Fullscreen HTML5Med Res (??MB)HTML4

Here’s one for all the model railway enthusiasts . . . DIGITAL D IGITAL COMMAND C OMMAND CONTROL C ONTROL PROGRAMMER for DECODERS DCC – Digital Command Control – is a widely-used method for controlling model railways, especially when running multiple locos/ trains on the same track(s). This DCC programmer is simple, cheap and easy to build – and operates from a computer’s USB port. by Tim Blythman DCC is a great innovation, allowing many model locomotives to be addressed and operated independently on the same track at the same time It has been embraced as a standard by the NMRA (National Model Railroad Association, based in the USA), so equipment from different manufacturers can inter-operate without issue. And since the standards are public, anybody can create DCC-compatible devices. But the big downside to DCC (especially for beginners) is the cost of a base station. Even the cheapest base stations cost several hundred dollars and each locomotive also needs a decoder. Usually, even the simplest base 38 Silicon Chip stations include the option to program decoders but this can be a bit fiddly to use and can interfere with the operation of other trains on the “main line”. In this article, we describe how to build a standalone programmer based using an Arduino microcontroller module with a custom shield. It connects to a computer USB port to provide a convenient interface for programming. It allows you to read and write the Configuration Variables (CVs) on DCC rolling stock, customising their operation and performance. We have also designed the shield to be compatible with the DCC++ Arduino software. DCC++ is an open-source hardware and software system for the operation of DCC-equipped model Australia’s electronics magazine railroads; see https://github.com/DccPlusPlus/ When programmed with the DCC++ software, the programmer can be used with the Java Model Railroad Interface (JMRI), which provides a way to control DCC-based model trains from a computer. Its most relevant functions for this project are the ability to load and save locomotive (decoder) configurations. There is also another graphical user interface (GUI) which is compatible with DCC++, written in the “Processing” language. See the DCC++ web page for details. As well as being able to use the DCC++ software, we have written a small Arduino program which allows siliconchip.com.au CVs to be read and written via commands on the Arduino Serial Monitor. You don’t need JMRI or DCC++ to use it in this mode. By the way, we have previously published two other DCC system related designs: a 10A DCC Booster in the July 2012 issue (siliconchip.com.au/Article/614) and a Reverse Loop Controller in the October 2012 issue (siliconchip. com.au/Article/494). DCC hardware interface A typical DCC system requires 1215V to operate (see the panel below for an explanation of how DCC works). You can power the Arduino from a voltage in this range but it isn’t necessary; we’re using a small boost regulator module so that you can also run it off a 5V USB supply. In this case, the boost regulator provides the 1215V DC to power the tracks and DCC decoder(s). The DCC signal is a square wave at several kilohertz and the locomotive/ decoder could draw a few hundred milliamps, so our programmer needs to be capable of rapidly switching the track voltage and supplying sufficient current. Luckily, this can be achieved using a low-cost, bog-standard 556 dual timer IC. This IC is basically two 555 timers glued together in a 14-pin package. The outputs on the 556 can deliver up to 200mA at 500kHz, so it is perfectly suitable for this project. The Arduino module generates the DCC signals with the correct timing, which the 556 converts into the correct voltage levels. We’re also using a current sense (shunt) resistor so that the Arduino can detect how much current the attached locomotive is drawing. The DCC decoder can send data back to the programmer by varying its current draw. It sends a response by briefly drawing at least 60mA from the tracks. This is important as it is the only way to read data back from the decoder. Circuit description Fig.1 shows the circuit of the DCC Programmer. It’s based around dual Why DCC? If you’re running more than one loco/ train on a layout, the only logical way to do it is with DCC, which gives control to each one without affecting any other(s). And because the power and control signals are separate, the locos are still powered even when not moving, so their lights and any sound effects can still be operated. The downsides of DCC are mostly to do with cost and there is the added complexity of adding decoders to your locomotive. DCC allows manual control over each train, just as you normally have with a normal controller – but DCC has the big advantage of being able to take its operating commands from a computer, allowing completely automated model layouts, if you wish. This is especially useful in larger layouts; DCC also allows automatic point switching, level crossing boom gates and so on, obviously with suitable motors. The JMRI software mentioned in the text allows control over all these functions. Fig.1: besides the Arduino itself, the other components are mounted on a plug-in shield. It’s based around dual timer IC1, which acts as two power inverters with some built-in logic circuitry, forming a basic low-power full-bridge driver under the control of the Arduino. DC/ DC boost converter MOD1 provides the ~13V DC supply for IC1 when the Arduino is running from a 5V USB supply. siliconchip.com.au Australia’s electronics magazine October 2018  39 556 timer IC1 and boost converter MOD1. Pins 4 and 10 of IC1 are the reset inputs of the two timers and the timer outputs are disabled if these pins are low. They are held low initially by the 1kΩ resistor from pin 10 to ground, so output pins 5 and 9 are low. These pins connect to either side of the track, either via CON6 (a header socket) or CON7 (terminal block), depending on which suits you best. So when both timers are in reset and both outputs are low, there is no voltage across the track. The enable signal from the Arduino is fed to CON3 so when this goes high, the reset input at pin 10 is pulled high via the 100Ω series resistor. Schottky diode D1 is then reversebiased, so the other reset input at pin 4 can also be driven high by the Arduino, via CON2 and its 10kΩ series resistor. The trigger input of the first timer (pin 6) is pulled to ground while the threshold input of that same timer (pin 2) is tied to VCC (pin 14), so its output at pin 5 is high by default. However, if the pin 4 reset input is held low by the Arduino (via CON2), then this timer is in its reset state, so output pin 5 is low. Therefore, during operation, the output at pin 5 follows the signal at pin 4. Output pin 5 is also connected to the trigger input of the second timer, at pin 8, while that timer’s threshold input (pin 12) is also tied to VCC. So output pin 9 is low when output pin 5 is high and vice versa, and thus the second timer operates as an inverter once it is enabled. Thus, when IC1 is enabled, there is always a voltage across the track, with the magnitude being close to VCC and the polarity depending on the signal from CON2. A 2.2Ω resistor between IC1’s ground pin and the main circuit ground is used to sense the current flowing between the track connections. Regardless of the direction of current flow through the track connections, that current must ultimately flow back through IC1 and to the power supply ground. The result is a voltage across that 2.2Ω resistor. That voltage is fed to CON5 via a 1kΩ series resistor and on to one of the Arduino’s analog pins where it feeds the internal 10-bit analog-to-digital converter (ADC) and is converted into a digital value in the range of 0-1023 by the software. The combination of the 10-bit ADC and the 2.2Ω current sense resistor means the resolution of this reading is around 2mA. JP1 is a three-pin header which allows you to choose whether the voltage fed to the 556 IC is from the Arduino’s DC input socket (via the VIN pin) or from MOD1, an MT3608 2A boost DC/DC converter module. If you are powering the Arduino from a 12V plugpack, then you can put JP1 in the VIN position, so the plugpack provides the track voltage. But if you are powering the Arduino from a computer or a 5V USB charger, this voltage is not sufficient to power the tracks. In this case, you can place JP1 in the VOUT+ position and adjust the trimpot on the MT3608 module to provide 13V to IC1. The stand-alone Arduino program has a very simple interface, and allows direct reading and writing of CVs. This is sufficient to fully manipulate any parameters, but may not be as intuitive as the advanced roster settings available with DecoderPro. 40 Silicon Chip While not vital, as MOD1 has onboard bypass capacitors, the shield has provision for 100µF input bypass and output filter capacitors. We recommend that you fit the input bypass capacitor but leave the output filter capacitor off (as described in the Construction section). There is also an onboard 6.8Ω resistor between the 5V supply and the module’s VIN+ terminal, to reduce the inrush current from the 5V supply when it is first connected and to limit the maximum current drawn during operation (eg, if the tracks are accidentally shorted). The shield is driven using two of the Arduino’s digital pins and feeds the current sense voltage back to one of the analog pins. Rather than make these fixed, and risk interfering with the function of any other shields plugged into the Arduino, all three pins can be selected using jumper shunts for maximum flexibility. The digital pins are selected by placing one jumper between CON1 and CON2 (for the polarity signal) and another between CON1 and CON3 (for the enable signal). By default, our software is set up to use digital output D5 for the polarity signal and D11 for the enable signal but you can change the software if you place the jumpers in other locations. Similarly, the analog pin is selected by connecting a pin on CON4 to the associated pin on CON5 and by default, we’re using analog input A1. The software The primary job of the software is This is the Simple CV Programmer interface in DecoderPro, and it allows CVs to be directly written and read. Using the Roster feature allows the locomotive to be given a name, and CVs to be saved to a configuration file, as well as grouping the CVs into logical groups. Australia’s electronics magazine siliconchip.com.au Fig.2: use this PCB overlay diagram and the photo at right as a guide when building the shield. Make sure that MOD1 is orientated correctly, with its VIN pins towards the electrolytic capacitor. IC1, D1 and the electrolytic capacitor are also polarised. The jumper locations shown here suit our sketch as well as the DCC++ code, when being used as a DCC programmer. to produce a signal across the tracks which carries the required DCC packets. It does this using an interrupt software routine (ISR) which is triggered every 58µs by a timer. 58µs is the halfperiod of a ‘1’ bit. To send a ‘1’ bit, we toggle the DIR pin each time the interrupt fires. To send a ‘0’, we should ideally have a half-period of 100µs but 116µs (2 x 58µs) is within the limits that are recognised by the decoder. So we merely wait for two timer interrupts to occur before toggling the DIR pin to transmit a zero bit. The interrupt handler steps through the DCC data array as each bit is transmitted, then sets a flag to indicate that the complete packet has been sent and moves onto the next packet. Thus, most of the real work is done inside the interrupt routine. During most programming sequences, the programmer sends several reset packets to the loco, followed by multiple ‘write’ packets and this is accomplished by placing the appropriate commands in the array to be transmitted. The detection of acknowledgement pulses from the loco is done by continually sampling the voltage across the current sense resistor, at the selected Arduino analog input. The quiescent sample value is used as a baseline value. If an acknowledgement is expected, the current sample is compared with the baseline and if the threshold is reached, a flag is set indicating an acknowledgement has been received. As well as being able to use the DCC++ software, we provide a basic serial terminal interface. This allows you to send DCC commands straight to the locomotive, assuming that you know what is required. These are inserted into the array of packets to be sent, but they are not transmitted until siliconchip.com.au you indicate that they are ready. The software then sets a flag to start the transmission. MOD1 is optional We are using boost regulator MOD1 for convenience, so that you can run the programmer off a 5V USB supply. However, if you plan to power the Arduino from a 12-15V DC supply, you could omit MOD1. In this case, you would need to place JP1 in the “VIN” position. Keep that in mind as you assemble the shield. Construction The DCC Programmer Shield is built on a PCB measuring 68.5 x 53.5mm (the size and shape of a standard Arduino shield) and coded 09107181. Use the overlay diagram, Fig.2, as a guide during construction. Fit the smaller resistors first, confirming the values with a DMM before soldering each in place. Follow with diode D1, taking care to orientate it as shown in Fig.2. Then install the two larger 1W resistors. You can now fit the 100µF input bypass capacitor for MOD1. This should be laid over on the PCB so that you can later stack another shield on top if you need to. Make sure its positive (longer) lead goes in the pad nearest the adjacent edge of the PCB. Next, solder in pin headers CON1CON6 and JP1 where shown. You can then fit MOD1, by first soldering four component lead off-cuts to the pads so that they stick out the top of the board, then lowering MOD1 over those leads and soldering them to the pads on the top of the module. You can then trim off the excess lead length on top of MOD1 and on the underside of the PCB. Now solder CON7 to the board, making sure it has been pushed down so it is sitting correctly on top of the Parts list – Arduino DCC Decoder Programmer 1 double-sided PCB, code 09107181, 68.5 x 53.5m 1 set of Arduino stackable headers (1 x 6-pin, 2 x 8-pin, 1 x 10-pin) 1 Arduino Uno or equivalent 1 MT3608 2A boost module [SILICON CHIP Online Shop Cat SC4437] 3 14-pin headers (CON1-CON3) 2 6-pin headers (CON4,CON5) 1 2-pin female header socket (CON6) 1 2-way terminal block, 5/5.08mm pin spacing (CON7) 1 3-way pin header (JP1) 4 jumper shunts/shorting blocks Semiconductors 1 556 dual timer IC (IC1) 1 1N5819 1A schottky diode (D1) Capacitors 1 100µF 25V electrolytic Resistors (all 0.25W, 1% unless otherwise stated) 1 10kΩ 3 1kΩ 1 100Ω 1 6.8Ω 1W 5% Australia’s electronics magazine 1 2.2Ω 1W 5% October 2018  41 PCB and its wire entry holes are facing towards the outside edge of the board. Fit the Arduino headers next. We’ve specified stackable headers but you could potentially use standard headers if you don’t plan to attach any shields on top of this board. Either way, make sure the long pins project out the bottom of the board. You need to solder the stackable headers carefully to avoid getting solder on the pins except near where they connect to the pads. Note that the stackable headers give more clearance for the components on the board underneath (eg, the Arduino). If you use standard headers, you may find that CON7’s pads short to the shield of the USB connector below, which is connected to ground. Adjusting MOD1’s output voltage Before installing IC1, you must adjust MOD1’s output voltage. Plug the shield into your Arduino board and then apply power. Use a DMM set to measure DC volts to probe the VOUT+ and VOUT- pads on MOD1. Adjust its onboard trimpot screw to get a reading just below 15V; counter-intuitively, the voltage is decreased by turning the adjustment screw clockwise. We set our track voltage to 13V and it works well. You can then remove power, unplug the shield and fit IC1, ensuring that its pin 1 notch faces into the middle of the board, as shown in Fig.2. You could potentially use a socket but given that the IC is supplying significant current to the tracks, it’s better to avoid that. Now fit the jumpers to select your desired track voltage source and to configure which Arduino pins are used for control. If you are unsure, insert the jumpers in the positions shown in Fig.2. The shield assembly is then complete. Software setup and testing Now you need to upload the Arduino sketch code to the board. This is done using the free Arduino IDE (integrated development environment). It is available for Windows, macOS and Linux and you can download it from www.arduino.cc/en/Main/Software The IDE is used to compile and 42 Silicon Chip Monitor indicating the locomotive’s short address (CV 1). Other CVs can be written to and read from with simple commands like this. To write a CV, use the format “w1:3” That command writes the value 3 to CV 1. Using it with DCC++ and JMRI DecoderPro If you want to run DCC trains from a computer, you can use the open-source JMRI The DCC software. It comes with Deprogrammer shield coderPro, which has a commounts on top of the Arduino prehensive “Roster” feature UNO board, as seen here. which that be used to save and restore locomotive programming upload the software to the Arduino. Now download the software pack- parameters. If you have a fleet of similar locomoage from the SILICON CHIP website. This contains a “standalone” Ardui- tives, this is a convenient way of manno sketch called “DCC_Programmer_ aging their various performance CVs. So that our Programmer can operShield_V2”, which is the easiest way to test your shield. Before you can ate with JMRI, we can use the openuse this sketch, you need to install a source DCC++ Arduino sketch. This was designed to work with other hardlibrary called TimerOne. TimerOne can be installed from the ware but our shield has been designed IDE’s Library Manager (Sketch menu to be compatible with that hardware, → Include Library → Manage Librar- so you can run DCC++ on it without ies…). Open the Library Manager and any modifications. Our software download package for search for “TimerOne” and then when you find it, select it and click the “In- this project includes the DCC++ software. To use it, extract and open the stall” button. In case you have trouble with that DCCpp_uno sketch. Then check the method, we also include a zipped “Config.h” tab and make sure that this copy of the library in the sketch down- line is set correctly: load package. This can be installed us#define MOTOR_SHIELD_TYPE 0 ing the Sketch → Include Library → You need to set up the jumpers on Add .ZIP Library menu option. Now extract and open the DCC_Pro- CON1-CON5 for your shield to match grammer_Shield_V2 sketch from the those shown in the overlay diagram, download package, eg, using the IDE’s Fig.2. This configuration is required File → Open menu option. Connect to emulate this motor shield type. Upthe Uno to your computer and select load the sketch to the Uno board, as the “Uno/Genuino” option from the described above. The DCC++ software also uses a seTools → Board menu. Also, check that the correct COM rial interface, so you can use the serial port listed under the Tools → Port monitor to examine its output. The DecoderPro program has the menu matches the one assigned to option to use a DCC++ programmed your Uno board. Select the Upload option under Uno, so our reprogrammed board can the Sketch menu and then open the now be used with DecoderPro. JMRI serial monitor (Tools → Serial Moni- is available for Windows, Mac and tor) at 115,200 baud. You should get Linux (including a Raspberry Pi vera message describing how to use the sion). However, note that you need to sketch. If you have a DCC locomotive, have Java installed to use it. Get JMRI from http://jmri.sourceplace it on a length of track wired to either CON6 or CON7 and type “r1” forge.net/download/index.shtml and in the Serial Monitor, followed by the open the DecoderPro program. Go to the Edit → Preferences menu and Enter key. If everything is working, you should under Connections, choose DCC++ see some text appear on the Serial as System Manufacturer and DCC++ Australia’s electronics magazine siliconchip.com.au Serial Port as System connection. Ensure the Serial port matches that of the Uno. Save the configuration and close DecoderPro so that it can re-load the new settings. Open DecoderPro again and under Edit → Preferences choose defaults, and ensure that DCC++ is selected for Service Programmer. Unless you have other hardware, you should select DCC++ for all options. Save, close and re-open DecoderPro again. Click the red power button and ensure that it turns green. You should see “Service Mode Programmer DCC++ Is Online” in the bottom left corner of the screen. Now use the Actions → Single CV Programmer menu option to open the Simple Programmer window. This window allows you to read and write single CVs via a basic interface. If you find you are getting the “No acknowledge from locomotive” error then the threshold that DCC++ uses for detecting acknowledge pulses from the track may need to be adjusted. This is a parameter in the DCCpp_ uno sketch. Close DecoderPro (so that it no longer has control over the Arduino) and navigate to the “PacketRegister.h” tab in the DCCpp_uno sketch. Near line 20, there is a value called ACK_SAMPLE_THRESHOLD which is 30 by default. We found that reducing this to 12 gave consistent results with three different decoders. Save the sketch with the changes, then upload it to the Arduino board again. Ensure that the Serial Monitor is closed before reopening DecoderPro and try the Single CV Programmer option again. siliconchip.com.au While the default value of 30 corresponds to the specified value of 60mA for the acknowledge pulse, the DCC++ sketch also applies some smoothing to the sensed current, so this may change the actual detected current threshold. To use the Roster feature, click the “New Loco” button in the top left corner of the main DecoderPro window. Click “Read type from decoder” and DecoderPro will read a number of the CVs to identify parameters such as the manufacturer. If this does not work, you can also choose Roster → Create entry to enter this manually. Once a Roster entry has been created, double-click it to open the Comprehensive Programmer, allowing more detailed and complex programming to occur. JMRI is a vast program with many features and we can’t pretend that we’ve covered a fraction of them here. There is comprehensive documentation online. The article at siliconchip.com.au/ link/aal3 introduces DecoderPro and is a good place to start. JMRI also has a layout editor (PanelPro) so that you can create track diagrams and these can be animated with information from the layout if you have the correct sensors installed. Other uses for this project You may not think that having a DCC Programmer is all that useful but this device can also be used as a minimal DCC base station. With the DCC++ software, the Programmer can act as a low-power booster. If you set the board jumpers for EN on pin D3, DIR on pin D10 and Current Australia’s electronics magazine Sense on A0 then the DCC++ sketch can use the Programmer as its “main line” output. This means that the output can be connected to some track and DecoderPro’s throttle window can be used to control a DCC decoder-equipped locomotive. Because its current capacity is limited, you’re not going to be running a fleet of trains but it could be handy for testing and experimentation. In our trials, we found a small H0 scale locomotive was able to be driven at low speed, including operating the lights. What about driving a small motor? Another possible use for this shield is as a small 12V motor driver board. If you need to drive a small 6-15V DC motor using USB power or some other source of 5V then this shield can be used as a reversible motor driver, without the need for an external power source. It’s suitable for motors drawing up to about 2W. Note that your 5V supply must be able to deliver enough current. Useful links Setting up DCC++ with JMRI DecoderPro: https://github.com/DccPlusPlus/BaseStation DCC Standards Page: www.nmra.org/ index-nmra-standards-and-recommended-practices OVERLEAF: I I I I I I I I I I I I I I I HOW DCC WORKS October 2018  43 How DCC works We published an article in the February 2012 issue explaining how DCC works and showed some typical DCC decoders (see siliconchip.com.au/Article/769). But that article didn’t go into much detail regarding the DCC protocol, so read on for a more detailed explanation. If you aren’t into model railways then the advantages of DCC may not be obvious. To understand why it is so useful, we’ll explain how model railway systems worked before DCC. For a standard DC model railway, a single locomotive is fed power through the tracks, with one rail being negative and the other positive. Varying the average voltage between the rails changes the loco’s speed while swapping the rail polarities reverses its direction of travel. A typical H0 scale locomotive (1:87 scale) runs from 12V DC, drawing from a few hundred milliamps up to an amp or so. Such a system only allows a single locomotive to be controlled as multiple locomotives would receive the same track voltage. And direct control of accessories such as headlights, steam or sound effects is not possible. With clever use of diodes, it’s possible to have directional headlights, but a battery is still needed for lights if the locomotive is stopped as there is no voltage on the track. The most common way to allow multiple locomotives to operate is to divide the track into individual “blocks” or sub-circuits which can be switched between controllers. As you can imagine, the switching rapidly becomes very complicated as the track layout expands or more controllers are added. DCC transmits both power and control signals using the tracks. The locomotive’s onboard decoder interprets the control signals and commands the motors, lights and other accessories. Power is supplied continuously, allowing accessories to operate even while the locomotive is stationary. Beyond lights, features such as smoke generators and sound effects modules are quite common (if not especially cheap). The DCC signal is a square wave with a 50% duty cycle and varying pulse width. The data is encoded in the pulse widths while the square wave, once rectified, provides DC power for the decoder, motor, lights etc. The track voltage is typically 30V peak-to-peak, with cycle times of around 100µs. A typical DCC base station consists of a microcontroller feeding an H-bridge of some sort, usually with some form of current detection to shut it down in the event of a short circuit. With exposed conductors in the form of rails, it’s bound to happen sooner or later. A simplified circuit diagram of a typical DCC decoder is shown in Fig.3. This shows some of the circuitry to drive two lamps (eg, headlight and marker LEDs) as well as the motor. The DCC protocol encodes a binary 1 as a pulse that is nominally 58µs high and 58µs low, with a binary 0 having high and low periods that are 100µs or longer (see Fig.4). Because the locomotive can be placed on the track either forwards or backwards, the absolute polarity cannot be determined, so the high pulse may come before or after the low pulse. Fig.5 shows what these ‘0’ and ‘1’ bit pulses look like when they are arranged back-to-back, forming a continuous AC waveform. While the standards are specific as to what should and should not be accepted as valid data, merely sampling the “high” period of adjacent bits is sufficient to decode the bitstream and in any case, each command has a checksum byte, so errors caused by timing inaccuracies can be detected and the corrupt command ignored. The peak frequency of this signal is around 8kHz, which is higher than the 5kHz that the widely-used L293 fullbridge motor driver IC can deliver. So DCC base stations designed around the L293 IC probably won’t work reliably. ®: Registered DCC trademark of the National Model Railroad Association (USA). Fig.3: a simplified version of a typical DCC decoder circuit, showing the bridge rectifier and filter capacitor used to convert the AC voltage on the tracks to a DC voltage, with the raw track signal also being fed to the micro so it can be decoded. The micro then controls the H-bridge motor driver transistors, lamps, sound effects module etc. 44 Silicon Chip Australia’s electronics magazine siliconchip.com.au Fig.4: in a DCC system, the voltage across the tracks has a more-or-less constant magnitude but its polarity is continuously reversing, creating a 5-8kHz square wave. This is rectified by the DCC decoder(s) and is used as a power source for their logic circuitry as well as driving motors, lights, audio amplifiers etc. But the frequency modulation also encodes digital commands, addressed to individual locos. DCC protocol The bitstream is broken up into packets, with each packet containing one command, for tasks such as setting the motor to a particular speed or turning a light on and off. The loco’s state is stored by the decoder and kept consistent until it is updated by a future packet. There are also dedicated programming commands, which set configuration values (CVs) in non-volatile EEPROM. These values are used to determine which address the locomotive responds to, how the speed changes with throttle position, how lights and other accessories react to inputs etc. It’s these CVs that our programmer is designed to read and modify. Each valid packet starts with at least 14 sequential ‘1’ bits. This is referred to as the preamble and allows the decoder to synchronise itself with the start of the data packet, which is indicated by the first following ‘0’ bit. The data for the first byte of the command (which is an address) follows this, with all bytes sent most-significant-bit (MSB) first (see Fig.6). Subsequent bytes are prefixed with a zero bit so each byte transfer requires nine bits to be sent. The last byte to be transmitted is the aforementioned checksum byte. After this, a ‘1’ bit is sent, indicating that the packet is complete. The final ‘1’ can also count as the start of the next preamble if packets are sent back-to-back. Since each byte in the packet is separated by a ‘0’ bit, the only place that more than eight ‘1’ bits can appear in a row in a valid DCC sequence in is the preamble, so decoders can’t be fooled into thinking a new packet is starting in the middle of a valid packet. Each packet is a minimum of three bytes so the minimum packet transmission length is 40 bits, giving a minimum packet time of just over 5ms. So it’s possible to send close to 200 commands per second using DCC. Once the decoder has received a packet and the checksum is correct, it checks the address byte. If it matches the address stored in the decoder’s EEPROM, it can act on the command and if necessary, send a response. That is generally only necessary for programming commands as the programmer needs a way to read the configuration from the decoder. In this case, the packet sent by the base station is equivalent to asking “is this bit of this configuration variable set?”, to which the decoder replies either with an acknowledgement or not. The acknowledgement is performed by the decoder by placing a load of 60mA or more across the tracks for 6ms. In practice, this is usually done by the decoder briefly powering the motor, often resulting in the locomotive inching forwards during programming. The acknowledgement is one of two reasons why programming usually occurs on a dedicated programming track. Firstly, it would be difficult to accurately detect the acknowledgement pulse in the presence of other loads (such as other locomotives) on the rails. Secondly, many programming packets are broadcast to all decoders on the track. So the only way to guarantee that only the correct locomotive receives the programming packets is to have a dedicated section of track for programming. The DCC standards are very detailed and make for interesting reading. It could potentially be used in situations other than model railways, where power and commands SC need to be sent and received over two wires. Fig.6 (right): the structure of the shortest possible valid DCC command, containing one address and one data (command) byte. The preamble always consists of at least fourteen sequential ‘1’ bits and each byte is separated by a ‘0’ bit, with the command being terminated by a checksum byte (to detect errors) and then a final ‘1’ bit to indicate that there is no further data in this command. siliconchip.com.au Australia’s electronics magazine Fig.5 (left):the sequences of ‘0’ and ‘1’ bit pulses are strung together to create a continuous AC waveform across the tracks, effectively forming a frequencymodulated square wave. The RMS voltage is very close to the peak voltage, providing a similar voltage and current to the motors in the locos, while the average voltage is effectively zero since the waveform is symmetrical. October 2018  45