Saturday, May 15, 2010

Types of Traffic Signal Cabinets

Many people don't see the large boxes near the intersection that hold the signal equipment. There are several different flavors of signal control equipment. These are NEMA TS-1, NEMA TS-2 and the CalTrans TEES style control environments.

Originally, the cabinets had electro mechanical switches and a rotary drum. They worked by having the rum rotate, and cams would engage and disengage mechanical points that would lead to circuits running the greens, yellows, and reds. Many agencies still use these types of controller assemblies. If you are standing next to a controller cabinet, and you hear it whirring, and when the lights change, you hear a clunk sound from the box, the controller may well be one of these old electro mechanical systems.

In the early 1960's, the first electronic controllers started being fielded. Everyone seemed to be doing their own thing without respect to a standard. Control equipment was produced by Singer (the same Singer as sewing machines), and at times, major companies like IBM, Raytheon and others were trying to figure out how to make better machines.

In the 1970's, two basic standards were developed. The California Department of Transportation developed a stander with a bunch of other states, and produced the Traffic Engineering Electrical Specification (TEES). About the same time, the National Electric Manufacturers Association (NEMA) produced the TS-1 specification. In the 1980's, NEMA published the TS-2 specification. Each of these specifications has matured over time, with new, updated specifications for the equipment.

In the late 1990's, several different organizations updated a standard specification for electronic communications and controller operation that is known as the National Transportation Communications Infrastructure Protocol (NTCIP). This is supposed to provide a national standard for how all of the parts and pieces within a traffic signal and ITS system.


NEMA TS-1

NEMA TS-1 is one of the older forms of control environments. Essentially, everything is communicating via a series of large wire harnesses within the cabinet. The controller has multiple milspec cannon plugs on the front and it talks with the equipment via a dedicated pinout / wire combination to the specific equipment. This is a reliable form of signal system, with some caveats. First is that the controller tells the equipment what to do, but never gets a check back from the equipment. This means that the signal controller assumes that the signal is doing what it has been told to do. There is a conflict monitor that makes sure that the signal does not malfunction.


Another issue with this type of controller is that while there are three of the cannon plugs defined by the NEMA TS-1 standard, any additional connectors are not defined. This is important, since the fourth connector plug includes how extra detection inputs are mapped, how the emergency vehicle preemption inputs are mapped, and other factors.

The basic NEMA TS-1 spec has 16 defined detection inputs. Depending on the generation of the equipment, there may only be between two and twelve phases available, and as few as 8 detection inputs. Some of the more modern NEMA TS-1 controllers can handle up to 64 detection inputs, but only 24 are really available.

The reason why this is important, is that if a cabinet is configured for one particular brand of controller, and the engineer wants to change the controller to a different brand of controller, some care needs to be given as to how the controller cabinet needs to be rewired to accommodate the new controller's inputs and outputs. This is especially true if the controller's manufacturer has a really unique hardwire mapping in the controller. For example, the Traconex CJ-32 controller is a modified NEMA TS-1 controller that has 32 unique detection inputs, however, 16 of these detection inputs are remapped to the phase hold and phase omit input pins. As long as you trade one Traconex CJ-32 for another CJ-32, it is ok, but change out the CJ-32 for a straight CJ, or and Econolite ASC/2, then you have your detection inputs causing the signal to omit and hold vehicle movements. That may not be what you are intending.

There are modern NTCIP controllers that are configured to fit in a NEMA TS-1 cabinet environment, and allow the pin inputs to be reassigned by the programming.

NEMA TS-2

NEMA TS-2 cabinets essentially refer to a controller that has a serial communications system referred to as "SDLC". This is a fancy term for a RS-485 communications system that connects the controller, Malfunction Management Unit (a fancy conflict monitor), the detection inputs, and the load bay.

There are other devices that can be added to the SDLC communications, including a frame grabber. The frame grabber is a product produced and sold by ATSI, that monitors all of the serial communications inside the system, and when an error occurs, the frame grabber will allow the engineer or technician to diagnose what was going on in a relatively simple manner.

The SDLC communications allow all of the equipment to talk back and forth, ten times per second, to send an command, and get a response that the command was executed.

The back and forth communications allow the controller to make sure that the signal is working properly. The downside is that once in a while the SDLC communications skips a beat or two. Depending on what type of communications parameter was skipped by the communications system, this can either be logged, or actually send the signal into flash.

The Malfunction Management Unit for the NEMA TS-2 environment also monitors pedestrian WALK / Flashing DONT WALK, which the Conflict Monitor for the NEMA TS-1 environment does not.

The NEMA TS2 cabinet is usually capable of 64 unique detection inputs. The specification allows for up to 128, but I do not know of any manufacturer who has allowed for all 128 yet. 64 may sound like a lot of detection inputs, but when you start mixing and matching video, radar and loop inputs, you can eat up 64 detection inputs really fast.

One of the primary considerations that the traffic engineer needs to understand with the NEMA TS2-1 is that the detection comes in groups of 16 on a Bus Interface Unit (BIU). This means that if you are going to need 17 detection inputs for a TS2 video detection system, you need to reserve all 32 detection inputs on 2 BIU's, which reduces the number of loop channels available. Alternately, the engineer could use a TS1 video detection card, and then each camera would have 4 unique detection inputs.

In the picture above, the controller is a 2070 controller, configured to operate in a NEMA TS2-1 environment.

One of the primary advantages of the NEMA TS-2 cabinet is that the internal communications are pretty standard. This allows for a very easy swap out of one brand / make / model of a controller for another. There should be very limited wiring differences between manufacturers. One of the few that may need to be addressed is if the signal is operating with FSK communications, the connector for the inter-cabinet communication may need to be modified. Where with the TS-1 cabinet, there are approximately 65 wires that have to be modified on the D connector, and if the controller uses another connector, maybe another 35 or so on the fifth connector.

Since the TS-2 cabinet talks via the serial RS-485 communications bus, it is very easy to swap out equipment.

Hybrid NEMA TS2-2

There is a hybrid NEMA TS2 cabinet where the load bay runs off cannon plugs and the detection inputs run off the SDLC communications. This can be handy if you have an older cabinet, and just want to add a video detection system without messing around with the cabinet wiring.

Some agencies also use this type of configuration because they want hardwired outputs from the controller but want the flexibility of the SDLC detection.

332 (CalTrans TEES)


The CalTrans TEES cabinet is a family of cabinets that are sometimes erroneously referred to as "332" cabinets. A 332 cabinet is one of the CalTrans style of cabinets. There are a bunch of different types of cabinets, within the family.

The following photo is a modified 332 cabinet.

In general, most 332 style cabinets are specifically wired to allow 26 detection inputs, with up to 46 detention inputs. This is done via a standard wiring harness and bridging of the detection inputs so that the there is a very standard pattern of where the detection comes in, and what the controller sees.

This type of standard wiring can be modified, and has in some cases, to allow each detection channel on the input files to have a unique detection channel on the controller. This can allow 46 unique vehicle detection channels, 4 emergency vehicle preemption channels, 4 pedestrian isolator switches and 2 channels of railroad preemption just by reassigning the wires on the existing C1 plug, assigning new inputs to the C11 plug, removing the bridging of the detection channels, and using a 2070 controller with an NTCIP style controller..

The advantage to the 332 style cabinet is that they are simple, robust, and cheap. They are essentially clones. The engineer is held by the rigid control of the 19 inch rack mount structure, and the limited choices that system holds. If anyone wants to arguer, please try putting a fiber distribution unit, a couple of fiber switches, a video detection system, a video encoder and some DIN rail mounted radar units into a standard single wide cabinet. It gets really tight.

The CalTrans Tees type of cabinets can be configures as a double wide cabinet, which does provide a very nice solution for having the traffic signal equipment on one side, and the Intelligent Transportation Systems gear on the other.

Blah, blah, blah. What does that mean? Essentially, the 332 style of cabinet can be a very powerful device.

There are some unique aspects to running a 332 cabinet that the engineer and tech need to know about.
  • The cabinet will run normally with the monitor removed and the door open. The NEMA TS-1 and NEMA TS-2 cabinets should drop to all-red flash if the monitor is removed. This can be a problem on the 332 style cabinets, if you remove the monitor, and don't secure the door open. If the door is unsecured with the monitor out, and opens and shuts, the drivers will see the signal go from normal operation to all-red flash, back to normal operation and flash every time the door closes and opens.
  • Putting the cabinet into all-red flash does not apply stop time to the controller. This means that if the cabinet is put into flash, then out of flash, the controller has continued to cycle away. If the cabinet is put into all-red flash, then taken out a few seconds later, the signal may have very different indications lit than what was operating when it went into flash.
  • Many 332 style cabinets to not have a hardwired stop time switch. Some controllers have the ability to assign the AUX switch on the front of the controller as a stop time switch.
  • The cabinet may be wired with a detection display panel. Many agencies have the display panel configured such that there are a bunch of detection switches, and if the switch is broken, or is not flipped to the correct position, the detection amplifier will place calls, but the switch will prevent the call from going into the controller.
  • If you have the cabinet set to normal operation (not switched into all-red flash), and you turn off the controller power, or remove the controller, all of the signal indications will go dark. To remedy this, manually set the cabinet into into FLASH and it should begin all-red flash.



Clearance times

Change and clearance interval are what traffic engineers call the yellow and all-red times.

There are a lot of different thoughts about how to set the yellow and all red times. This posting only addresses how they are done. There is a lot of controversy about yellow and all-red times with regard to red light cameras.

Most states adopt the Manual of Uniform Traffic Control Devices (MUTCD) as the standard for signs, markings, signals, and so forth. Most states also have laws that define how the requirements of the MUTCD are modifies for that state. The 2003 MUTCD states the following regarding the yellow and all red.

Section 4D.10 Yellow Change and Red Clearance Intervals
Standard:
A yellow signal indication shall be displayed following every CIRCULAR GREEN or GREEN ARROW signal indication.
The exclusive function of the yellow change interval shall be to warn traffic of an impending change in the right-of-way assignment.
The duration of a yellow change interval shall be predetermined.
Guidance:
A yellow change interval should have a duration of approximately 3 to 6 seconds. The longer intervals should be reserved for use on approaches with higher speeds.
Option:
The yellow change interval may be followed by a red clearance interval to provide additional time before conflicting traffic movements, including pedestrians, are released.
Standard:
The duration of a red clearance interval shall be predetermined.
Guidance:
A red clearance interval should have a duration not exceeding 6 seconds.

This leaves a lot of open area for interpretation.

Yellow Change Interval

Some agencies have a set the yellow times as a fixed time, say 3 seconds. Some agencies do some field adjustments on the settings.

Many engineers follow the Institute of Transportation Engineers kinematic model. This considers the driver's perception-reaction time, prevailing speed of traffic, grade of the road and other factors, and calculates a yellow time. The kinematic model can mathematically calculate the yellow at less than 3.0 seconds.

Most traffic signal controllers have a default setting that will not allow less than 3.0 seconds of programmed yellow time. Most conflict monitors (and malfunction management units) will send the signal into flash if the signal provides less than 2.7 seconds of yellow time. Some controllers and CMU/MMU's can be programmed to defeat this minimum setting for the yellow. I have worked on well over 100 different traffic signals, and have never had a reason to reduce the yellow below 3.0 seconds.

All Red Clearance Interval

All red rimes can vary widely by agency. Some agencies use zero seconds everywhere by policy. Some agencies use one second everywhere by policy. Some agencies use a calculated value.

The ITE kinematic model calculates the all red time with respect to the width of the intersection, speed of traffic, and the average length of a vehicle. In many states, it is legal to enter the intersection during the yellow. The red time calculation assumes that an average length car (typically 20 feet) enters in the last instance of the yellow time, and calculates the time required to have the rear bumper clear the conflict area at the posted or prevailing speed.

This may be excessive when the signal has timed out, and the signal has no one in the intersection during the all-red time, but it is near impossible to have fail safe detection, where the lack of vehicles can be sensed.

There are a couple of controllers that have the ability to provide variable all-red time. These make assumptions as to how the controller has terminated the phase (Gap Out vs. Max / Force Off) and applies an extra amount of all red. The problem with this is it is unpredictable and will not work while in coordination.

There is also a method using external logic circuits (and, or, set / reset flip flop cards) to have special vehicle detection apply an Omit All-Red by Ring. This can work in coordination, however if the detector fails, the signal will default to never provide any amount of all-red time. The applications of these that I have REMOVED from service have been buggy.

So why put in yellow and all-red time?

It is done for safety. It may be aggravating to wait 4, 5, or 6 extra seconds, but they are provided to let the drivers know that the signal is changing and they have to decide to stop, or to go through the signal, and then clear the signal before the next phase is given a green.

Yellow and all-red times can provide an interesting view into the human mind. The 5 or 6 seconds that you wait seem like an absolute eternity when you are waiting. On the flip side, the same time may seem really short if you are rolling through the intersection. The same 5 or 6 seconds will seem like an instant flash of time when you are doing something enjoyable.

Friday, May 14, 2010

Basic traffic signal operation Pretimed Operation

Pretimed operation is the most basic type of operation. In this mode of operation, the signal will alternate the green, yellow and all-red time without regard to the traffic flow.


A signal may have one, some, or all phases on pre timed mode. For many downtown grids, most of the signals are on pre timed mode, and in many cases, the signals are in pedestrian recall.


Some signals are temporarily essentially set to pre timed mode if the vehicle detection fails.


The amount of time for each movement may be programmed to change by time of day.


The minimum amount of time for each phase should equal the sum of WALK, Flashing DONT WALK, Yellow and Al-Red. In some cases, some agencies allow the flashing DONT WALK to extend through the Yellow time. This can shave a few seconds of time off the cycle length of the signal.


Pretimed operation with the signals in pedestrian recall can be a very effective way of dealing with a grid network of signals. Many agencies don't put pedestrian pushbuttons in downtown grid signal networks, as the typical pedestrian volumes would have the signals constantly serving pedestrians anyway.

A pretimed signal operates on a user specified amount of time for each specific movement. For instance, if the signal is a 2-phase signal (2 specific signalized movements), like at the intersection of two one-way streets, and is operating in pretimed mode, the signal will serve one movement for a specified amount of time, followed by the second movement.

In this example, the first movement may be programmed for 20 seconds of green, followed by 3 seconds of yellow, followed by 3 seconds of all-red. The second movement would then commence, programmed for 30 seconds of green, then 3 seconds of yellow and 3 seconds of all-red. The total cycle time would be 62 seconds (20+3+3+30+3+3=62).

There are advantages, and disadvantages to this type of operation. The advantage is, it is pretty straightforward to set up, and the specific amount of green assigned to each phase (unique signalized movement) can be varied by time of day.

The variation by time of day allows for the cycle length to be adjusted to lengthen when necessary, and shorten when necessary.

The downside to this type of operation is that the signal is essentially a dumb box, without any way for it to modify operations by what is actually happening with the traffic. If the second movement has no cars, why serve it? If the 2nd movement has only 1 car waiting, why serve it for the entire time?

The fact of the matter is that with some situations, pretimed works pretty well, like in a downtown central business district grid. It also is a lot cheaper than installing vehicle or pedestrian detection systems in the intersection. It is also a pretty good way to deal with failed detection, as a temporary measure until the detection can be fixed.

It is possible that specific movements are timed, and other movements have vehicle detection, which would make that type of a signal semi-actuated.

Terminology

All professions have technical jargon. Traffic engineering is no exception.

This post will be a running commentary with an attempt to define some of the technobabble that will be in this blog. I will attempt to keep it in some sort of order... Maybe grouped in some way, maybe alphabetical... No promises.

DEFINITIONS

MAX (Max1, Max2, Max3) - The max time is the maximum amount of time that the signal will serve a phase after an opposing call is served.  For example, if a signal has served the MIN green time and is resting green on the main street, and there is constant traffic on the main street, the MAX timer starts timing after the first conflicting movement has a valid vehicle call.  If the traffic on the main street reduces, and the signal gaps out before the MAX timer times down to zero, then the signal will gap out and serve the side street.  This is similar, but different than forcing off in coordinated mode.
MAX (Dynamic) - NTCIP controllers have a dynamic max ability.  This allows the controller to be set with a short max (20 or 30 seconds) for each movement, but also with a higher dynamic max time and a step interval.  Essentially, if the traffic is low volumes, the signal will gap out, and the normal MAX time will apply.  As the traffic volumes increase, some of the phases of the signal will begin maxing out, and the signal will increase each phase's MAX time by the step interval up to the Dynamic Max time.  As traffic volumes decrease, the Dynamic Max time will reduce automatically.  This means that the main street could have a 20 second MAX time, but a 70 second Dynamic Max Time with a 10 second step interval.  This allows the signal to breathe for the uncoordinated cycle time based on the past 2 to 10 minutes of actual use.
Overlap - an overlap is a special type of signal operation that is allowed to operate across one or more vehicle phases. There are a lot of uses for overlaps to be described later
Phase - a phase is a distinct movement at a signal. Generally, the through movements are each given a phase number that is even. Generally, the protected lefts are given a phase number that is odd.
Recall - in general, all of the recall types can be assigned to any or all of the vehicle phases. When a signal has one or more phases in some form of recall, the signal will cycle between the movements with the recalls. Many signals are programmed to have the signal rest on the main street by putting the main street thru movements in min recall.
Recall (max) - this is where the phase is constantly being serviced for the maximum amount of time, either MAX1, MAX2, MAX3, Dynamic Max, the coordinated operation split time, or for a coordinated phase, the currently available bonus time plus split division time.
Recall (min) - this is where the signal will serve the minimum green time programmed, plus the variable initial green time.
Recall (pedestrian) - where the signal automatically will serve the pedestrians each cycle. This may be for any, or all phases.
Recall (soft) - Soft recall is not used very often. Essentially, this mode of recall allows the signal to cycle to different movements when there are no active vehicle calls on the movements in soft recall. This can be useful for low volume operations, to allow the signal to cycle between some movements before going back to the main street. The signal needs to have good stop bar detection in order to make this work. Care needs to be taken to make sure that the signal can not create a yellow trap with the soft recall settings..

First post



Welcome to my blog. I will be discussing how traffic signals work, why they do what they do, and why you may be getting more reds than greens.

Traffic signals can be very complicated devices. Traffic signals may be pre timed, semi actuated, fully actuated, coordinated, uncoordinated or running with an adaptive algorithm. The vehicles can be undetected, Or detected by a variety of methods. Vehicle detection can be accomplished via pressure plates (very rare now), or detected by a variety of inductive loop wires, video detection and radar detection.

There are a variety of brands, makes, models, software versions and standards. I have extensive experience with Naztec Apogee, Econolite ASC/3, ASC/2 and 2/S and 8000, Traconex J-9, CJ and CJ-32, Peek 3000 and 9200, and LMD 8000 controllers. I also have worked on Eagle Epac, Wapiti 170's and VS-Plus controllers. These range between NTCIP, NEMA TS-1, NEMA TS-2, and CalTrans TEES Control environment.

What does all this mean? Well, the devil is in the details.

The photo at the top of this page is of a NEMA TS2 Type 1 traffic signal cabinet.  The cabinet is a pole mount cabinet, and is quite full of hardware.  The signal cabinet has an Econolite ASC/2S-2100 controller, with Reno 1600G Malfunction Management Unit, and Reno E1200 induction loop detectors. Not shown in the controller is an ATSI Frame Grabber - which was installed after the picture was taken. 
This cabinet is configured to operate with up to 16 load switches, for a total of 16 specific movements, which can be programmed for any combination of vehicle, pedestrian and overlap movements totaling up to 16.