buffer (Q1, Q2) and switching MOSFET, because I noticed that the chip got pretty warm

when pulling 1A continuously. So i over-dimensioned the switching stage to keep the temperature of the chip down. Probably it was not a problem at all, but I prefer to keep things safe,

especially because the magic lamp is connected to the power supply day and night, and sometimes one of our cats adds enough capacitance to switch on the magic lamp on.

Use a fast recovery Schottky diode for D1 and use a low ESR capacitor for C1. Also check that L1 can handle the current that you want to draw from the converter. The efficiency of the converter largely depends on the quality of D1, L1 and C1, because these components are responsible for storing and delivery of the power ...

Wireless power transfer transmitter based on Mazilli ZVS oscillator

See Wireless power transfer for the explanation around the Mazilli oscillator and the start-help for the oscillator.

Oscilloscope screen of the voltage at one of the MOSFET gates : 

Oscilloscope screen of the voltage between the 2 drains of the MOSFETS :

Click here to download the LTSpice simulation file for the Mazilli oscillator

Wireless power transfer receiver

See: Wireless power transfer for the explanation of the wireless power receiver.

RGB LED controller with PIC12F683

The RGB-LED controller is build around a Microchip PIC12F683.

​The PIC12F683 uses a semi-random generator to change the color of the RGB-LED randomly and very slowly, so you don't really notice the change when you look for a few seconds.

​The firmware controls the RGB-LED using a software PWM of 3 digital output pins (red, green and blue) ...

​The PWM value is compensated with an exponential correction, so the LED brightness perception appears to change linearly with smooth color changes ...

Click here to download the PIC12F683 slow colour changer firmware

Power delivery of the wireless power system

I could draw a maximum of 250mA from the wireless receiver before the ultra low drop regulator dropped out, with the wireless receiver coil a few centimeters straight above the wireless transmitter coil.

Maybe with precise tuning of the resonance frequencies of the transmitter and receiver LC networks, the power delivery can be optimized and increased. But for this project, the power delivery is more than sufficient.

Total power consumption of the magic lamp

The total current consumption of the magic lamp is about 150 mA at 15V with the wireless power transmitter switched on and the RGB-LED lit up.

When removing the wireless power receiver, the current consumption is about 94 mA, so the receiver consumes about 60 mA at 5V for the RGB-LED and the PIC12F683.

When the transmitter is switched off, only the touch sensor is connected to the power supply and draws about 5 uA, so virtually nothing.

Pictures

Videos

Demonstration of the wireless power transfer and capacitive proximity/touch sensor