Scribd
Upload
85 views68 pages

Elektor USA - October 2014

rtrtr

Uploaded by

siammo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
85 views68 pages

Elektor USA - October 2014

rtrtr

Uploaded by

siammo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
[Link] mag azine October 2014 lektor Extremely Low Frequency (ELF) Receiver | PWM Control for Flashlight Temperature Sensor Board | Dot Display Driver | Microcontroller BootCamp (6) IoT & the Search for a Protocol | Lux Meter | Chip Tip: Magl2C-VDRM Visual Basic on the Raspberry Pi | DesignSpark Tips & Tricks | Weird Components: Magnetrons 3D Printing Sure Can Be Useful | USB Fix @ Review: Atmel ICE Debuggger/ Programmer @ Retronics: a 1965 Telefunken Carphone (ektor Projects Extremely Low Frequency (ELF) Receiver ‘You can in fact receive some extremely interesting signals between 0 Hz and about 20 Hz. Using the receiver described here, an ADC module, an Arduino and some free PC software it is possible to receive and make recordings of, these signals. PWM Control for Flashlight “The main function of this unit is to reduce the brightness of an LED at the user's command. ‘An additional function is also provided: the LED can be flashed at full intensity, which can come in hhandy for example when you are walking at night. ‘Temperature Sensor Board ‘This board is equipped with an ‘ATtiny microcontroller and an RS 485 driver, and itis possible to connect several sensors in paralle! to one board. In addition we present some ‘example firmware which ‘communicates temperature readings using the ElektorBus protocol. Dot Display driver ‘An indicator for four ranges, based ‘on opamps, with LED readout. Microcontroller BootCamp (6) We delve into serial ‘communication—specifically, using the SPI bus and associated protocol. ToT & the Search for a Protocol Calling engineers and software designers collaborate on a solid protocol for IoT devices. 42. Lux Meter Don’t believe the hype or the manufacturer—with this instrument you can reveal the real light intensity produced by lamps. 56 Chip Tip: Mag?°C-VDRM “There's no end in manufacturers honing the performance af the voltage regulator. Here's a very ‘advanced one. 68 Visual Basic on the Raspberry Pi If Python is not up your street, try something a little easier—say, Visual Basic. 62. One for Al A technical look at the latest ICE Debugger/programmer Atmel says ‘covers all of their AVR, Xmega and ARM-Cortex devices. © Industry 72 News & New Products {A selection of news items received from the electronics industry, labs and organizations. © DesignSpark 48 DesignSpark Tips & Tricks Day #14: The Autorouter This month we drop manual PCB ok Tension batt = Tension batt basse - Tension hysteresis —'6V-0V = 6.0V If Adeval < Tension batt Then ‘ADC measurement and switching to low battery mode Etat_led = 2 end Tf 2: ‘Low battery +t = Tension_batt_dechargee - Tension_hysteresis '5,5V-8V = 5.5V If Adeval < Tension_batt Then "ADC neasurenent and go to discharged battery * mode - cut the load Etat_led = 2 ‘Tension_batt_chargee + Tension_hysteresis Tension batt Then case 3: ‘IF low battery/ cutoff LED Tension_batt = Tension batt_chargee + Tension hysteresis °7,2V + ov = 7.2 If Adeval => Tension batt Then'ADC neasurenent and return + "0K mode "if battery is charged Etat_led = 1 end Tf End Select continuously the state of charge of the battery. This state is indicated using D2 as follows: ‘© green = battery charged; + orange = battery low: ‘red = battery almost fat. It snot good for the battery to continue to dis- ‘charge it when its almost flat, and so in this last ‘case the microcontroller turns off all the LEDs in the flashlight. The LEDs automatically light again ‘when the supply voltage increases sufficiently. If the battery supply Is at 8 V then 1.5 V will ‘appear on pin 3 of IC1 and D2 will glow green. If the supply falls to 6 V then the microcontrol- ler will measure 1.14 V and D2 will glow orange. ‘At 5.4,V, with a measured voltage of just 1.0 V, the LED glows red and the flashlight turns off |When designing the cut, care was taken to min- Imize its power consumption: it would be rather ‘a pity to throw power away needlessly in a LED ‘dimmer! It is for this reason that the battery level Indicator LED flashes in all three of its states. Schottky diode 01, type BAS70 or equivalent, protects input PB4 from possible excess voltage. ‘Capacitor C1 (10 nF) serves to debounce switch 2. Should your switch be ofa particularly inde- ‘sive nature you may find that this isnot suffi- ‘lent and hence that the dimming action is not ‘smooth: before replacing the switch, try increas- Ing the value of C1, say to 100 nf, and see ifthat solves the problem. ‘Atype AP1117 voltage regulator provides power for the circult at 3.3 V. although the LED chain ‘operates at a rather higher voltage (typically 7.2 V) the microcontroller Is isolated from its supply by the MOSFET that switches its power. Software ‘The program [1] was written using BASCOM-AVR. Its easy to modify the code, and only the demon- stration version of the compiler is needed. Having ‘set up variables with the thresholds for the various battery states, we initialize Timer®, which han- des PWM signal generation, and Timer, which is responsible for the flashing of the battery status. Indicator LED. The code spends most of Its time walting for a command from the user. Further down the code are the routines for indicating the charge status of the battery and for applying the thresholds to the A/D converter results to deter- mine the state of the switch and pushbutton, In ‘flash’ mode (when S1 Is pressed) the LEDs are driven using the maximum possible pulse width. A brief time-out filters out any contact bounce that might lead to unexpected operation. ‘The comments provided inthe listing should help guide you if you decide to make any changes to It. For example, you may wish to increase the speed at which the brightness increases or decreases. This can be done by reducing the value ofthe variable Pas or by changing the PWM. period from the value provided. The unit can be ‘made to respond more quickly to S1 by reducing the debounce timeout in variable Tenpo_bp_on_ ‘value. Finally, you might want to alter the intial PWM value (and hence brightness at switch-on) from the default (Pwn = 100). Construction ‘The printed circult board design Is shown in Fig- ure 2. The compact layout Is achieved using ‘surface-mount components, and this allows the board to be fitted easily into a cylindrical-style flashlight. However, we did encounter some unex- pected problems with certain models where the transparent tube was so firmly glued to the other parts that it proved Impossible to separate them ‘without damage. ‘Once the flashlight Is disassembled the only elec- trical modification to be made is to disconnect the ‘common ground connection of the LED chains (which are often arranged as thirty series pairs all wired in parallel). The red cross In the circutt diagram (Figure 1) shows where the connection Is broken. A wire must be soldered at this point ‘connecting the bottom of the LED chain (or chains, as the case may be) to the terminal on K3 which In tum connects to T1. Two further wires connect tthe flashlights ground connection and the positive battery terminal to K1. Then all that remains to bbe done Is to make two holes in the flashlights body where S1 and S2 can be mounted. (40019) Internet Link [11 www [Link]/140019 Figure 3. If luck ison your side you will find a flashlight model ‘ike this one that can easly ‘be disassembled and then ‘reassembled withthe ‘immer circuit inside. Design: André Goldberg, Mauk van der Laan and Ton Giesberts Ted: Jens Nickel Temperature Sensor Board _ with RS-485 interface ‘Temperature sensors are needed in many automation applications, and the RS- 485 interface allows reliable communication of data even over long distances. Our compact temperature sensor board is equipped with an ATtiny microcontroller and an RS-485 driver, and it is possible to connect several sensors in parallel to one board. In addition we present some example firmware which communicates. temperature readings using the ElektorBus protocol, and software to display the ‘Our series of articles on the ElektorBus in 2011 ‘generated a lot of interest among readers. We received hundreds of e-mails containing useful hints and tips or describing homebrew projects Using the bus. One notable fan of the Elektor- Bus Is André Goldberg, who bullt an ambient temperature measurement and control system. ‘An Important aspect of the project Is the use ‘of ElektorBus nodes with temperature sensors, which transmit readings to his PC. Part of the appeal of the ElektorBus protocol Is its simplicity. For example, each message Is ‘exactly 16 bytes long. The payload from a sen- sor board can carry four Integer readings (In the range -1023 to +1023) to a central control unt: these might represent the outputs of four con- nected temperature sensors. A message going in tthe opposite direction can, for example, instruct the sensor unit whether to report temperatures {In Celsius or Fahrenheit, or specify the interval between consecutive readings. The ElektorBus protocol provides commands for al of the above, see [1]. ‘An RS-485 interface Is used for communication. In hhalf-duplex mode this requires two signal wires, and a separate ground must also be provided (see [2]). RS-485 Is in theory highly immune to Interference, and at the data rate used on the ElektorBus (9600 baud) communication over dis tances in excess of 30 m Is possible. Mini bus nodes We have previously published circult designs and printed circutt boards for ElektorBus nodes [3][4]. However, for many applications these boards are physically too large. André Goldberg gave some thought to the question of how small the a bus node board could be made: the design that he delivered to our labs measures just 18 mm by 26 mm. It includes an ATtiny microcontroller in ‘an SMD package, which offers six GPIO pins. The RS-485 driver Is of course also an SMD device. Solder pads are provided to connect the bus lines, ‘and likewise the in-system programming pins of the microcontroller and two GPIOs are also only brought out to solder pads, all in the interests, of saving space. The external oscillator was also dispensed with, and the temperature sensor is a 10S18S20 one-wire device. This consumes only ‘one GPIO pin on the microcontroller as it uses ‘a special asynchronous protocol for communt- ‘cation that removes the need for a clock signal to accompany the data signal, The data signal ‘even provides power to the sensor: see the data~ sheet [5] for more details. Furthermore, several ‘one-wire sensors can be connected to a single pin on the microcontroller: each sensor includes ‘a preprogrammed unique 64-bit 1D code, and the microcontroller can use the 1D to address a particular sensor and receive a reading from it. ‘The DS18S20 outputs 9-bit readings with a res- olution of 0.5 °C. Challenges ahead ‘The proposed hardware design creates several challenges on the software side. ‘© The ATtiny45 has no hardware UART that can be used to drive the RS-485 interface IC. This means that the UART (both transmit and receive parts) needs to be implemented In software. An advantage Is that the two GPIO pins used can be chosen arbitrarily. ‘© A library Is needed to read the one-wire sensor devices, using one further GPIO pin. Since we may have several sensors con- nected to the same pin, the microcontroller will also have to store the IDs of the individ- ual devices so that the results can be output In the correct order. ‘© The ATtiny does not have enough program memory to contain the whole ElektorBus protocol library that we have described pre- viously [6]{7]. The required parts of the pro ‘tocol will need to be reimplemented by hand. André Goldberg decided to take advantage of ‘two open source libraries, one for the software UART and one for the one-wire bus. During the configuration process the addresses of the indi- vidual sensor devices are read out and stored in the microcontroller’s EEPROM. The readings from the four sensors are multiplied by ten so that the ‘temperatures in Celsius with a resolution of 0.1 °C are represented as integers in accordance with ‘Component List Resistors (0805) Ri= 2.20 foaa~ ana. Ra'= 1200 RS,ae = tka Capacitors C1,C2,C4 = 100nF 25V, 10%, X7R, SMD 0805 (C3'= 10uF 25V, 10%, XSR, SMD 0805 C5 = 100uF 16V, 20%, tantalum, SMD Case F ‘Semiconductors D1 = PMEGZO1OAEH D2 = LED, yatow, SMO 0805 D3 = LED, green, SMD 0805 ICL = ATiny85-20SU, SMD S0-852 IC2= 14785, SMD S0-8 14 = 78L05 Miscellaneous ‘Ki, Ka =2-pin pinheader, 0.4 inch pitch, right angled 3 = 3-way PCB screw terminal bloc, 0.2 inch pitch PCB ret 130468-1 V2.0 13 = Sihe quartz oscilator, Se7mm, SMD (LF SPKOO19079) ‘k2'= 6-pin (2x3) pinneader or baxheader, 0.1 inch pitcn Figure 1 “The single-sided circuit board measures just 1.14 by 1.5 inch (29 mm by [Link]). The headers are ‘mounted on the underside. Figure 2. The crcl centers around the Any. Oly one in (PE) i needed by the sofrare UART as transmission and reception ‘never occur simultaneously. the ElektorBus protocol. The remaining bytes in the ElektorBus protocol packet are ‘manually’ assembled into the code. The circuit In the Elektor Labs old hand Ton Glesberts imme- diately set about thinking how to ‘Elektorize’ the board. We decided to allow for the possibility of soldering in header pins or screw terminals for all the connections if desired, and we added a crystal oscillator for more reliable communication: this Is particularly important if the circult Is to bbe subject to extreme variations in temperature. Nevertheless Ton managed to keep the board ‘small: the final design measures about 31 mm by 32 mm (see Figure 1) [8]. The board Is dou- ble-sided, with the ElektorBus screw terminals, the two-by-three programming header and the two two-pin headers for the power supply and for connecting the one-wire sensors located on the back of the board. The SMD components can be soldered by hand. Our freelance colleague Mauk van der Laan found a significant improvement that could be made to the circult. Instead of using two of the six GPIO pins forthe transmit and receive connections for the software UART, we use only one pin, switching ‘ts function between input and output as needed. Because communication on the RS-485 bus Is. half duplex, a node never needs to speak and lis- ten simultaneously; collisions between messages have to be avoided. The advantage of saving a (GPIO pin is that we can use It to drive an LED to Indicate the status of the node. ‘The circult diagram Is shown in Figure 2. The central component is the ATtiny85 [9], which has ‘more flash memory than the ATUny45. All sx port pins are used, in some cases for more than one purpose. The one-wire temperature sensors are connected to PB4 via header K1. PB3 Is driven by the SMD crystal oscillator module. The SPI Interface on PBO, PB1 and PB2 and the reset pin PBS form the ISP interface that allows code to be loaded into the AVR microcontroller. In normal operation PEO drives the status LED. The level on PB1 determines whether the RS-485 transceiver Is In ‘transmit’ mode (PB1 high) or ‘receive’ mode (PB1 low). PB2 Is the pin which, ‘as described above, carries the UART data being transmitted or received. ‘The RS-485 signals are connected to K3. The 120-9 termination resistor can be fitted or omit- ted as required. Configuration We provide standard firmware to run in the ‘ATtiny85: its of course open source and is avail~ able for free download at [8]. Mauk van der Laan hhas incorporated in the code a software UART library, a one-wire Interface library (modified from an Arduino library) and a small ElektorBus Interface. The whole thing Is controlled by a kind of operating system that provides rudimentary ‘multitasking (see the text box for more details). More advanced users will find studying the com- mented C++ code very rewarding. Up to four DS18S20 temperature sensors can be connected in parallel to Ki. In this configuration the datasheet instructs us to tie the VDD pin of each device to ground so that each sensor derives ‘its power from the data line. The software uses an area of EEPROM ('slot one’ to ‘slot four’) In the ATtiny to store the IDs of the temperature sensors. When power is applied to the node the microcontroller extracts the ID from each sensor ‘ver the one-wire bus. It then compares them with the stored IDs. If @ stored ID is not found fn the bus, then that ID Is deleted and the stor- age slot becomes free. If an ID is seen on the bus that does not match a stored ID, It s stored In the first free storage slot. It Is now clear how we configure the system. First attach Just one one-wire sensor and apply power to the board. The ID of this sensor will be stored Inthe first storage slot. Its a good idea to label this first sensor ‘0’ (assuming that programming In C has taught you to count from zero; if you are 2 BASIC programmer you may prefer to label it ‘11. Then disconnect the power supply, add the second sensor in parallel at K1 and power tthe board up again. This will store the 1D of the ssecond sensor In the second slot. Repeat the process for the third and fourth sensors. “The progress of the configuration process is neatly Indicated by the status LED. When the software starts running the LED blinks rapidly. It then flashes four times, each flash being etther short or long. A short flash corresponds to an unused stor- age slot, a long flash to a recognized ID address. If its desired to remove a temperature sensor (even if ts storage slot is not known) the sys- tem can simply be powered down, the sensor disconnected, and the system powered up again. ‘The corresponding slot will be freed. The process described above can then be repeated to add a new sensor. o one of 0 Fo | + RRA] ae tf [+ [olo)o| 0) 0 |e 0 op fan A ean sexe ElektorBus interface ‘The four temperature sensors correspond to four ‘subnodes’. Our standard firmware periodically sends the four temperature readings packed into a single ElektorBus message, taking advantage of the ‘channels’ within the message payload. Each channel can carry a value from -1023 to +1023 [1], and each channel occupies two bytes within the message payload. The four channels ‘occupy bytes 6 to 13 of the payload, counting from zero. The transmitter address in bytes 4 Figure 3. Structure of a message to set the interval between readings on the temperature sensor board (only bytes i to nine shown). “Interval scale’ is 9, which is the code for tenths of a second. The ‘channel bits ‘in principle could allow diferent intervals to be set for different temperature ‘sensors, but they are ‘not used in the standard and § Is fixed at ‘S’ in our node. Bytes 2 and 3 contain the receiver address, fixed In our stan- dard firmware at ‘10’. Bytes 14 and 15 (check- sum/CRC) are not used. ber of sensor boards on a single bus. In so-called ‘direct mode’ collisions are avoided very simply by having a fixed pattern of transmissions. The sensor node sends the temperature readings at specified intervals to the ‘master’ (normally the control software running on a PC). If the master wants to send a command to the sensor node It Figure 4. The sensor board sends its readings tothe PC over the The standard firmware is designed for one-to-one S-485 bus. ‘communication rather than for use with a num- waits until it receives a message containing tem- perature readings and then sends its message in the pause immediately thereafter. At the nor- ‘mal bus speed of 9600 baud a message repeat Interval of under 100 ms is possible; however, the conversion time of the sensor devices is con- siderably longer than this (It can be as high as 750 ms) and so a message repeat interval of a few seconds Is a better choice. ‘Mauk’s firmware sends temperature readings (in Celsius, multiplied by ten) from all of the sen- sors it has identified at a default interval of 1 s. ‘The node also listens for commands to change the interval, the other possibilities being 5 s and '500 ms (see Figure 3). The units used for the temperature values can also be changed to Fahr- enhelt and back to Celsius. When the sensor ‘The standard firmware was written by Mauk van der Laan, @ freelance software and electronics developer. Mauk’s software is modular and includes a specially-designed operating system based on state machines to ensure the correct timing of the execution of tasks: see below. The file ‘Onewire.h’ contains the one-wire interface library. “[Link]/-h’ is the state machine that is responsible for detecting the devices’ IDs and for reading temperature values from them. ‘The software UART driver ‘[Link]/.h’ which is used for half-duplex RS-485 communication is derived from ‘an open-source RS-232 library. The ElektorBus interface “[Link]/.h can, however, also work with @ hardware UART. The line Haefine ELEKTORBUS_ORIVER_INCLUDE “ElektorBusSwlw.h” in the file ‘config.h’ causes ‘[Link]’ in the standard firmware to access the software UART library. “[Link]/.h’ is the state machine that controls the bus interface. Seeker ‘The operating system offers so-called ‘cooperative ‘multitasking’. In contrast to preemptive multitasking a task is never interrupted by another: instead, it yields control ‘voluntarily to the scheduler when itis ready. This means that no interlock mechanism is needed to prevent a task being paused at an inappropriate point. Also, unlike more heavyweight embedded operating systems such as FreeRTOS, ‘the tasks do not have their own stacks. Each task just has a single state variable, and so memory usage is minimal: the ‘operating system can run on an ATtiny with just 8 Kbyte of flash memory and 2 Kbyte of RAM. ‘The individual tasks are implemented using state machines. ‘The entire code for each task is expressed in a switch-case block: depending on the value of the state variable a different section of the code will be executed. When its actions are complete, the task calls the method nextState() which returns control to the main task (also called the ‘runtime’. ‘Alternatively the task can call nextDelay() with an argument specifying a delay in milliseconds. This constitutes a request ‘ot to call the task again until the specified time has elapsed. ‘The scheduler in the runtime maintains a list of running tasks, each of which is derived from the “Task’ class. The node receives a command message of this type, the next regular message it transmits will con- firmation of its current units and scaling (and @ ‘message containing temperature readings will be lost). Subsequent messages will contain read- Ings as normal. Since the reading in Fahrenheit can easily exceed 102, the sensor automatically switches from a scaling value of ~1 (tenths of a degree) to 0 (whole degrees) when a command to use Fahrenheit units Is recelved. PC software ‘An ordinary terminal emulator program Is suff- cent for an initia test, at least to check that the board is indeed outputting readings at one-sec- ond intervals. We have also written some cus- ‘tom demonstration software for the PC. The user Interface Is as usual written in HTML and JavaS- Cript. The software download [8] contains the software [Link] and the folder UIBus, which should be dragged to your desktop. ‘The sensor board Is connected using Its RS-485, Interface and the Elektor RS-485-to-USB con- verter [3], whose USB port Is plugged Into the PC (Figure 4). On launching the ElektorBusBrowser ‘you must set the number of the COM port that Is allocated to the RS-485-to-USB converter at the top of the screen, and then click on the ‘Connect’ Figure S. ‘The user interface on the PC is as usual based on HTML, Using the inter button. Ifthe sensor node is running and cor- fe posse te adja the rectly configured the readings should now appear measurement interval and fon the screen. Commands can also be sent to the temperature unts used. ‘scheduler calls the execute() method of each task in turn (unless it has requested a delay which has not yet expired). Each task should run for just a short time (perhaps a few hundred cycles) to maintain the illusion of full multitasking. The foll 19 example code shows a task that flashes an LED. class BLinkTask + public UserTask { ‘enum States (Idle, BLinkingon, BLinkingort}; pubtte: void start() nextState(BLinkingon) ; 3 void stop() { nextState(Idte) ; d private: void turnon() ord turnorr(); virtual void execute(); virtual byte getTaskid() { return BLINKTASK1D; } votd BLinkTask: :execute() { switch(state) { ‘case BLinkingon: 500 ms ‘turnon() 5 nextDelay(BLinkingort, 506); retur case Blinkingorf: 500 ms turnort(); nextDelay(BLinkingOn, 500); returt derautt: pante (1 11 turn the LED on and wait 11 turn the LED off and wait // Anvalid state: give up Figure 6. ‘Simulation using an ‘Arduino Uno and the Elektor ‘extension shield. This node behaves exactly lke the ‘temperature sensor board ‘onthe EiektorBus. the node, but itis necessary to tick the ‘Direct Mode’ check box first. ‘The user interface is straightforward and should be self-explanatory (Figure 5). The labels “°C’ and ‘°F’ next to the temperature values only ‘change when a message is recelved from the sen- sor node confirming the receipt of a command to ‘change units. The change In scaling factor is also taken into account. To take a look at the HTML and JavaScript code, cick on the ‘Source’ button. If you also have the Andropod Android bridge board, you can couple the temperature sensor board with an Android smartphone or tablet [10]. ‘The HTML and JavaScript user interface works equally well in the ElektorBusBrowserForAndro- ‘pod, which can be downloaded for free from Goo- le Play. ‘Simulation ‘Since the standard firmware for the board was ‘not available at the time we were writing the PC control software, we simulated a tempera- ture sensor node using an Arduino Uno board fitted with an Elektor extension shield [11] and ‘an RS-485 module [12] (see Figure 6). The sim- ulated temperature values are generated using ‘a potentiometer and are shown on the display in tenths of a degree Celsius. The Arduino Uno sends the simulated value to the PC in channel 0, and of course this happens at fixed intervals. It also ‘responds to control commands issued by the PC In exactly the same way as the standard firmware running on a real sensor node. The source code for the ATmega328 Is also avallable for down- load [8]: itis based on the EFL 7] and uses the ElektorBus library. The ‘Hardware’ directory of the Atmel Studio project contains the code files for the Arduino Uno, the Elektor extension shield and the RS-485 ECC module. These make the higher layers of the software independent of the hardware and they are in tum independent of one ‘another. Further software for the new shield will bbe presented in our next issue. Of course the standard firmware and the Elek- torBus protocol are not set in stone: you are free to write your own software and design your own protocols. Different sensors, switches and even ‘actuators can be connected to K1: port pin PBS can easily be used to generate digital signals or to measure voltages. If you do some up with a ‘new use for the board, please drop us an e-mail to let us know or tell us at [Link]. (120468) Web Links [1] www [Link] [2] www [Link]/ 110225 [3] wow [Link]/ 110258 [4] www [Link]/ 110727 [5] http:/[Link]/en/ds/[Link] [6]_wwew [Link]/ 120582 [7] [Link]/120668 [8] [Link]/130468 [8] [Link]/atmel-2586-avr-8-bit-microcon- troller-attiny 25-attiny4S-attiny85_datasheet. pdf [10] [Link]/110405, [11] [Link]/140009 [12] www [Link]/130155 8y Clemens Valens (Giektor Labs) Dot Display Driver ‘There are situations when it is useful to scale the dynamic range of a signal down to a few coarse sub ranges. As an example you can think of an indicator that ‘Several solu- tions to this design chal- lenge exist, like the ever popular LM3914 (also avallable as NTE1549, and its siblings LM3915 and LM3916) with Its convenient ‘moving dot mode, or a microcontroller with an analog Input. The LM3914 Is easy to use but It does not offer much control over the way the Input signal is quantized. A microcontroller gives you all the flexibility you can wish for, but it requires programming. The circutt presented here allows full control of how the input signal is quan- tized but It does not require any programming apart from calculating a few resistor values. ‘The craut s straightforward. A multi-output volt- ‘age divider is dimensioned to cover all the ranges that matter. Each voltage divider output Is com- pared to the input signal; when the latter exceeds the former, the corresponding comparator out- put swings High. This is all standard stuff, but not good enough, since when the Input exceeds 2 certain level, all the comparators below this. level will go High instead of just the last one. Our {goal is to have only one active output for each level, not several. To paraphrase the LM3914's datasheet, our circuit must be in ‘dot’ mod, not “bar graph’ mode. ‘Adding (moving) dot mode can be achieved by making the higher level comparators con- trol the comparators below them in some way, does not show the exact speed of a motor, but only a few values like Off, Slow, Medium, Fast and Too Fast. Such signals can easily be obtained by low-pass filter- ing) amplitudes, frequencies or pulsewidths of signals produced somewhere in the system that you want to keep an eye on. for instance by forcing their outputs Low or by disabling them. In this circuit every com- parator controls the output of the comparator directly below it with a PNP transistor. So how does this work? When two consecutive comparator outputs are Low, the base and emitter of the transistor between them will both see the same voltage. Consequently the transistor will switch off and 50 wll be the LED driven by t. When the lower com- parator output goes High while the upper stays Low, the transistor is switched on because its emitter voltage will be high enough with respect to ts base voltage, hence the corresponding LED vill ight up. If now the upper comparator output becomes High too, the base and emitter voltages are again the same, causing the transistor to block and the LED go out. In short, an LED can only shine ifthe comparator output straight above it is Low while the comparator output straight below itis High. This assures that only one LED at a time can be on: the circuit is in dot mode. “There is a subtlety to this circuit that you have to be aware of. It lurks in the comparator out- put or, to be more precise, in its impedance. My ‘experiments involved a single-supply rallto-rall (SS-R2R) opamp type TS924, and everything worked perfectly fine. However, when I checked the availabilty ofthis opamp, I discovered that Ithas become obsolete and that is replacement, the TS924A, Is hard to get in a DIP package. So T ‘switched to another SS-R2R op-amp, an LMC464 that I happened to have handy. To my surprise, with this chip, my circult no longer lit just one LED, but It also the LED right below It. Not too bright, but more than enough to be noticed. The cause Is its rather high impedance making the ‘output of the new op-amp drop as a quite steep function of the current it has to deliver. Suddenly the high level was not so high anymore, allowing the transistor driven by this output to turn on slightly because the transistor’s base voltage no longer equaled its emitter voltage. Checking the datasheets of both opamps explained it all. The TS924A, a pretty cool opamp that I heartily recommend, remains close to R2R with loads as low as 600 whereas the LMC464 Is specified for loads of 25 k2 or higher. Okay, so the LMC6464 was a bad idea, but what about the good old LM324? With this chip the circuit worked almost fine even though the LEDs were a bit less bright, the LM324 high output level being about 1.5 V below its supply voltage. ‘This created a problem for the lowest LED (Off) because I had tied the emitter of its controlling transistor to 5 V, too high for the opamp out- put to make the LED shine continuously. Adding two diodes In series with the emitter fixed this problem. ‘The Input stage of the circult Is experimental; 1 just plugged some random component values Into the schematic because It all depends on your input signal. I did my experiments with a 3-V 22-kHz PWM signal and got excellent results with C1 = 220 nF and R15 = 2.2 kA. The PCB was designed in such a way that capacitors of, up to 10 pF should fit without problems. High er-value parts should work too, but hovering over R1S and R14. ‘Adjust P1 to scale the quantization levels. Note that the scale in the schematic Is linear thanks to resistors R5-RB having Identical values, but this Is by no means obligatory. Instead of the LEDs you can mount a pinheader to connect the board to, say, another one with relays on it. Do not forget to buffer the outputs If you have to drive heavy loads (to avoid the problems described above). Think ULN2003, for Instance. Web Link [Link] Figure 1. Dot Display Driver (000) schematic. ‘Component List Resistors (0.25) R9,RL0,R4,R42,R13 = 3900 AR2,F3,R4,R,R6,R7,RB = 47K 1,02 = 1na148 Ici = uMa24 {LEDI,LED2,LED3,LED4,LEDS = LED, red, 3mm Ti-Té = BCSS7C Miscellaneous Ki = 3:pin pinheader, 0.1" pitch Pea iaotti-t Figure 1. ‘Connecting a type 4094 shift Microcontroller BootCamp (6) The SPI interface Serial communication with each unit of information following the previous one is actually the normal situation. That’s how we talk to each other in person or by phone, read and write text—or in the case of Retronics: send telegrams. In many cases all you need is a single line to transmit data. However, adding a clock line makes things more reliable. Let’s find out. ‘On the Serlal Peripheral Interface (SPI) bus, the actual data travels bit by bit over one line—for ‘example from a microcontroller to a display, an EEPROM or an SD card. In most cases It's also desirable to be able to read data, so you need a ‘second line to provide a return channel for the data. There's yet another part to the picture: a ‘lock line. The clock signal always clearly indl- ‘cates when the next bit Is available on the data line. That eliminates the need for precise agree- ment on data timing, which both parties have to supervise with timers. With these three lines, data transmission Is fairly bombproof. Port extension with a shift register ‘The first thing we want to try out is actually not an SPI interface, but instead something entirely different. Shift registers have been around for ‘a long time (before microcontrollers were even invented) and are good for understanding how serial data transfer works. They can also be put to good use in combination with a microcontrol- ler. That’s because port lines are always scarce, especially on Arduino boards. A port extension with a shift register can help ease the scarcity. With a type 4094 8-bit shift register, you need three lines to talk to It and you end up with eight new outputs. You can increase this to 16 by connecting a second shift register, or even 80 if you connect ten shift register ICs in series. If you need a lot of outputs, that’s an especially low-cost way to meet the requirement. Figure 1 shows the connections to the Uno board. The 8-bit shift register has a clock input (CL) and a data input (D). The data Is applied to the D input one bit at a time, starting with the most significant bit, and a rising edge is applied to the clock input CL for each bit. The data is shifted through the individual flip-flops of the register step by step with each clock pulse. There Is also the strobe input STR. When a pulse Is applied to the strobe input, all the data present in the shift register is transferred to the type-D flip- flops connected to the output pins. You could also tle the strobe input to the supply voltage \Vce, but then all the intermediate results of each shift operation would appear on the outputs. By contrast, if you apply a strobe pulse after all the bits have been shifted in, you only see the final result on the outputs. The software for all this (Listing 1) is simple. To ‘output a byte D, the code first copies the most significant bit to the bit variable B (B = D.7) and puts it on the corresponding port pin. It then generates a positive clock pulse on CL with a length of 1 millisecond. A microsecond would also be sufficient, but the slower output Is easier to see on an oscilloscope. After the clock pulse, a shift instruction (shift 0 , Left) causes all bits of D to be shifted left by one position. This puts what used to be bit 6 on the output. This process is repeated until all eight bits have been shifted out. At the end comes the strobe pulse, and then all eight bits are present at the outputs of the 4094. The code continuously increments the data byte to be transferred so the outputs of the shift regis- ter change while the program Is running. The current value Is shown on the LCD if the Elektor Extension shield is fitted, and it is output to the terminal emulator. This makes it easy to com- pare the output levels of the shift register to the digital value being output. If you need more than eight outputs, you can use the Qs output of the 4094 IC. Each bit that is clocked into the shift register appears at the Listing 1. Output using a shift registe ‘UNO_shift.BAS Shift Register 4094 Sregfile = "[Link]” Serystal = 16000000 $baud = 9600 Dim Dat As Byte Dim D As Byte Dim N As Byte Dim B As Bit Sr Alias Portb.4 "4994 pin 1 Da Alias Portb.3 '4994 pin 2 CL Alias Portb.2 "4994 pin 3 Config Portb = Output cs Led Dat lca" Print Dat D = Dat For N=1T0 8 B=D.7 Da=8 Wattms 1 asl Wattns 1 case Wattns 1 shift D, Left Next N sr=1 Wattms 1 sr=0 Waitms 1 yt = Dat + 1 Waitms 500 Loop End Qs output eight clock pulses later. The D input of the next shift register can be connected to this output. In this way you can connect as many 4094 ICs In series as desired, with the clock and strobe lines connected to all of them in parallel. Of course, the software will have to be modi- fied accordingly. First it shifts out all of the bits necessary to fill the chain of shift registers (e.g. 16 with two ICs or 80 with ten), and then it out- puts the common strobe pulse. Manual data transmission Although you only need one line for the data, you also need a clock line when you use a clocked serial interface, as in the above example with a shift register or with the SPI bus. If you compare this with Morse telegraphy, for example, you can see the difference. There both parties have to agree on the transmission rate, and no breaks are allowed within an information unit. For example, if a radio operator wants to send an “x” (dash dot dot dash) and stops in the middle to scratch his head, the two characters “N” (dash dot) and "A" (dot dash) are sent instead. The situation Is exactly the same with an asynchronous serfal data Interface, where both parties have to agree on the baud rate. After the transmission of a byte has started, all of the bits must be sent within a precise time frame. By contrast, with SPI the timing Is not critical and any desired delays are allowed. The additional clock signal makes the transfer entirely independent of the speed. No matter whether the data rate Is just one bit per minute or a million bits per second, the data will Listing 2. SPI master and slave (very slow). 'UNO_spi1.BAS Shift in/out $regtile = Serystal Sbaud = 9600 Dim D As Byte Dim B As Bit Dim N As Byte S1 Alias Pinc.0 Portc.@ = 1 $2 Alfas Pinc.1. Portc.1 = 1 Ledl Alias Portc.2 Darc.2 = 1 Led2 Alias Portb.2 Ddrb.2 = 1 Do ‘D = Rnd(255) cs Led D Print D Locate 2, 1 For N=@To7 B=D.7 Led B Print B; Led2 = B Waitns 300 Ledl = 1 Wattms 200 Ledl = 0 Waitms 500 Shift D , Left Next N Ledi = 0 Led2 = @ waitms 2600 cs Print D=0 Do Loop until Si = 1 For N= To7 shift D , Left Do Loop Until $1 = 0 If S2 = 0 Then B= 1 Else B D=0+8 Led B Print B; Wattms 100 Do Loop Until Si wattms 160 Next N Print Locate 2, 1 Led D Print D Wattms 2600 Loop End be transferred properly. On an SPI bus there is always an SPI master and an SPI slave. Data can travel in both directions, but the master ways generates the clock signal. ‘You can try this for yourself manually, where you (@s the user) assume the role of master. You can transmit a byte by pushing the two buttons S1 and S2 (Figure 2). This Is not the usual way of doing things, but it helps you understand exactly how it works. One of the buttons Is for the data, and the other Is for the clock. Aside from that there’s nothing new you have to learn, since you already know how a byte Is put together. Here's, how it works: First you send bit 7. If it is a‘1', ‘you press and hold button S2; otherwise you don't. Then you press button S1 briefly without changing the state of $2. The receiving end (the slave, which in this case is the Arduino board) then knows when it should read the bit from the data line. Now you repeat the process for bit 6, bit 5, and so on until bit 0. ‘The program in Listing 2 displays the data in both directions. At first the microcontroller is the master and you are the slave. A byte with a random value is sent, with the clock signal indi- cated by LED1 and the data indicated by LED2. If you watch carefully, you can read the trans- mitted byte. However, that’s not easy, so the byte Is also shown on the LCD and sent to the terminal emulator. The individual bts also appear one after the other on the LCD and the terminal emulator screen: 83 e1019011, ‘Then the roles change. Now you are the master, and your job Is to send exactly the same byte back to the microcontroller. Here you can see that the ability to send data at any desired speed is a big advantage, since you can take all the time you want to decide which bit value to send next. For example, suppose you want to send the decimal number 100. Bit 7 corresponds to decimal 128, which is more than 100, so it is ‘0’. Next comes bit 6 with a value of 64, so it’s a‘1', and you're left with 36 still to send. This means that bit 5 (32) Is a1, which leaves 4. The next two bits (bit 4 = 16 and bit 3 = 8) are ‘0’, bit 3 (4) isa 1’, and the last two bits are ‘0’. Now you have sent the binary number 01100100, with each bit ‘marked by a clock pulse. It may not have been all that easy, but the microcontroller had no problem reading the data. You can regard this as a test of your concentration, and if the LCD shows the right result, you pass the test, If you look closely at Listing 2, you will see that. the bits are inverted when they are read. That's because pressing the data button yields a zero bit value. This Is not especially intuitive, so the result Is inverted when the bit Is read to make things easier for you. Another interesting aspect Is using the instruction D = Rnd (255) to generate a pseudo-random number. In fact, this always generates the same sequence of numbers, but the Bascom Help glves some suggestions for what you can do about this. From microcontroller to microcontroller In this example, data is sent over the SPI bus from one microcontroller to another. The data Is this case consists of 10-bit readings from the ‘A/D converter. This shows another advantage of SPI, which is that the data width is not fixed. No matter whether you send 8, 10, 12 or 16 bits, the procedure is always the same. If the only objective were to connect two microcontrollers together, it would actually be less effort to use an asynchronous serial interface with the TXD and RXD lines. The SPI bus, by contrast, is better for controlling and communicating with external Figure 2. Manual input and output. isting 3. SPI master. *UNO_sp12.BAS SPI Master cts Cursor off $regftle = "[Link]" $crystal = 16000000 wattms 200 Sbaud = 9600 Do Dim B As Bit Dout = Getadc(3) "Pot Dim Dout As Word Locate 1, 1 Dim N As Byte Led Dout Dim I As Byte led cs=0 Sck Altas Portb.5 Wattms 26 Darb.5 = 1 Most Alias Portb.3 Darb.3 = 1 Cs Altas Portb.2 Darb.2 = 1 Cs=2 wattms 1 Most = 6 Shift Dout , Left Sck = 0 Next N csel Config Adc = Single , Prescaler = 32 , Wattms 100 Reference = Avcc Loop Start Adc End Listing 4. SPI slave. Portb.2 = 1 *UNO_sp13.BAS SPI Slave $regfile = "[Link]” Do Scrystal = 16000000 Do ‘$baud = 9600 Loop Until Cs = 6 Din=0 Dim Addr As Byte Dim B As Bit Dim Dout As Word Dim Din As Word Dim N As Byte Dim I As Byte S1 Alias Pine. Portc.@ = 1 2 Alias Pinc.1. Portc.1 = 1 Sck Alias Pinb.S Portb.5 = 1 Mosi Alias Pinb.3 Portb.3 = 1 Cs Alias Pinb.2 For N= 1To 10 Shift Din , Lett Do Loop Until Sck = 1 Din = Din + Most Do Loop Until Sck = 0 Next N Do Loop Until Cs = 1 Locate 1, 1 Led Din tea" Print Din Loop End Master ee se hardware. Here the main purpose of the exercise Is to illustrate the transmission protocol. As previously with the 4094 shift register, a third line Is involved here—In this case the chip select line /CS. The slash (/) means that the signal on this ine Is active Low. The chip select line allows you to connect several slave devices to a single ‘master. In that case they share the data and clock lines, but each one has its own chip select line. When that line is low, the corresponding slave knows that it is selected. There's also another benefit from using a chip select line. If there is any delay in enabling the slave, there may be some confusion about which bits have already been transferred. However, ifthe slave waits until It sees a falling edge on its CS input (high to low signal transition), It knows that the transfer is starting. And If a noise pulse Is read as a clock signal, the rest of the data for that transfer Is trash, but on the next access everything Is again as It should be. ‘The ATmega328 also uses the SPT bus for pro- gram download from an external programming device. The following lines are therefore avall- able on the six-pin programming connector on the Arduino board and on the Elektor Extension shield (ICSP in Figure 3): the clock line Serial ‘Clock (SCK) on 85, the write data line Master ‘Out Slave In (MOSI) on B3, and the read data line Master In Slave Out (MISO) on B4. There Is no chip select line, but the Reset line has the same effect because programming takes place with the Reset line pulled low. Now we want to Use these lines exactly as intended. This has the ‘advantage that we can use the hardware SPI unit of the microcontroller, if it has one. With hard- ware SPI we do not have use program code to Put each bit individually on the data line as in the previous examples, and everything Is a lot faster. However, we still need a chip select line, and in this case we use the B2 line for this purpose. ‘The master uses the MOST line as the output ‘and generates the clock and chip select signals (Listing 3). The process Is slowed down a bit by three 1-millisecond delays so that all the signals can easily be seen on the oscilloscope. Besides, we don’t want to make things too difficult for the slave. If you wish, you can test the boundaries by reducing the delays until transmission errors start to occur. The three lines are inputs for the slave device (Listing 4). It constantly waits for specific signal ‘edges on the /CS and SCK lines and then reads In a bit from the MOST line. Since everything Is. handled by software here, the code must walt for ‘each edge in a Do loop. This takes a bit of time, ‘so data transmission must be slower than with hardware SPI implementation. The recelved data Is shown on the display and on the terminal emu- lator. When you turn the potentiometer on the master board, the change is visible on the slave. Figure 2. ‘An SPI connection between ‘wo microcontrollers. SPI EEPROM 25LC512 ‘There Is a wide range of ICs available with an ‘SPI interface, including A/D converters, memory devices and display drivers. Serial EEPROMs from Microchip and other companies are available at low cost and are widely used. The 25LC512 (not tobe confused with the 24C512, which has a I2C bus interface) has a capacity of 64 KB, and It isa ‘good solution when the 1-kllobyte capacity of the ‘ATmega328's internal EEPROM Is not sufficient. I7C EEPROMs are more widely used in the hobby Figures. Connecting a serial EEPROM. Listing 5. Reading and writing data over MOSI and MISO. ‘Sub Spioutin Dino For N=0To7 ‘Shift Din , Left Din = Din + M1s0 ‘If Dout.7 = 1 Then Wost Wattus 3 sek = 1 Wattus 2 Sck = 0 wattus 2 Shift Dout , Left Next End Sub 1 Else Most realm, but SPI types are generally preferred in the professional realm because they offer espe- ally high operational reliability. Figure 4 shows the connections to the Uno board. ‘The pins of the original SPI interface (2x3 pin header) are again used here. That makes it easy to bulld a convenient plug-in memory module by fiting the IC in a socket soldered to a small 6-pin socket header. You only have to connect ‘one additional line. Ths isthe chip select line, which is again assigned to 82 because the Reset line present on the connector cannot be used for this purpose. Here there are two data lines. MOSI (Master Out ‘lave In) is the data output line and Is connected to the Serial Input (SI) pin of the EEPROM, while MISO (Master In Slave Out) Is used to read data from the Serial Output (SO) pin. The microcon- troller is always the master, and it generates the dock on the SCK line. To go with this exercise, we have written a subroutine (subroutines are called “Sub” In Bascom; see the inset) that transfers. data in both directions in a single go (Listing 5). Before the subroutine is called, the data to be ‘sent must be placed in the global variable bout, and after the call the received data Is located In the variable Din. Of course there are situations in which data Is only written, but In that case zero bits are usually sent in the other direction. ‘The relatively complex data sheet for the 25LC512 tells you what has to be sent to the device, as well ‘as when and how. After the chip select line has been pulled low, the memory chip first receives a simple byte command that specifies what action is to be performed. To read data from the mem- ‘ory, you have to send a3" command followed by two address bytes forming the high-byte and low-byte portions of the address. after that as many data bytes as desired can be read out with ‘automatic address incrementing (see Listing 6). ‘The program displays the sequential addresses and the data bytes that are read out. A brand-new or fully erased EEPROM always delivers only the value 255. Now let’s try to program some data. For that we use the byte command ‘2’. However, 2 bit of preparation Is necessary fist. Writing must be enabled by sending the command ‘6” (Listing 7). To check whether writing Is enabled you can read the EEPROM status register, which requires sending the command ‘5’. Each action Is only effective if you pull /CS low at the start and then return it to the high level at the end. Listing 6. EEPROM readout (excerpt). cs=0 Dout = 3 Spioutin Dour = @ Sptoutin Dout = 0 sptoutin 1-0 Do Locate 1, 1 Lea I Print I; Print " Tete. Sptoutin Locate 2, 1 Lea Din Print Din Wattas 200 Loop tread "a8. ..ALS *A0...AT7 If the value in the status register Is 2, the IC Is enabled for write operations. Complicated? Yes, but that makes it especially foolproof. Now we are allowed to write data to the mem- ory, but even then there's something we have to consider. The memory space is divided into pages, which In the case of the 25LC512 have a size of 128 bytes. Each transfer Is limited to a maximum of 128 bytes of data, or as many bytes as it takes to reach the next page bound- ary. After this you switch /CS high and then give the EEPROM enough time to actually store the data, According to the data sheet, 5 ms Is enough Listing 7. Enabling write access (excerpt). cs=0 Dout = 6 Sptoutin essa Wattns 20 cs=0 Dout = 5 Spioutin Sptoutin essa Locate 1, 1 Led Dtn Print Din ‘write enable *read status ‘status ‘Subroutines are called “Sub” in Bascom, and they are used to allow 1a block of code to be called repeatedly from different locations in a program. To make this possible, each subroutine must be declared at the start of the program. This way the name of the subroutine is known to the compiler, so the it can be called using this name in the same way ‘as Bascom functions. Declare Sub Sptoutin() Dout = 6 Sptoutin Sub Sptoutin oin= 0 Shift Dout , Left End Sub ‘You always have to be very careful with the variables used by ‘subroutine. In the case of the Sptoutin subroutine, all of the variables it uses are “global” variables, which are dimensioned at the start of the ‘main routine and are therefore valid in the entire program. However, you could do things differently and transfer the data to the subroutine when it is called: Declare Sub Sptoutin (Byteout as Byte) In that case the variable Byteout would not be valid globally, but only “within the subroutine. The subroutine call would then take the form: Sptoutin 6 ‘This saves one line compared to the previously shown call. ‘You can also transfer a group of several variables to a subroutine in a ‘subroutine call, as can be seen from the example of the Bascom Sptout ‘subroutine: Spiout Dout , 1 For novice programmers, it's generally safer to use only global variables in your own subroutines. For advanced programmers, on ‘the other hand, selecting the optimal form of data transfer is a major ‘consideration. Listing 8. Using the software SPI (excerpt). Contig Spi = Soft , Din = Pind, Ss = None , Clock = Portb.S , Node = @ sprint €s Alias Portb.2 barb.2 = 1 cse2 cs=0 Dout = 4 ‘Sptout Dout , 1 sea Wattns 1 cs=0 Dout = 3 ‘Sptout Dout , 1 Dout Spiout Dout , 1 ‘write disable ‘read "a8. als » Dout = Portb.3 , Dout = @ Spiout Dout , 1 I Do Locate 1, lear lca" sptin Din, 2 Locate 2, 1 Led Din lea" Print 1 Print" Print o1n Wattas 200 T=tel If T >= 65535 Then Exit Do Loop time for storing the data. If you exceed the page boundary (I tried it), the result Is chaos. Then the data you find in memory is totally different from what you wanted to write to memory. For this reason, the example program carefully obeys the rules and writes 128 bytes to the first page from address 0 to address 127, with the data val- es In ascending order from 0 to 127. When the data Is read back, that's exactly what you see. ‘As usual, the entire program code (UNO_spiEE1. bas) can be downloaded from the Elektor website [1]. It performs the following actions in sequence: © Command 6, write enable ‘© Command 5, read status register; display for 1 second Listing 9. Data storage in the timer interrupt routine (excerpt). Ting_isr: Dout = 3 14000 4s Addr = High(seconds) Tiner® = 6 Dout = Adar 18, .AIS Tacks = Ticks +1 ‘Sptout Dout , 1 If Ticks = 25 Then ‘Addr = Low(seconds) Ticks Dout = Addr *0...A7 U = Getadc(4) Sptout Dout , 1 veusa End If ‘Addr = Seconds bout = u Addr = Addr and 127 Spiout Dout , 1 If Addr = @ Then ‘start of page ‘Adar = Seconds cs=0 ‘Addr = Addr And 127 Dout = 6 twrite enable If Addr = 127 Then "End of page Sptout Dout , 1 cs=2 cs=a End If Wattus 100 Print Seconds cs=0 Seconds = Seconds + 1 Dout = 2 narite If Seconds = @ Then Seconds = 65535 Sptout Dout , 1 End IF Wattus 100 Return cs20 IoT & the Search for a Protocol By Jens Nickel ‘The ‘Intemet of Things’ (I0T) is set to change our ives. In order {or the dverse equipment to understand each other a consistent, application orientated protocol is needed. For example it's neces- ‘sary to define a general purpose representation of measurement ‘values from sensors and positional information for actuators. Ideally the protocol will not present too much of an overhead for ‘a small, low cost microcontroller. At the same time it needs to be versatile enough to cater for the maforty of applications that ‘you might find, for example in a Smart Home environment [1]. This won't be a cakewalk, that's why in our March 2014 edition ‘we appealed to the Community to help us out and send their Ideas to our competition web page at [Link], In the mean time we have already recelved many Interesting ‘suggestions that got us thinking, as well as pointers to suitable existing protocols (2][3]. Now we would like to provoke some discussion by outlining a framework for a new protocol that is stll‘work In prog- ress’. This Is your chance to make a difference—if you think the Ideas are {good then say so, if ‘you can think of a bet- ter solution we want to hear it. This Is going to be a collaborative effort and we think, with your help we will arrive at an optimal solution. Get involved and {90 to [Link]! (140298) 1] [Link]/index. php xapautomation Eifesceee neues Messoges are passed between devices, for example es HTTP- POST data (this need further discussion). Each message consists of a header together with blocks of information. The type of information block is defined by two key words (these are described as ‘Case’ or ‘Mode’, see below). A few examples: “Set Value’: Sets an actuator to a value “Current Value’: A measurement value is transferred “Query Value’: Request a measurement value to be transferred (Polling) “Set LimitMin’, ‘Set LimitMax’: Set end limits to an intelligent sensor ‘Set Interval’: Set up sensor to send measurement at intervals ‘Option Min’, ‘Option Max’: The sensor indicates its range ‘Option Value’ (followed by physical units and size): “The sensor indicates what itis measuring and the ‘measurement units. ‘An outline of the message structure (optional information is bracketed), the // symbol means ‘or’): Message: Header + Block (+ Block (+ Block ..)) Block: Case + Mode + (Subnode +) (Location +) (Time +) Case: ‘Set’ // ‘Current’ // ‘Query’ // ‘Option’ Mode: ‘Value! // ‘Min’ // “Max! // ‘Interval’ // ‘LimitMin’ // “LimitMax’ Subnode: Integer Location: LocationString: String Time: TimeString: String Data: Value (+ Value (+ Value + ..)) Value: Number + (Accuracy +) Unit + Quantity Number: Float // Integer // Binary Accuracy: Float // Integer Unit: Unitkey: String ‘Quantity: Quantitykey: String In the end all elements of data types Float, Integer, Binary Und String will be returned. Exactly how the elements (in ASCII format) are represented must be defined. Amongst ‘other things we also need suggestions for the representation ‘of lacation and time (LocationString, TimeString). [A versatile protocol will allow the messages to be made ‘simpler or more complex. For example where Presets, (predefined actuator positions) are not required or where ‘sensor alarm information such as (‘limit exceeded’) must be ‘accommodated in the message. ‘© Command 2, write 128 bytes starting at address 0 ‘© Command 3, read memory starting at address 0; endless loop Data logger ‘One practical application for the 64-KB memory sa data logger. The objective here Is to acquire ‘measurement data from the ADC4 analog Input ‘once per second and store the data. The memory will be full in approximately 18 hours. You don't always have to program everything ‘yourself, since Bascom has a lot of ready-made functions for many situations. In this case you have the option of configuring an SPI interface as ‘a software interface using any desired port pins or {as a hardware interface using the microcontroller pins designated for this purpose. The hardware ‘SPI is especially fast and is commonly used for tasks such as driving graphical displays. However, this involves a whole lot of parameters that must ‘be configured for each specific application, which requires a detailed study of the ATmega328 data sheet. Things are a bit easier with the software ‘SPI function, and it provides a reasonably high transmission rate. Although writing your own SPI procedure from the ground up is not a bad idea because It allows you to implement the timing diagrams in the data sheets very clearly, the ready-made software SPI is more convenient and faster, which is why we use It here. The interface configuration specifies which lines are to be used. For Din, Dout and Clock we use the familiar MISO, MOST and SCK lines, which are already avallable on the ICSP connector. The 8S line corresponds to the /CS line. In this case this line should not be operated automatically by Bascom because itis usually necessary to trans- fer a lot of data during a single active chip select phase. Consequently, this line (on port pin 82) will still be operated “manually”. The Wode = © setting Is also important, because there are four different SPI modes. ‘The program excerpt in Listing 8 shows how the software SPI Is used to read dats from the serial EEPROM. The instruction Spiout Dout , 1 sends ‘exactly one byte, which is transferred in the vari- able bout. In the other direction, the instruction sptin Din , 1 reads one byte, which is then avaliable in the variable Din. The entire program reads all the data from the EEPROM and shows the contents on the display and on the terminal ‘emulator screen. As usual, the entire program code (UNO_spiLog~ [Link]) can be downloaded from the Elektor website [1]. It is too large to be listed fully here. Pressing SI starts a measurement run. It can be stopped at any time by pressing S2, after which the stored data can be read out. ‘A timer interrupt routine (excerpt in Listing 9) Is used to control the timing during data acqui- sition. The voltage on ADC4 Is measured and stored once per second. The 128-byte block size of the EEPROM Is taken into account. At the start of each block, a write access Is started and the current address Is transferred, followed by 128 bytes of data. At the end of the block, the /CS line is pulled high to allow the EEPROM to store the entire block. Since the /CS line Is connected to port pin PB2, LED2 on the shield is lit when the line Is high. The LED therefore flashes each time ‘a block of data has been transferred to memory. (1an24s-t) Web Link [1] www [Link]/140245 Ree sa nee ued eet Many readers who are using the original Arduino boot loader have encountered the following problem: if a program performs serial output, the Arduino programmer in Bascom does not work properly the next time and hangs. Some readers have discovered that this problem can be resolved by using the Arduino IDE before using the programmer again. Simply transferring the Blink program, for example, is sufficient. However, there's an easier solution. After you launch the programmer in Bascom, briefly press the Reset button on the Uno board three times (or more) at roughly 1-second intervals. After this the programming function will work again, even when ‘the previous program included a Print output. This has been discussed intensively in various topics on the Elektor forum (now at [Link]). If you use the MCS boot loader on the Uno board, this problem does not occur. Another alternative is to use an external programmer. By Karl-Anton Dichtel (Germany) Lux Meter With 1 Ix to 100 kIx measuring ranges Conventional incandescent lamps have been phased out in many areas in the world, including the EU and Canada. What's left is halogen lamps, energy-effi- cient lamps and LED lamps, which in fact have better efficiency thai ndescent_ lamps. However, if you want to know how much better they are, you basically have to trust the manufacturer's data. Although trust is good, measuring it your- self is better. The lux meter described here can help. Particularly with low-cost LED lamps from low- wage countries, some of which are alot cheaper than brand-name products, you can hardly be blamed for being a bit sceptical. For example, among the 500,000 Items available on a well- known auction site there are exactly five 3-watt spot lamps with an asking price of 1 euro. The aimed light output ofthese lamps is 210 lumen. “That sounds realistic, but is it actually true? ‘To avold being forced to rey on such claims, you need an instrument that can reliably measure brightness. Such an instrument is useful for more than just checking whether a lamp works. For ‘example, I's good for checking whether a partic- tular workstation is adequately litand in many countries there are regulations governing this. ‘Aux meter also offers other benefits with lamps. For example, for a variety of reasons halogen lamps are popular for certain purposes. The brightness of halogen lamps tends to decrease over time, in part due to the accumulation of ‘metal vapor deposits on the glass envelope. AS a result, at some point you are getting more heat than light from the lamp. Even energy-efficient lamps get darker with age. Measurements can help you decide when It’s time to replace a lamp that is no longer up to snuff. Measuring light ‘The key element of every instrument is a sensor that converts the desired physical quantity into an easily measured voltage, preferably with a reasonably linear conversion characteristic. Of course, nowadays a fair amount of electronics is integrated into many sensors to enable them to output digital data directly. However, the tradi- tional sensor for light is a photodiode. Along with types intended for the transmission of infrared signals, there are special, more accurate types for measurement purposes. The BPW21R used here is an example of the latter type, and it is suitably housed in a sturdy metal TOS package with 2 glass window (Figure 1). ‘This hermetically sealed photodiode has a rela- tively large active area of 7.5 mm2, which gives It high light sensitivity. This sensor is also very linear, as can be seen from the data sheet [1]. Figure 2 shows the Incredibly linear relation- ship between the short-circult current and the illuminance. The log-log scale is necessary here because the sensor covers an extremely wide brightness range of more than seven decades. Even though this is a conventional discrete com- ponent, Its specs are nothing to be ashamed of, which makes it the Ideal light sensor for use in 2 DIY lux meter. A photodiode generates a current that is propor- tional to the illuminance in lux, rather than the luminous flux in lumens typically used in lamp ‘specifications. As 1 lux is equal to 1 lumen per square metre, tis theoretically easy to convert between the two units, but the area illuminated by a lamp depends on the distance and on the beam angle. Thanks to the Internet, you don’t need a pencil and paper for this and you don’t have to solve any algebraic equations. Various websites (e.g. (2]) have ready-made calcula- ‘tors where you can simply enter a value in lux or lumens and convert it into the opposite unit. Measurement method As already mentioned, the output from the pho- todlode Is Its short-circult current. A highly lin- ear relationship between illuminance and signal level can only be obtained by operating the pho- todiode under virtually short-circuit conditions, which means that the circult for measuring the current must have a very low input impedance. However, the quick and dirty approach of con- nected a 1-ohm resistor across the sensor and ‘measuring the resulting voltage drop Is not very effective because the voltage at 1 lux would only be about 9 nV. Although microcontrollers with Integrated A/D converters are usually the pre- ferred choice for measurement tasks, the A/D converters of commonly used microcontrollers are not sultable for such low voltages. With the 10-bit resolution of a typical AVR microcontroller and a reference voltage of 1 V, the resolution Is around 1 mv, which Is five orders of magnitude greater than the full-scale value of the lowest ‘measuring range. The idea of using an interme- diate voltage amplifier to boost the voltage by a factor of 100 million (to obtain roughly 1 V at 1 Ix) is unrealistic, since the signal would be com- bad '* Lux meter with 6 measuring ranges: be, 10 bx, 100 bx, 1 kx, 10 kbc and 100 kbc ‘* Measurement resolution in 1 Ix range: 0.01 be '* Autoranging '* Automatic power down after 1 minute '* Measurement data output over serial interface ‘© All components readily available '* Low power consumption ‘* Can be calibrated over the serial interface using a terminal emulator pletely buried in the noise (among other prob- lems). We therefore need a different approach, The circuit in Figure 3 shows how it can be done. The sensor D1 Is connected across the Inputs of an opamp (IC3), which has its non-in- verting input connected to a reference voltage of approximately 0.65 V at the junction of R3 and D2. When light shines on D1, it wants to supply a current, which would cause the voltage on the Inverting input of IC3 to drop. This Is countered by negative feedback through Ri or R2. Conse- quently, the voltage at the output of the opamp rises to the level necessary to cause the current flowing through the feedback resistor to offset the current from the diode, so the voltage between the inputs of IC remains at zero. As a result, D1. {s virtually short-circuited and therefore oper- ates with a very linear characteristic. To obtain the highest possible measurement accuracy, we chose a Microchip MCP6061T for 1C3 instead of a standard opamp. With extremely low input cur- ‘ents inthe picoamp range and an offset voltage cof only 150 pv, this opamp will not degrade the ‘measurements. The converted sensor signal from 1C3 Is routed to connector K2 for test purposes. let Cheuk Cara (A) AOE 10" 10% 10" 108 102 108 108 10° ,-mumnarce Figure 1 ‘Schematic depletion ofthe [BPW2IR package. Figure 2. ‘The short-circuit current (i) versus illuminance (Ey) characteristic ofthe PWR. Figure 3. ‘The circuit of the x meter essentially consists of a sensor, a signal amplifier, precision A/D converter, ‘2 microcontroller and a display module. Circuit description ‘At 1 lx (with RE1B in the Ri position) the volt- age at the output of 1C3 Is appreximately 1.8 mv (9 nA x 200 kQ) That is still not enough for the. AID converter of a normal microcontroller. Tis could be solved by adding another gain stage, but if you want a precise instrument it’s better to-do the job right. Instead of using the so-so AADC integrated in the ATmega328, we opted for an external A/D converter (IC2) with an inte- grated precision input amplifier and a resolution of 18 bits, That may sound lke overall, and in fact we only use 15 bits, but there are good rea- sons for this approach. One of the main reasons Is the measuring range, since the working range of the human ‘eye extends from 1 Ix (full moon) to 100 kix (full sunlight). This corresponds to five full decades, and it requires circuity with a very large dynamic range. This requirement is met a follows: Swftch- Ing between R1 and R2 changes the signal level by a factor of 100. This means that the subse- quent circuitry only has to handle a dynamic range of 1000:1. Since the internal gain of C2 can also be switched between 1x and 8x, the resulting dynamic range Is sufficient to obtain accurate measurements over the complete instru- ‘ment measuring range. For the detals of switch- ing the gain levels and measuring ranges, see the firmware code. In any case, the accuracy of the circult Is approximately 0.5%. C2 is controlled by the microcontroller (IC1) over the IC bus. It is used to read data and configure C2. The relay Is also controlled by two outputs. of IC1, which allows the current through the coll to be reversed. This direct drive arrangement Is possible because the relay (RE1) can man- ‘age with a coll current of about 24 mA. REI Is, ‘hat Is called a bistable relay, which only needs a short current pulse in one direction to switch over and remains latched in the new position without any current. It can be switched back by simply applying a short current pulse in the opposite direction. This method prolongs battery life, Relay contacts are preferable to any sort of MOSFET for this application. C1 also perform three other tasks. Some of its pins drive the LCD in nibble mode, and another pin switches the backlight on via 73 during a measurement session if necessary. Two other pins handle serial data transfer with TTL signal levels. The pin assignments of K3 are compatt- ble with the Elektor 808 USB to serial converter. This gives you a serial interface over USB if you need it. ‘Another function of IC1 Is to maintain the supply voltage (via pin 24 and transistors T2 and T1) for one minute after the device is switched on by pressing button 1. After this the Instrument switches off automaticaly to prolong battery life. ‘Another remark about power: with a battery-pow- ered instrument such as this, it’s worthwhile to take measures to save energy. One of them is the previously mentioned control of the LCD backlight via T3, or disabling the backlight with 3P2, Illumination is only activated automatically and adapted to the ambient light level when the ambient light level is less than 64 Ix. This smart measure by itself saves about 20 mA under bright Conditions, which is fairly significant considering that the current consumption of the rest of the Instrument Is 14 mA. To avoid the power dissi- pation of a standard voltage regulator, a special low-power, low-dropout regulator (1C4) Is used to generate the regulated 5 V supply voltage. ‘The regulator can work down to a battery voltage of approximately 5.5 V. at lower input voltages its output voltage collapses. The display stops working below 5 V. Choke Li provides additional filtering for the analog portion of 1C1. Construction When fiting components on the PCB you need a steady hand, a fine soldering tip, small tweezers, soldering paste or thin wire solder and a magni- fying glass. Although the sensor is an old-fash- loned leaded component, nearly all of the other components are SMDs— mostly in the especially popular 0603 package In the case of resistors. and capacitors. If you are newcomer to solder- Ing SMDs, we strongly advise that you draw on the assistance of an experienced colleague or friend. Figure 4 shows the component side of a prototype circuit board, which should be enough to give you an idea of what you're up against. ‘A magnifying glass is helpful when you're look- Ing for shorts between SMD pins, which can be eliminated with a bit of desoldering braid and flux. That works fairly will with the pin spacings. of the ICs used here. ‘The display Is mounted on the other side of the board (see Figure 5). If you want to mount the LCD module so it can be removed, you can imple~ ‘ment the connection using socket headers and long pin headers, However, if you don’t need to have a dismountable display you can simply use sultable lengths of wire (16 in total) I's import- ant to keep the LCD module spaced away from board enough to avoid short circuits between the rear of the display module and the PCB. Firmware ‘The software for IC1 was written In C using the free AVR Studio IDE (V4.18) and Is available free Of charge (Including the hex file) from the Elektor ‘web page for this project [4]. A 200 kHz clock Figure 4. ‘The component side ofthe ‘assembled prototype board. [Neary all ofthe components ‘are mounted on this side, ‘even the tiny” SMDs in type 603 packages. Figure. Circuit board with mounted ‘display module. The photodiode and the power- ‘on button are atthe top. (5 ms period) for the relay drive pulse is derived from the 10.24 MHz clock signal, and from it a 2 Hz clock is derived for sequence control. The code for driving the LCD was taken from an arti- dle on the [Link] website [5], where It is extensively commented. The IC routine Is, recycled from a project for a weather station, which Is why it contains unused (leftover) code for a rotary encoder and other things. If you are interested in the details, see the comments in the code. ‘The standard ISP connector K1 is used to down- load the program code to IC1. To power the microcontroller during the download process, the battery must be connected and JP1 must be ‘Component List Resistors Default ratings: 1%, 0.1W, SMD0603 Ri = 200k2 R2 = 2k2 R3 = 3.9k2. Rd, RS,RE,RB,RL0,R13 = 10k2 R7 = 822 RO = 33k Rit = 47k 5% R12 = 330 5%, 0.2W R1g,RAS = 2200 5%, 0.2W PL = 10k 20% trimpot, SMD, Vishay TSS3Y}103MRi0 Capacitors Default ratings: 10%, 10V, SMD0603 Ci = 1nF 5% SOV, COG/NPO C2 = 3.3uF 20V, SMD case A, tantalum, €3,C5 = 470nF (C4,C13 = 4.7UF 20V, SMD case A, tantalum C6,C12 = 10uF, SMD Case A, tantalum’ C7,C14 = 10nF SOV. (C8,C9 = 22pF SOV, COG/NPO C10,C11 = t00nF 25V Inductors LL = 10pH 20%, 250mA, 1.05, SMD 0603 fitted. Alternatively, you can keep S1 pressed for the duration. The program starts running for the first time after the download. Since all bytes in the EEPROM memory of a new microcontroller are set to $FF and are restored to that value when the device is reprogrammed, the microcontroller can easily see whether the Instrument Is starting up for the first time. If the value FFpex Is found in two EEPROM bytes, a calibration factor of 1 is provisionally stored in those two bytes. This corresponds to a conver- sion ratio of 9 nA = 1 lx. More precise calibration can be performed later via the serial interface. The next step Is to measure the offset of the ‘opamp and the A/D converter without the pho- ‘Semiconductors ‘BPW2IR, TO-S D2 = 754148 RY, SMD 0805 D3 = PMEG2010AEH, SMD SOD-123F Ti = IRLML6402PBF, SMD SOT-23 2 = BCB48B, SMD SOT-23 3 = 2N7002, SMD SOT-23 IC = ATmega328P-AU, programmed, SMD TQFP 328 IC2 = MCP3421A0T-£/CH, SMD SOT-23-6 IC3 = MCPEOGLT-E/OT, SMD SOT-23-5 C4 = LP2951-50D, SMD SO-8 Miscellaneous K1 = 6-pin (2x3) pinheader, 0.1” pitch BT1,.P1,P2,K2 = 2-pin pinheader, 0.1” pitch (K2 } optional) 43P3 = 2-pin pinheader, 0.1” pitch, right angled k3 = 5-pin pinheader, 0.1" pitch, right angled 31, P2,3P3 = jumper, 0.4" pitch ‘Si = pushbutton, 6x6 mm, vertical, make contact BTi = 9-V-battery clip with wires, optionally with re- cceptacles for pinheader pins REI = relay, bistable, 2 contacts, coll 4.5V / 202.52, 41A/L10VDC contacts (Panasonic AGN2104H) Xd = 10.24MHz quartz crystal, 18pF; HC-49/4H LCD1 = LCD module, 2x8 characters, backit, 58x32 ‘mm, TCO8028-01YA0 (Elektor Store # 120061-75) PCB # 130109-1

You might also like