Fullscreen HTML5Med Res (??MB)HTML4

12V Automotive Variable Speed Fan Controller This 12V speed controller could be used in any vehicle with an intercooler or one with inadequate fans – or indeed in any application where there is a need to control the speed of a low voltage DC fan or fans in response to changes in temperature. Simple to wire up, it can control up to 120W of fans. W We deliberately kept the design e designed this Speed as simple and low-cost as possiController to run the ble, while satisfying a long list of intercooler fan on a perrequirements: formance vehicle. We looked for • It had to be easy to wire up, bepre-built units on ebay and AliExcause chasing wires and messing press but nothing really suited the with a packed fuse box in a motor application. vehicle can be a nightmare. Simple 12V on/off thermostats • It must not flatten the vehicle suitable for automotive applicabattery if left unattended for long tions are available but surprisingly periods. expensive given their simplicity. • It needed to be able to run a We found very few which could powerful fan, able to keep a large actually vary the fan speed and engine cool. these were both expensive and • It needed to be easy to set up highly complex, with dozens of and tweak. And so on. wires. Our design fits all the above criWhy do variable speed controllers need to be so complicated? Assuming the fan and battery teria – and can do the job anywhere you need to run a 12V are earthed to the vehicle, all you really need is one wire DC fan to control temperature for power and one for connection to the fan, a temperature sensor and maybe a few adjustments to allow you to set the How it works The general concept is shown in the simplified circuit temperature threshold and so on. Of course, some fans may not be earthed – and there are of Fig.1. In essence, it is a PWM (pulse-width modulation) controller doubtless many non-automotive applications which will with inputs for battery voltage and temperature. A compararequire extra connections – but overall, it’s pretty simple! tor monitors the battery voltage against But we couldn’t find a suitable controlby Nicholas Vinen a 4V reference. This stops the fan from ler, so we decided to build one. 64 Silicon Chip Celebrating 30 Years siliconchip.com.au Features & specifications running if the battery voltage is below a preset value. • Pro portional fan control (PW M, 1% increments) Trimpot VR1 allows the switch-on threshold • Runs from 12V DC voltage to be set between 8.4V and 16.8V. For automotive applications, you would normally set it • Compact, light and easy to build to switch on for voltages above 13.5V, so that the • Designed to survive in the har sh automotive electrical environment controller will switch on when the alternator is • Fan power up to 120 W (maximum current 10A ) running and switch off once the engine (and thus • Fan soft start and gentle spin-dow n alternator) stops. • Under-voltage lockout (UVLO) with hysteresis A comparator feedback resistor adds around 0.5V of hysteresis so that once the supply voltage • Adjustable UVLO threshold (8.4-16.8V) has risen high enough for the fan controller to be • Ultra-low quiescent current when shu t down (<20µA) activated, the voltage must drop by a further 0.5V • Fan switch-on temperatu re adjustable between -7. 5°C and +100°C below this threshold before it will switch off. • Maximum fan speed tem perature also adjustable This prevents the fan from “chattering”, or being • Sea led lug-mount thermisto r can be used for temper rapidly switched on and off. The PWM controlature sensing • Minimum and maximum fan duty cycle can be cha ler includes a two-second switch-on delay which nged (default: 25%/100% • PWM frequency can be ) also helps prevent this. set from 50Hz - 1kHz (de fau lt: 1kHz) Temperature is monitored by an NTC thermistor • Fan speed compensation applied for variations in supply vol tage which is connected in series with trimpot VR2, with the two components connected between the 5V supply rail and GND (0V). This provides a voltage which varies with temperature, rising as the thermistor gets hot- IC1 to measure the battery voltage. ter. This is the control voltage input for the PWM controlThis is a power-saving measure; Q2 is held on while the ler so that the fan duty cycle, and thus speed, rises as the fan is operating but if the low-voltage cut-out is engaged and temperature increases. the fan is switched off, Q2 is also switched off, so no current flows through this divider. It is only energised for around The circuit 1ms every two seconds, when the unit re-checks the supNow have a look at the full circuit of Fig.2. Both the com- ply voltage to see if it is high enough to continue operation. parator and the PWM controller functions are provided by Thus the 0.3mA which would flow through this divider is a PIC12F675 microcontroller. Compared to a discrete de- reduced to an average of just 0.15µA. That’s important when sign the micro gives a lower component count and lower the quiescent current of the rest of the circuit is below 20µA. quiescent current; important when the fan and motor is off. Otherwise, the divider current would swamp it, increasing The PIC does three main jobs: it monitors the battery the quiescent current by a factor of 15 times. voltage, reads the thermistor temperature and drives the We’ve done something similar with the other two dividgate of Mosfet Q1 to control the fan speed in response to ers formed by the NTC thermistor and trimpot VR2, as well these readings. as trimpot VR3. The upper ends of both dividers are shown connected to Soft start and power saving +5V in Fig.1 but as you can see from Fig.2, they are driven The micro provides a soft-start feature where the PWM from output pin GP0 of IC1 instead. duty cycle will only change by 1% every 100ms. This pin is brought high, to +5V during normal operation So if the unit is switched on while the sensor is hot, the but is brought low to 0V when the supply voltage is low, refan will ramp up to maximum speed over about ten seconds. ducing the quiescent current by a further 1mA or so. And This limits the current drawn because GP0 drives the base from the supply and should of NPN transistor Q3 which in also reduce its tendency to turn drives Q2, bringing GP0 “hunt” for a particular speed high enables all three dividers (ie, varying up and down pesimultaneously. riodically). IC1 checks the battery voltOne particular difference age every two seconds if it’s between the full circuit of inactive (due to low battery Fig.2 and the simplified vervoltage) or every 100ms if sion of Fig.1 is that the voltit’s active. The 1nF capacitor age divider which allows IC1 from pin 6 to ground provides to monitor the battery voltage a small amount of filtering for is not connected directly to the this battery voltage, rejecting 12V supply. noise and also reducing the Instead, current flows from source impedance for IC1’s the 12V input, through fuse internal analog-to-digital conF1 and the 470Ω series resisverter (ADC), which can affect tor and then to transistor Q2’s the accuracy of its readings. emitter. Q2 must be switched We stated earlier that the Fig.1: the circuit concept is a comparator to monitor the on in order for current to flow battery voltage and a thermistor to monitor temperature. range of low-battery cut-out to the divider, thus allowing voltages is from 8.4V to 16.8V. siliconchip.com.au Celebrating 30 Years January 2018  65 Fig2: micro IC1 monitors the battery voltage, the air temperature and sends a PWM signal to drive the mosfet, which in turn controls the fan speed. You can verify this by calculating the division ratio of the UVLO divider with trimpot VR1 at both extremes and then multiplying this by the pin 6 threshold of 4V, set in the software. Actually, the threshold is 4.0V for the unit to switch on and 3.8V for it to switch off, ie, there is a 0.2V hysteresis. This translates into a supply voltage hysteresis of around 0.4-0.8V, depending on the setting of VR1 (because of the voltage divider feeding pin 6). This reduces the chance of the unit constantly toggling on and off because of the voltage drop caused by the fan switching on. Temperature sensing When the voltage at pin 6 is high enough for the unit to become active, it measures the voltages at input AN2 (pin 5) and input AN3 (pin 3) every 100ms. The voltage at AN2 is determined by the resistance of the NTC thermistor (which is connected via CON3) and the setting of trimpot VR2. The thermistor has a nominal resistance of 10kΩ at 25°C while VR2 can be varied between about 0Ω and 10kΩ. As trimpot VR2 is turned clockwise, its resistance drops and therefore the NTC thermistor resistance must drop further to achieve the same voltage at pin 5. Since by definition, an NTC thermistor’s resistance drops as its temperature rises, it follows that turning VR2 clockwise increases the required temperature to achieve a certain voltage at pin 5. Analog input AN3 is simply connected to the wiper of VR3, which is connected between GP0 and GND, thus varying the voltage applied to AN3, providing a convenient way to set the temperature required to achieve maximum fan speed. Since IC1’s ADC is configured to use the 5V rail as its ref66 Silicon Chip erence, and the dividers feeding both AN2 and AN3 are effectively between 5V and 0V, the readings it takes at both AN2 and AN3 are ratiometric. Thus, variations in the 5V supply voltage will not change either of these readings, assuming that output GP0 is close to 5V when high; it should be, given the relatively light loading. The software compares the reading at AN2 to a fixed 1V (nominal) reference and the reading from AN3 and uses these values to compute the required duty cycle for PWM output GP5. If AN2 is below 1V, the target duty cycle is zero. If it’s equal to or above the reading for AN3, it will be close to 100% and anywhere in between will result in a duty cycle value between the programmed minimum and maximum values (25% and 100% by default). So as described above, VR1, VR2 and VR3 allow easy adjustment of the three main settings: the switch-on supply voltage, fan switch-on temperature and maximum fan speed temperatures respectively. There are actually three additional settings but these are not set via trimpots (at least, not directly). These are the PWM frequency, the minimum fan duty cycle and the maximum fan duty cycle. They default to 1kHz, 25% and 100% respectively. There is a procedure to go through if you want to change any of these, and the altered setting is stored in EEPROM inside IC1. See below for details. Fan drive The GP5 output (pin 2) drives the gate of Mosfet Q1 directly. Q1 is a low on-resistance, logic-level type with a low gate capacitance. As such, it is reasonably efficient when driven in this manner (without a dedicated Mosfet driver Celebrating 30 Years siliconchip.com.au or even series resistor), although we have purposefully kept the frequency low (50-1000Hz) in order to keep switching losses under control. Basically, there are two types of losses in the fan drive system, both of which contribute to heating in Q1 and if the total is excessive, Q1 could be damaged. These are moreor-less fixed losses due to Q1’s on-resistance and switching losses due primarily to the fact that Q1 is in partial conduction (ie, higher than normal resistance) while it is in the process of switching on and off. The faster Q1 switches, the lower the switching losses but this fast switching requires a high current to be sourced/ sunk to the gate terminal, to rapidly charge and discharge it. Hence, with a relatively low drive strength available from the general purpose output pin on the micro, we can expect higher switching losses. Switching losses are proportional to the drive frequency since the more gate transitions there are per second, the more time it spends in partial conduction. Hence, keeping the frequency relatively low helps. The only real disadvantage is that, since 1kHz is an audible frequency, you may hear some whine from the fan motor when the duty cycle is between 0% and 100%. In our test vehicle, the fan noise is drowned out by the V8 engine. In fact, it’s hard to tell from behind the wheel, whether the fan is running at all (this is not true of the factory-fitted radiator fans!). It may be more problematic if you’re controlling a fan to cool a desktop PC or some other domestic situation, but we have provided a way to minimise this, as we shall explain later. By the way, Q1 is an automotive-rated Mosfet and typical dissipation can be expected to be under 1W for loads up to 10A, so no extra heatsinking is required. 4A schottky diode D2 is connected across the fan motor output terminal, to absorb back-EMF when Q1 switches off and inject it back into the 12V supply. Q1 is avalanche-rated and should survive without D2 but we decided to add it as a “belts ‘n’ braces” measure; you don’t have to install it if you are sure it’s unnecessary but it certainly doesn’t hurt. Battery voltage compensation Our description of how the duty cycle is calculated above omitted one detail. While fan speed is related to the duty cycle applied to Q1, it will also vary depending on the supply voltage. In order to provide a consistent fan speed based on temperature, we apply some supply voltage compensation. This means is that when you set the control voltage for 100% fan duty cycle, we consider this to be full speed at the minimum supply voltage as set by VR1. As the supply voltage rises above this minimum, the fan duty cycle is reduced proportionally. So for example, if the switch-on voltage is set to 13V but the actual supply voltage is 14.4V when the control voltage reaches the maximum setting (as determined by VR3), the actual duty cycle will be reduced to 90% (100% x 13V ÷ 14.4V). This means the fan speed should not vary (much) as the supply voltage varies. However, that does not mean the unit will never exceed a duty cycle of 90% when the supply is at 14.4V. It will still increase the duty cycle if the control voltage (ie, temperature) increases further beyond the “maximum” setting. It will continue increasing duty cycle linearly until Q1 is fully siliconchip.com.au switched on (ie, 100% duty cycle). You can think of this as a bit of a “turbo” mode for your fan when the supply voltage is high enough. Power supply Because this unit can be used in automotive (or even marine) applications, where you can expect all sorts of spikes and dips and other nasties on the supply rail, we have included protection measures to prevent the unit from being damaged. Power coming into the unit first passes through 10A blade fuse F1. This is mainly to protect against a shorted fan motor. In a motor vehicle, the unit should always be connected with an external fuse between the unit and the battery (either in the fuse box, or inline with the wiring) but it’s a good idea to have an internal fuse, just in case. The fan connects directly between the fused 12V rail and the drain of Mosfet Q1. Q1 is designed for automotive use and has an avalanche rating of 450mJ, which is relatively high. This, in combination with the inductance of the fan motor, should allow it to handle the typical brief (but high voltage) spikes which can occur in an automotive DC supply. But the rest of the circuit has separate protection, with a series 470Ω 0.5W resistor feeding reverse-polarity protection diode D1 and transient voltage suppressor TVS1, which is bypassed by a 2.2µF ceramic capacitor. These feed REG1, which is an automotive-rated ultra-low quiescent current linear regulator. The 470Ω resistor and 2.2µF capacitor form an RC lowpass filter to reduce the severity of the spikes, while TVS1 clamps the larger ones to a maximum of about 40V, which is the upper limit to the operating voltage rating of REG1. The 470Ω resistor also acts to limit the maximum current that TVS1 must clamp. REG1 is a low-dropout linear regulator and these tend to have stability issues depending on the output filter capacitor used. That’s because they have an internal feedback loop with significantly more phase shift than a traditional linear regulator. We have carefully chosen the output filter capacitor to have an ESR in the required range for stability. We would have preferred to use a ceramic capacitor, as these tend to be more reliable but they almost universally have too low an ESR to suit the LM2936 regulator. We could have added a series resistor but that would be another component on an already packed board. The Vishay 293D-series tantalum capacitor has an operating temperature range of -55°C to +125°C, with suitable voltage derating. In fact, we’ve provided a sufficient voltage rating for the capacitor to be OK up to temperatures of +150°C and Vishay’s reliability calculator suggests this part in our application should have a mean time between failures (MTBF) of 17 million hours at a constant 125°C. So it should be OK for, oh, just on 2000 years! The only additional components in the circuit are the 100nF supply bypass capacitor for IC1 and the 1kΩ pullup resistor at its MCLR input, to prevent spurious resets. Construction The Fan Speed Controller is built on a very small double-sided PCB, just 49.5 x 30.5mm and coded 05111171. Almost all the components are through-hole types and are fitted to the top side; there are just two SMDs, both on the bottom side and both easy to solder. Celebrating 30 Years January 2018  67 Fig.3: the PCB component overlays for the top side (top diagram, [a]) and underside (bottom diagram [b]), both shown life size. There are only two components to solder on the underside – both are SMDs but both are quite large and easy to solder. [a] [b] One SMD is the 22µF tantalum capacitor in a B-size case (3.2 x 2.6mm) and this is soldered in place under the mounting location for REG1. It’s a polarised component and will have a stripe to indicate the positive side. This must go towards IC1; see bottom side overlay diagram Fig.3(b). This also shows the location for schottky diode D2, with its cathode (striped end) towards the top (near) edge of the PCB. The main thing to watch for with these components is to make sure that the solder forms a good fillet between the rectangular lead on the end of the component and the pad on the PCB. If you spread a little flux paste on the PCB pads before soldering, it will help the solder flow down and make good contact with the PCB. With those in place, flip the board over and start fitting the through-hole components, using top side overlay diagram Fig.3(a) as a guide. Start with the resistors, checking the resistance of each with a DMM before soldering, followed by diode D1, with its cathode stripe orientated as shown. TVS1 is also polarised and this can be fitted now. IC1 is next but make sure it is programmed before soldering it in place. It’s difficult to re-program once on the board and we strongly recommend that you don’t use a socket since the IC could vibrate loose or corrosion could form over time, causing intermittent contact and failure. Double-check that its pin 1 dot is towards the corner of the board before soldering the pins. The next job is to mount Q1 on the board by bending its pins and then attaching its tab using a short M3 machine screw, shakeproof washer and nut. Once it’s firmly secured, solder and trim the three leads. You can now fit the three non-polarised ceramic capacitors in the locations shown. Now crank out the leads of transistors Q2 and Q3, and regulator REG1 and solder them as shown in Fig.3(a). Don’t get the parts mixed up since they look almost identical and are only distinguishable by their labels. You can then solder the three identical trimpots, VR1-VR3, with their mounting screws located as shown. That just leaves fuse holder F1 and the three connectors. If you are wiring in the unit with an inline fuse (strongly recommended for automotive applications), you could replace F1 with a wire link. However, we opted to keep the onboard fuseholder and we fitted a 10A fuse, with a 7.5A inline fuse. The idea behind this is that the inline fuse is 68 For clarity, we’ve shown the topside and underside views of the PCB a little larger than life size. Note the polarity of the 22µF tantalum capacitor and the schottky diode (D2) on the underside pic. There are some minor differences between this prototype and the patterns at left. Silicon Chip easier to replace and so should blow first but the onboard 10A fuse has been kept as a last-ditch protection measure. Assuming you are fitting F1, you will either have two separate blade fuse clips or a single assembly with both clips fitted to a plastic base. Either way, you will need to insert the clips as shown and push them down fully onto the PCB before soldering. But if you are using the individual clips, you will have to be careful because it requires quite a lot of heat to get good solder adhesion and the solder can unfortunately run down through the middle of the clip, preventing a fuse from being inserted. We certainly don’t recommend you solder the clips with a fuse inserted since this can result in the fuse being soldered to the clips! So it’s a balancing act; you need to use enough solder and heat it sufficiently for it to adhere to the clips but not so much that it runs through. If you do get solder inside the fuse clips, you will need to use a solder sucker and probably also some flux paste and solder wick to remove the excess. Note that we didn’t fit any of the connectors to our prototype because we were concerned that the wires could vibrate loose and contacts could corrode, so we decided to solder the wires directly to the PCB. If you do fit the connectors, make sure the wire entry holes of the terminal blocks face to the outside of the PCB (ie, to the left as shown in Fig.3). There’s no need to dovetail the terminal blocks as they are spaced apart slightly. If you aren’t fitting the connectors, we strongly recommend that you make sure the wires will fit through the holes before going any further. Since the holes are sized to suit connectors and thus are too small to admit high-current wires, you will probably be better off soldering PCB stakes to the board and then solder the wires to the stakes later. You could drill out the holes for CON1 and CON2 to accept wires but then we suggest you solder them to both sides of the board, so you can take advantage of the parallel copper tracks top and bottom. Fitting it in its case We chose an IP65 flanged polycarbonate case for this automotive application because the unit needs to be waterproof Celebrating 30 Years siliconchip.com.au Testing and set-up Fig.4 shows an easy way to test and set up the Fan Speed Controller. The LED and series resistor take the place of the fan and show you when it will switch on and how fast it will be running (ie, how bright the LED is). The 1kΩ potentiometer allows you to vary the supply voltage to the board and the 10kΩ potentiometer simulates the NTC thermistor and allows you to simulate changes in temperature. If you have an adjustable DC bench supply, you can do without the 1kΩ potentiometer and simply connect the supply up directly to CON1. Insert fuse F1, wind the 1kΩ resistor fully anticlockwise, switch on the supply and advance the 1kΩ pot to about half-way. Check that you have at least 7V across CON1. If you have much less than that, there could be a short circuit on the board, so switch off and check it carefully. Now measure the voltage across the 470Ω resistor next to D1 on the board. The quiescent current in this condition should be around 18µA, giving an expected reading of 8.5mV. If you get a reading above 15mV or below 5mV then something is wrong so check your work. Depending on your meter, you may see the reading jump up every two seconds; this is IC1 waking up to check the supply voltage. If you want to alter the PWM frequency or fan minimum/maximum duty cycle, now is the best time to do it. See the panel titled “Advanced setup” for instructions. The first main setting to make is the low supply cut-out voltage. Set the 10kΩ off-board pot to about halfway, then wind VR1 and VR3 fully clockwise and VR2 fully anti-clockwise. Adjust the 1kΩ potentiometer (or your DC supply) to the desired switch-on voltage. Adjust VR1 clockwise until the test LED switches on. You can now test it by reducing the supply voltage until the LED switches off. If you measure the Fig.4: a convenient test jig to set up your fan speed controller, as explained below voltage across CON1, it should be around half a volt lower than the switch-on threshold that you just set. Next, we set the switch-on and maximum speed temperatures. First, refer to the table at right and write down the thermistor resistance at the desired minimum and maximum temperatures. For temperatures between those shown, you can simply estimate the value (it’s all pretty approximate anyway). For example, for 38°C, we know the resistance will be somewhere between 6.5kΩ and 5.3kΩ and probably closer to the latter, so we could take a guess at 5.8kΩ, which turns out to be spot on. Now adjust your off-board 10kΩ potentiometer while measuring the resistance between the two pins that are wired to the board, until you reach your computed switch-on threshold value. Then rotate VR2 clockwise until the test LED lights. Now, re-adjust the 10kΩ potentiometer to get a resistance reading that corresponds to your maximum speed temperature, and rotate VR3 anti-clockwise until the test LED starts to dim, then slowly rotate VR3 clockwise again until it is back at maximum brightness. That completes the set-up; you can now connect the NTC thermistor to CON3 and apply a heat source to it and verify that the LED behaves as expected. (to handle rain, car washes, etc) and also able to handle temperatures up to 100°C (eg, engine coolant) without damage. At 64 x 58 x 35mm (not including flanges), this case is nice and compact, making it easy to mount in the engine bay. While it’s available in a beige and dark grey, unfortunately, the dark grey version is only rated for temperatures up to 85°C and the beige version would look out of place in an engine bay. So we painted the outside of the beige case with a layer of etch primer and then several coats of matte black Jaycar’s HB1022 engine spray paint (intend- IP65 case is ideal ed for painting rocker cov- for this project because it has ers and such). We were careful to avoid both a sealing getting paint into the chan- gasket and a mounting nel where the waterproof flange. And gasket is fitted as this may it’s just big affect its sealing properties. enough to Even though the case is house the PCB! siliconchip.com.au Temperature Resistance (°C) (Ω) -10 55.3k -5 42.3k 0 32.7k 5 25.4k 10 19.9k 15 15.7k 20 12.5k 25 10.0k 30 8.1k 35 6.5k 40 5.3k 45 4.4k 50 3.6k 55 3.0k 60 2.5k 65 2.1k 70 1.8k 75 1.5k 80 1.3k 85 1.1k 90 900 95 800 100 700 waterproof, due to the harsh environment of an engine bay it would also be a good idea to spray both sides of the PCB (avoid the top of the preset pots) with a conformal coating, such as HK Wentworth’s Electrolube HPA. This makes the PCB and components virtually impervious to liquids. Do this after verifying that the PCB assembly is working properly. We glued the PCB into the bottom of the case using neutral cure silicone sealant. Don’t use acid cure silicone; it can corrode metal parts. It was then just a matter of drilling holes into the case for the wiring, feeding it through, soldering it to the board and then using silicone to seal the areas where the wires enter the case. We chose to solder the wires to the board, rather than using terminal blocks and headers, because we were concerned that vibration could work the wires loose over time. Be careful if you do this Celebrating 30 Years January 2018  69 Parts List – 12V Fan Speed Controller 1 double-sided PCB, coded 05111171, 49.5 x 30.5mm 1 10A ATO/ATC blade fuse with matching blade fuse clips (F1) 1 M3 x 6mm machine screw, shakeproof washer and nut 2 mini 2-way terminal blocks (CON1,CON2) [optional] 1 2-way polarised pin header (CON3) [optional] 1 NTC thermistor [to suit application] Semiconductors 1 PIC12F675-E/P microcontroller programmed with 0511117A.HEX (IC1) 1 LM2936-5.0 5V 50mA ultra low quiescent current regulator (REG1) 1 IPP80N06S2L-07 N-channel automotive Mosfet in TO-220 package (Q1) 1 MPSA92 200V 500mA PNP transistor (Q2) 1 BC546 100mA NPN transistor (Q3) 1 1.5KE30A 30V 1500W unidirectional TVS (TVS1) 1 1N4004 1A diode (D1) 1 SK4200L 4A 200V SMD schottky diode (D2) Capacitors 1 Vishay 293D226X0016B2T OR 293D226X9016B2T 22µF 16V SMD tantalum capacitor, Case B 1 2.2µF 50V multi-layer ceramic 1 100nF 50V multi-layer ceramic 1 1nF 50V multi-layer ceramic Resistors (all 0.25W 1% metal film unless otherwise stated) 2 100kΩ 1 22kΩ 1 10kΩ 1 1kΩ 1 470Ω 1/2W metal film 3 10kΩ 25-turn vertical trimpots (VR1-VR3) Additional parts for radiator fan control 1 radiator fan drawing up to 7.5A <at> 14.7V (eg, SPAL VA09-AP8/C-27S) 1 radiator fan mounting kit 1 64 x 58 x 35mm IP65 polycarbonate enclosure with mounting flange (Jaycar HB6211) 1 SAE plug to battery terminal 7.5A fused lead [Jaycar PP2012] 1 SAE inline socket with 1.8m 16AWG automotive twin lead [Digi-Key Cat 839-1349-ND] 2 M6 brass nuts (or size to suit battery terminals) 2 M6 beryllium copper crinkle washers (or size to suit battery terminals) [element14 Cat 2770730] 1 2-pin Nylon Molex plug to suit radiator fan [Jaycar PP2021] 1 1m length figure-8 10A automotive rated cable (for fan wiring) 1 10kΩ 1% lug-style NTC thermistor [eg, Altronics R4112] 1 1m length figure-8 light-duty automotive rated cable (for thermistor wiring) 1 2-way waterproof plug and socket set (optional, for thermistor wiring; [eg, Jaycar PP2110]) 1 adhesive thermal pad or a small tube of thermal paste heatshrink tubing petroleum jelly neutral-cure silicone sealant a few small pieces of high-density foam a selection of large and small cable ties 70 Silicon Chip This cheap radiator mounting kit sourced from ebay has four ties, four springs, four plastic discs and eight adhesive foam pads. though since you will probably have to fit PC stakes to the board and then solder the wires to those. The mounting holes are too small for anything but the thinnest copper wire to be fed through. All the wires soldered to the board had external connectors to make removing the module easy (for maintenance). The two battery wires go via a water-resistant SAE plug and socket, the NTC thermistor wires via a 2-pin waterproof plug and socket set and the fan wires were crimped and soldered to a Molex socket, to match the existing plug on the fan. We placed heatshrink tubing over the thermistor and fan wiring and after shrinking it down, injected some silicone into the back of the Molex plug and socket to improve their ability to withstand a soaking. The silicone was also forced into the ends of the heatshrink tubing to stop water getting inside and possibly entering pinholes in the wire insulation. Where possible, fit these connectors to the wires after you have figured out where you’re mounting the unit and cut the wires to length. Otherwise, you will be left with a lot of excess cabling to bundle up. Installation procedure We used our prototype to control a 300mm fan for a water-to-air intercooler radiator on a supercharged V8 engine. This is a worthwhile upgrade for any vehicle with an intercooler which will be driven in traffic. See the separate panel for an explanation of the benefits. However, this project is just as applicable for normal radiators in vehicles which do not have adequate cooling, for whatever reason and the installation details will be virtually identical. As you can see from our photos, the new radiator is a “pusher” style which is mounted at the front of the radiator stack. We chose this type for two reasons; one, the intercooler radiator is in front of the main radiator and we wanted fresh air to be forced over it and two, there was already a pair of “puller” radiators mounted at the back of the radiator stack, which you may be able to see if you examine the photos carefully. In extreme conditions, the front “pusher” and back “puller” fans will work in concert to force fresh air into the front of the first radiator, through the air conditioner condenser and engine radiator and then over the engine itself, where it will tend to be forced out from under the engine bay. Fan mounting The first step was to mount the fan on the radiator. This was done using a cheap but effective mounting kit compris- Celebrating 30 Years siliconchip.com.au ing four ties, four springs, eight adhesive foam pads and four plastic discs (see photo opposite). The ties are a bit like cable ties but they have a flat plate at one end. You thread one of the adhesive foam pads over the tie (the pads have a hole in the centre) and then force the plastic tie between the fins of the radiator, from the opposite side where you want to mount the fan. You then slip a second foam pad over the tie shaft so it’s in contact with the opposite side of the radiator. The tie then goes through the radiator mounting flange and you slip the spring (small end towards fan) and plastic disc over the tie. The plastic disc has a hole in the middle with little teeth which grab the bumps on the tie, giving a one-way ratchet effect. As you pull the tie through the disc, it compresses the spring and foam pads until the radiator is held firmly in place. The foam pads on either side of the radiator prevent the force holding the fan onto it from damaging the delicate fins. Once the ties have been installed on the four corners of the fan and tensioned appropriately, it’s held in place very well and won’t budge under normal acceleration, braking and cornering forces. In our test vehicle, we had very poor access to the back of the radiator; there was around a 10mm gap between it and the radiator behind it at the top, reducing to around 5mm at the bottom. As such, were only able to attach the fan to the radiator at its two upper mounting points. To compensate, after slipping the two adhesive foam mounting pads between the fan’s two lower mounting points and the radiator, we then forced a highly compressed block of closed-cell foam into the gap between the front of the fan motor housing and the plastic cross-member which sits behind the vehicle’s front grille. This holds the fan firmly against the radiator, preventing it from moving forward under heavy braking and takes some of the gravitational load off the two upper mounting points thanks to the resulting friction at the two lower mounting points. So far, this arrangement seems to have stood up to the abuse which results from driving on Sydney’s pothole-filled streets. By the way, when mounting the fan, we made sure it wasn’t resting on the oil cooler below; a small piece of foam was inserted between it and the oil cooler while the fan was being mounted and then finally removed, giving around 5mm of clearance, so that it doesn’t bounce up and down when going over bumps and damage the oil cooler fins. Wiring With the fan in place, we then found a suitable location to mount the control box itself, next to the headlight housing. It was then secure in place by routing some large cable ties through the holes in the box flanges and around nearby anchor points. A piece of foam was wedged under the unit to reduce the vibration transmitted to it while driving. The fan wiring is simple; having plugged the fan plug into its matching socket, we simply tied both cables to convenient anchor points to stop the wires flapping around. We then used a cable tie to clamp the NTC thermistor lug onto the intercooler radiator right next to the water inlet pipe. While not shown in these photos, we later wedged an adhesive thermal pad between the two to ensure good heat conduction. With the thermistor wire plugged into the matching siliconchip.com.au Advanced set-up Normally, the unit operates with a PWM frequency of 1kHz, a minimum duty cycle of 25% and a maximum duty cycle of 100%. These defaults are stored in the EEPROM of IC1 and so they can be changed if necessary. The most common reasons to change these are if you are controlling a fan or fans that use brushless (electronically commutated) motors, such as most computer fans, or if your fan won’t run properly with a duty cycle of just 25%. In these cases, you might want to drop the PWM frequency or raise the minimum duty cycle respectively. The set-up for these parameters takes advantage of the fact that normally VR3 is adjusted to give a maximum fan speed control voltage above 1V. That’s because the minimum (fan switch-on) control voltage is fixed at 1V so it doesn’t make much sense to have the maximum voltage be lower than this. So if VR3 is set to apply a voltage of 1V or below at pin 3 at start-up, this will activate the advanced set-up mode. If for some reason you want the fan to switch on at full speed, you can set VR3 to give a reading just above 1V at pin 3. However, you will need to be careful to make it high enough to avoid triggering this set-up mode. The actual threshold is close to 1/5th of the supply voltage, so check the output of REG1 and divide by 5 before setting VR3, to be safe. Selecting the parameter Follow these steps, based on which parameter you want to adjust. 1) PWM frequency – adjust VR3 to give a voltage at TP3 which is equal to the desired PWM frequency, where 1mV = 1Hz. So for example, adjust for 500mV if you want 500Hz PWM. Connect a 10kΩ resistor across CON3 (or if you have a 10kΩ pot wired across CON3, as described in the testing procedure, rotate it fully anticlockwise) and apply power. Wait for at least one second and then rotate VR3 clockwise until TP3 is well above 1V. Then adjust VR2 to give a PWM waveform at pin 2 and check the frequency with an oscilloscope or frequency counter, to verify that it has been set to the correct frequency. 2) Minimum duty cycle – adjust VR3 to give a voltage at TP3 which is equal to the desired minimum duty cycle, where 10mV = 1%. So, for example, adjust for 330mV if you want a minimum duty cycle of 33%. Disconnect the NTC thermistor (or anything else) from CON3 and apply power. Wait for at least one second and then rotate VR3 clockwise until TP3 is well above 1V. Then re-connect the NTC thermistor or pot to CON3 and adjust VR2 fully anti-clockwise. Wind it slowly clockwise until you get a PWM waveform at pin 2 and check the duty cycle with an oscilloscope or DMM with duty cycle measurement, to verify that it has been set to the correct minimum. 3) Maximum duty cycle – adjust VR3 to give a voltage at TP3 which is equal to the desired maximum duty cycle, where 10mV = 1%. So, for example, adjust for 750mV if you want a minimum duty cycle of 75%. Short out CON3 and wind VR2 a few turns clockwise, then apply power. Wait for at least one second and then rotate VR3 clockwise until TP3 is well above 1V. Then re-connect the NTC thermistor or pot to CON3 and adjust VR2 fully clockwise. Wind it slowly anti-clockwise until you get a PWM waveform at pin 2 and check the duty cycle with an oscilloscope or DMM with duty cycle measurement, to verify that it has been set to the correct maximum. Having finished making any or all of the above changes, re-verify that TP3 is set above 1V and you can then go through the normal set-up procedure to adjust VR1-VR3. Celebrating 30 Years January 2018  71 Adding a fan to an intercooler An intercooler is a radiator which cools the air going into an engine. It’s normally fitted between a (turbo)supercharger and the engine or in some cases, between multi-stage turbocharger compressors. It may cool the air directly or there may be a liquid coolant which transfers the heat energy to a second radiator which is cooled by ambient air (which is the case in our test vehicle). This is beneficial because the (turbo) supercharger has the side-effect of heating the intake air as it compresses it and forces it into the engine. That increases the chance of the fuel detonating, which could damage the engine and it also limits the effectiveness of the supercharger because the hotter air is less dense, partially negating the benefit of compressing it to fit more into the cylinders. Most vehicles can benefit from improved airflow past the intercooler radiator and that’s certainly true in this case. Our test vehicle greatly benefited from fitting a fan on the intercooler radiator, despite the fact that it is mounted in front of the main radiator which already has two high-performance cooling fans on the back. That’s because an intercooler radiator operates at a much lower temperature compared to the main engine radiator; the intercooler is typically around 10°C above ambient compared to around 90°C (absolute) for the main radiator. Because of the proximity of the two, when the vehicle is stopped (eg, at a red traffic light) or moving slowly (eg, in a queue of vehicles), especially uphill, there is a tendency for heat from the main radiator to “soak” the intercooler, leading to increased intake temperatures, reduced performance and a louder exhaust note. socket, again we tied both wires to mounting points on the bumper and chassis to keep it tidy. That just left the battery wiring. This was routed under the cross-member which supports the radiators and tied to it and the clamp which holds the battery in place. It was then just a matter of removing the inline fuse, fitting the eyelets over the bolts holding the battery clamps on and then fixing them in place using a beryllium copper crinkle washer and brass nut for each terminal. We made sure these nuts were done up tight, crushing the washers and forming a good electrical contact between the eyelet lugs and battery terminals, before smearing both terminals with petroleum jelly to prevent water from encouraging electrolytic corrosion due to the dissimilar metals used. It was then just a matter of re-inserting the fuse and the unit was ready for testing and tweaking. One final comment regarding installation. You will notice that we went to a fair bit of trouble to waterproof our control box, the wiring and the sensor. Once the traffic clears and the vehicle starts moving again, the intercooler gets back to normal temperatures after a couple of minutes but performance suffers until then. And in some cases, you could hit another red light or more traffic before the intercooler is back to its normal operating temperature. That’s solved by fitting an extra “pusher” fan on the front of the intercooler. It only switches on in situations where the normal airflow is not adequate to keep the intercooler in its ideal temperature range and the extra fan-forced air helps cool both the intercooler and also the normal radiator in this situation. Fitting an electric fan to a normal radiator You may be aware that most modern cars have electric radiator fans while older vehicles tend to have belt-driven or clutch-coupled fans driven directly from the engine crankshaft. Some of these older vehicles have a tendency to overheat and in that case, adding an electric radiator fan to replace or complement the existing mechanically-driven fan is an easy solution. Part of the reason for this is that older vehicles just weren’t as well-engineered but also they may not have been designed to sit in traffic for long periods because they didn’t have the sort of traffic that we have to deal with these days! The added electric fan will have little effect most of the time but certainly will give you peace of mind in the summer months, especially if you’re stuck in a bad traffic jam. Keep in mind though that if you have an older or classic car that’s overheating, it could also be due to blocked radiator coolant passages, a stuck thermostat or some other mechanical ailment. In that case, it would be better to fix it than to add an electric fan as a band-aid (despite the fact that this may well solve the problem). Keep in mind that you could easily get a high-pressure jet of water at the front of the radiator when washing the vehicle and that quite a bit of water will enter when driving in a rainstorm at speed. You don’t want your electronics to corrode should that water find its way inside. That’s why we earlier suggested also spraying the PCB with a conformal coating – just in case! Final adjustments For automotive applications, we recommend setting the low-battery cut-out voltage to between 13.5V and 14.0V. This way, the fan will only run when the engine is running and the alternator is charging the battery. If you set it too close to 13V then you might find that the fan will try to run sometime after the engine has been shut off, as the battery voltage can “rebound” to a little over 13V once the load on it has dropped to minimal levels – it takes a while for the voltage to settle to the expected 12.9V of a fully-charged, unloaded lead-acid battery after a long drive. Resistor Colour Codes Qty   2   1   1   1   1 72 Silicon Chip Value 100kΩ 22kΩ 10kΩ 1kΩ 470Ω 4-Band Code (1%) brown black yellow brown red red orange brown brown black orange brown brown black red brown yellow purple brown brown Celebrating 30 Years 5-Band Code (1%) brown black black orange brown red red black red brown brown black red brown brown black black brown brown yellow purple black black brown siliconchip.com.au If the fan does try to switch on in this condition, chances are it will immediately turn off again because the extra load on the battery will pull its voltage below 13V. This can result in the fan trying to spin up every couple of seconds, despite the hysteresis built into the battery voltage monitoring. It won’t do any harm but it could be a bit annoying if you can hear it happening. In that case, all you need to do is rotate VR1’s screw clockwise a little (say half a turn to one turn) to increase the threshold until that no longer happens. The temperature settings will probably require tweaking too. In our application, we set the switch-on threshold to 40°C and the maximum speed temperature to 55°C, on the basis that we didn’t want the fan running all the time on an average summer day (where ambient temperature could easily exceed 30°C) and that if the intercooler is above 50°C, engine performance would start to suffer. We ended up dropping those temperatures slightly, to around 38°C and 50°C, as this seemed to keep the engine operating in an optimal manner. If you’re fitting the fan to the engine radiator, you will want to use much higher temperatures. You can expect the coolant exiting a fully warmed-up engine to be around 90°C; remember, its boiling point should be above 100°C because of the antifreeze mixed into the water and because virtually all vehicles use a pressurised cooling system to keep the boiling point as high as possible . So if your temperature sensor is at or near the entry hose for the radiator then you will want to set the fan switch-on temperature somewhere around 90°C. If the sensor is at or near the exit, it will need to be significantly lower than this. How much lower depends on how efficient your radiator is. Chances are you will need to take a guess at the initial setting and then adjust it based on your observations while driving. If the fan is running full speed after a normal drive then you need to increase the temperature setting. On the other hand, if the fan doesn’t switch on at all even after a hard drive on a summer’s day, you need to lower the setting. In general, it’s probably a good idea to keep the maximum fan speed temperature close to the switch-on temperature because the difference between the coolant temperature in a properly working cooling system and one which is overheating is not huge. We would suggest setting it around 10°C higher than the switch-on threshold. You can increase it a bit if you notice the fan speed “hunting” (oscillating) or reduce it if the fan switches on but the coolant temperature still rises above what you would consider ideal, despite having an appropriate fan switch-on temperature. Controlling computer fans or other fixed installations While we designed this project with automotive applications in mind, it would also be quite suitable for controlling “muffin fans”, as used in computers, or to cool various pieces of industrial equipment, etc. You could even consider using it to control a fan which ventilates your home, basement, roof cavity, etc, or forces underfloor airflow to help prevent stale air and mould buildup. As long as the fan runs off 12V DC and draws no more than 10A, it will work OK. And you can connect multiple 12V fans in parallel, up to that 10A limit. The one issue that you will need to keep in mind is that these muffin fans normally use brushless (electronically commutated) motors which do not respond well to highfrequency PWM control. So you will probably need to drop the PWM frequency to somewhere in the range of 50-200Hz. See the panel titled “advanced set-up” for information on how to do this. If you’re lucky enough to have four-wire computer fans, one of the four wires (the blue one) can be used to provide PWM control. So in this case, wire up the red, black and blue wires in parallel. Connect red to +12V, black to GND and blue to the negative terminal of CON2. Connect a 1kΩ resistor between the pins of CON2 and the fans should then operate normally at the default PWM frequency of 1kHz. SC The new intercooler fan was added in front of the radiator since the existing radiator fans were already mounted at the back (just visible near the top of the photo). You need to use a “pusher” fan if it’s being mounted on the front. Here you can also see how we mounted the temperature sensor and tied the wiring to nearby structural elements to prevent it from moving while the vehicle is in motion. siliconchip.com.au Celebrating 30 Years January 2018  73