Traffic Control Systems Handbook: Chapter 7. Local Controllers

Source: Eagle Products

Figure 7-1. Model 2070 Controller.

7.1 Introduction

This chapter provides detailed information on intersection traffic signal controllers so that the user can:

• Understand the principles of controller operation,
• Become familiar with various controller types, and
• Select controllers for specific applications.

Table 7-1 presents some basic definitions used throughout the chapter, while Table 7-2 summarizes functions performed by a local controller. Table 7-3 summarizes the two distinct modes of traffic signal controller operation - isolated and coordinated. A signal operating in isolated mode can also be said to be operating free or uncoordinated.

Table 7-1. Definition of Controller Terms

Terms	Definitions
Controller Assembly	The complete electrical mechanism mounted in a cabinet for controlling signal operation. The controller assembly generally includes the cabinet.
Controller Unit	Portion of a controller assembly which selects and times signal displays.
Intersection Controller Unit	The traditional and original usage, most commonly referred to as traffic signal controller.
Special Controller	Includes units for lane use control and other applications not involving the traditional assignment of right-of-way for vehicles and pedestrians at intersection or midblock locations.

Table 7-2. Traffic Signal Controller Functions

Can Control: single intersection closely spaced multiple intersections midblock crosswalk Electrically switches signal indications: red yellow green WALK DON'T WALK other Assures appropriate right-of-way assignments in accordance with pretimed or actuated intervals or phases Times fixed clearance intervals such as: flashing DON'T WALK yellow red clearance Times greens and green arrows for: fixed-duration (pretimed control) variable duration (up to a predetermined maximum) according to traffic demand (actuated control) Times special function timed intervals such as: lane controls turn controls blank out signs

Table 7-3. Isolated Versus Coordinated Signal Modes

Mode	Definitions
Isolated (Free)	The signal controller times right-of-way assignments independently of other signals. If one or more phases are actuated, the cycle length may vary from one cycle to the next.
Coordinated	The signal controller timing is coordinated with that of one or more adjacent traffic signals to avoid stopping approaching platoons of cars. Traditionally, this involves operating this and adjacent signals at the same, fixed-duration cycle length. Adaptive coordination techniques can achieve coordination while still allowing the cycle length to change from one cycle to the next.

A subsequent section of this chapter discusses controller units for applications other than traffic signals. See also Chapters 3 and 4 of this Handbook for additional information on some special control concepts.

7.2 Types of Operation

Despite the many variations in their design, traffic signals can be classified according to operational type as:

• Pre-timed (or fixed time),
• Fully actuated, and
• Semi-actuated.

Table 7-4 describes characteristics and applications of each of these types.

Table 7-4. Types of Signal Operation

Operation	Characteristics
Pretimed	Occurrence and duration of all timing intervals, both vehicle and pedestrian, in all phases are predetermined.
Fully Actuated	• All phases are actuated (i.e., use vehicle or pedestrian detectors). • Phases are skipped (not served) if no vehicles or pedestrians are detected. • If vehicles are detected but not pedestrians, only the vehicle portion of the phase may be served. • The Green interval of phases can vary in duration, between minimum and maximum values, depending on detected traffic demand. When a vehicle leaves a detector, the green is extended by a few seconds known as passage time or green extension. The phase terminates if all detectors for the phase remain unoccupied for duration longer then the ‘gap’ time. • The Walk interval is usually of fixed duration, but if the signal is coordinated, the Walk interval may be allowed to extend to make use of predictable additional green time, especially for main street phases. • Other intervals (e.g., yellow, red clearance, flashing Don't Walk) are of fixed duration.
Semi-Actuated	• At least one phase is guaranteed to be served while others are actuated. • This phase receives a guaranteed, or fixed, minimum amount of time. • If there is no demand for actuated phases, the guaranteed phase remains green longer than its "fixed" green time. • If the signal is coordinated, a guaranteed phase is usually the main street through phase. If actuated phases terminate before using all their split allocation, the spare time can be reassigned to the guaranteed phase, causing it to receive more than the "fixed" amount of green.

An actuated traffic signal is one that employs vehicle or pedestrian detectors to activate a particular phase (change it from red to green) only when vehicles or pedestrians are present. Once activated, the duration of the green display may vary depending on the number of vehicles detected.

Pre-timed, or fixed-time, phases are served for a fixed duration every cycle regardless of the number of vehicles or pedestrians present. A signal is pre-timed if all phases are fixed, and is fully actuated if all phases use detection. A semi-actuated signal has a mixture of pre-timed and actuated phases.

Coordinated signals are often operated in a semi-actuated mode. In this case, the main-street through phases need not have detectors, and are served every cycle regardless of demand. A coordinated signal must operate with a fixed-duration cycle. In a typical semi-actuated signal, if one or more actuated phases do not require all their allocated portion of the cycle, unused time is automatically re-assigned to the main street, non-actuated phases, which always terminate (turn yellow) at the same point in the cycle regardless of how early they commence (turn green).

Most modern traffic signal controllers support all of these types of signal operation. Even though a signal controller may provide actuation features for all phases, any or all phases may be made to operate as pretimed by use of the "call to non-actuated" input, or by using phase parameters such as recall, minimum green, and coordinated phase designation.

7.3 Range of Applications

Types of Signal Operation

Table 7-5 summarizes applications of the above-described types of signal operation, for each of the following three commonly encountered intersection environments:

• Isolated - a signalized intersection that is physically remote from other signalized intersections and therefore does not benefit from signal coordination.
• Arterial - a signalized intersection that is one of a series of adjacent signalized intersections along an arterial roadway, and benefiting from coordination during at least some times of the day - commonly found in suburban areas.
• Grid - a signalized intersection that is one of a series of adjacent signalized intersections in a grid of fairly short blocks - commonly found in older, high density urban areas and central business districts.

Table 7-5. Application of Signal Control Types

Type of Operation	Isolated	Arterial	Grid
Pretimed	Usually not appropriate.	Appropriate only if always coordinated and the side street volumes are high and consistent.	Appropriate
Semi-actuated	Appropriate only if main street traffic is consistently heavy.	Appropriate if always coordinated.	Appropriate to actuate left turn phases and other minor movements, and mid-block pedestrian signals.
Fully Actuated	Appropriate	Appropriate if not always coordinated.	Usually not appropriate.
Volume Option for actuated phases (see Section 7.5)	Appropriate for phases with only detectors set back more than 40 meters (125 feet).	Appropriate for phases with only detectors set back more than 40 meters (125 feet).	Usually not appropriate because slow speeds mean less detector set back.
Density Option for actuated phases (see Section 7.5)	Appropriate if high speeds, as higher initial gap can reduce stops.	Appropriate if high speeds, as higher initial gap can reduce stops.	Usually not appropriate due to low speeds.

Pretimed control best suits locations where traffic proves highly predictable and constant over a long period of time, and adjacent signals need to be coordinated at all times. These situations are commonly encountered in dense grid street networks (1).

Fully actuated control usually proves the most efficient operation at isolated intersections. On making the decision to install a traffic signal, first consider fully actuated control. Its traffic-responsive capability adjusts cycle and phase (split) lengths to fit changing demands from cycle to cycle. Rarely do approach traffic volumes at an isolated intersection remain predictably constant over a long period. Because all phases usually do not peak simultaneously, it should not be assumed that a full-actuated signal operates on a fixed cycle length even with high traffic demand.

Fully actuated control applies to a variety of signal phasing and detection schemes ranging from a simple two-phase operation to an 8-phase dual-ring configuration. Because of its skip-phase capability, the 8-phase dual-ring controller may operate as a basic two-phase controller under light traffic conditions; in the absence of demand, the controller unit ignores that phase and continues around the ring seeking a serviceable phase (1).

If an actuated signal is always coordinated, the cost of signal construction and maintenance can be reduced by using semi-actuated signal operation, with the main street through phases as pre-timed phases without vehicle detectors.

Protected, Protected / Permissive, and Permissive Operation

Traffic operations should aim to eliminate unnecessary delays at signalized intersections. Appropriate use of protected / permissive and permissive only traffic operation provides one means of reducing left- turn movement delay.

Provide separate left-turn phases only where needed, because unnecessary separate left-turn movements increase cycle length and traffic delays. Traffic control without separate left-turn operations can minimize delay for all movements including left-turns. However, conditions exist that require protected / permissive operation or justify protected (only) operation. Asante, et al. provides a set of guidelines for left-turn protection (2). The report provides guidance on:

• Justification of some form of protected left-turn phasing,
• Selection of type of left-turn protection, and
• Sequencing of left-turns.

Permanent changes from one type of operation to another may prove appropriate as traffic volumes change over time. Traffic operation can also change from protected to protected / permissive or permissive operation as traffic patterns change during the day and / or week.

When addressing left turn movement issues, it may be important to provide a left turn pocket for permissive left turn movements. However, in some cases, this will require the elimination of parking near the stop line in order to make room for the additional width needed for the left turn pocket.

Special Controls

A number of applications use special-purpose controller assemblies with electrical switching of signal indications akin to intersection controllers. Some of these applications include:

• Flashing beacons for various applications such as:
  • Roadway hazard identification,
  • Enforcement time definition for speed limits,
  • Intersection hazard identification with stop control, and
  • Use of visual-attention device with individual stop signs.
• Lane control signals (e.g, reversible lanes),
• Changeable lane use signs at intersections,
• Movable bridge signals and one-lane, two-way operation signals,
• Overheight vehicle controls to avoid structural damage by overheight commercial vehicles, and
• Audible pedestrian signals (3, 4, 5) that emit a buzzer or chirp sound for the initiation of a walk interval or phase to the visually impaired.

7.4 Controller Evolution

The evolution of traffic signal controllers parallels the evolution in related electronics industries. Signal controller unit hardware has evolved from the days of motor-driven dials and camshaft switching units to the adaptation of general-use microprocessors for a wide variety of intersection and special control applications.

In the early years of traffic signal control, virtually the only commercially available controller units were the electromechanical type. Later, several manufacturers introduced semi- and full-actuated controllers equipped with vacuum tube circuits for timing functions. The traffic engineer adjusted interval and phase timing via knobs on a control panel. Transformers and vacuum tubes in these analog units generated considerable heat, requiring forced-air circulation and filtering in controller cabinets. Some manufacturers retained solenoid-driven camshafts for lamp switching, while others used stepping relay-driven stacked rotary switches and encapsulated relays. Short component life and timing drifts characterized these controllers.

Replacement of the vacuum tube with the transistor introduced low-voltage circuitry with only a fraction of the former heat generation. The high-amperage heater circuits and high-voltage B plate circuits once required for vacuum tubes passed from the scene. The mid-1960s saw transistorized circuits first used for timing and phasing functions. Lower operating temperatures increased component life, and digital timing ensured timing accuracy and eliminated fluctuations. During this period manufacturers also introduced the solid-state load switch for lamp circuits. Wide variations in component and equipment arrangements from manufacturer to manufacturer also prevailed during the 1960s. Designs varied from those in which all timing and phasing components were placed on a single circuit board to those that used modular, plug-in phase and function-oriented designs.

The integrated circuit (IC) proved the next major step in controller evolution as microchip technology significantly reduced component size. These very small chips were linked together in circuits and sealed within an IC envelope to form the microprocessor. This development led to microcomputers - small, lightweight, low-cost units used practically everywhere today.

The traffic control industry quickly incorporated microprocessors into new signal controller designs. They are used in all modern traffic signal controllers.

The functionality and characteristics of a modern signal controller are determined by software more than hardware. The same physical controller may operate quite differently when loaded with a different software package.

Different standards have evolved for modern traffic signal controllers, including those developed by the National Electrical Manufacturers Association (TS 2), and Caltrans, New York DOT and FHWA (Model 170). These standards, and the Advanced Transportation Controller (including the ATC 2070) are discussed in Section 7.6.

7.5 Controller Characteristics

Signal Timing and Coordination

Traffic signal controllers alternate service between conflicting traffic movements. This requires assignment of green time to one movement, then to another. If left turns have separate controls, and at complex intersections, there may be more than two conflicting movements. The length of time taken to complete one round of service for all conflicting movements is called the cycle length, and the allocation of the cycle length between the conflicting traffic movements is called the split.

To minimize traffic delay, it is desirable that a platoon of vehicles leaving one intersection arrives at the next intersection during a green display. This is called platoon progression and is achieved by coordinating the operation of adjacent signals. Signal coordination is most commonly achieved by operating adjacent signals at the same cycle length, with a pre-determined offset between the start of the cycle at one intersection and the start of the cycle at the next. See Chapter 3 for further discussion of coordination timing parameters.

The cycle length, split, and offset may need to change during the day as traffic volumes change. Controllers, therefore, allow the user to establish multiple sets of these basic coordination timing parameters. Each such set is referred to as a timing plan or timing pattern, and one timing plan or timing pattern is in operation at any given time. The timing plan or timing pattern in operation can be changed either by a time-of-day schedule stored in the controller or by a command from a master device.

Interval Control versus Phase Control

Traffic signal controllers available today can be categorized as interval controllers (also called pretimed) or phase controllers (also called actuated). The former allow the user to divide the cycle into any number of intervals, with the duration of each interval being set by the user. The user then defines which output circuits are switched on during which intervals. For example, a particular interval may be used to time part of the green for one vehicle movement, part of the flashing don't walk for a pedestrian movement, the yellow for another vehicle movement, and part of the red and steady don't walk for others.

The cycle length equals the sum of the interval durations, and all intervals are timed sequentially. The user can also specify a start-of-cycle offset for signal coordination. The interval durations, output definitions, cycle length, and offset can all be varied from one pattern to another, and therefore can be varied during the day.

Modern interval controllers typically also allow a degree of actuated operation, whereby selected intervals may be skipped if there is no demand, or the duration of selected intervals can vary dynamically by detector actuations. If an interval does not use all of its allocated time, the spare time can be assigned to a following interval. Some controllers allow the user to create quite elaborate customized logic for controlling interval occurrence and duration.

Phase controllers take a different approach to signal timing. They divide the cycle into phases, with each phase having five pre-defined intervals - green, yellow and red clearance for vehicle control; and walk and flashing don't walk for pedestrian control. The user specifies the duration of each of these intervals, or in the case of the green interval, the minimum and maximum duration. If the signal is coordinated, the user also specifies a split time for each phase, and a start-of-cycle offset.

The user assigns a phase to a set of compatible vehicle and pedestrian movements. If coordinated, the split times for all phases in a ring must sum to the cycle length. Each phase is assigned to a timing ring (Figures 7-2 and 7-3). Phases assigned to the same ring time sequentially, but rings time concurrently. Therefore, if the controller is using two rings, two phases can be timing simultaneously and independently.

Phase controllers use barriers or phase concurrency groups to define conflicts between phases in different tings. Within a concurrency group (between two barriers) the phases in different rings can time independently, but all rings must cross the barrier (move to a different phase concurrency group) simultaneously.

Within a concurrency group (between two barriers) the user can specify the desired order (sequence) in which phases in the same ring are to be served. From one pattern to the next, the user may vary the cycle length, offset, split, and phase sequence.

Phase control is particularly well suited to actuated control of normal intersections, especially those with protected left turn movements. Two actuated left turn phases on the same street can time independently, with say the westbound turn phase receiving less time than the eastbound in one cycle, and the opposite occurring in the next cycle. For this reason, and their ease of setup and additional actuation features, phase controllers have become the dominant type.

Figure 7-2. Three-phase Controller Phase Sequence for Single-Ring Controller.

Figure 7-3. Phase Sequence for Dual-Ring Controller.

For many years, phase controllers were limited to eight phases allocated to two rings in a fixed arrangement. This works very well for most intersections, but does not provide the flexibility needed for unusually complex intersections. Also, if fixed-time control is sufficient and left turn phasing is not prevalent, such as often occurs in the central business districts of large cities, the interval controller is adequate. Interval controllers therefore have remained in use, although their numbers are dwindling as phase controllers have expanded to accommodate more phases and rings, and have added features such as redirection of outputs. Each phase in a phase controller can be operated either pretimed (fixed time) or actuated.

The National Electrical Manufacturers Association (NEMA) TS 2 standard specifies minimum functional standards for both interval and phase controllers. Most modern controllers meet most or all of these minimum requirements and most controllers also provide additional functionality not yet standardized.

Controller and Cabinet Components

Most modern traffic signal controllers have the following basic hardware components:

• User interface (keypad and display)
• Central processing unit (microprocessor, memory, etc.)
• External communications connectors (serial ports, Ethernet, USB, cabinet wiring, etc.)
• Power supply (converts 110v AC to 24v, 12v, 5v DC for internal use)
• Optional additional serial communications processor (FSK modem, RS 232)

Serial communications ports are often used for establishing a link to a master control unit or computer. Such connections may be permanent to a remote master or computer, or temporary to a laptop computer used by field personnel. Ethernet is increasingly being used instead of serial communications. As special serial port may be used to communicate with in-cabinet equipment in the case of a serial-bus cabinet (see NEMA TS 2 and ATC sections below).

Within the signal controller cabinet, and connected to the controller, are the following basic auxiliary components that interact with the controller:

• Malfunction management unit (also referred to as a conflict monitor)
• Vehicle and pedestrian detectors (sensor units, circuit isolators)
• Output circuit drivers (load switches driving signal displays)
• Optional external communications devices (external FSK modem, fiber transceiver, wireless transceiver, Ethernet switch, etc.)

Detectors are used only for actuated signals. A load switch uses a low voltage direct current output of the controller to switch a 110v AC circuit on or off, thus turning on or off a signal display viewed by motorists or pedestrians. For a particular phase, one circuit is switched off just as another is switched on.

The malfunction management unit (MMU) can be configured to check for conflicting signal indications and various other malfunctions including absence of an OK status output from the controller (watchdog output), short or missing clearance intervals, and out-of-range operating voltages. If a malfunction is detected, the MMU automatically places the signal in an all-red flashing state, overriding the outputs of the controller. Modern controllers can sense this condition and report the malfunction state to a master or central computer.

Pattern Selection

Modern controllers offer the following three alternative methods of determining which pattern or plan to operate:

Internal time-of-day schedule - the user configures a schedule that tells the controller when to change the pattern or plan, based on the day of the week and time of the day. Special schedules can be created for holidays or other dates on which traffic conditions are unusual. The controller's clock, which keeps track of date, day of week, and time, is regularly compared to the entries in the schedule. No external communications are required. This mechanism is often used as a backup when an external pattern selection method fails. This method is commonly used.

Hardwire interconnect - multiple electrical wires (typically seven) installed between the controller and a master unit, have a steady voltage applied or removed to indicate which pattern or plan is to be used. When the combination of active (voltage on) and inactive (voltage off) wires changes, the combination is used by the controller to look up which pattern or plan to change to. Traditionally, this method was used to independently select which of several pre-defined cycle lengths, offsets, and splits to use, thus emulating the selection of dial, offset, and split keys in an electromechanical controller. Use of this method is declining.

External command - using digital communications (typically via a serial or Ethernet port on the controller), a master unit or computer sends a command message to the controller, instructing it to change to a particular pattern. This method is commonly used. If the controller loses communications with the source of pattern commands, it can automatically revert to using its internal time-of-day pattern selection schedule. The same communications link is typically used to receive status information from the controller, and to enable remote changes to controller parameters.

It is also possible for the user to manually lock a controller into a particular pattern, such that any of the above pattern selections is ignored.

Synchronization for Coordination

Signal coordination requires all controllers in a coordinated group to have a common time reference, so that start-of-cycle offsets are applied accurately. Before controllers had internal clocks, this was typically achieved by connecting the controllers to a master unit using the hardwire interconnect method described above. Once each cycle, one of the input wires changes its state for a second or two (called a pulse), thus signaling the commencement of the background cycle to all connected controllers simultaneously. Each controller then times its own offset from this common reference point. Use of this hardwire interconnect method is declining, in favor of time base coordination.

Today, controllers have internal clocks capable of keeping reasonably accurate time for at least several days. All controllers in a coordination group can be configured to use the same time of day (say midnight) as the reference point for offset calculation. The common background cycle is assumed to start at this time of day, and each controller can time its own offset from this common reference point. This is called time base coordination.

Eventually, however, the controller's clock will drift and need to be reset to standard time. Clocks can be reset using any of the following techniques:

Manual - periodically, a user goes to the controller in the field and resets the time according to an accurately set watch or other source of standard time (e.g., cell phone time display, telephone call to voice time, etc.). This method is not favored as it is laborious, error-prone, and subject to neglect. Depending on the model of controller, operationally significant drift can require manual reset after only several weeks of operation.

Hardwire pulse - a master unit pulses a hardwire input to the controller at a pre-defined time of day. When the controller senses this pulse, it sets its clock to the pre-defined time of day. As long as all controllers in the coordinated group receive the same pulse, it doesn't matter if the clock of the master unit is not entirely accurate.

External command - using digital communications (typically via a serial or Ethernet port on the controller), a master unit or traffic signals management computer sends a command to the controller (say once each day), instructing it to immediately set its clock to a time specified in the message. Even signals under the command of different central computers can be coordinated as long as each central computer has its clock set accurately.

Third-party time source - a standard time source, such as a WWV radio receiver, cell phone time monitor, or Internet connection, is installed in the cabinet and the controller either listens for periodic broadcast time updates or periodically initiates a request for a time update from a time server.

Actuated Controller Operation

Regardless of the hardware standard a controller complies with (NEMA, ATC, or Model 170), the functionality of the resident software is similar, and generally operates as defined in the NEMA TS 2 standard.

The basic timing characteristics of actuated controller units are as follows:

• Each phase has a preset minimum green interval to provide starting time for standing vehicles.
• The green interval extends for each additional vehicle actuation after the minimum green interval has timed out, provided that a gap in traffic greater than the present unit extension setting does not occur.
• A preset maximum limits green extension. Controllers provide two selectable maximum limits (commonly referred to as MAX I, and MAX II).
• Yellow change and red clearance intervals are preset for each phase. Red clearance is not always needed.

In addition to detector inputs, each phase is provided with a means for the user to permanently place a call for vehicle service (minimum or maximum green recall), or for pedestrian service (pedestrian recall). Maximum green recall places a call for the phase and when served prevents it from terminating prior to expiration of the maximum green timer.

The maximum green timer on a respective phase does not begin timing until a serviceable opposing phase detector call. Therefore, a phase with continuing demand may remain green for some time before a conflicting call is registered that starts the timing of the maximum green.

Phase control concepts related to rings and barriers are described in Table 7-6, and basic actuated timing parameters are described in Table 7-7.

Table 7-6. Actuated Controller Definitions

Feature	Description
Single-Ring Controller Unit	Contains 2 to 4 sequentially timed and individually selected conflicting phases arranged to occur in an established order or sequence. Phases may be skipped in 3 and 4-phase controllers. The phases within a ring are numbered as illustrated in Figure 7-2.
Dual-Ring Controller Unit	Contains 2 interlocked rings arranged to time in a preferred sequence and allow concurrent timing of respective phases in both rings, subject to the constraint of the barriers (compatibility lines). Each ring may contain up to two phases in each of its two barrier groups, for a total of eight phases. Each of the respective phase groups must then cross the barrier simultaneously to select and time phases in the phase group on the other side. The phases within the 2 timing rings are numbered as illustrated in Figure 7-3.
Multi-Ring Controller Unit	A controller supporting more than eight phases and two rings. Any number of phases, up to the maximum supported by the controller, can be arranged in any number of rings. Conflicts between phases in different rings are specified using either barriers inserted between groups of phases, or phase concurrency lists This document has not been validated in the field. I would not recommend its inclusion here unless disclaimer are clearly included.
Barrier (compatibility line)	A reference point in the designated sequence of dual-ring and multi-ring controller units at which rings are interlocked. Barriers ensure conflicting phases will not be selected or time concurrently. At a barrier, rings terminate the current phase and cross the barrier simultaneously, as illustrated in Figure 7-3.
Dual Entry	A mode of operating in a dual-ring and multi-ring controller units in which one phase in each ring must be in service. If a call does not exist in one of the rings when the barrier is crossed (from the other phase group), a phase is selected in that ring to be activated by the controller in a predetermined manner. For example, referring again to figure 7-3 in the absence of calls on Phases 7 and 8, Phase 2 and Phase 6 terminate to service a call on Phase 3. Programming for dual entry determines whether Phase 7 or Phase 8 will be selected and timed concurrently with Phase 3, even though no call is present on either Phase 7 or Phase 8.
Single Entry	A mode of operation in a dual-ring and multi-ring controller units in which a phase in one ring can be selected and timed alone when there is no demand for service of a non-conflicting phase in another ring. For example, referring to figure 7-3, after the termination of Phase 2 and Phase 6, the controller unit will service a call on Phase 3 in the absence of calls on either Phase 7 or Phase 8. While Phase 3 is selected and timed alone, Phases 7 and 8 (in Ring 2) will remain in the red state.

Table 7-7. Actuated Controller Basic Timing Parameters

Setting	Description
Minimum Green	The absolute minimum duration of the phase's green indication. The phase cannot gap out or be forced off during this interval.
Variable Initial Green	A time calculated from the number of approach detector actuations during red. In the absence of a stopline detector, it allows sufficient time to service vehicles queued between the stopline and an advance detector. The phase cannot gap out or be forced of during this interval. The duration of this interval is affected by related parameters including Added Initial (amount of green added per actuation) and Maximum Initial.
Pedestrian Walk	The minimum duration of the Walk indication for pedestrians. The phase cannot gap out or be forced off during this interval.
Pedestrian Clearance	The fixed duration of the Flashing Don't Walk indication for pedestrians. The phase cannot gap out or be forced off (except for railroad or emergency vehicle preemption) during this interval.
Green Extension	The amount of time by which the green is extended after a vehicle is detected. If the minimum green, variable initial green, Walk, and FDW have all expired, and no approach detector input is currently On, the phase green can terminate (gap out) if the time gap between consecutive vehicles exceeds the green extension time plus the time the detector input remains On while the vehicle is being sensed.
Maximum Green	Even if vehicles are still approaching, the phase green will be terminated (forced off) after this amount of total green time following a call for service on a conflicting phase. This parameter overrides Green Extension, but none of the other parameters above.
Yellow Clearance	The fixed duration of the yellow indication that always follows the green indication.
Red Clearance	The time during which both the terminating phase, and the following conflicting phase(s) about to start, simultaneously present a red indication.

One or more actuated phases may also use the volume and / or density options, each being an add-on to basic actuated operation, as follows.

• The "volume" option increments an initial green interval timer each time a vehicle is detected while the phase is red. The minimum green is timed as the greater of the normal minimum green and this computed initial green, up to a maximum. In the absence of stopline detectors, it can be used to count the number of vehicles waiting in front of the advance detectors and increase the minimum green, if needed, to clear this queue.
• The "density" option reduces the gap time while the phase is green, if vehicles or pedestrians are waiting (have been detected) on other phases. The gap is reduced gradually over time, requiring a progressively greater density of approaching traffic to avoid termination of the green.

A dual-ring actuated controller allows different sequencing of left turn phases. Table 7-8 and Figure 7-4 describe phase sequence options for a signal with odd numbered phases serving left turns, and even numbered phases serving their opposing through movements. Typical left turn sequence options are leading lefts, lead-lag lefts, and lagging lefts. One such sequence can be used on one street (one barrier group), while a different sequence is used on the other street.

Table 7-8. Phase Sequence Options

Sequence	Description
Leading Left Turn	Sequence begins with Phase 1 and Phase 5, the opposing turns moving together. As demand ends or maximum green is reached on either Phase 1 or Phase 5, the respective left-turn is terminated after the proper change and clearance intervals, and the opposing thru movement (Phase 2 or Phase 6) is given a green indication concurrent with its accompanying left-turn. As demand ends or maximum green is reached on the remaining left-turn movement, it is terminated after the proper change and clearance intervals, and its opposing thru movement is released. Phase 2 and 6 then run together until demand ends or maximum green time for both phases is reached. The phases then, after display of proper change and clearance intervals, terminate simultaneously at the barrier line. As shown in figure 7-4, the above phase sequence also applies to the phases beyond the barrier line (Phases 3, 4, 7 and 8) in the other phases group.
Lead-Lag Left-Turns	Sequence begins with Phase 5, a left-turn, and its accompanying Phase 2 moving concurrently. As demand ends or maximum green is reached on Phase 5, that left-turn is terminated after the proper change and clearance intervals. The opposing thru movement, Phase 6, is released to run with Phase 2. As demand ends or maximum green for Phase 2 is reached, it is terminated after the proper change and clearance intervals, at the barrier line. As shown in figure 7-4, the above phase sequence also applies to the phases beyond the barrier line (Phase 3, 4, 7 and 8), in the other phase group. Also, it must be noted that either of the opposing left-turns in each phase group may lead the phase sequence.
Lagging Left Turns	Sequence begins with the opposing thru movements, Phases 2 and 6. As demand ends or maximum green is reached on one of the thru movements, that phase (2 or 6) is terminated after the proper change and clearance intervals, and its opposing left-turn (Phase 1 or 5) is released to run concurrently with the accompanying thru movement, that phase (2 or 6) is terminated after the proper change and clearance intervals, and its opposing left-turn (1 or 5) is released. Both left-turns run together until demand ends or maximum green on the latest released phase is reached. Phases 1 and 5 then terminates simultaneously after the proper change and clearance intervals at the barrier line. As shown in figure 7-4, the above phase sequence also applies to the phases beyond the barrier line (Phases 3, 4, 7 and 8), in the other phase group.

Figure 7-4. Dual-ring Basic Phase Sequence Options

Any of these sequences can operate at all times, or can change during the day as the timing pattern changes. However, phase sequence needs to be chosen with care if the left turn movement can be made both protected and permissively, and a traditional five-section signal head is used (two left turn arrows and three balls). In this case, a phase sequence involving a lagging left turn phase, either lead-lag left turns or lagging left turns, can result in a potentially dangerous situation known as the "left turn trap." A motorist turning left permissively and waiting for a gap in opposing traffic sees the green ball change to a yellow ball. The driver assumes the on-coming traffic also sees a yellow ball and will stop, when in fact the on-coming traffic may continue to see a green ball and not stop. This problem is eliminated by the flashing yellow arrow display for protected / permissive turn control. In this case the permissive indication (the flashing yellow arrow) tracks the opposite-direction through phase instead of the same-direction through phase.

The TS 2 standard specifies various external control inputs to the controller that modify its normal behavior. They are grouped into three categories:

• Inputs per phase (see Table 7-9)
• Inputs per ring (see Table 7-10)
• Inputs per controller unit (see Table 7-11)

Phasing Other Than Eight-Phase Dual-Ring

Many modern controllers, or controller software packages, offer sixteen or more phases in four or more rings, and eight or more overlaps, allowing control of numerous traffic movements needing separate phases or overlaps and more than normal eight-phase, dual-ring logic. Some examples of non-standard phasing used to control two closely-spaced intersections are discussed in Section 3.9 and in the following section on diamond interchanges.

Even intersections using only eight phases and two rings may have non-standard logic applied. One example is conditional re-service of a leading left turn phase following its opposing through phase (see Figure 7-5) - the left turn phase appears twice in the cycle, both before and after its opposing through phase, but only if the through movement is sufficiently light. Another example is "separated phases" logic, which can be used, for example, to prevent a leading left turn phase from operating concurrently with a lagging left turn phase from the same street if the two turning movements physically conflict in the middle of the intersection.

Table 7-9. Inputs Per Phase

Input	Description
Vehicle Detector Call	Enters a vehicle demand for service into the appropriate phase of the controller unit.
Pedestrian Detector Call	Enters a pedestrian demand for service into the associated phase of the controller unit.
Hold	Command that retains the existing right-of-way and has different responses, as follows depending upon operation in the vehicle non-actuated or actuated mode: For a non-actuated phase, energization of the hold input maintains the controller unit in the timed out walk period with green and walk indications displayed. Energization of the hold input while timing the WALK portion of the green interval does not inhibit the timing of this period. De-energization of the hold input and with the WALK interval timed out causes the controller unit to advance into the pedestrian clearance interval. Re-application of the hold input while timing the pedestrian clearance portion of the green interval neither inhibits the timing of this period nor the termination of the phase. For an actuated phase, energization and de-energization of the hold input operates as follows: (a) Energization of the hold input allows the controller unit to time normally but inhibits its advance into the yellow change interval. Energization of the hold input inhibits the recycle of the pedestrian service unless the pedestrian recycle input is active and a serviceable call exists on the phase. The rest state signal indications for that phase are green and DONT WALK. (b) De-energization of the hold input allows the controller unit to advance into the green dwell / select state when all green periods are timed out. (c) De-energization of the hold input with all intervals timed out allows the controller unit to recycle the walk interval if there is no conflicting demand for service and a pedestrian call exists for that phase. However, if there is any serviceable demand on an opposing phase with the hold input de-energized, and with all intervals timed out, the controller unit advances into the yellow change interval and does not recycle the walk on that phase until those demands have been served.
Phase Omit	Command which causes omission of a phase, even in the presence of demand, by the application of an external signal, thus affecting phase selection. The omission continues until the signal is removed. The phase to be omitted does not submit a conflicting call to any other phase but accepts and stores calls. The activation of Phase Omit does not affect a phase in the process of timing.
Pedestrian Omit	Command which inhibits the selection of a phase resulting from a pedestrian call on the subject phase, and it prohibits the servicing of that pedestrian call. When active, the Pedestrian Omit prevents the starting of the pedestrian movement of the subject phase. After the beginning of the subject phase green, a pedestrian call is serviced or recycled only in the absence of a serviceable conflicting call and with Pedestrian Omit on the phase non-active. Activation of this input does not affect a pedestrian movement in the process of timing.

Table 7-10. Inputs Per Ring

Input	Description
Force-Off	Command which provides for the terminations of green timing or WALK hold in the non-actuated mode of the active phase in the timing ring. Such termination is subject to the presence of a serviceable conflicting call. The Force-Off is not effective during the timing of Initial, WALK or pedestrian clearance. Force-Off is effective only as long as the input is sustained.
Red Rest	Requires the controller unit to rest in red in all phases of the timing ring(s) by continuous application of an external signal. The registration of a serviceable conflicting call results in the immediate advance from Red Rest to green of the demanding phase. The registration of a serviceable conflicting call before entry into the Red Rest state results in the termination of the active phase and the selection of the next phase in the normal manner, with appropriate change and clearance intervals. The registration of a serviceable call on the active phase before entry into the Red Rest state even with this signal applied, results (if Red Revert is active) in the continuation of the termination of the active phase with appropriate yellow change interval and Red display for the duration selected in Red Revert. The formerly active phase is then reassigned right-of-way.
Inhibit Maximum Termination	Disables the maximum termination functions of all phases in the selected timing ring. This input does not, however, inhibit the timing of Maximum Green.
Omit Red Clearance	Causes the omission of Red Clearance timing intervals
Pedestrian Recycle	Controls the recycling of the pedestrian movement. The operation depends on whether the phase is operating in the actuated or non-actuated mode: In the actuated mode, if a serviceable pedestrian call exists on the subject and the Hold input is active, the pedestrian movement is recycled when the Pedestrian Recycle input is active, regardless of whether a serviceable conflicting call exists. In the non-actuated mode, if the subject phase has reached the Green Dwell / Select state, the Pedestrian Omit is not active on the phase and a serviceable conflicting call does not exist, the pedestrian movement is recycled when the pedestrian recycle input is active.
Stop Timing	When activated, causes cessation of controller unit ring timing for the duration of such activation. Upon the removal of activation from this input, all portions which are timing, will resume timing. During Stop Timing, vehicle actuations on non-Green phases are recognized; vehicle actuations on Green phase(s) reset the Passage Time timer in the normal manner, and the controller unit does not terminate any interval or interval portion or select another phase, except by activation of the Interval Advance input. The operation of the Interval Advance with Stop Timing activated clears any stored calls on a phase when the controller unit is advanced through the green interval of that phase.
Maximum II (Selection)	Allows the selection of an alternate maximum time setting on all phases of the timing ring

Table 7-11. Inputs Per Controller Unit

Input	Description See section 3.5.5.5 of NEMA TS2 Standard (6)
Interval Input Advance	A complete On-Off operation of this input which causes immediate termination of the interval in process of timing. When concurrent interval timing exists, use of this input causes immediate termination of the interval which would terminate next without such actuation.
Manual Control Enable	Places vehicle and pedestrian calls on all phases, stops controller unit timing in all intervals, and inhibits the operation of the Interval Advance input during vehicle change and clearance intervals
Call to Non-Actuated Mode (Two per Controller Unit)	When activated, causes any phases appropriately programmed to operate in the non-actuated mode. The 2 inputs are designated Call to Non-Actuated Mode I and Call to Non-Actuated Mode II, respectively. Only phases equipped for pedestrian service are to be used in a non-actuated mode.
External Minimum Recall to All Vehicle Phases	Places recurring demand on all vehicle phases for a minimum vehicle service
External Start	Causes the controller unit to revert to its programmed initialization phase(s) and interval(s) upon application of the signal. Upon removal of this input, the controller unit commences normal timing.
Walk Rest Modifier	When activated, modifies non-actuated operation only. Upon activation, the non-actuated phase(s) remain in the timed-out WALK state (rest in WALK) in the absence of a serviceable conflicting call without regard to the Hold input status. With the input nonactive, non-actuated phase(s) do not remain in the timed-out WALK state unless the Hold input is active. The controller unit recycles the pedestrian movement when reaching the Green Dwell / Select state in the absence of a serviceable conflicting call.

Figure 7-5. Example of Special Phase Sequence for Conditional Service of Left-Turn Phase

Diamond Interchange Operation

Some actuated controllers provide a special mode of operation derived from the Texas Department of Transportation's historical approach to diamond interchange operation. Modern controllers can provide similar functionality without the need for a special mode of operation, as described in section 3.9.

Two particular phasing arrangements and logic for diamond interchange operation have been used in Texas (7). These are referred to as the 3-phase and 4-phase sequences and are described in Table 7-12. The operation can change between the sequence options in response to external commands. The City of Dallas provides for four sequence variations. The two sequence variations shown in Figure 7-7 are used by the Texas Department of Transportation. Typical detector locations for operation of the controller unit in 3-phase, lag-lag, or four-phase (with overlaps) sequencing, with locally produced external data, are shown in Figure 7-8. Software also provides the option for use of any compatible combination of phases at the ramp intersections, in response to computer-issued command data, as shown in Figure 7-9.

The 3-phase sequencing shown in Figures 7-6 and 7-7 can provide a shorter cycle length than the 4-phase sequencing shown in Figure 7-7. For example, Texas DOT conducted a study in which the two phase sequences shown in figure 7-7 were compared at a number of intersections during isolated full-actuated control. The cycle lengths for the 4-phase sequence were 40 to 80% longer than for the 3-phase sequence. Expect similar reductions in cycle lengths at locations in other isolated and interconnected systems, as long as the left-turn movements remain within reasonable limits, and storage is available between the off-ramp (frontage road) connections. Where turning movements are high onto and / or off of the ramp connections (frontage roads), the 4-phase sequence provides the best operation.

One of the three phase sequences shown in Figure 7-6 can also apply when certain turning movements prove heavy. If the controller includes more than one phase sequence, the sequences can be changed to accommodate operational requirements.

Table 7-12. Special Phase Sequencing

Operation	Description
Left-Turn Restoration	In the operation of a standard 8-phase controller unit, the service of a left-turn can be restored without first cycling through the barrier line. In this operation, the controller unit monitors the time remaining on any thru movement phase which is opposed by a thru phase which has gapped out. If the time remaining on the non-gapped phase is sufficient for at least a minimum service of its associated (parallel) left-turn phase, the controller unit terminates the gapped-out phase and reservices the left-turn. Figure 7-5 illustrates the phase sequence.
Full Diamond Interchange	The operation of 1 standard 8-phase controller unit with modified software for signalization of a full diamond interchange. Figures 7-6 and 7-7 show 4 sequence variations: A 3-phase lead-lag operation in which traffic on both ramp approaches begins simultaneously (Phase 1). Phase 2 follows Phase 3 if there is a demand (detector activation) for the phase. Phase 3 follows Phase 2 if there is a demand for the phase, and Phase 1 follows Phase 3 if there is a demand for that phase. A 3-phase operation in which traffic on both cross street approaches begins simultaneously and is followed subsequently by Phases 2 and 3 if there is a demand for each of these phases. A 3-phase, lag-lag operation in which the traffic on both ramp approaches is released simultaneously (Phase 1). Subsequent vehicle actuations and / or maximum green time-outs determine diagram flow from Phase 1 to 1 of the 4 overlap phases, or directly to Phase 2. Depending upon demand registered and which Phase 1 overlap was previously served, the controller unit will move to serve 1 of the 2 Phase 2 overlaps or go directly to Phase 3. In the absence of demand from either ramp approach, the controller unit may proceed from Phase 3 back to Phase 2, or to 1 of the 2 Phase 2 overlaps. A 4-phase operation with 2 overlaps, in which traffic on one of the ramp approaches is released simultaneously with thru and left-turn traffic (on the intersecting arterial) at the other ramp intersection, thereby clearing any possible internal queue for the traffic turning left from the ramp (Phase 1). As shown in the diagram in Figure 7-6, several optional flow paths are available, any of which could be followed, based upon registered demand and / or maximum green time-outs on certain approaches. For purposes of illustration, the following flow sequences assume continuing demand on all detectors. From Phase 1, the controller unit moves to Phase 1 overlap, in which the opposing traffic on the arterial (at the, as yet, unserved ramp intersection) is released while the ramp approach green continues. The Phase 1 overlap phase must be of fixed time duration since the running ramp green must be terminated to accommodate the progressive movement of the arterial traffic released at the start of the overlap phase. This fixed time period is determined by the travel time of accelerating arterial traffic from a stop at one ramp intersection, and through the other ramp intersection. The controller unit proceeds then to Phase 2 green to accommodate the above described approaching arterial traffic (thru and left-turns). For Phase 3 initiation, traffic on the arterial (at the, as yet, unserved ramp) is cleared and terminated for release of traffic on the ramp approach. As the diagram shows, flow continues to Phase 3 overlap and on to Phase 4, serving remaining traffic movements. The phase sequence described for each of the 4 sequence variations assumes that there is demand for each phase. Since the controllers are full traffic actuated, it is possible for phases to be skipped. The array of flow lines and arrows on the diagram represents all possible sequence paths the controller unit can take. The different sequence variations shown in Figures 7-6 and 7-7 are applicable and depend on the traffic patterns at the interchange. The software for 2 or more of the sequences can be provided in the same controller unit and changed by time-of-day or on a real-time basis as traffic patterns change.

Figure 7-6. Diamond Interchange Phasing (3-phase).

Figure 7-7. Diamond Interchange Phasing (3- and 4-phase).

Figure 7-8. Typical Detector Configuration for 3-phase, Lag-lag, and 4-phase (with overlap) Special Sequences.

Figure 7-9. Computer Controlled Diamond Interchange Operation.

Single Point Freeway Interchange Operation

The single point urban interchange (SPUI) shown in Figure 7-10 has been installed at a number of freeway locations. The design provides a basic six movement operation as shown in Figure 7-11. It is similar to typical five-phase control at normal intersections, except that pedestrians and right turns may require special treatment. It is difficult to efficiently allow pedestrians to cross the cross-street, and pedestrians crossing the ramps may require separate controls at left and right turn slots.

The Texas Transportation Institute studied the single point design, which resulted in warrants and guidelines (8). The SPUI and the tight urban diamond interchanges with a distance of 250 to 400 ft (76 to 122 m) between ramp connections (or frontage roads) were judged viable competitors.

The study recommended the following guidelines for the SPUI:

• Equivalent left-turn volumes exceed 600 v/hr as large truck volumes are anticipated from off-ramps having left-turn volumes exceeding 300 v/hr
• SPUI becomes a good candidate with:
  • Restricted right-of-way,
  • High volumes with major congestion,
  • High incidences of left-turns and large truck volumes (see above), and
  • High accident incidence locations.
• SPUI is not a candidate at sites with:
  • Severe skew angles,
  • A wide overcrossing roadway,
  • Adverse grades on the cross street,
  • Moderate-to-high pedestrian crossing volumes, or
  • A combination of high through-volumes and low turning volumes on the cross-street.

Figure 7-10. Single Point Urban Interchange (SPUI)

Figure 7-11. Typical SPUI 3-phase Sequence.

Freeway incident management often makes use of continuous frontage roads. Due to longer cycle lengths and increased delays, the SPUI is not recommended where continuous frontage roads exist when the SPUI and the frontage roads are grade-separated with one elevated above the other.

System Capabilities

The actuated controller, when used as a local unit in a traffic signal system, can provide additional functions other than previously described. Through the use of communications to a supervising master or central computer, the controller receives and implements a variety of commands. In closed-loop systems or central computer control systems, a two-way communications system returns information from the local unit to the central facility. The control status of the local controller and timing plan in effect exemplify returned local-oriented information. In many systems using two-way communications, system detector information is also returned to the supervising master unit or central computer.

A user at a central management computer can upload and examine the controller's data set (timing parameters). A copy of the controller's data can be stored in a central database, modified, and downloaded to the controller in whole or in part.

Implementation of downloaded interval durations and phase sequences may be subject to local minimums, maximums, or other checks, or the downloaded data may overwrite existing data with no checks. Methods vary from system to system, and traffic engineers must remain aware of the resulting impacts on traffic flow and operations safety.

An in-the-field master unit may also store a copy of controller timings.

7.6 NEMA, Advanced Transportation Controller, and Model 170 Standards

The National Electrical Manufacturers Association (NEMA) maintains the TS 2 standard (6) for traffic signal controllers and related equipment. This standard defines functionality, interfaces (physical and logical), environmental endurance, electrical specifications, and some physical specifications, for the following components:

• Traffic signal controllers,
• Malfunction management units,
• Vehicle detectors,
• Load switches,
• Bus interface units,
• Facilities for signal flashing and related control transfer, and
• Cabinets.

The TS 2 standard does not specify the physical size, shape, or appearance of most components except where standardization is necessary for physical interchangeability of whole components from different manufacturers. Although maximum dimensions are specified for the controller, a manufacturer is free to make a unit of any smaller size from any material, in any shape, with internal sub-components of any type, as long as it meets the other requirements of the standard. There are no requirements that enable interchangeability of sub-components or software between controllers from different manufacturers. It is assumed that the whole controller and its software will be swapped out when a change is made. The standard specifies a range of alternative cabinet sizes, all having shelves, and a door on one side only.

The TS 2 standard includes basic specifications for interval controllers (called "pretimed" in TS 2), but provides far more detail for phase controllers (call "actuated"). Signal phasing and timing functionality discussed above applies only to phase (actuated) controllers, the predominant type in use today.

Hardware requirements for controllers are specified by NEMA TS 2 in the following areas:

• A, B, and C connectors for cabinets using the older TS 1 standard
• Serial bus for communications with MMU, detectors, and load switches
• Serial ports for communications with computers and master units (RS 232 and FSK modem)
• User interface (keypad and display, required but details not specified)
• Maximum dimensions

The NEMA TS 2 standard defines two alternative types of input / output interfaces for the controller. One consists of binary (on or off) logic wires (analog) connected to the controller via three round connectors designated as MS-A, MS-B, and MS-C. This interface was originally standardized in a prior NEMA standard - TS 1. It is still widely used, and remains an option within TS 2. It is common for NEMA-compliant controllers to provide additional input / output control wires via a non-standard connector MS-D.

The other type of input / output interface specified in TS 2 is a serial bus. This option reduces the amount of wiring in the cabinet by providing an analog-to-digital converter and aggregator close to the detectors or load switches that are the source or destination of the inputs or outputs. Then a simple serial communications cable connects these bus interface units to the controller. Each bus interface unit supports multiple detectors or load switches.

A controller built to the physical requirements of the NEMA TS 2 standard is typically referred to as a NEMA controller. It is intended to operate in a "NEMA" cabinet meeting the NEMA TS 2 specifications, and can use either the A, B, C connectors (often called the TS 1 interface), or serial bus interface (often called the TS 2 serial interface) for cabinet inputs and outputs.

For actuated traffic signal controllers, the TS 2 standard defines functionality, primarily in the following areas:

• Phases arranged in a particular sequence in rings with barriers
• Overlaps (green outputs that can span multiple phases)
• Single and dual entry logic (what phase to select in the second ring if no call there)
• Pedestrian recycle (allowing pedestrian Walk to start other than at the start of green)
• Phase intervals and their timing (including minimum and maximum green times, yellow clearance, red clearance, and pedestrian timing)
• Coordination timing (cycle, offset, split, permissive period, time base)
• Phase selection points (when "phase next" is selected)
• Phase call storage (locking calls)
• User-specified vehicle and pedestrian recalls
• Automatic recall at forced phase termination
• Conditional re-service of a phase within a barrier group
• Simultaneous gap out
• Start up process
• Red revert time
• Preemption
• Flashing operation, dimming, diagnostics
• Remote communications (including NTCIP requirements)

The same functionality applies to NEMA controllers using either of the cabinet input / output interfaces (A, B, C connectors or serial bus).

The Advanced Transportation Controller family of standards is maintained by a consortium composed of NEMA, ITE, and AASHTO. Two standards are currently in place:

• The Advanced Transportation Controller 2070 (ATC 2070)
• The ITS Cabinet for ATCs (9)

The ATC 2070 standard (10) is based on the Caltrans Model 2070 controller specification (11) (12) (13) (14). Unlike the NEMA TS 2 standard, the ATC 2070 standard specifies every detail of the controller hardware and internal sub-components, but does not specify any application software functionality It requires the OS-9 operating system, a minimum of 4 MB of dynamic random access memory (RAM), 512 KB of static RAM, and 4 MB of flash memory. It also specifies the form and function of the following modules and a standard chassis and card cage into which card modules from any manufacturer can be inserted:

• Power supply
• Central processor unit module
• Field input / output interface module
• FSK modem module
• RS232 serial ports module
• Fiber transceiver module
• Front panel (user interface)

In addition to the standard modules, some manufacturers offer proprietary communications modules such as Ethernet switches, and a VME card carrier, that plug into the controller's standard card cage. The original Model 2070 specification included provision for an auxiliary five-card 3U VME cage within the chassis with the central processor being on a VME card. This option is retained in the ATC 2070 specification, but has not proven popular. The VME cage and processor is rarely specified or supplied. A controller without the VME cage is often distinguished as a "2070 lite," and has its central processor located on a module in the main 2070 card cage.

Anyone can develop software for an ATC controller, for any purpose (e.g., traffic signal control, field master unit, ramp metering, count stations, dynamic message sign control, reversible lane control, etc.) knowing that it will operate on controllers from any manufacturer. Most ATC controller software for traffic signals adheres to the functionality specified in NEMA TS 2, and is functionally similar to a NEMA controller.

The ATC 2070 standard includes options for input / output interfaces that enable its use in any of the four standard traffic signal cabinets - TS 1, TS 2 serial, ITS Cabinet, and Caltrans Model 33x cabinet. The TS 1 cabinet input / output interface module includes a standardized fourth connector, called the D connector.

The ITS Cabinet standard (10) combines the best features of the Caltrans Model 33x cabinet and the NEMA TS 2 serial cabinet, while providing for additional inputs and outputs, more distributed and flexible fault monitoring, and reduced cabinet wiring. It is a rack cabinet, with optional sizes, one or two racks, and doors in both front and back. The standard includes specifications for all cabinet components except the controller, detector cards, and load switches. It can be used with the ATC 2070 controller and TS 2 detector cards and load switches.

Instead of a single Malfunction Management Unit, the ITS Cabinet standard calls for a Conflict Monitor Unit and multiple Auxiliary Monitor Units - one in each input or output rack. Instead of a Bus Interface Unit, it calls for a Serial Interface Unit that integrates the serial interface into the input or output connector, and uses a different protocol than that used in the BIU. This protocol is the same as used internally in the ATC 2070. It is a new standard and it will take some time before compliant components are readily available and large numbers of ITS cabinets are deployed. ATC 2070 controller software needs some modification to operate in an ITS Cabinet.

The ATC standards working group is developing additional controller standards that will give more flexibility for both controller hardware and software. A new version of the ATC controller will allow the use of different physical forms, different central processing units, and perhaps different operating systems. Additional communications ports and memory are also planned. An application program interface standard will facilitate the portability of software applications between controllers using different processors and operating systems, and will allow sharing of system resources between multiple applications (from different suppliers) operating simultaneously on the same controller.

Specifications jointly developed by the states of California and New York describe the Model 170 family of traffic control components (11). These standards cover the hardware for cabinets and all components, including the controller. As with the ATC standards, the Model 170 specifications do not specify software functionality. These specifications date back to the 1970s. The Model 170 controller is based on the Motorola 6800 processor, which is no longer manufactured. Processing power and memory are severely limited and software written for the Model 170 controller cannot be readily expanded to add features such as support for more than 8 phases and two rings, or full NTCIP communications.

The Model 170 controller is widely used and will continue to be used for some time to come. As replacement parts are no longer manufactured for some components, they will have to eventually be replaced. Caltrans developed the Model 2070 controller as its replacement.

The Model 33x cabinets used with the Model 170 controller are supported by an optional Model 170 style field input / output module in the ATC 2070 standard, and it is therefore relatively easy to replace a Model 170 controller with an ATC 2070. However, Model 170 software does not automatically run on an ATC 2070.

Some manufacturers provide variations of the Model 170 controller which include:

• Improved front panel user interface,
• More capable central processor, and
• Additional memory.

Although not standardized, such enhancements provide another means of prolonging the life of the Model 170 family.

The New York State Department of Transportation uses a similar controller, the Model 179 (16). Although using a somewhat more powerful microprocessor, the Model 179 has not achieved the same acceptance as the Model 170.

Controller Selection and Migration

The selection of controller and cabinet should be based on an analysis of the agency's requirements.

For typical applications, any of the three standard controller types - NEMA, ATC, Model 170 - is adequate. However, the Model 170 controller has limited capacity for supporting advanced software applications, such as full NTCIP support or use of more than eight phases in two rings. Obsolescence of the hardware also makes the Model 170 controller a poor choice for long term applications.

Traditionally, NEMA controllers have been made to operate only in NEMA cabinets, although the latest NEMA controllers will also operate in the ITS Cabinet. The ATC controller can be used in any type of cabinet, with the appropriate field input / output module, but NEMA controllers provide a more compact and simpler option in NEMA TS 1 cabinets. Agencies often have a preference for one type of cabinet based on factors such as training of field personnel, existing inventory of spare components, aesthetic considerations (mainly size of cabinet), and cabinet placement policies.

If an agency wants to use a small one-door cabinet (e.g., in a central business district), it needs to use a NEMA controller with a size and shape suited to that cabinet. If a large NEMA cabinet is used, either the ATC or NEMA controller may be suitable. If a rack-mount cabinet (e.g., Model 33x or ITS Cabinet) is preferred, then the ATC controller (or Model 170 if feasible) is needed.

Some manufacturers offer hybrid controllers that provide some of the features of a NEMA controller (e.g., small size and shelf-mount) and some of the features of an ATC 2070 (e.g., standard processor and operating system able to operate anyone's software, slots for ATC communications modules, and standard interfaces). Some manufacturers offer a small cabinet and integrated controller. This is often referred to as a CBD cabinet. Some such products are based on the ATC specifications but don't adhere to the ATC 2070 standard for physical size and modularity.

As more and more cabinets with traditional parallel wiring between the controller and cabinet inputs and outputs (NEMA TS 1 and Model 33x cabinets) are replaced with serial bus cabinets (NEMA TS 2 and ITS Cabinet) the distinction between NEMA and ATC controllers will be less significant. The latest NEMA and ATC controllers can operate in either of the standard serial bus cabinets, and allow the user to operate any software compatible with the ATC 2070.

The choice of software operating in the controller is often an overriding consideration. If the software that comes with a NEMA controller provides unique features that are needed, that controller may be the best choice. If that software is also available, or only available, for use on an ATC controller, then the ATC controller may be preferred. An ATC controller, and some NEMA controllers, can be purchased separately from its software, allowing more competitive procurement if a particular software package is needed.

Another consideration is the need for spare parts and user training to support different types of controllers and cabinets. It is usually preferable to limit the number of different controller and cabinet types in use.

An agency may wish to migrate from one type of controller to another, either as part of an upgrade program or to take advantage of benefits of a particular controller type. Most agencies cannot afford to perform a wholesale replacement of all controllers overnight, but do the changeover gradually.

Any consideration of controller replacement must take into account the existing cabinet and any changes planned or needed to the cabinet. If cabinets are being replaced for other reasons, this presents an opportunity to also replace the controller, and it may be appropriate to change to a different type of cabinet.

A NEMA controller generally cannot operate in a Model 33x cabinet designed for the Model 170 controller, and a Model 170 controller cannot operate in a NEMA cabinet (either TS-1 or TS-2 serial). However, an ATC can operate in any type of cabinet of sufficient size, if it has the appropriate interface module. An ATC that does not conform to the removable Field Input / Output module part of the ATC standard does not have the flexibility to be reconfigured to operate in a different parallel cabinet, but will usually include a serial port for use in a serial cabinet (e.g., NEMA TS-2 or ITS Cabinet).

Software written for the Model 170 controller will not operate on an ATC, and vice versa. Traditional NEMA controllers cannot operate software written for either the Model 170 or ATC. Therefore, a change between these types of controllers will invariably involve different software and user training for the new software.

It is common for an agency to have two types of cabinets or controllers in use at any point in time, as it migrates from one type to another. Most agencies try to avoid having more than two different types in use at the same time.

1. Tarnoff, P.J., and P.S. Parsonson. "Selecting Traffic Signal Control at Individual Intersections." National Cooperative Highway Research Program Report 233, Transportation Research Board, Washington, D.C., 1981.

2. Asante, S.A., S.A. Ardekani, and J.C. Williams. "Selection Criteria for Left-Turn Phasing, Indication Sequence and Auxiliary Sign." HPR Research Report 1256-IF, University of Texas at Arlington, Arlington, Texas, February 1993.

3. Oliver, M.B. "Guidelines for Audible Pedestrian Signals." Public Roads, pp. 33-38, September 1989.

4. Oliver, M.B., J.C. Fegan, and S.A. Ardekani. "Audible Pedestrian Signals-Current Practices and Future Needs." Institute of Transportation Engineers Journal, pp. 35-38, June 1990.

5. "Committee for the Removal of Architectural Barriers, Audible Pedestrian Traffic Signals for the Blind, Intersection Evaluation Procedures." San Diego Council Policy Number 200-16. San Diego, CA, May 1985.

6. "Traffic Controller Assemblies with NTCIP Requirements." NEMA Standards Publication TS2-03, National Electrical Manufacturers Association, 2003.

7. Haenel, H.E., A.H. Kosik, and B.G. Marsden. "Innovative Uses of Traffic Responsive Control in Texas." Presentation Paper, Engineering Foundation Conference, Henniker, New Hampshire, State Department of Highways and Public Transportation, Austin, TX, July 1983.

8. "Single Point Urban Interchange Design and Operations Analysis." National Cooperative Highway Research Program Report 345, Washington, DC, December 1991

. 9. "Intelligent Transportation Systems (ITS) Standard Specification for Roadside Cabinet (ITS Cabinet)." AASHTO, ITE, NEMA, Washington, DC, 2003.

10. "ATC 2070 - Advanced Transportation Controller (ATC) Standard for the Type 2070 Controller." AASHTO, ITE, NEMA, Washington, DC, 2001.

11. Quinlin, T. "Development of an Advanced Transportation Control Computer." CALTRANS Report.

12. "Model 2070 Advanced Transportation Management System Controller Concept Description, Final Draft." CALTRANS, August 2, 1993.

13. "Transportation Electrical Equipment Specifications." California Department of Transportation, October 1994.

14. Bullock, D., and C. Hendrickson. "Software for Advanced Traffic Controllers." Transportation Research Record 1408, Washington, DC, 1993.

15. "Traffic Signal Control Specifications." (as amended), California Department of Transportation, January 1989.

16. "Traffic Control Hardware Specifications." Division of Traffic Engineering and Safety, New York State Department of Transportation, Albany, NY, June 1990.