Fullscreen HTML5Med Res (??MB)HTML4

Sometimes measuring the temperature is just not enough; you need to see how the temperature changes over time. This simple, compact and inexpensive thermometer provides a low-cost solution. Graphing Thermometer Andrew Woodfield’s O ver the past year or so, I’ve found myself designing an increasing number of circuits with surface-mount devices (SMD). That required a set of tools for building prototype SMD printed circuit boards (PCBs) in my workshop. One of those tools is a hot plate reflow soldering system, made from a 500W electrically heated metal plate measuring about 100 × 200mm. The controller adjusts the heating of the plate so it matches the recommended temperature profile for reflow soldering of SMD components. To design the controller and build the reflow plate system, I needed a way to measure temperatures as high as 300°C. As Lord Kelvin once said: When you can measure what you are speaking about, and express it in numbers, you know something about it. When you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the stage of science. This certainly applies to SMD reflow soldering. Without careful and accurate temperature Photo 1: a typical low-cost non-contact infrared temperature measurement ‘gun’. 78 Silicon Chip low-cost measurement directly on the PCB, ideally as close as possible to the solder paste and components, and a method to observe the temperature variations as changes are made to the reflow hot plate hardware and controller software, it is very difficult to understand what is going on. Temperature measurement devices Initially, I purchased an inexpensive infrared (IR) ‘gun’ for these measurements, shown in Photo 1. It displays the average temperature of 3-10cm diameter sections of the hot plate. This is the approximate area that the gun’s sensor measures. The measured area also depended on the distance between the gun and the hot plate. It was challenging to get consistent temperature measurements. The reflow controller I was designing measured the hot plate temperature with a thermocouple mounted inside the hotplate. This temperature was shown on the controller’s LCD screen. Thermocouples are made from two dissimilar metal alloy conductors welded together at one end. The measurement point looks like a tiny metallic ball. This sensor generates a tiny voltage that is proportional to the temperature of that ball. Commonly available low-cost K-type thermocouples can be used to measure temperatures from about -200 to +1370°C. However, results sometimes differed between the infrared gun and the hotplate thermocouple. Which value was correct? Given that they were Australia's electronics magazine measured at slightly different locations, they could both be correct. As the saying goes, “A person with two clocks never knows the right time”. I needed another thermometer, ideally using a small, low-mass temperature sensor to precisely measure the temperature at the desired location. Thermocouples come in many forms. Some are enclosed in a protective cover or are physically large. These can have a significant ‘thermal inertia’, taking some time to report the temperature, so I wanted to avoid using those. Also, I wanted a thermometer that was small enough to be shifted easily around the workbench to measure temperatures at different locations in the reflow equipment. I also recognised that it was essential to graph the measured temperatures during the entire reflow process. These ‘temperature profiles’ (the temperature variation over time) can vary dramatically with each change in configuration, equipment, control software and reflow materials. That made it all the more important to view the overall result for each four to five-minute reflow test run. It is possible to buy a low-cost thermocouple-based thermometer interface for use with a laptop or tablet. However, such arrangements are quite unwieldy in many situations. All the required power cables add to the muddle on my already untidy bench. I tried to find a ready-made graphing thermometer, ideally a compact, portable, battery-powered device. You might think there are many such devices for sale. But surprisingly, at siliconchip.com.au least when I searched, I could not find anything suitable aside from several rather expensive meters from US retailers. This led me to design this simple and very low-cost thermocouple graphing thermometer. Design objectives As I mentioned, I need to measure temperatures up to 300°C. A measurement accuracy of ±10°C is acceptable in this case, provided the results are repeatable. When reading 300°C, that equates to a ±3% error. This appears to be in line with many commercial thermometers. Since there are other applications for this type of thermometer, I felt other ranges would be useful. The Thermometer’s design therefore allows for maximum temperatures on the LCD screen vertical axis ranging from 50°C to 600°C. While some K-type thermocouples can handle temperatures up to 1300°C, the materials used in the least expensive sensors appear only suitable for temperatures up to about 600°C. However, health and safety risks (and my anxiety) substantially increase above 600°C, so that sets the upper limit. Temperature samples taken at rates from once per second to, say, once every 5-10 seconds seemed suitable. Looking at a number of potential LCD screens, and allowing for graph axes and labels, 100 samples plotted along the horizontal axis of each graph appeared to be the practical limit. That gave an equivalent horizontal time axis span of between 100 and 1000 seconds (less than 2 minutes to a little over 15 minutes) per screen. Most reflow profiles run for 3-6 minutes (180 to 360 seconds), so this was well within this measurement range. To allow for other uses, I extended the sampling rate range further. The final range of 1-900 seconds per sample (up to 15 minutes) provides a maximum screen x-axis span of up to 25 hours or just over one day. Hardware considerations The typical approach for temperature measurements is to use a lowcost K-type thermocouple, an Analog Devices (originally Maxim) MAX6675 or MAX31855 8-pin IC as the interface device, and a microcontroller. The display, whether a 7-segment LED display, LCD screen or OLED screen, is also driven by the microcontroller. siliconchip.com.au Parts List – Graphing Thermometer 1 single- or double-sided PCB coded 04102261, 70 × 56mm 1 ATtiny85-20PU 8-pin, 8-bit microcontroller programmed with 0410226A. HEX, DIP-8 (IC1) 1 8-pin DIL IC socket 1 1.7-inch 128×64-pixel LCD screen with 3.3V UC1701X controller (LCD1) [AliExpress 32215047945] 1 0.9-3.3V to 3.3V boost regulator module (REG1) [AliExpress 4000252822321] 3 4-pin PCB-mounting tactile switches (S1-S3) [Altronics S1122, Jaycar SP0603] 1 DPDT slide switch (S4) [Altronics S2010, Jaycar SS0852] 1 AA or AAA cell holder (BAT1) [Altronics S5026/S5054 or Jaycar PH9203/PH9261] 1 AA or AAA cell (BAT1) 1 2-way barrier screw-type terminal block connector (CON1) [RS 144-8151, AliExpress 4000170287617] 1 8-pin header socket and matching header strip (CON2) 1 2-pin header socket and matching header strip (CON3) Capacitors 1 47μF 16V electrolytic 2 100nF 50V ceramic Resistors (all axial ¼W ±5% or better) 3 10kW 2 4.7kW 1 150-1000W (depending on desired backlight brightness; see text) I was keen to minimise the component count, along with the overall size and cost of the device. I wrote some software a few years ago for a soldering iron temperature controller. Originally designed for cheap thermistor-monitored soldering iron handpieces, some builders had encouraged me to try to modify it for the less common thermocouple-based handpieces (see www.zl2pd.com/ SolderingStation.html). It proved possible to do this with the 8-pin ATtiny85 that I’d used for the original controller. I didn’t develop that option any further – I already had the thermistor-type soldering station I’d designed. That software used an unusual analog-­to-digital converter (ADC) feature available on a few members of the ATtiny microcontroller family. This is a two-input balanced ADC interface that also includes an optional integrated 20× gain stage. There is also a 1.1V internal ADC reference available. Those features make these ATtiny microcontrollers suitable for thermocouple sensors. The ATtiny85 was the smallest in this family, coming in an 8-pin package. Would it be possible to connect everything else that would be necessary for a graphing thermometer within that 8-pin package limit? Australia's electronics magazine To answer this question, I turned to the display. I wanted to graph the results on a small graphics-­ capable display. One set of possibilities was the compatible and well-known 128×64 pixel 0.96in (24mm) or 1.3in (33mm) OLED screens. While the text on these is perfectly readable, my eyesight is no longer ‘fighter pilot’ quality. Tests confirmed that the individual graphed pixels were just a bit small. I did not want to be trying to peer at a display located in close proximity to something at 300°C. I value my eyebrows! Another option was one of the larger legacy KS0108B or T6963Ctype 128×64 pixel LCD screens. While low in cost, these demand a cluster of processor pins for data and control. A 1.8in (46mm) colour TFT LCD screen also looked possible, but the software necessary to drive its larger 160×128pixel display looked likely to exceed the limited 8kiB firmware space in the ATtiny85. I found the solution in a low-cost UC1701X-based 128×64 pixel 1.7in (43mm) LCD. It’s large enough to be readable, includes an excellent integrated LED backlight, and is controlled over a simple three-wire SPI serial peripheral interface, along with chip select (CS) and reset inputs. March 2026  79 Fig.1: the Graphing Thermometer is based on ATtiny85 microcontroller IC1. Its eight pins are just sufficient to handle thermocouple sensing, LCD updates and pushbutton sensing. Three pushbuttons would also be required for the Up, Down, and Next user control pushbuttons. When I added all this up, it would require ten I/O pins. The ATtiny85 only has six. However, with a little software and hardware effort, it proved possible to squeeze everything onto that device. The LCD screen also determined the power supply design, as it requires a 3.3V supply. Tests during development showed that this voltage is surprisingly critical. If the LCD supply voltage falls below 2.7V, it starts to misbehave, reversing and inverting text and graphics in quite disconcerting ways. So, in case you are tempted, this is not a design to power from a pair of 1.5V cells. Circuit description Fig.1 shows the circuit of the graphing thermocouple thermometer. You will see there’s very little hardware in this meter! The Microchip (Atmel) ATtiny85 microcontroller handles the measurements, drives the LCD screen and senses button presses. It’s clocked from the ATtiny85’s internal 8MHz RC oscillator divided down to 4MHz to reduce current consumption. A tiny boost regulator module increases the AAA cell’s 1.5V output to a reliable 3.3V supply, as required for the LCD screen. Most of these modules at the time of writing use the very efficient ME2108A switching regulator. However, the prototype’s module used a TPS61201. These operate similarly. These regulators are rated up to 1A, but the current drawn by this meter is a miserly 6mA at 3.3V. The 80% regulator efficiency results in about 15mA being drawn from the 1.5V battery. This gives a reasonable battery life for typical intermittent use. Since omitting the typical MAX6675 thermocouple interface chip results in the need to use one of the balanced analog-to-digital converters in the ATtiny85, two input pins are required on the ATtiny85 for the thermocouple. These are both internally configured as ADC inputs. One of these pins (pin 2) is then tied to ground. This may seem to be a significant waste of I/O resources, but that’s the only option when using the balanced ADC mode. The user pushbuttons connect to a single pin on the ATtiny85 via four resistors. Each pushbutton produces a different DC voltage on pin 1 of the ATtiny85. This voltage is read by the internal 10-bit analog-to-digital converter, allowing the pushbutton status to be determined by the software. Unusually, this pin is also used as an output. It drives the active-low chip select line for the LCD’s UC1701X on-glass controller. The UC1701X controller inside the LCD only sees a voltage that falls below 0.7V as a valid CS low signal. The pushbutton resistors are therefore selected to avoid this voltage range. The prototype Graphing Thermometer (shown at actual size here) was built on a singlesided PCB. The AA or AAA-cell holder on the back props it up at a useful viewing angle for the LCD. I used a 1kW backlight resistor. 80 Silicon Chip Australia's electronics magazine siliconchip.com.au Even if a pushbutton is pressed during a display update, the selection is (briefly) ignored. The resistor values ensure that the LCD only sees the processor’s commands. Once the display is updated, the processor can go back to detecting the pushbutton status. This novel arrangement ensures there is no disturbance to the operation of the display, yet key presses can be detected as needed. The LCD’s SPI connections use the three remaining pins. The LCD’s remaining pin, the reset input, is only toggled when the LCD’s power is applied. It’s not otherwise used. A resistor and capacitor are used to generate this power-on-reset signal. This avoids using another processor pin. The LCD has an integrated LED backlight. A 150-1000W resistor supplies 2-10mA current for this from the 3.3V rail. Higher values provide extended battery life – with a 1kW resistor, the backlight current is just 2mA. Alternatively, the backlight supply resistor may be removed (using reflected light only to see the screen). Another option is to connect an external toggle switch connected in series with it to save power when making use of the longer sampling intervals that are possible, while still being able to use the backlight when required. A 1kW backlight resistor gives a relatively modest light level. If brighter light levels are required, this can be changed, for example, to 560W (moderate brightness) or 150W (high brightness). voltage and calculates the temperature from this. This value is reported on the LCD screen, in the lower lefthand corner, and the value is plotted on the graph. After 100 samples are plotted, the screen is erased and a new plot begins automatically. Accurate thermocouple measurements require cold-junction compensation. This method measures the temperature at the thermocouple meter terminals (the ‘cold junction’) using another thermometer and corrects the thermocouple measurement accordingly. The ATtiny85 has an internal temperature sensor, but I am not using it. Instead, a simple ‘thermocouple offset’ value is entered in the meter, a method suggested by Horowitz and Hall in the reference textbook “The Art of Electronics”. As the authors suggest, it works well when measuring over these temperatures (see below). Construction The graphic thermometer is built on a single-sided 70 × 56mm PCB that’s coded 04102261 (see Fig.2). All the components are mounted on this, including the AAA cell holder. No additional wiring is necessary, which makes for a fast and easy build. Start by soldering the resistors and capacitors. Then, using some of the cut-off leads of those resistors, install the three PCB links located at the top-centre of the board (if you’re using a commercially made board, it might not need links fitted). Next, gently bend the leads of the 47μF capacitor at right angles and then mount it as shown in Fig.2. This polarised component must be mounted so that the shorter negative lead is closest to the edge of the PCB. The three standard tactile 6×6mm pushbuttons can be fitted next. The pushbuttons come in a variety of shaft Cold-junction compensation for thermocouples The meter software is written in Bascom, the Basic-like compiler for the AVR processor family. This allows for relatively quick software development. The compiler generates surprisingly efficient code, and the resulting software occupies about 90% of the available program space. After initialisation and user selection of settings, the device spends most of the time waiting for the current sampling period to expire. During this time, the pushbuttons are checked for a user ‘Next’ command. That will result in the meter exiting the display mode and returning for new user settings and a new graph plot. After each sampling period has passed, the ATtiny’s analog-to-digital converter samples the thermocouple A conductor generates a voltage proportional to the temperature difference across it (Fig.a). This is called the Seebeck effect. We described this in detail on page 51 of the November 2023 issue (siliconchip.au/Article/16013). This tiny voltage cannot be measured directly because it is cancelled out by the voltage generated as a result of the connections required by the measurement circuit. However, by joining two conductors, each made from a different material, the difference between the voltages generated by this pair of conductors can be measured. These wires are welded together at one end to form the ‘thermocouple’ (Fig.b). K-type thermocouples made from Chromel and Alumel are the most common. These produce about 41µV/°C. This value varies slightly due to slight manufacturing and materials differences. They are typically used to measure temperatures from -200°C to +1350°C. However, the thermocouple’s voltage is proportional to the temperature difference between the hot Fig.a: a temperature junction (the measurement point) difference across a conductor and the cold junction (the connecgenerates a small voltage. tions to the measurement circuit). The temperature at the cold junction must therefore also be known to calculate the actual temperature at the hot junction. The best way to do this is to integrate another type of temperature sensor into the measurement chip. However, since an instrument like Fig.b: a thermocouple temp sensor. this is normally used over a relatively narrow range of ambient temperatures (eg, the indoor temperature may normally only vary between 20°C and 30°C), and the required accuracy may not be high, a simpler cold-junction calibration method may be employed. By measuring the thermocouple’s output voltage at a known temperature with another sensor, it is possible to determine the hot junction temperature. If calibrated this way, as long as the ambient temperature doesn’t vary much, the readings will still be reasonably accurate. siliconchip.com.au Australia's electronics magazine Software March 2026  81 Fig.2: the single-sided PCB holds all the parts required for the meter, including its power supply. If a doublesided PCB is used, the wire links are not required. The 2-way socket above S1 provides a solid mounting for the LCD. lengths but, in this case, any length will work just fine. Next, mount the 8-pin DIL socket for the ATtiny85 IC and install the 12-way and 2-way socket strips for the LCD connections. In the prototype, I only installed eight of the 12 pins (LCD pins 5-12), since pins 1-4 are not actually used by the display. The matching pin-strip connectors should then be fitted to the LCD. The pins on the prototype had a conical taper and a flat pin face (see Fig.3). These were soldered with the flat pin face mounted against the LCD. This ensured the longer pin shafts are free to mate with the matching pin-socket strip. Do not plug in the LCD screen yet. This will be done after completing some initial tests. Next, solder the DPDT slide switch and the two-way 7.62mm spacing screw connector for the thermocouple. Don’t connect the thermocouple itself just yet. It tends to get in the way of the remainder of the assembly. The boost regulator module can now be mounted on the PCB. It should be installed close and parallel to the board, to allow the LCD screen to be plugged into place. Mount the AA or AAA cell holder on the solder side of the PCB. If you have a commercially made PCB, you can solder this in place easily from the component side of the PCB, since the pads are almost always throughplated. If you have made the PCB at home, you will need to leave a 3-5mm space between the base of the battery holder and the PCB so you can carefully solder these pads with the tip of your soldering iron. You’ll find that when the thermocouple thermometer is operating, the battery holder gives the user a convenient viewing angle for looking at the LCD screen. Initial testing Do not plug in the ATtiny85 IC or LCD screen yet. Insert a fresh AA or AAA 1.5V cell. Zinc-carbon or alkaline types may both be used. Connect a DC voltmeter between pin 8 (positive meter lead) and pin 4 (negative meter lead) of IC1’s socket. Switch on the power and confirm that the meter reads between 3.0 and 3.3V. Next, making sure no buttons are pressed, check that the voltage on pin 1 of IC1’s socket is also 3.0-3.3V. Press the Up button only, then the Down button only, and finally the Next button only. The voltmeter should read 2.1V, 1.5V and 1.0V (±0.1V) for these tests, respectively. If these are not correct, check that the resistors are fitted in the correct locations, soldered correctly, and that there are no solder bridges. Programming the ATtiny85 Fig.3: the pin-strip connector should ideally be fitted as shown. 82 Silicon Chip You may have purchased a preprogrammed ATtiny85. In this case, you may ignore this section. If not, you will need to program your ATtiny85’s flash program memory with the project’s HEX file, then program the ATtiny85’s fuse settings. These configure some internal ATtiny85 settings. The complete details are shown in Australia's electronics magazine the panel titled “Programming the ATtiny85”. Final assembly Carefully insert the programmed ATtiny85 into its socket, then plug the LCD screen into its sockets. You may wish to add a couple of small self-­adhesive rubber feet to the lower underside edge of the PCB, to avoid scratching your test bench. The battery holder provides the upper edge footing and allows the meter to rest on the bench or table at a useful viewing angle. If you wish, the meter may also be mounted in an enclosure, but this is not essential. Finally, connect your K-type thermocouple to the screw connectors (CON1). Thermocouples are polarised. It must be connected correctly or the meter will not operate properly, so if you get strange measurements, try swapping the leads. Operation After switching on the power, a ‘splash screen’ graphic will briefly appear on the LCD. This is cleared, then a prompt asks the user to select the maximum temperature for the vertical axis graph display. Use the Up and Down keys to change the value (in 50°C steps), or press the Next key to continue with the default value of 300°C. The second and final prompt asks the user to select the required sampling rate. There are various possible values for this period, from 1 to 900 seconds. Use the Up and Down keys to change the period, or press the ‘Next’ key to use the default period of three seconds. 100 temperature measurements are then taken and plotted on each screen. The value of each measurement is also briefly written in the LCD’s siliconchip.com.au Programming the ATtiny85 Unless you purchase a programmed ATtiny85, it is necessary to program your blank ATtiny85 before using it. A programmer like the USBasp (www.fischl. de/usbasp) is required. It can be purchased online from many suppliers often for less than $10. Such programmers are used with a PC or laptop. Suitable software is available for Windows, Linux and macOS online. This description will focus on the Windows platform. You will also need an adaptor to connect the appropriate DIL IC pins to the programmer. My 8-pin adaptor was published in the September 2020 issue (on page 47; siliconchip.au/Article/14563) and the PCB is still available (from siliconchip.au/Shop/8/5642). The drivers for the chosen programmer must be installed prior to using it. The drivers for the USBasp can also be obtained from the link above. Programming software is required to actually program the ATtiny85 from Windows, Linux or macOS. Suitable free software for Windows includes eXtreme Burner (siliconchip.au/link/ab3m), AVRDUDESS (siliconchip.au/link/ab3n) and Khazama (siliconchip.au/link/ac9e). There are a number of websites and YouTube videos describing the setup and use of these programs. Here is a summary of the procedure required to program the ATtiny85 for this project: O Load the USBasp drivers onto the Windows PC O Plug in and complete the installation of the USBasp programmer. If the option is present on the USBasp programmer, and some boards support this feature, select 5V operation rather than 3.3V for programming the ATtiny85. O Download the programming software and install it. Once running, select “ATtiny85” as the target device. O Download the HEX file for this project (siliconchip.au/Shop/6/3578) and select it as the file to be used to program the ATtiny85. Note: Some versions of the Extreme software require the replacement of the chips.xml and fuselayout.xml device files to program the ATtiny85. These two files are found (in a typical Windows install of Extreme) under C:\Program Files\Extreme Programmer AVR\Data. Rename the original file called chips.xml to oldchips.xml and fuselayout. xml to oldfuselayout.xml. Then unzip the new files from the file extremeXML­ update.zip into that directory. Restarting Extreme will allow the programming of this and several other AVR devices. lower left-hand corner. When the graph reaches the right-hand side, the ATtiny85 clears the display, redraws the graph axes, then proceeds to plot another 100 measurements. Holding the Next key down for a second at any time during the graph display will cause the meter to exit the measurement and display process and then prompts the user for new settings again. Calibration Holding the Next key down when the meter is switched on enters a special routine to calibrate the meter with its low cost K-type thermocouple. These thermocouples vary slightly due to materials and manufacturing. This routine allows an offset to be entered to better match the display to the actual temperature, particularly in the 0-200°C range. siliconchip.com.au Connect the thermocouple to the meter and check the current room temperature from another thermometer. Let’s say it reads 18°C. Switch on the meter while holding the Next button down. After the splash screen, it will then report the temperature as seen by the thermocouple, and prompt the user to adjust this to match the correct value. Use the Up and Down keys to do this. Then press the Next key again. The meter will then save the required offset in the ATtiny’s internal EEPROM for future use. With a 10-bit ADC, the temperature resolution is around 4°C. Thus, you may not be able to get the calibration exact; within about ±2°C is good enough. This usually only needs to be done once. From this point onward, the meter will use these settings. They Australia's electronics magazine Screen 1: one of the graphs produced by my Thermometer from my reflow hotplate. The thermocouple was placed on top of a PCB, and the temperature was sampled once every three seconds. The latest measurement is reported in the lower-left corner of the screen. Screen 2: a domestic oven was set to 170°C and the internal temperature was monitored with the thermocouple mounted centrally above a wire tray. The graph shows the oven reaching temperature after about eight minutes, rising and falling slightly as the oven holds that temperature. remain set at these values until you decide to reprogram them. Powering down does not alter them. Final comments I found that this meter met practically all of my requirements. I’ve used it extensively to plot numerous runs of my reflow hot-plate system. I’ve also used it, for example, to check the operation of our oven. A feature I would have liked to include in the design was a method to save the results shown on the LCD screen for long-term documentation. Sadly, there were no free pins available to support that function. Taking a photo of the LCD screen is probably the easiest solution. In the meantime, it’s proving very handy. No doubt, other uses will come to mind now it is in my workshop. I hope you find it equally useful! SC March 2026  83