Fullscreen HTML5Med Res (??MB)HTML4

RGB LED ‘ANA This colourful and unique clock tells the time with LEDs arranged in a circle that light up in different colours to represent the hour, minute and second ‘hands’. There’s even a light chaser effect you can optionally enable that has the LEDs racing around each second. It’s a straightforward build that looks great when finished. By Nicholas Vinen T his RGB LED Clock is simple and stunning, using very few components outside the LEDs. I came up with the idea of this clock being vaguely aware of the somewhat similar “Mesmeriser” clock by Grantronics that was published in the June 2005 issue and available as a Jaycar kit at the time. I knew it used single-colour LEDs; I thought it’d be much nicer to make my Clock from multi-coloured RGB LEDs. This is a hybrid digital/analog clock. All the principles involved in timekeeping and the display are digital, but it mimics an analog clock in how it shows the time. Rather than two or three physical hands that rotate about the centre, pointing to the numbers, 60 LEDs are arranged around the outside of the face. They light up where an analog clock’s hands would point. A real analog clock uses different hand sizes to distinguish the hours, minute and second hands. Instead, this clock lights up the LEDs in different colours. If they ‘overlap’, the colour is a mixture! For example, if the hour hand is red and the minute hand is green, when they point to the same number, the LED lights yellow (which is what you get if you combine red & green light). Funnily enough, if you are familiar with that clock, you will see that I’ve had some similar ideas to its designers, the main one being that it incorporates an optional ‘chaser’ that runs around the clock face once per second. It’s very eye-catching and also serves to indicate to the beholder the passage of each second. Having said that it’s similar to the Mesmeriser, my chaser does operate a bit differently. You can see a video of the clock in operation with this feature enabled at siliconchip.au/Videos/ RGB+Clock If you don’t like or want the chaser, it’s easy to turn off with a button press. Includes all the parts in the parts list except the power supply. The microcontroller is pre-programmed. Choose a BZ-121 GPS module or Pico W (that you need to program) for the time source. One major difference from the Mesmeriser is in terms of complexity. I wanted this clock to be as simple as possible (“keep it simple, stupid!”). So besides the LEDs, there’s just one microcontroller, 20 resistors and a few other bits and pieces. That earlier design used way more parts and would have been a lot of work to build. I would say this one is elegant in its simplicity. I also wanted to keep the parts relatively cheap so that it could be made into an affordable kit. I considered using SMD RGB LEDs, but the 5mm through-hole types I ended up using work extremely well and are quite inexpensive in bulk. The only expensive part is the PCB because it’s quite large at 200mm in diameter. Still, you wouldn’t want a wall clock much smaller than that. I think the KISS (keep it simple, stupid) principle worked out quite well because this design is considerably easier to build than the almost 20-year-old Mesmeriser design. That one had a digital display as well, but I thought most people would want either a digital or analog like readout, not both. One concession I made to Australia's electronics magazine siliconchip.com.au RGB LED Analog Clock Complete Kit (SC7416, $65.00) 66 Silicon Chip ALOG’ CLOCK ● 200mm black clock face with 60 RGB LEDs that light different colours for the hour, minute and second ● 12 different colour schemes to choose from ● Optional ‘subsecond’ hand chaser ● Adjustable, automatic PWM-based dimming using an LDR to sense the ambient light level ● Two optional single-colour LEDs (any colours) to indicate AM or PM ● Time source: GPS module or NTP time via the internet using WiFi ● Accuracy: typically within one second ● Time zone: from GMT−14 hours to GMT+14 hours in 15 minute increments ● Daylight saving: manual one hour toggle ● Power supply: 5-12V DC <at> 50mA from plugpack or USB ● Time source baud rate options: 4800, 9600, 19200, 38400, 57600, 115200 digital clocks is to add optional AM/ PM indicator LEDs to make the time unambiguous. While I think this clock is nicer than the Mesmeriser, the kit is less than half the cost (even less if you consider 20 years worth of inflation!). It is also a bit larger than the Mesmeriser, and since the LEDs are right at the edge, the display is larger again. I didn’t think it was realistic to power a clock that uses LEDs for time display from a battery, so it’s simply powered from a DC supply between 5V and 12V (ideally in the range of 6-9V). That could mean a plugpack or USB supply. The simplest way to connect it is to have a thin figure-8 wire hanging down from the bottom of the clock to the nearest power point. If you really wanted to power it from a battery, with the average current draw of around 25mA for our prototype, you could expect four AAs to power it for around five days. So you’d want to use high-capacity rechargeable (eg, NiMH) cells to avoid spending a fortune on alkaline cells. In terms of timekeeping, I have offloaded that to your choice of either a GPS module or a WiFi module that siliconchip.com.au fetches NTP time via the internet and your WiFi network. That makes the clock extremely accurate, with no drift, while also helping to keep the circuit simple. The Clock is designed to hang on the wall as a PCB assembly. It might get a bit dusty, but you don’t normally touch clocks often, so it doesn’t strictly need a case. If you want to put it in a case, you can likely find a suitable one. The easiest solution is probably to buy a cheap clock in a plastic case that’s large enough, gut it and install this PCB in that case. Just make sure it’ll fit first! Circuit description The full circuit is shown in Fig.1. The key to keeping it simple is the use of Charlieplexing to allow all 60 RGB LEDs, containing a total of 180 junctions (plus the two optional AM/PM LEDs) to be driven from a 20-pin IC. This technique involves keeping most of the pins connected to the LEDs in a high-impedance state (eg, configuring them as digital inputs). One output is driven high (to +5V) and the other is pulled low (to 0V). A different LED junction is connected across each possible pair of pins, so depending on Australia's electronics magazine which pin is high and which is low, only one lights. Due to the way the connections are arranged, we can actually light any combination of the red/green/blue junctions in a single LED at any time. By multiplexing them (switching quickly between different states), we can make it appear that multiple LEDs are lit at once. In the case of this clock, we need to light up to five LEDs at once: for the hour ‘hand’, minute ‘hand’, second ‘hand’, the optional ‘subsecond hand’ (which goes around the face once per second) and the optional AM or PM LED. That means each LED is lit for a maximum of 20% of the time. To compensate for that, we drive them pretty hard, so they still look quite bright. By lowering the amount of time they are lit for (ie, having all LEDs off sometimes), we can control the brightness, too. The formula for the number of LEDs (ℓ) that can be driven by a Charlieplexed arrangement for a certain number of pins, n, is ℓ = 2(n−1) + 2(n−2) + 2(n−3) + ... + 2, which can be simplified to ℓ = n(n−1). This is a quadratic equation, so we can solve it for n and get the formula n = (1 + √1 + 4ℓ) ÷ 2. May 2025  67 Plugging in our value of ℓ = 182, we can determine we can do it with exactly 14 I/O pins. However, our circuit uses 15 because that allows us to use a much simpler arrangement where figuring out which pins to drive high and low to light any given LED is trivial. You can see the arrangement we used in the circuit diagram. The RGB LEDs are grouped in sequential sets of four, having their common anodes all tied together and connected in turn to I/O A (LED1LED4), B (LED5-LED9) etc, up to the 15th I/O, O (LED57-LED60). For the first group with their anodes driven by I/O A, the cathodes are driven by B, C, D, ... K, L and M. For the second group with the anodes driven by I/O B, the cathodes are driven by C, D, E, ... L, M and N. After the third group, which ends with the final cathode being driven by the 15th I/O, O, it wraps around to A again. The critical thing is that no I/O pin appears twice in the same group. The AM and PM LEDs are connected to spare combinations of pins that are not used for any of the RGB LEDs. To light one LED, all we have to do is figure out which group it is in and drive the corresponding shared anode pin high. We then pull one, two or three of the cathode pins connected to that LED low to light it with a particular colour. The nets designated A through O connect to pins on microcontroller IC1 via 68W current-limiting resistors. These were calculated with the microcontroller’s absolute maximum current limit per pin of ±25mA, as well as the typical limit for an LED being 20mA (although, with duty cycle always being under 50%, it isn’t really a concern). Each LED will have two resistors in series when it is lit, one in the anode circuit and one in the cathode circuit. Assuming the lowest LED forward voltage at 20mA is 1.8V and the supply is exactly 5V, that means there will be 3.2V across the resistors, allowing a maximum of 23.5mA to flow (3.2V ÷ [68W × 2]). However, the microcontroller’s output transistors also have an inherent resistance that we can calculate as being close to 68W from information in the data sheet. This means that the series resistance for each LED is effectively around 200W, so the actual current limit is closer to 16mA, comfortably under the 25mA limit. 68 Silicon Chip Brightness adjustment We want the LEDs in the clock to be bright in a well-lit room but not so bright at night so they don’t sear your eyeballs. Thus, there is a light-­ dependent resistor (LDR) near the middle of the clock face that senses the ambient light level. It forms a divider with the 100kW resistor in series wtih it, across the 5V supply rail. At higher light levels, the LDR’s resistance is low, so the RA5 analog input of IC1 will be close to 5V (probably around 4.5V). As the light level drops, its resistance will increase to 100kW and above, so that voltage will drop to 2.5V and lower. The microcontroller can thus use its analog-to-digital converter (ADC) to read the voltage on that pin and adjust the LED brightness. That is done using pulse-width modulation (PWM). We’ll explain how it’s implemented in the software section. Cheekily, we also use pin 2 of IC1 to sense presses of pushbutton switches S1 and S2. They are used to change the clock’s configuration, set the time zone, compensate for daylight saving, adjust the LDR sensitivity and so on. When one is pressed, it pulls pin 2 either almost all the way up to 5V or all the way down to 0V. The LDR’s resistance doesn’t vary enough to allow the voltage to get that close to either rail, so the microcontroller knows a button has been pressed. The 220W series resistors are low enough not to interfere with that, but high enough to avoid damage in case both buttons are pressed (an invalid condition). Timekeeping While IC1 has an internal oscillator, it isn’t super accurate, so it’s only used for timekeeping from second-to-­ second. For longer intervals, we rely on the time source: either a GNSS (eg, GPS) module, which gets its time from atomic clocks in satellites, or a module that fetches the time via internet NTP servers using a WiFi network. Either way, we’re relying on that module to have a crystal for reliable timekeeping, and we simply get its updates (once per second or more frequently) and display whatever time it gives us. The GNSS or NTP module is connected via six-pin header CON3. Some modules have four or five wires, in which case some of these pins are not connected. Actually, only three are required: Vcc (5V) and GND (0V) to Australia's electronics magazine supply the module with power, and TX, which is the pin it uses to send serial data to our microcontroller that includes the time. We provide a pad for soldering the module’s RX wire that pulls it high via a 10kW resistor so that the module doesn’t get spurious serial data due to EMI. We never actually need to send it data. The 1PPS pad is provided as a place to anchor a 1PPS wire if the module has one; we don’t need that either. The EN pad will pull up the module’s EN wire, if it has one, to enable it. Remaining circuitry IC1 has a 100nF supply bypass capacitor for stability, plus a 10kW pull-up resistor on its MCLR (reset) pin to prevent spurious resets. Optional in-circuit serial programming (ICSP) header CON2 provides a way to reprogram IC1. While some LEDs will light dimly while doing this (as pins 18 & 19 of IC1 are used for both programming and driving LEDs), we didn’t find this interfered with programming the chip. All that remains is the simple linear power supply. REG1 is a low-dropout 5V regulator that allows you to feed in a higher voltage (6-12V DC) and it will provide a nice, stable output to run the clock. As the current draw is usually less than 50mA, it will only dissipate 350mW at most ([12V – 5V] × 0.05A). It’s in a medium-sized SOT223 package soldered to the board, so it will handle that easily. The AMS1117 regulator requires a 1µF ceramic capacitor on its output, plus an input bypass capacitor of at least 100nF, so we have provided those. They are both 1µF to simplify construction. Mosfet Q1 provides reverse polarity protection, as the power input is via a simple two-pin header or soldered wires. If the supply polarity is connected correctly, Q1’s gate is pulled up, and it switches on, connecting the supply negative wire to circuit ground. If the supply wires are swapped, Q1’s gate is pulled down and its body diode is reverse-biased, so no current will flow. ZD1 protects Q1’s gate in case a negative voltage exceeding 16V is applied to the board, however unlikely that is. Options While we are specifying common-­ anode RGB LEDs, it will actually work siliconchip.com.au Fig.1: the circuit is dominated by the 60 RGB LEDs that connect to microcontroller IC1 via fifteen 68W series resistors. The micro can light any element of those LEDs (or the two extra ones) by bringing one of the connected pins high and the other low, while the remainder are kept as high-impedance digital inputs. siliconchip.com.au Australia's electronics magazine May 2025  69 with common-cathode RGB LEDs too. All that needs to change if CC LEDs are used is for the drive polarity to be reversed, ie, instead of pulling a pin high, it is pulled low, and vice versa. The only catch is that if you fit the AM & PM LEDs (LED61 & LED62) and are using common-cathode RGB LEDs, they need to be reversed (rotated by 180°) so that their polarities match the other LEDs. PCB layout You would think with all these LED connections, the PCB layout would be a nightmare, but actually, it was straightforward. We have kept it as neat as possible, and quite symmetrical, since the PCB also forms the clock’s face – see Fig.2. We could have hidden most of the circuitry on the back, but we thought it’d be more Fig.2: the 200mm diameter PCB forms the clock face, with the 60 RGB LEDs arranged in 6° increments around the outer edge. interesting to have some on the front! The microcontroller, IC1, is right at the centre of the face, which seems appropriate since it’s also logically at the centre of the circuit. Most of the resistors are to its left and right. The bypass capacitor is above it, while the LDR is centred below it. The pushbuttons are on the left and right sides, lined up with IC1. The power supply components mount on the back of the PCB, towards the bottom. We initially thought of using the auto-router to make the 244-odd connections to the LEDs, but came up with a better (and much neater!) idea. The 15 I/O lines (labelled A through O on the circuit diagram) are assigned to 15 bottom-layer circular tracks that run at fixed intervals inside the circle of RGB LEDs. Top-layer radial linear tracks run from each RGB LED pad partway towards the centre of the PCB, all terminating just above the innermost bottom-layer ring track. With this arrangement, all we needed to do was place one via on each of those radial tracks at the appropriate location to join it to the correct circular track (A-O). One of each of the radial tracks connected to the A-O lines is then routed to the series resistor that connects to the appropriate microcontroller pin. (Thanks to Tim Blythman for his help in suggesting this neat arrangement!) While there are pads for the three connectors to be inserted from the front side of the board, we siliconchip.com.au thought it would be a bit ugly having them stick out. These pads don’t have exposed copper on the front of the board, to keep it looking clean, but it is still possible to solder vertical or (ideally) right-angle headers on the back side. Most constructors would not need to fit the ICSP header. Since GPS modules usually come with plug-in wire assemblies, there’s no real need to fit a header for CON3. Instead, we suggest you simply solder the wires to those pads on the back, as we did. We stuck the GPS module on the back of the PCB with double-sided tape to hide it, as otherwise the wiring will look a little messy. That just leaves the power input, CON1. You have a few options there. You could solder a right-angle polarised header to the back of the board and use a plug to connect it. You could also just solder bare wires (either directly from a plugpack, or to an inline barrel socket) to those pads. There’s also the option of using a small, separate board we developed that can be soldered to the main board using a pin header, which has an onboard USB socket. You can then use a 5V USB supply to power the clock. While that will probably mean REG1 will be in dropout and not regulating, and the LEDs might not be quite as bright, we haven’t found it to make that much difference. That small add-on board will be useful in many applications, so we’re presenting it as a separate project. Software The software is simple in principle, although it is actually quite involved when examined in detail. The microcontroller runs with a 16MHz main clock (‘Fosc’) that results in a 4MHz instruction clock. Three hardware timers are used: TMR0, TMR1 and TMR2. The other peripherals we need are the analog-to-digital converter and the UART for serial reception. TMR1 is used for timekeeping and to control the main loop rate. It is a 16-bit timer running from Fosc ÷ 4, and it uses a 1:1 prescaler, meaning it overflows at a rate of 61.035Hz (16MHz ÷ 4 ÷ 65536). Happily, this is almost exactly what we want. Assuming the subsecond hand/chaser is enabled (and it looks cool, so why wouldn’t you?), the subsecond hand makes 61 steps each siliconchip.com.au second. That’s because it has to go around the clock face once (60 LEDs) plus advance one step with the second hand (plus one LED). This is pure coincidence, but it works out really well! While we advance the clock to the next second after 999.4ms (61 ÷ 61.035), we expect the GPS/NTP module to give us a time update after exactly 1000ms, so that event is used to reset the timer. That way, as long as we are getting time updates every second, there are no glitches in the time display. TMR0 and TMR2 are used for LED multiplexing and PWM, respectively. They are both 8-bit timers that run from Fosc ÷ 4, using a 4:1 prescaler, so they run at 3.906kHz (16MHz ÷ 4 ÷ 4 ÷ 256). Each time TMR0 overflows, it triggers an interrupt routine that switches to lighting the next LED (using port masks precalculated by the main loop). That means each LED is lit for 51μs at a time, giving a duty cycle of approximately 20% at full brightness. TMR0 and TMR2 run in lockstep. At full brightness, TMR2 never triggers. The PR2 register controls how soon the TMR2 interrupt occurs after TMR0 overflows. As the value in PR2 is reduced, the TMR2 interrupt occurs earlier and when its interrupt service routine (ISR) is called, it simply switches all LEDs off. The earlier that happens after switching on, the dimmer the display becomes. This means we can use the PR2 register as an 8-bit PWM control for all LEDs, since it occurs at the same interval after each multiplexing step. Besides calculating which LEDs to light based on the current mode and time, the main loop also receives serial data from the GPS/NTP module, decodes it and uses it to update the current time. After getting the UTC/ GMT time, the configured timezone offset and DST offset (if enabled) is applied before updating the clock face. It reads the voltage from the LDR divider on each run through of the loop, filters it, and uses it to calculate the new brightness level. It also checks if button S1 or S2 has been pressed, performs debouncing, looks for long or short presses and changes the mode as appropriate. Construction The RGB LED Clock is built on a large circular PCB measuring 200×200mm (ie, 200mm in diameter) with a black Australia's electronics magazine solder mask, coded 19101251. We recommend you start by fitting the topside SMDs, with IC1 being the first. You should have a temperature-­ controlled soldering iron, fine-tipped tweezers, solder wire, a syringe of flux paste and a roll of solder-­wicking braid on hand. Work in a well-ventilated area, either using a fan, next to an open window or outdoors as flux smoke is not good for you if you inhale it. If you bought a kit or programmed microcontroller from the Silicon Chip Online Shop, your chip will already be programmed and you can just solder it to the board. If you have a blank chip and the ability to program it offboard, do so now. We described how to do that in our article on the PIC Programming Adaptor from the September 2023 issue (siliconchip.au/ Article/15943). If you don’t have such a programming rig and your chip is blank, solder it now and you can use CON2 to program it later. Before soldering it, make sure you identify pin 1 and line it up with the markings on the board and in the top-side component overlay diagram, Fig.2. Fixing a reversed SMD chip is not fun! Tack one pin to the board using a little solder, then inspect the leads to verify they are all over their pads. If not, remelt that solder and gently nudge the chip into position. Verify again that the IC is orientated correctly, then solder the diagonally opposite pin. Apply flux paste down the pins on both sides, then solder all the remaining pins. You can do it individually, by adding a little solder to the iron tip and touching it to the junction of the pin and pad, or you can drag-solder them several at a time. If any bridges form (which is likely with drag soldering), add a little flux paste and then use solder-wicking braid to remove the excess solder. Inspect all the solder joints and pins under magnification to ensure all pins have been soldered to their pads with good fillets and no bridge remain, then move onto the passives, using Fig.2 as a guide. It’s best to fit the 220W, 10kW and 100kW resistors first, since all the remaining resistors will be the same value (68W). The sole capacitor on this side is the 100nF type. You can use a similar technique as for the IC: tack the part to one pad, check its orientation, adjust if necessary, then solder the other pad. Add a May 2025  71 tiny bit of flux paste to the first one and heat it with the iron to reflow the joint. With all these parts in place, clean the flux residue off the board using alcohol or a flux solvent. It’s much easier to clean the board before any through-hole parts are in place. This will be the clock face, so you want it to be nice and presentable! You’ll probably have to clean it two or three times, using a fresh section of lint-free cloth each time, to get it looking nice. Now solder the two tactile switches using the same technique. This will probably leave a little flux residue, but we don’t want to soak the switches in solvent as it can damage them, so apply Fig.3: there aren’t too many components on the back – six SMDs for the power supply near the bottom, up to three connectors (depending on your preference), the time source (a GNSS module shown here) and... some wires to hang it on the wall. The wiring shown suits the BZ-121 module we used but virtually any 5V-powered TTL module should work. You can also use a Pico (2) W programmed as an NTP Time Source. a little solvent to a section of the lintfree cloth and wipe off the flux residue using that. Flip the board over (rest it on a cloth so its face doesn’t get dirty or scratched), then solder the parts on the reverse side, as shown in Fig.3. Don’t get Q1 (possibly coded XORB) and ZD1 (likely coded T12 or Z3) mixed up, as they are in the same type of package. When finished, clean up the flux residue, although thoroughness is less crucial since it will be against the wall later. You can now solder the LDR on the front side; you may need to bend its leads in to make them fit the pad pattern. Make sure it’s straight and close to the board so that it looks nice. Soldering the LEDs Now for the big job. This will probably take you at least a couple of hours. Don’t rush it as it’s harder to fix problems like bridged pads than it is to do them right the first time. Make sure you fit each of the 60 RGB LEDs the right way around. Pay attention to the orientations of the flat faces on the PCB silkscreen and in Fig.2. For each LED, insert it and then rest the board on your work surface such that the weight is on that LED, so it’s pushed into the board. You could place something like an eraser between the lens of the LED you’re soldering and your bench to ensure that. Then solder one of the outer leads, with the soldering iron on the outside, using a minimal amount of solder. siliconchip.com.au Don’t use too little; you need to be able to see that a fillet has formed. Still, you should use just enough solder to get that fillet. Next, check if the LED is flat and straight. If not, you still have a chance to push it into the PCB with your finger while you remelt that initial solder joint. Once you’re satisfied, solder the opposite lead, again with the iron coming in from the outside. The critical part here is that the pads are very close together, with large holes and thick leads, so it’s quite easy to bridge the pads and somewhat difficult to fix it if a bridge forms. For the two remaining leads, bring in the soldering iron from the middle or outside of the board so it touches the pad and lead, then feed in a minimal amount of solder from the outside until it melts, as shown in Fig.4. Again, use just enough to form a fillet. Repeat for the final lead/pad, then use sharp side-cutters to trim all four leads reasonably close to the PCB and evenly. When trimming, use eye protection and/or hold onto the leads so they don’t fly into your eye! They can attain quite a high velocity when cut through. Now check to make sure the pads have not been bridged. If you have followed our instructions carefully, they should not be. If they are, add some flux paste and use solder-­wicking braid to remove the excess solder until the bridge is clear. You can then add a tiny bit of solder back to the pads to ensure there’s enough. Repeat until you have soldered 5-10 of the LEDs. Testing the LEDs It’s best to test the LEDs as you go, because a short circuit at any point on the board can cause the whole display to go haywire (after all, there are only 15 separate tracks connecting to all 60 LEDs). When first powered up, the firmware runs a display test where LED1-LED62 are lit up white in sequence at about 4Hz. This should allow you to quickly spot problems. I found the easiest way to power the board was to get a regulated 5V supply (eg, a bench supply) and use clip leads to attach two male-male jumper wires to its outputs. With the clock face towards you, right-side up, rest the black (negative) wire in the right-most terminal of CON1 (the one labelled GND in Fig.2). The weight of siliconchip.com.au Fig.4: no doubt there are many ways to solder the RGB LEDs without accidentally bridging the very closely-spaced pads, but this is what we found worked best. The soldering iron tip and solder should come in from 180° opposite positions along the long axis of the pads. That minimises the chances of accidental bridges, which are difficult to clear. The final PCB will feature more widely spaced pads to make this a little less tricky than our prototype (although we managed to do it). the cable will bring it contact with the plated through-hole. Similarly, insert the positive jumper wire into the side of CON1 labelled + and ensure the two wires are not shorted, then switch the supply on. Ideally, it should be current-limited to around 50mA (the board will draw less than 20mA in this configuration). Power it up and check that LED1 lights, then LED2 etc. If you don’t have a current-limited supply, you could use a fixed-voltage DC supply with a series resistor to prevent damage in case there is a fault. For example, a 12V supply with a 150W 1W series resistor would easily deliver the ~20mA needed to power the circuit for testing but would be limited to 80mA in case of a fault (with the resistor dissipating just under 1W). When powered, if any of the LEDs don’t light up white, or they light at the wrong time, switch it off and check the most recent LEDs you’ve soldered for bridges or bad joints. Once they all light up OK, go back and solder a few more LEDs. Keep testing until all 60 are fitted and they are lighting up nicely. You can then solder LED61 and LED62 if you are going to use them. Remember to reverse them if you used common-cathode RGB LEDs. They will light up individually during the test, after all 60 of the RGB LEDs. While the LED solder joints are on the back of the board, if you have a lot of flux residue left behind, you may want to clean it up to avoid getting sticky hands when handling the clock. A proper flux remover is better than alcohol here, as alcohol dries fast, but you can use it if that’s all you have. Because of the sharp leads, you will have to dab it off, rather than wipe it off. Australia's electronics magazine It took a little while, but we cleaned this flux residue up. Try not to let any get onto the face (ie, the opposite side of the PCB), or you’ll have to clean it again so it looks nice. Final assembly There isn’t much left to complete the clock. If you’re using CON2, fit it now; most constructors will not need it. We suggest you stick the GNSS or NTP module to the back of the clock using double-sided foam-cored tape. Cut it to size and stick it on. The GPS module will usually have a flat, non-­ conductive ceramic antenna that you can stick somewhere without too many tracks. The NTP module has conductive parts on the back, so make sure none of them can short to the PCB. Most of the back of the Clock PCB is covered with a solder mask, so short-circuits are unlikely, but you should still check. If you’re using an NTP module like the WiFi Time Source for GPS Clocks (June 2023; siliconchip.au/Article/ 15823), you will need to configure it so it can connect to your WiFi network. That is usually best done before wiring it up; refer to that article for instructions. Now wire the GPS or NTP module to CON3. You can use a header and plug(s), if you want, but we think soldering wires directly to the pads is fine. As mentioned earlier, you won’t necessarily have all six wires to connect to CON3. It doesn’t matter; the vital ones are VCC, GND and the TX signal from the module (which goes to the pad marked TX, not RX; ie, the labels are from the module’s point of view). If your module has an EN wire, check that it’s active high. If so, you May 2025  73 There’s nowhere suitable on the PCB to mount a USB connector, so if you want to use USB power, you can attach the add-on board like this. can solder it to the EN pad. Otherwise, solder it to GND. Most modules will have an RX pin, which should be soldered to the RX pad, pulling it high to disable it. Power supply For power, you can solder a lightduty figure-8 lead to the pads for CON1 on the rear of the board and either hard-wire it to a 6-12V DC plugpack or similar, or have an inline barrel socket or other connector at the end of that wire to connect a DC supply. Try to get the polarity correct (refer to Figs.2 & 3), although the board should not be damaged if you do accidentally reverse it. Another option is to build our USB Power Breakout Board, described in the accompanying short article. This is a small PCB that can accept a USB Type-C or USB Type-B Micro/Mini socket and a pin header. It supplies ~5V DC to that pin header when a USB supply is connected. This may not run the clock at full brightness, but it won’t be far off, and USB supplies are common and convenient. We didn’t put pads for a USB socket on the clock PCB itself because there was no room to do it along the edge (the whole edge is filled with LEDs).If the socket wasn’t near the edge, there wouldn’t be room for a cable to plug in, so the PCB would need an ugly cut-out. Spacing this small PCB off the back of the clock PCB on a header provides enough room for the cable to plug in. The only components you need on 74 Silicon Chip that PCB are the USB socket, a 0W SMD M3216/1206 resistor and the pin header. Once that’s assembled, you can then solder it to the back of the Clock PCB as shown in the photo above. Note how the USB Power Breakout Board PCB has four through-hole pads; this allows a two-way pin header to be fitted with multiple different polarities. The header position shown in Photo 1 is required to match the polarity of CON1. If you’re unsure, refer to the separate article and its PCB overlays to see how the header position affects the polarity. Finally, you will have noticed several large rectangular pads on the back of the Clock PCB. These are provided to solder a loop of wire for hanging the clock, with other pads near the bottom to solder wire loops so it hangs vertically on the wall. The photo overleaf shows how we attached the wires to our Clock for hanging; you can use a similar arrangement. You can bend the wires into different positions to suit your hanger (whether it’s a nail, screw, hook or whatever). If possible, do that before soldering the wires to avoid stress on the soldered joints. Final testing Presumably if you’ve gotten to this point, the LED testing went well, so the microcontroller and LEDs are working. There isn’t much else to go wrong, apart from the GPS/NTP module wiring and the power supply itself. We suggest you use the same Australia's electronics magazine procedure you used for testing the LEDs and check that the current draw is under 100mA. With the BZ-121 module we used, our Clock drew around 45mA. If that’s the case, it’s unlikely you have anything really wrong. You can test the LDR-based dimming now. Place a small opaque object like a credit card over the LDR and observe the LEDs. They should dim when you block the light to the LDR and brighten again when you remove the obstruction. You will have to cover the LDR thoroughly, as even a bit of light leading around its edges is enough to make the LEDs quite bright. If that doesn’t work, you probably have a soldering problem with the LDR or its series resistor, a button is stuck down or the LDR is the wrong type. If you’re using a GPS module, we suggest you put the clock near a window and set the baud rate to the correct one for your module, as described below. Leave it powered for around 30 minutes, then come back and check if it’s showing the correct time for London (ie, GMT). If so, that means it has acquired the signal and is decoding the data properly. You can then complete testing by setting it up. If using an NTP module, you should have set it up earlier, so once you have set the correct baud rate (as explained below), it should connect and show you UK time within a few seconds. In that case, proceed to the following section. Setting it up The first setting is the baud rate. This can only be changed after power-­ up when the initial LED test has completed and the spinning chaser (at roughly one ‘rotation’ per second) is operating, to indicate it’s waiting for GPS/NTP data. If you see a clock face instead, it’s likely the initial baud rate was already correct and it’s getting data, so you can skip this bit. The chaser will initially be red if it isn’t getting any valid data, changing to green if there is data but no valid time yet. Once it’s green, it’s usually just a matter of time before it switches to telling the time (assuming you have a strong enough GPS or WiFi signal). During this time, one of the digits 1-6 will be lit blue, indicating the baud rate: 1. 4800 baud 2. 9600 baud 3. 19,200 baud siliconchip.com.au 4. 38,400 baud 5. 57,600 baud 6. 115,200 baud (default) Pressing A will go to the next lower baud rate, while pressing B will go to the next higher one. If you don’t know the correct baud rate, try each one for a few seconds until the chaser changes to green. If it doesn’t for any baud rate, switch off and check your wiring. Remember that TX from the module should go to the TX pad on the PCB. All settings, including the baud rate, are stored in EEPROM, so you shouldn’t have to do this again. The remaining settings that can be accessed in time display mode are: 1. the time zone offset and optional DST (+ 1 hour when activated) 2. the colour scheme 3. the second and sub-second hand modes 4. the LED dimming calibration (minimum & maximum brightness) Once you’re in clock mode and the time has been acquired, you can set the time zone offset. Hold down the A button for a second and release it. The time display should remain, but it will now flash at 1Hz with a 50% duty cycle. If daylight saving is active in your area, hold down B for one second to enable DST mode. A short press of the A button will make the time 15 minutes earlier, while the B button will make it 15 minutes later. Use this to set the correct local time, then hold down the A button to return to the normal display. The time zone you set will be stored in EEPROM. After this, if your area has DST and the time changes, you just need to go into this mode and hold down B for one second, then hold down A for one second to switch between DST and non-DST. Alternatively, you could just change the time zone offset until the time is correct. the second ‘hand’ visible, plus the ‘sub-second hand’ in the same colour. The sub-second hand is a chaser that runs around the clock face each second. It starts where the second hand is, goes all the way around, and ‘pushes’ it over to the next second on the tick. A short press of B will cycle through the four possible second-hand modes: 1. the second hand and sub-second hand are visible and matching colours (the default) 2. the same as #1 but with a dimmer sub-second hand 3. the second hand and sub-second hand are visible, with the sub-second hand being white 4. the same as #3 but with a dimmer sub-second hand 5. the second hand is visible but the sub-second hand is not 6. there is no second hand, only the hour and minute hands Dimming adjustments A long press on the B button in time display mode will switch to the brightness/dimming adjustment mode. In this mode, you can control both the maximum brightness and how the brightness reduces at lower ambient light levels. By default, the maximum brightness is 100%, reducing to a low level, but not the lowest possible, in total darkness. Upon entering this mode, you are adjusting the maximum possible brightness. Pressing A will reduce the maximum brightness and you will see the display dim. Pressing B will increase it (if it is not at its maximum). While making this adjustment, the LDR reading is ignored; you are seeing the brightness level that will be used at the highest possible ambient light level. Keep in mind that, depending on where your clock is positioned, it may not normally ‘see’ a very high ambient light level. For this reason, you can actually set the maximum brightness above 100%. This will not make the display brighter, but it will mean that the ambient light level has to drop further before the brightness starts reducing. The clock face shows a continuous chaser in this mode to help you see the brightness level you’ve set, which spans a portion of the clock face related to the possible brightness range. The portion from six o’clock to twelve o’clock shows the maximum brightness setting, so the chaser gets Parts List – RGB LED Analog Clock In time display mode, short presses of the A button will cycle through the six possible colour schemes for the hour, minute and second hands. Each can be red, green or blue, but they must all be different. Use whatever scheme is easiest for you to remember. The default is blue for the hour ‘hand’, green for minutes and red for seconds. A short press of the B button will cycle through the six possible second-­ hand modes. The default is to have 1 black double-sided PCB coded 19101251, 200×200mm 1 5-12V DC 100mA power supply (6-9V DC recommended) 1 2-pin vertical or right-angle header (CON1; for power – see text) 1 5-pin right-angle header (CON2; optional – for ICSP) 1 6-pin right-angle header (CON3; optional – for GPS module) 1 5V GPS module or compatible NTP time source (MOD1) [BZ-121 GPS module recommended, Silicon Chip SC7414] 2 2-pin SMD black tactile pushbutton switches (S1, S2) 1 20×20mm piece of foam-cored double-sided tape (to affix GPS module) 1 200mm length of tinned copper wire (to make hanger/standoffs) 1 USB power supply module (optional; see text and accompanying article) Semiconductors 1 PIC16F18146-I/SO micro with 1910125A.HEX, wide SOIC-20 (IC1) 1 AMS1117-5.0 15V input low-dropout linear regulator, SOT-223 (REG1) 1 AO3400 or equivalent logic-level N-channel Mosfet, SOT-23 (Q1) 1 BZX84B5V6 or BZX84C5V6 5.6V zener diode, SOT-23 (ZD1) 60 frosted-lens 5mm RGB LEDs (LED1-LED60) * 2 5mm high-brightness LEDs with diffused lens of various colours (LED61 & LED62; optional AM & PM indicators) * the kit will come with common anode LEDs but common cathode types can also be used Capacitors (all SMD M3216/1206 size 50V X7R) 2 1μF 1 100nF Resistors (all SMD M3216/1206 size unless noted) 1 100kW light-dependent resistor (LDR1) [GL5528] 2 100kW 2 10kW 2 220W 15 68W siliconchip.com.au Australia's electronics magazine Colour scheme May 2025  75 shorter as you reduce it and longer as you increase it. A long press of A will return to the time display, while a long press of B in this mode will switch to adjusting the other end of the range, ie, how much it will dim at low light levels. Similar to the maximum brightness setting, you have a fairly wide range of adjustment here, as you may wish the clock to fully dim before it is in total darkness. Or you may wish for it to never go to minimum brightness, even when the LDR sees no light. The default is somewhere between those two extremes. While making this adjustment, the LDR is active and the display will dim based on the current ambient light level and the current setting so you can see its effect. To simulate the clock being in darkness, you will need to cover the LDR with something opaque, like a credit card. Small objects can easily have light leak around the edges, so make sure the object is touching the whole face of the LDR and extends beyond it in all directions if you want to simulate total darkness. Being in a dimly lit room for this adjustment will help. In this mode, a long press of button A will return to time display mode, while a long press of button B will go back to adjusting the maximum brightness, as described above. Using it We attached thick wire to these solder pads on the rear of the Clock, so that it can be hung on a wall. The clock face is designed so it doesn’t take attention away from the LEDs. 76 Silicon Chip Australia's electronics magazine Once you have set your time zone offset, confirmed that your time source is working, adjusted the brightness levels and chosen your preferred colour scheme and mode, the clock is set up and ready for use. If it loses power, when it regains it later, it will go back into exactly the same mode. It just might take a while for your time source to resume, especially if it’s a GNSS/ GPS module. All that remains is to hang it on the wall and connect the permanent power supply arrangement. We recommend soldering an inverted-U-shaped piece of tinned copper wire between one pair of pads on either side at the top of the clock. Bend it so that it will comfortably hook around the head of a screw in your wall, or a wall hook. A couple of smaller loops soldered across the two pairs of pads near the bottom of the clock will stop the bottom of the clock from touching the wall. However, you may wish to have it angled down, as it could make it easier to read. So you could omit those loops, or make them stick out less than the upper one. If there is no power source under where you want to hang the clock, you could run a thin figure-8 cable from CON1 up behind the clock, then along the wall and then down to a more convenient location. The wire will be less visible if it’s the same colour as the wall (you could paint it). While few homes have picture rails on the walls anymore, if you’re lucky enough to have them, they are an excellent way to hide such wire runs! SC siliconchip.com.au