In recent months, there has been a move to integrate capacitive-touch and capacitive-proximity user interfaces (UIs) into lighting applications. The simplicity of the UI, the ability to use irregularly-shaped sensors and the ability to seal the UI is advantageous for creating pleasing and low-maintenance LED interfaces. Unfortunately, differences in technology and techniques between lighting and touch-sensing can create conflicting design goals, particularly in the cost-constrained arena of architectural lighting. Let us see how we can bridge this gap.
Getting started with capacitive touch
Let us start with a basic overview of capacitive touch. A capacitor is essentially two conductors separated by an insulator. Depending upon the type of insulator, the area of conductors and the distance between these, the capacitor will have more or less capacitance. The following equation shows the basic relationship between various factors. C represents capacitance of the capacitor and A represents area of the overlap between the two conductors, the two physical constants ε0 and εR represent permittivity of free space (ε0) and relative permittivity (εR) of the insulating material, and D represents distance between the conductors.
While the space between the plates usually contributes majority of the capacitance, there are also electric field lines connecting opposite sides of the capacitor that contribute to the capacitance of the capacitor. See Fig. 1 for an example plot of the electric field lines for a typical two-plate capacitor. In a capacitive-touch system, it is the electric field lines that project out from the capacitive, rather than the lines between conductors, that touch sensors utilise to detect a touch.
How it works
Human beings, and in fact the majority of carbon based life, are composed of a variety of chemical compounds mixed with a large quantity of water. Water is a fluid composed of electrically-polarised molecules, which means, an electric field can polarise molecules in water very easily. As a result, water based and carbon based life have a very high relative permittivity (εR>60), so there is a very pronounced effect on electric fields. That is what allows the usage of capacitive-touch interfaces (Fig. 2).
Typically, human beings show up as an increase in the capacitance of a sensor. So, all we need for a capacitive-touch or capacitive-proximity interface is a method for measuring the capacitance of the sensor to a sufficient resolution and a conductive sensor pad. (Note: The other conductor of the sensor is typically the electrical ground of the circuit, and the actual conductor is the ground of the circuit. For line-operated systems, this ground is typically earth-ground. So, line-operated systems actually enjoy greater touch sensitivity because we are in close proximity to a lot of earth.)
So, all we need is a means to measure capacitance, right? Well, in a perfect world, yes! Unfortunately, we live in an imperfect and somewhat noisy world, so we actually have to add a few qualifiers to that statement. What we actually need is a method for measuring capacitance that is both low-impedance and has low-susceptibility to noise. The low-impedance part prevents external electric fields (conducted noise) from affecting the capacitance measurement, and the low-susceptibility to noise prevents external RFI (radiated noise) from affecting the capacitance measurement.
Conducted noise. If the capacitance-measurement system has high levels of conducted, common-mode noise on its power supply, it will look like noise is being injected into the touch-sensor. Remember that the circuit cannot tell the difference between it moving up and down electrically, and the sensor being drug up and down electrically. So, conducted noise looks to the circuit like noise on the sensor. By using a low-impedance measurement system, we reduce the effects of conducted noise by dragging water molecules of the user up and down in time with circuit-ground, and limit the effect of the ambient-ground pulling on our sensor through the user.
Removing external noise. To subtract out the external noise, we typically use a differential-measurement method, if possible. While it would be great if we could tap into the user’s ground for this function—it is generally problematic to connect to the user’s ground. So, instead, we try to do two measurements, one with a positive charge on the sensor and one with a negative charge. When we subtract the two, we get an approximation of a differential measurement that is good for most low-frequency noise.
Limiting radiated noise. To limit the effects of radiated noise, we do two things—we limit the amount of time that the sensor is connected to the conversion circuitry, providing a pathway for the noise, and we dither the timing of the sample to prevent a beat frequency between the sample rate and the radiated noise. Actual mechanics of the conversion process also determine the susceptibility of the system to radiated noise, so some conversion methods will be better suited to the rejection of radiated noise.
All these techniques help reduce the amount of noise coupled into capacitance conversion. However, no matter how carefully we make the conversion, some noise will sneak through in the sample. In addition, the amount of shift we see due to a touch, and particularly proximity detection, is quite small.
To handle the noise that does come through and to improve sensitivity, we take multiple samples and average the results together. This increases the amount of shift we see due to a touch, helps average out the noise and actually limits the rate of change in the measured value. After all, the speed of a user’s touch is significantly lower than the typical frequency of noise in the system, so we can live with a system that has a slower response time if it helps to cancel out the noise.
Another useful function is to incorporate a slew-rate limiter on the data. Basically, this function looks at each new sample; if the sample is above average, the average is increased by 1-5, and if it is less, the average is decreased by a similar amount. This prevents large noise spikes from dragging our average up and down, while still passing slower changes in the samples.
Together, these functions allow a capacitive-touch system to operate, even in noisy environments. It turns out, that is exactly the environment in which lighting systems have to be able to survive. After all, lighting systems typically share power with some prolific noise sources, including HVACs, computer systems, inductive loads (motors and pumps) and other lighting systems, which all put out plenty of conducted noise. On top of all this, we also live in a wireless world, which includes mobile phones, Wi-Fi and broadcast radios/televisions.
So, any lighting system that wants to include capacitive touch, and especially proximity interfaces, will require capacitive systems that can operate in the presence of both radiated and conducted noise sources. Fortunately, most capacitive-touch systems currently on the market can tolerate noise levels that are typically encountered in home and office environments. As designers, we just have to make sure we verify that the pre-fabricated capacitive-touch system is rated for the level of noise we are likely to encounter.
Implementing this into the interface
Now that we have a noise-resistant capacitive touch/proximity system, how can we use it for our interface? Well, in the simplest system, we just have to turn the lighting on and off. However, most higher-end systems are also going to need dimming, so we really need an interface with some level of gesture-recognition. Further, because handing out a users’ manual with a light switch is impractical, gestures used will have to be intuitive for the user. Finally, whatever system we use, it has to be reasonably immune to false triggering.
The main requirements are:
1. Simple and intuitive method for turning lights on and off
2. Simple and intuitive method for dimming the lights
3. Reasonable immunity to false triggering
4. Minimal power consumption when not in use
5. Low material cost
From requirements one and two, we know we will probably need some kind of legend explaining the operation of the interface. Given that we are also talking about lighting control, our legend will either have to glow in the dark or have some kind of tactile information. From requirement three, we know we cannot just toggle the light on and off when a user passes in proximity to the switch.
Requirements four and five indicate that we will have to limit both power and cost. Fortunately, most modern microcontrollers can operate on very little power. These also have all necessary peripheral functions required for both lighting control and capacitive touch.
Therefore, given all these requirements, a reasonable interface would use capacitive proximity to turn on the backlighting for the button legends and a basic buttons-and-slider interface for turning the lights on and off, and dimming. All necessary functions to implement the design should be available in a variety of single-chip microcontrollers, and the necessary capacitive UI software should be readily available from the microcontroller manufacturer.
The proximity sensor on the button legends allows the user to find the switch and turn on the legends, even in the dark. The lighted legends provide him or her with basic instructions for the use of the system. If we also add a software lockout on the button-and-slider controls, such that these cannot be activated within two to three seconds of initial proximity detection, this should prevent the accidental setting change if the user brushes past the interface.
We also minimised power and cost, because we can use a less expensive power supply to only supply the energy for backlighting the legend for the 10-20 seconds the user will need to set the lighting level. Our interface also minimises cost because the capacitive proximity and touch-sensor design can be implemented using a low-cost printed-film sensor.
Now, some may ask, why not make the complete interface capacitive-proximity based? For example:
1. If the user moves his or her hand from left to right, this could turn the light on, and right to left to turn it off.
2. Moving the hand up/down past the sensor could control dimming of the light.
While this is possible with the technology currently available, the question becomes how does it affect user experience? How will the user find the interface in the dark without potentially turning the lights on at full brightness? What happens if the user brushes past the sensor and turns off the lights? Or, what happens when the family dog brushes its tail past the sensor?
While these may sound somewhat contrived, the designer does have to keep in mind that capacitive touch is susceptible to this kind of environmental noise, and should consider how sensitivity of a capacitive-proximity system can affect operation. It is also possible to handle problems in this kind of system using a more robust gesture-recognition system. However, processing requirements for implementing even a simple two-to-three gesture pattern-recognition system are typically beyond the capability of small, low-cost microcontrollers. So there is a cost trade-off to be considered.
Some things to keep in mind
Capacitive proximity and capacitive touch are exciting new technologies, but the designer must remember that these bring not only new freedom in design but also new challenges that have to be considered when designing a UI. Noise-susceptibility, both electrical and environmental, as well as complexity of the UI experience, has to be considered in design. Remember, it is not just a case of substituting one switch for another. Instead, it is a completely new technology with its own set of advantages and challenges. After all, the novelty of a new interface fades quickly if it is difficult to use and is susceptible to new factors from which the previous system was immune.