In triac dimming, a control circuit in the light switch provides power to the lamps AC mains input for short periods of time each mains cycle (Fig. 16.1). Each half cycle has a period of 10 ms (for 50-Hz AC mains supply) and, in the example shown, the triac turns on after about 6 ms and off at the end of the half cycle. Thus the lamp is off at other times and so produces light with 100% modulation at a repetition frequency of 100 or 120 Hz. Old-style incandescent lamps produce light by heating a tungsten filament, so the 100% electrical modulation gives much less light modulation; this because the filament remains hot for some time and emits light until it cools down. However, unless an LED lamp has a large-storage capacitor that keeps the current flowing through the LED P–N junction, the LEDs turn off completely during the triac off periods.

The advantage of using a triac is that there is no additional wiring besides the AC mains supply, so triac dimming can be added to existing circuits with very little effort. The disadvantage is that the dimming range is limited and all lamps on the same circuit are dimmed together. It is still commonly found in residential lighting circuits.

The 1–10 V dimming system has an advantage in that the AC mains can be permanently wired to all lamps in a building without any concern regarding the control. The control wires are connected separately, as shown in Fig. 16.2.

But there are disadvantages because it uses single-wire point-to-point control, with a common ground return. This means that if lamps need to be controlled separately, control cables with multiple wires must be installed, one for each lamp. If the control needs changing, such as if a partition wall has been installed in an office, the control system needs to be rewired. Single wire control also means that it is susceptible to interference, so in electrically noisy industrial environments, the lamps could flicker when high-power equipment is used.

One method of interfacing a 1–10 V control system to an LED lamp driver is to use a ramp generator and a comparator, to create a PWM signal for controlling the light level. When the ramp voltage exceeds the 1–10 V signal, the comparator outputs a high-level output. This is shown in Fig. 16.3.

As the control bus is a balanced pair of wires and Manchester encoding is used for the signals, it is very reliable and not susceptible to external interference.

The DALI protocol uses logarithmic dimming, to match the sensitivity of the eye. Switching to half brightness means that the lamp actually appears half as bright to the eye and not that it produces half the amount of lumens. In fact, there are 256 steps of brightness, with the lowest level being 0% (off) and the highest level being 100% (full power). Due to the stepped levels, DALI is not suitable for theatre lighting or other lighting where the light level must increase smoothly in low-ambient light conditions. Theatre lighting needs a dimming range greater than 10,000:1.

Data is transmitted over a twisted pair cable using EIA-485 (RS-485) differential signaling. For speed reasons, there is no error checking and correction of the transmitted data, but because it uses a very robust medium, which is not affected by external interference, errors are rare. Data lines must be terminated with 120 Ω, which is equal to the cable’s characteristic impedance, to prevent signal reflections that would otherwise occur in an unterminated transmission line.

RS-485 allows bidirectional half-duplex operation, with one driver connected at each end of the cable pair. Driver and receiver pairs can be located at various points along the bus, but only one driver can be active at any one time. The unused driver is put into the high-impedance state, so that it does not affect the data signals transmitted from the other end of the link.

The RS-485 bus should be one continuous pair with a 120-Ω terminating resistor at either end. An alternative “fail-safe” bus termination has a 130-Ω terminating resistor, a 750-Ω “pull-up” resistor from the A-wire to +V_s, and a 750-Ω “pull-down” resistor from the B-wire to ground. This ensures that the receiver is never in an indeterminate state; this fail-safe termination is shown in Fig. 16.5. Note that some RS-485 transceivers have a fail-safe mechanism built-in and therefore do not need the external bias circuit.

The RS-485 bus is a transmission line, so correct termination is important to prevent signal reflections. Spurs off the bus should not be allowed, unless the system operates at low speed. This is because reflections from an unterminated spur will affect the data pulse shape and cause errors. The RS-485 bus can be up to 1250-m long, at data rates of up to 100 kbps. At higher-data rates, the maximum line length is reduced. For example, at 10 Mbps, the maximum line length is about 30 m.

The RS-485 logic and voltage levels are +2 V to +6 V (Space) and —2 V to —6 V (Mark). Drivers are commonly powered from either ±3.3 V or ±5 V; for example, Intersil’s ISL32458E can be powered from either of these voltages; it is very robust against 20-V common mode voltages and overvoltage conditions on the bus, with 60-V continuous voltage protection.

Like DALI, DMX also has 256 levels. However, two channels can be used together, so that one channel is used for coarse control and the other channel can be used for fine control. Thus the dimming range with two control channels is 65,536 levels.

The bus itself is single wire (LIN bus), with a ground return (GND) and a power rail (V_Bat). There is one master node and up to 15 slave nodes. The LIN bus voltage is in the range 9–16 V (the automotive battery voltage range). As the LIN bus is not a balanced transmission line, it has limited range and speed capability. The data rate is up to 20 kbps, transmitted over a distance of 40 m maximum. The master is identified by having a 1-kΩ termination on the LIN bus, whereas the slaves have a 30-kΩ termination. In practice, transceiver ICs have an internal 30-kΩ termination, so the slave nodes do not need any further termination on the LIN bus and only the master needs an addition 1-kΩ external termination. This is shown in Fig. 16.6.

There are many suppliers of LIN bus driver ICs. Examples are: ON-Semiconductor with NCV7382, Texas Instruments with TPIC1021A, and Microchip with MCP2021A. These all have a sleep mode, where the quiescent current is just a few microamperes, and slope control to limit the rise and fall time of the data pulses, to keep the EMI very low. The MCP2021A has the useful feature of an internal low drop-out regulator, to provide up to 70 mA for a microcontroller from the V_Bat supply.

Note that each supplier uses slightly different IC pin descriptions. Microchip use V_BB for the V_Bat supply pin, LBUS for the LIN bus pin, and V_SS for the ground (GND) pin. Texas Instruments use V_sup for the V_Bat supply pin, LIN for the LIN bus pin, and GND for the ground pin. ON-Semiconductor uses V_s for the V_Bat pin, BUS for the LIN bus pin, and GND for the ground pin. Fortunately, all three suppliers use the same RXD for the receive data pin and TXD for the transmit data pin. Other suppliers may have slightly different names.

As there is a common bus, with the voltage pulled low during signaling, it is possible to have a transmission clash. This is where two slaves attempt transmission at the same time. A clash is detected because a checksum is transmitted at the end of each message; if the checksum does not agree with the data received, the system (via the master node) asks each slave node to respond separately. Thus a clash causes several messages to be transmitted, with the consequential delay. As the LIN bus is not very fast anyway, it cannot be used for critical applications requiring high-speed control.

The most common use for the LIN bus is mechatronics, so in a vehicle this could include seat control, window control, etc. Another low-speed application could be turning on lights and adjusting their brightness.

CAN bus drivers are powered from a 5-V supply. The maximum differential voltage across the bus is 3 V. If the differential voltage is greater than 1 V, it is said to be “dominant” and if it is less than 0.5 V it is said to be “recessive.” The corresponding input and output data lines to the microcontroller are logic 0 (dominant) and logic 1 (recessive). The terms “dominant” and “recessive” will be described later. There are a number of manufacturers of CAN bus driver ICs, including Microchip (MCP2561), NXP (TJA1050), Maxim (MAX3058), Texas Instruments (SN65HVD232D), and many others. For those working on space applications, Intersil make radiation-hardened CAN bus drivers!

The CAN bus driver is connected to the bus, as shown in Fig. 16.7. Note, that the bus itself has 120-Ω line terminations, reducing reflections on the transmission line and thus helping to maintain speed.

Nodes are not allowed to transmit until there has been a period of inactivity on the bus. Once a quiet period has passed, every node has an equal opportunity to transmit. This is carrier sense multiple access. When a node is transmitting a logic 1, it would expect to receive a logic 1, as the same pair of wires are used for send and receive. However, if it transmits a logic 1 and receives a logic 0, it knows that its transmission has been corrupted by another node transmitting at the same time. This is collision detection. Collision resolution is when the importance of the message is used to decide which node transmits first.

Earlier I mentioned the terms “dominant” and “recessive” are given in the CAN specifications. A recessive bit is a logic 1, and can be a level from an active or passive pull-up. The dominant bit is a logic 0 and is actively driven to ground by the transmitter. The idle state is the recessive level (logic 1). If one node is transmitting a dominant bit, while another node is transmitting a recessive bit then there is a collision. The dominant bit wins and the node transmitting the recessive bit identifies the collision and stops transmitting. Thus the higher-priority message is transmitted immediately and the node transmitting the lower-priority message automatically attempts to retransmit later, pausing for six clock cycles after the end of the dominant message.

Messages are addressed, not to the node, but to a function. A physical node carries data to a microcontroller, which processes the signals. A node and its microcontroller may support more than one function. Every node receives the messages and the microcontroller checks them for errors. If no errors are detected, the node will send an acknowledgment (ACK). Then the attached microcontroller decides if it supports the function being addressed. If no, the message is ignored, but if yes the message is acted upon. For example, one node could be located inside a car door may support the functions “unlock door” and “detect open door.” If the car door has not closed properly, the “detect open door” function would signal that the door was “open,” which would then be processed by the node controlling the display panel in the dashboard. The display panel node would light a warning lamp in the control panel.

The CAN bus is used to control LED lamps in automotive applications (and transport in general). Due to its speed, the CAN bus is used in some industrial applications. Although most industrial applications are mechatronic, some controlling lighting is also possible. For example, some elevators use a CAN bus for door control, motor control, and light control (including backlighting the push-button controls).

Regardless of the data transport mechanism, once a signal is received and decoded, it is converted into a physical medium like RS-485 for transmission to the lighting control system. The transmitted radio signal contains the lamp address and the light level setting required. Both DALI and DMX can be provided with radio interfaces. Examples of suppliers are Virtual Extension, which produces a DALI radio extender, and Wireless Solution Sweden AB, which produces a wireless DMX system.

Magnetek produces a wireless CAN bus transceiver, WIC-2402, which operates in the 2.4-GHz band. It has a 500-m range and a data speed up to 500 kilobaud, and so could be preferable to long-range wired connections.