User:Bahammond55/sandbox

TWACS®
This page is intended to provide a history of the TWACS Power Line Communcation Technology

The TWACS communication technology was invented around 1972, and has kept many people gainfully employed for over 50 years.

The core technology has a data rate of a 15 bits per seconds on a 60 Hz power system.

The Original Invention of the TWACS®  Technology by Arthur D. Little Corp

TWACS® started with a Request for Proposal (RFP) by New England Power Services Company (NEPSCO) in 1972. They were interested in determining if any technology available at the time was adequate to perform Automatic Meter Reading over the AC power line. They had heard that such systems could improve responsiveness to customer inquiries and potentially reduce costs. They issued an RFP to investigate these possibilities.

Arthur D. Little Corp. (ADL) was one of the firms that received the RFP. ADL discussed the inquiry internally but, just three days before it was due,  decided not to respond to the RFP. However, an ADL researcher had some ideas that he wanted to investigate and intervened, and assembled a team of four to respond. They won the assignment.

Many shortcomings were recognized in the various schemes that were proposed. It seemed particularly difficult to have a cost effective system that would reach all of the customers. However, one concept that prevailed was the thought that the 60 Hz waveform reached every customer quite well, and implicit in the sine wave is the fact that there is no voltage on the line 120 times every second. Could these zero-crossings provide an opportunity for communication?

Quite a few advances were being made in the electronics industry at this time. The microprocessor had been invented in 1971. Large SCRs had been developed for switching power, and by this time, small inexpensive SCRs were widely used for dimming lights. Could these commonly available SCRs be used for communication? The team at ADL bounced ideas around. Could the SCRs create a waveform near the zero cross that would be suitable for communication? Could the SCRs handle large pulses of current for a few milliseconds provided their half-cycle ratings were not violated? (The device specifications didn’t specify this use of the product, and the manufacturer couldn’t answer that question without further testing.) If large, power-switching SCRs were used at the substation transformer, would there be a way to perturb the shape of the voltage sine wave? Could the SCR be used to inject a small modulating signal in series with the source voltage, and with it cause a small phase shift of the 60 Hz power voltage? If so, this could provide a means of communicating with customer equipment served by that substation.

These were difficult questions to answer at the time. The early 1970’s didn’t have the modeling and simulation tools that are available today. As the engineers at ADL studied their proposal, what started as a paper study quickly spilled over into an evaluation in the lab. Within three days of starting the study, a bank of light bulbs was flashing under the control of inexpensive SCRs. These tests answered some of the basic questions, but to really test the communication concept, further study would be required.

ADL submitted their report to NEPSCO, along with a proposal for further study. With this report in 1972, one could say that the idea for a Two-Way Automatic Communication System (TWACS) was born – even though it was perhaps just a “twinkle” in the eyes of the inventors.

The proposal by ADL for further study was accepted by NEPSCO. The first test was to simulate a feeder in the lab. Three four-wire, wye-connected  transformers were put under an electric load to simulate a feeder, and a fourth phase-shifting transformer was connected to this feeder along with a large SCR on its secondary. This provided a means for the utility to signal an “outbound” to all of the customers served by the substation bus. Similarly, at the “customer” end of the line, it was found that the firing of “cheap SCRs” created a current pulse that could be detected at the feeder neutral. Thus, an “inbound” message could be created from the customer end of the line into the “substation.” This test in 1974 established that voltage modulation could be used to communicate an outbound message (from the utility substation to the customer site) and current modulation could be used to communicate an inbound message (from the customer site to the utility substation). The hope was that if this were used in the field, it could communicate long distances without repeaters, and without bypass components. The next step would be to verify these results in the field.

In the 1976 time frame, a feeder at the West Methuen substation of Massachusetts Electric was chosen for field testing the concept. Positive results in the field kept the project moving to the next stage which was prototyping products. Further testing during this stage found a curious problem which came very close to causing the termination of the entire project. But with intervention once again from the original researcher, further analysis showed that the difficulty stemmed from the symbol encoding scheme. Unipolar symbols drove the modulation transformers towards saturation of their magnetic cores. This caused signals to not propagate very well. The ultimate remedy was to utilize bipolar symbol encoding. A Hadamard matrix provided the needed symmetry as well as the opportunity for multiple, orthogonal channels of communication. With an inbound symbol length of N=8 pulses, it provided N-2=6 independent channels of communication.

Figure 1 -- Portion of a typical TWACS inbound symbol (current waveform in red)

As thoughts of commercialization continued in the late 70’s, additional partners that could offer the financial backing necessary to bring the concept to the marketplace were sought out. Emerson Electric joined the group as the main partner, and the TWACS product was introduced to the marketplace in the Fall of 1980 in Emerson’s name.

Improvements were made in the product as the prototype moved towards commercialization. A considerable amount of discrete logic was eliminated from the design by the introduction of a microprocessor. For the AMR product, the introduction of non-volatile memory allowed the kWh pulse-count to be retained in the event of a power outage. A number of pilots were started as utilities tested the new technology.

As with many technologies, some problems were found during the commercialization phase. The outbound “series injection” technique originally developed by ADL was found to be costly to deploy and created power quality and light flicker issues at customer sites. The technique attempted to shift the location of the zero cross. In the 1983 timeframe, the Emerson team developed a simpler technique which merely changed the slope of the curve as the waveform approached the zero cross. This change was found to be much easier to implement and eliminated the issues found with light flicker and power quality.

The dissemination of the microprocessor allowed endpoints to be developed that served the AMR application that was originally envisioned, but the size of the numerous components would still not allow a solution that would fit under the factory-original glass of the revenue meter. This made it difficult to sell the volumes of AMR solutions that the company had hoped to sell.

In the 1980’s Florida Power & Light (FPL) began asking questions similar to NEPSCO, but FPL was interested in developing a load control system instead of an AMR system. FPL uses oil-fired power plants to meet the peak load during the summer time. The oil embargo of the 70’s caused FPL to search for alternatives to burning oil. The TWACS technology was found to be a perfect fit for load control. The power line communication technology worked wonderfully to broadcast messages from the utility substation and reach devices at the customer premises. The voltage sine wave from the substation travelled to all points downstream, and the TWACS outbound waveform rode along as part of that sine wave. The only challenge was the software. The TWACS master station had not yet been developed into a mature product with the robust features needed by a large investor-owned utility. So FPL contracted with Emerson to purchase the TWACS field equipment, and they contracted with Control Data Corporation for the development of the master station, with FPL subject-matter experts being closely involved in the detailed specification of all the products.

The sole focus of the TWACS effort during the mid ‘80s was the development of the load control product for FPL. The first units were installed in 1987, and over time, FPL’s load control program grew to become the largest load control program in the world. FPL has built out the system to contain more than 815k endpoints served by 460 substations. (These devices can disconnect hot water heaters, pool pumps, and other dispatchable load. In return, customers who sign up for this program receive a discount for energy, along with promises on the maximum number of times that load will be shed.) FPL’s load control program can control over 1,000 MW of power in normal operation, and 2,000 MW of power in an emergency situation.

Despite the success of the program, changes at Emerson Electric occurred. Emerson’s management team decided to change their product focus. A number of divisions (including Emerson’s TWACS division which at the time was formally known as “Chance Load Management System”) were spun off. In 1990 a holding company called “ESCO” was formed. The division carrying TWACS was renamed “Distribution Control Systems, Inc.”

The company continued to pursue both load control and AMR opportunities and make improvements in the product lines. In the late 90’s a transponder was developed that fit under the original glass of the meter. To make it happen, a number of the original product requirements had to be revisited. Even though a TWACS inbound pulse is only a few milliseconds long, was it really necessary to signal inbound at 35 Amps? Was it really necessary to use an inductive firing element, or could a resistive firing element work just as well? The introduction of an adjustable firing angle allowed the inbound signaling to be reduced to 17 Amps, the use of a “small” resistive firing element to signal inbound messages, and improvements in the search algorithm at the substation allowed the messaging capability of the product to be safely trimmed back and yet still meet the AMR market requirements. This greatly reduced both the size and cost of the solution.

System Eveolutiuon

TWACS-0

The first generation of the system was known as TWACS-0.

TWACS-5

The second generation of the system was know as TWACS-5

TWACS-10

This system was developed around 1987. as a replacement for the T-5 system as a result of a request from Florida Power and Light.

The substation equipment for this system was based upon a VME backplane architecture, with a numimum of 6 cards, SPA, SUA, CPA, CUA, RPA, and RAA. The SPA, CPA, and RPA each used the motorola 68000 CPU.

At that time the OMU was also re-designed using he motorola 68HC11 microprocessor.

Digital Inbound Receiver (CRPA)

Around 1989 two of the engineers at Chance Load Management Systems began researching using digital signal processing to improve the inbound performance of the TWACS system. One of the engineers, Dr. Sioe Mak wrote algorithms using basic programming language and came u with a concept that he referred to as "Pattern Recognition"  The other engineer developed algorithm using the DaDISP tools and programming in the "C Language.   The algorithms became known and the Correlation Detectors.

In the fall of 1990, a project was started to develop a prototype of a a receiver to implement and test the feaabilty of this. A Texas Instruments TMS320C30 DSP as chosen for this research. Over tthe next few months the algorithms was written and ported to run on the hardware, and then arounnd ??? the team rented a U-Haul truck and when to the Robertson Substation in Hazelwood, MO to test the new system.

The results of this testing were positive and showed that using these new techniques could improvement the sensitivigty of the receiver, and as a result, a new project was started in January 1991 to design a new receiver set (CRPA/CRAA) to replace the 68000 based reciever (RPA/RAA). The goal of this was for the new receiver set to be field replacement for the existing set.

Initially the concept was to have this new CRPA board have 2 CPU, a 68000 for the VME interface and the 'C30 DSP for the signal processing. But as the project developped it was clear that it would better to use only the 'C30, and that it could do all of the functions required. At the time of this development it was the common belief that DSP applications had to be written in assembly language. But the team was able to implement the algorithms almost entirely in ""C", and only reverted to assembler for a few critical functions.

One of the issues in the firmware team has to decige upon was 2 algorithms to use, Pattern Recognition or Correlation. The research did not provide a conclusive decision as to this, so it was decided to implement both. In the initial prototype it was determined that doing both would consume about 60% of the CPU cycles for the 'C30 running at 33 MHz. But as the firmware was developed it was found that this could be done with only 15% CPU utilization.

In the final implemention there were 4 decoders running simulataneous, 1) Pattern Recognition, 2)Correlation Rule, 3 Correlation Polarity,and 4) T-10 Emulation.  The last methos was intended to simulation the function of the T-10 RPA.  In many cased this was found to be the best method.

The firmware was impkemnente in such a way that the performance of each of the methods could be measured. this was originally intended to be used to device which methos to keep and witch to elimitate. Be siince all could run at the same time and there was plenty of CPU cycles available, they were all kept.

Field Testing at FPL

In April 1991, a team of 2 engineers traveled to FPL for the first field trials of the new receiver sets. Testing was done at a number of substation (Mallard. ???)

The test results were very good, and as a result FPL began purchasing these and upgrading there systems.

Bus Level Detection vs Feeder Level Detection

It was the goal of FPL to use bus level detection. This goal was mainly die to the cost of installation of feeder detection, but also due to the issue of "Searching".

The '93 CRU

In late 1991 and early 1992 a new project was started to design a replacment for the T-10 CRU. This goal was to cost reduce the SCE equipment. Much discusstion was had about how to do this and wherere to eliminate the VME bus. In the end the decision was made to keep the same basic architecture, and combine the SPA,SUA, CPA anmd CUA into a single board knwo as the SCPA. Also the interface to the OMU was redesigned and became the OFIA board.

The CRPA/CRAA were also changed, where the CRAA becasue the CRMA in the '93 CRU. One of the innovations for the CRPA, since it was very new was that it was able to detect if it was in a T-10 system or a '93 system and adapt to the system that it detected. This was very useful for FPL since they did not have to keep two sets of board.

Downloadable Firmware for SCE

In the '93 time frame the engineers has a vision of being able to download firmware the the SCE equipment, but the ...

One of the issues that has been experience at FPL was that problems that had been resolved in the firmware were not being deployed to the SCE equipment. As a result the same issues were being discussed as problems, which in fact had been corrected, sometimes years before.

This feature was finally implemented around 1998 or 1999?

ADLC

MCAA (Multi-Channel Analog Assembly)

In 1996 a project was started to design a repacment for the CRMA. This main gaol of this project was to allow a single recviever to process data from any of the 32 inputs, and do this concurrently.

The first application where this was needed was to speed up the search algorhtm. By being able to process inbound data streams from all phases and feeders concurrently it would be pssobel to search unitw much faster. Also, in the future, the concept of concurrent phase communication could be achieved by adding inbound receiver, or a more powerful receiver.

At the time, the company had been downsized making development of new products difficult, and also later in the project the main hardware designer deciced to lasve the company. As a result, the project last it momentum and was eventually cancelled, before it was completed.

A replacement for this was not done until around 2005. As a result of the failure to complete this project the advancement of the technology was much more difficult.

Concurrent Phase Communication

The TWAC system was originally only able toi communicaate on a single phase at a time.

Around 1999 a project began to add the feature to allow communication on all three phases at the same time. There were really two issues to be resolved. One was related to outbound signalling corrupting the inbound that was occurring at the same time. The other was>>>

It was known that on many systems, there was minimal outbound to inbound interference. So this would allow this to be doine withoufn solving the outbound cancellation problem, with the expectation that at some time in the future this would also need to be solved.

The initial proof of concept was done at Wisconsin Public Service (WPS) in 2000. This was done by installing two additional OMU's on one of the buses, and then configuring the substation bus as if there were 3 buses. This allowed the concurrent phase testing to be done without any firmware changes, but just to "trick" the system to run in parallel. This testing proved that this feature was technically feasible, and so the firmware project was started.

Adding this feature to the SCE only required firmware changes to the SCPA, but did also require adding two addition receiver sets to the substation, since a receiver was required for each inbound phase due to the limitation of the SCE equipment. (This problem would later be resolved using the MIRA). Since the SCE equipment used the VME bus, it was possible to add up to 4 receiver sets. Also the concept of IPU shadowing was invented. This allowed all three receiver to mapped to the same inputs, and only required the installation of some additional ribbon cables.

.

Key Contributors
 * Dr. Sioe Mak
 * Gordon Gregg
 * The authors: David Haynes, Gordon Gregg, and Sioe Mak; would like to thank all of those who contributed to the article. The information provided by Reed Johnston offered important insight into the conception of the idea and early development of the product. The interviews with (and reviews by) Bob Richardson, Bill Weber, Dennis Kelley, John Hessling, and Benjamin Hammond are also very much appreciated.