Hornby Zero 1
Zero 1, a Digital Command Control System
Hornby Railways Zero 1
- Return to the DCC History page
Hornby Railway's Zero 1 was one of the first command control systems to be implemented with Digital technology. It was designed and manufactured by Hornby in the United Kingdom. The Zero 1 master controller was built around a microprocessor. Even though it was discontinued after about ten years, the equipment was reportedly quite robust and could be easily repaired.
Zero 1 was developed by Hornby Railways, a major manufacturer of model railroad products, unlike many of the command control makers of the time. Prior to Zero 1, the only big brand name to be associated with command control was General Electric, with their ASTRAC system.
Development costs of the Zero 1 is said to have been about £300,000 (£1.7 million, or US$2.75 million in today's money (2012)), a substantial sum.
Introduced by the United Kingdom based manufacturer Hornby Railways in 1979, the United States in 1980, and Canada in 1981. It was a digital system based around a Texas Instruments TMS1000, a four bit microcontroller able to address and control up to 16 locomotives.
Development began in the mid 1970s. The name Zero 1 was chosen to communicate that it was a digital system. Hornby management had realized the declining price of digital electronics would soon make possible a digital command control system which could be manufactured and retailed at a reasonable price. They began the project and received technical assistance from Texas Instruments (TI), who made custom Integrated Circuits for the system. TI introduced the first microcontroller in 1971, which made a device like the Zero 1 Master Unit practical. Since everything was on one IC, design was greatly simplified. The IC contained RAM, ROM, a 4 bit microprocessor and other components of a computer all in one small package.
One design goal for the Zero 1 was a decoder that could be fitted into most of Hornby’s products.
The Zero 1 used a 32-bit code generated by the custom TI TMS1000 microcontroller, transmitted every third cycle of the square wave track voltage. For an 8.33-millisecond interval (60 Hz system), it replaced the track voltage. Because track power is off during transmission, the system was very resistant to interference. The code contains an identifying pulse, for up to 16 locos and 99 auxiliaries. Since Hornby was based in the United Kingdom, two different systems were made for 50 and 60 Hertz power systems which were not compatible.
The main system consisted of a master unit, R.950 (in the UK-240V 50Hz), Australian units were designated R.925 (also 240V/50Hz), US and Canadian units were designated R.944 (120V/60Hz), with up to three additional slaves (sold separately) which could be attached via a 15 pin edge connector. The throttles also offered inertia, and were capable of supplying up to 4 amps. The slaves made it easier to control multiple locomotives. Documents at the MERG Website show how one Master and 15 slaves can be daisy chained in order to achieve full independent control of 16 locomotives, each having its own controller. All sixteen engines could be in motion at the same time (if they didn't take more than 4A in total).
Upon power-up, the Master unit immediately assigned itself to address #1. Any slave units were assigned 2, 3, and 4, respectively. The master unit's keypad was used to assign decoder addresses for the slave controllers, or itself. Hornby also planned to offer the R.952 Hand Held Slave Controller, which could be used with the master/slave combination, limited to a single handheld controller, as the last in any daisy chain.
Other interesting features included user defined inertia. Since the system only had 16 speed steps, this feature made for more realistic operations. The use of a square wave, interrupted at regular intervals by data packets, resulted in noisy motor operation and poor low speed operations.
The Zero 1 system could operate 16 locomotives, and up to 99 accessory decoders. Due to the 4 bit microprocessor used, only 16 locomotives could be on the layout, or addresses would be duplicated. Any additional locomotives would have to be stored on an electrically dead track to prevent conflicts.
Decoder addresses were hard wired on the PCB. The user could change them, but disassembly of the locomotive was required to do that. By applying conductive paint in a specified manner, the address was set. (Soldering would damage the decoder.) A small vial of conductive paint was included with the decoder for this purpose. The early decoders (up to Revision C) employed small posts and wires that were used to create the connections for the address.
The Zero 1 decoder was known to be the smallest decoder on the market, making it easy to install in many locomotives. it retailed for about £15, about £60 (US $97) today. Even at that price, it cost almost as much as a locomotive.
Three wires were used, two connected to the track, and one to the motor. Three wires made installation very easy, as the motor did not need to be isolated from the frame. This would make installing the decoder in a split-frame locomotive (very typical of Hornby designs) easier by eliminating the need to isolate the motor. By making one brush more positive (or negative) than the other, direction of travel was established. Due to the thyristor circuit used to control the motor, the system could be very erratic. The UK (50Hz) version only used a single triac, compared to the two shown in the picture of a 60Hz North American receiver (The two triacs appeared on decoders made after 1980).
The early single Triac decoders were failure prone, and sensitive to noise. The two Triac type were originally marketed by Hammond and Morgan, which was later taken over by Hornby. Hornby adopted the twin triac decoder in 1981. The H&M digital controller was compatible with Zero 1.
Hornby announced the Revision D R.965 decoder (or Locomotive Module, as they called it) in 1981, which incorporated the conductive paint method of setting the address. The address could be changed by scraping off the paint.
Data was transmitted using FSK (frequency shift keying).
Data was transmitted every third pulse. The data frame consisted of a number of 4 bit nibbles. The first nibble, START, had an identification number to identify it. For accessory decoders the START nibble was 0100, for a locomotive it would be 01XX. After START, four more nibbles of data were sent, each being a locomotive instruction. Four transmissions were required to send instructions to all 16 locomotives. After the locomotive data was sent, the next four bits set the direction of each locomotive. A logical 1 meant forward, zero, reverse. Each locomotive had a bit in that nibble for direction. Then a PARITY nibble to insure data integrity. The final data sent was a STOP nibble.
Speeds were determined by a number from 0001 (stopped) to 1111 (full throttle). The value of 0000 was not allowed. If nibbles 2 and 3 were "0000 0000", this frame was identified as one for accessory control. The next two nibbles (4 and 5) were the device address of the accessory decoder. Nibble 6 (Direction) used only the most significant bit to set the state of the accessory.
Locomotive data was sent sequentially, where accessory decoder data was only sent when it was entered on the controller. That way dirty track would not cause problems as a locomotive instruction would be sent repeatedly.
The PARITY nibble was calculated by the master controller.
Example Locomotive Frame:
The START Nibble would be 0100, 0101, 0110, or 0111. Each 34 bit data frame contained the speed and direction information for four locomotives. Four complete cycles would be needed to send all 16 locomotives their instructions. This process was repeated to insure that locomotives had their instructions should a problem occur that interrupted data transmission. Accessory decoder data was only sent once, when the button was pressed, since intermittent data losses were less likely to occur.
For SPEED, a value of 0000 was invalid, as the accessory decoder frame used 0100 0000 0000 as the first three nibbles to identify it.
Example Accessory Decoder Frame:
Note that the first two nibbles after the START are each set to 0000, indicating that this frame of data was for the accessory decoders. The locomotive speed could never be less than 0001 as the 0000 value was reserved for the accessory decoder frame.
The decoder was built from serveral components. The main component was a micro controller, with the address set using four lines. The track signal was supplied to it, and it used a Gate Control output to switch on the two triacs that controlled the motor. A few transistors, diodes and some resistors finished off the parts list.
The triacs were Silicon Controlled Rectifiers, which can be switched on, and will stay in that state until the polarity on the terminals is reversed. GE's ASTRAC also employed SCRs to control the motor. Speed is controlled by when the SCR is switched on, a full cycle for full speed, and a partial cycle for lower speeds. As the track power was a 20VAC signal, the SCRs would be switched off 60 times a second (50 times in the UK.)
Track power was a square wave, at about +/-20VAC. Track power came in three phases: Forward, Reverse, and Data. By controlling which of the two TRIACS mounted on the PCB was 'on', and for how long, the motor speed and direction was determined. Data was transmitted on every third cycle. No locomotive without the Zero 1 receiver installed could be on the layout. The constant 20VAC track power would be harmful. The square wave was chosen for better low speed operation. A sine wave made low speed control difficult as the voltage was always changing. It was not possible to use a Zero 1 equipped locomotive on a DC layout, but it was possible to wire a switch into the circuit to bypass the decoder.
Track power was transmitted at 50 or 60 Hertz, and data was transmitted at 6.6 or 13.3KHz.
Hornby also sold wire clips, or Points De-Isolating Clips (M1185) to make switches "Zero 1 Friendly". ( Their version of a DCC Friendly Turnout.)
Double heading required both locos facing the same direction. The master controller merged the two locomotives into one, and was not able to determine direction individually.
A Hornby Railways advertisement in the March 1981 issue of Model Railroader listed the American prices for a Zero 1 system. The R944 Master Controller unit was $149.95, and the slaves were $49.95. A points/accessory module listed for $49.95, and a loco module was $24.95. (All prices in US dollars.) A later accessory was the Micro Mimic panel which could represent turnout positions using LEDs.
Hornby pricing in Canada was $219.95 for the R944 Master Controller, the R945 Slave Controller was $69.95, and the R947 "Loco Module" listed for $32.95. An R946 Accessory decoder was $69.95, which could control up to 4 switch machines. Exchange rates between the UK pound and the Canadian dollar/US dollar would explain the price differences.
As the Hornby brand was not a well known brand name in the United States, Hornby Railways had to invest a lot of money marketing the Zero 1 system in the US. Hornby Railways always had a larger presence in Canada, sold in major retailers such as Eatons, but delayed the rollout until 1981, a year after the US introduction. It was introduced in the UK in 1979.
Hornby's parent company Meccano Ltd. had set up manufacturing operations in the US beginning in the 1920s, but poor sales of Meccano and Hornby products doomed the enterprise and by the late 1930s Meccano abandoned the US market. The main competition was Erector sets which were made by A.C. Gilbert, who claimed his product was not a copy of Meccano. A.C. Gilbert also made the American Flyer train sets.
Meccano continued to sell Hornby trains and Meccano sets in Canada, through small hobby shops and major retailers such as Eaton's department stores.
Over time, after acquisition and reorganizations by other companies, the Hornby brand would become the name of the company.
Hornby advertisements stated you could model a large passenger terminal with up to 16 locomotives in operation at once. Poor marketing may have limited the Zero 1's appeal in the North American market, although some model railroads did use it.
It was the most popular of all the command control systems in use by the mid 1980s. (Model Railroader reader surveys).
In today's (2017) dollars, the master unit would $425, a slave unit would be $142, and a decoder would cost $93.50.
In 1980 Dunbee-Combex-Marx, the company that owned Hornby Railways, was liquidated, placing Hornby Railways in receivership. This effectively ended any further development of the Zero 1 system.
Hornby would discontinue the main controller unit in 1986, and by 1991 their catalog warned that supplies were limited of the R955 locomotive module. Due to parts issues, Hornby was no longer supporting Zero 1 by this time.
Other Zero 1 Suppliers
Hammant and Morgan
H&M was a British supplier, known for their analog DC controllers. They also offered a Zero 1 compatible system. Hornby would eventually purchase the company. H&M's Zero 1 decoders were known to be better than the original Hornby units, eventually Hornby introduced an improved decoder that addressed the problems with the early design. The decoders were about 38 x 13 x 8 mm, and could supply up to one amp.
Hammant and Morgan sold their HM5000 Advanced Power Transmitter, with twin controls, and the consort controller (HM5500) for even more trains running at once. The system demanded a 6A AC power supply. The 11th Edition catalog that lists the HM5000 Advanced Power Transmitter implies that the power supply is a separate item, but the production units had the power supply built into the steel enclosure. They even included a fast clock with 4 ratios (1:1, 1:6, 1:12 and 1:24), and an audible overload alarm. The specifications stated it had two microprocessors, and was completely compatible with the Hornby Zero 1. Unlike the Hornby offering, the master unit was in a metal enclosure. It featured LED displays for the locomotive addresses (and the clock) & accessories that were selected, plus an ammeter, in the catalogue it shows a horizontal moving coil ammeter, where production models had a bar graph display.
Features included the ability to display the address selected, the inertia, the last accessory decoder used, and the time, at the press of a button. Small LED indicators were lit to show what function was displayed, and were also used for indicating direction.
(Hammant and Morgan Ltd. Catalog, 1980)
Offered their Power Grid Systems command control system, which was compatible with the Zero 1 system, in 1996.
ZTC Controls in the UK still support the system and have made improvements to it.
ZTC also makes DCC decoders that can be programmed to work with a Zero 1 system, and their controllers have a mode that enables control of a Zero 1 equipped locomotive.
One of the founders of ZTC was employed by Hornby as a member of the Zero 1 development team. Legend says that ZTC stood for Zero Two Controller.
The TMS 1000 family of 4 bit microcontrollers was introduced in 1974. They were a very simple design with only two 4-bit general registers, 43 instructions, and no interrupts. They came with 1kB of ROM onboard, and 64 X 4 bits of RAM. It had a maximum clock speed of 400 kHz.
From the Texas Instruments press release (1974):
Lowest cost. Broadest support. PMOS, NMOS, CMOS.
You're to market faster, and more economically, using TI's Series TMS1000 microcomputers. They carry today's lowest price. Versions are available for less than $3.00 in volume quantities. And nobody matches the depth and breadth of TI's total support that speeds design-in.
Proven Microcomputer Family specific application requires flexibility. Which is what you get with TI's TMS1000 Series. All in the table are P-channel MOS/LSI circuits. A mature technology of proven reliability. In fact, TI has more than six years of experience in building millions of PMOS devices.
Reliability is enhanced by single-chip construction that cuts component count as much as 75% - which helps hold total system costs down.
Soon to come: NMOS circuits for faster throughout and CMOS microcomputers for applications requiring low-voltage, low-power operation.
The TMS1000 and it's variations powered thousands of calculators, as well as microwave ovens and other electronic devices, including the Speak and Spell toy
The $3 cost (in quantity) made the IC very attractive to Hornby's Zero 1 Design Team as it was a low cost device which eliminated a lot of components in the process, making the product cheaper. Today, a microcontroller can be had for less than a $1 in quantity.
Return to the DCC History page
Yahoo Groups has a repository of documents available on the MergZero1 page for those interested in Hornby Hobbies Zero 1 products over its short lifespan, which has all three phases of its offerings,
- Phase 1 Locomotive Control
- Phase 2 Accessory Control
- Phase 3 Micromimic Display
It seems that the extra ports on the rear of the R.958 Phase 3 Micromimic Display Console were never utilised, the Display console also had buttons marked ERASE, REWRITE, MEMORY and ROUTE, with a lid at the rightmost, on the rear were two ports marked RAILDRIVE and LIGHTPEN, these have no known function. Commentators regularly speculate that these ports may have been Phase 4 and 5, if developement continued.