Fullscreen HTML5Med Res (??MB)HTML4

600W DC-DC converter for car hifi systems Thinking of fitting high-power audio amplifiers to your car’s stereo system? This 600W DC-DC Converter steps up the battery voltage to provide the high-voltage split supply rails required by the amplifiers. PART 1: By JOHN CLARKE If you like lots of bass and high sound levels in your car then you will want to build this 600W DC-DC Converter. It is used in conjunction with one or more amplifier modules so that up to 360W RMS total (or 180W per stereo channel) can be delivered to the loudspeakers. With that sort of power, and provided your loudspeakers are up to the task, you will have the makings of a really first-class car hifi system. Not 32  Silicon Chip that we are advocating that you use this type of system to blow your brains out or to annoy other motor­ists. That’s not what a high-power car hifi system is used for at all. Instead, it’s used to provide good clean sound with plenty of bass and with plenty in reserve for those high-power tran­sients. What sort of amplifiers can be used with this converter? One that immediately springs to mind is the “Plastic Power” amplifier module described in Features • Output voltag e adjustable • High power ca pability • Fuse protected • Under voltage cutout • Overcurrent p rotection • Fan cooled • Over-temperat ure cutout • Power indicat ors the April 1996 issue of Silicon Chip. This module requires a ±57V supply and is capable of delivering 175W into a 4Ω load or 125W into 8Ω. Two of these modules (for stereo) would provide an excellent hifi amplifier system for your car. That said, the choice of amplifier module is not restricted to any specific type, as we have catered for a wide range of supply options. However, the two modules in a stereo pair must be Fig.1: block diagram of the DC-DC Converter. It uses a switchmode driver stage to produce pulse width modulated (PWM) signals and these are used to drive complementary Mosfet switching stages. These stages in turn drive step-up transformer T1. Its secondary output is then fed to a bridge rectifier and filter capacitor stages to develop the plus and minus DC output rails. capable of operating from a common supply voltage. The amplifiers can be rated for any power, provided that the total power drawn from the DC-DC Converter is less than 600W. This final restriction does not mean that a single 600W amplifier or two 300W amplifiers can be powered by the converter. We must take amplifier efficiency into account and all amplifiers draw more power than they can deliver into a load. In theory, the maximum efficiency of a class B amplifier stage is 78.5% but this does not include the power dissipated by the quiescent current. In practice, the average power amplifier module will be about 60% efficient at full power. This means that only 60% of the power drawn from the converter will be supplied to the load. This in turn sets the maximum amplifier power rating to about 60% of 600W, or 360W total. If two amplifiers are used, then each one should be rated at no more than 180W. Physical arrangement As can be seen from the photos, the 600W DC-DC Converter is quite large. It is built into a two-unit high rack-mounting case and would normally be installed in the boot or, if space permits, under a seat. The unit is fan cooled to keep the components within their heat ratings and this will have some bearing on the final mounting arrangement, as the air vents must be kept free of any restrictions. The only external inputs are from the battery and the igni­tion switch, while the unit provides +V, -V and GND connections to the power amplifiers. Heavy duty cables are used for the battery supply connections and these are a necessity since the unit can draw up to 63A. Heavy duty wiring is also used for the power supply outputs to the amplifier modules. Three front-panel LEDs (Power, Output + and Output -) are used to indicate the status of the converter. The power LED simply indicates when the converter is switched on, either via the ignition or a separate switch. The two remaining LEDs indicate that the plus and minus amplifier supply rails are present. Basic principle The basic principle of the DC-DC converter is really very simple. It works by alternately switching the 12V battery supply to each half of a centre-tapped transformer primary. The result­­ing AC waveform is then stepped up by the transformer secondary and then rectified and filtered to provide the plus and minus supply rails. To achieve high efficiency and reduce the number of bulky components, the circuit operates at a switching frequency of about 22kHz. This high frequency allows us to use a ferrite transformer rather than a bulky ironcored type. The circuit also uses highspeed power Mosfets to switch the transformer and fast recovery diodes for the rectifiers. Power Mosfets were used because they are very fast and have low switching losses. In addition, power Mosfets have a positive temperature coefficient which means that they automatically “throttle” back if the output stage starts to overheat. In addition, the circuit incorporates comprehensive protec­ tion facilities. These include low-voltage cutout, current over­ l oad protection and over-temperature cutout. The low-voltage cutout is a particularly useful feature. In effect, the converter circuit monitors the battery voltage and if it drops below a certain level, the converter switches itself off. This not only saves you from the inconvenience of a flat battery but is also necessary to protect the Mosfets. To explain, a Mosfet is turned on by applying a voltage to its gate. If this voltage is too low, the Mosfet will not fully conduct and this can lead to excessive power dissipation and device failure. The current overload protection circuitry operates at two levels. First, there is a 63A fuse in the supply line which will blow if there is a drastic fault in the converter itself. Second, the circuitry features inbuilt current limiting to provide pro­tection against output short circuits. The accompanying specifications panel shows the performance of the converter. Note that its efficiency is better than 80% at full rated output. Block diagram Fig.1 shows the block diagram of the DC-DC Converter. As mentioned above, it uses a centre-tapped step-up transformer which is driven by Mosfet transistors. The secondary winding is also centre-tapped and is fed to bridge October 1996  33 WHY A CONVERTER IS NEEDED FOR HIGH POWER OK, so why do we need a converter to boost the supply rails for the power amplifier in the first place? Why not simply power the amplifier directly from the 12V battery? To understand this, we need to consider some basic theo­ry. First, we know that the power delivered into a load is the output voltage squared divided by the load resistance (ie, P = V2/R). Now let’s assume that we have a 12V battery which is charged to 14.4V. An amplifier powered from this battery can typically deliver a maximum output of 11V peak-topeak or about 3.9V RMS – see Fig.2. Thus, the maximum power which can be delivered into a 4Ω load from a single-ended configuration is about 3.8W (3.9 x 3.9/4). This can be increased by wiring two power amplifiers in a bridge configuration. If that is done, the output voltage supplied to the load is doubled and so the power output will be four times great­er at about 15W (which is still quite modest). All this assumes that the battery is actually delivering 14.4V. In practice, this only happens if the motor in your car is running and has had time to fully charge the battery. So in practice, the power outputs from single-ended or bridge connected amplifiers will be even less than the above figures. As a result, if we want high power, we need to either reduce the load resistance or increase the supply rails for the amplifier. However, a very low load resistance is impractical because the current in the amplifier output stages becomes exces­sive. This in turn causes high losses in both the amplifier and loudspeaker wiring. The efficient way to increase the power is to increase the voltage, rather than reduce the load impedance. This is because the power is proportional to the square of the voltage and only proportional to the inverse of the load impedance. This “square law” effect means that if we double the voltage, we quadruple the power. By contrast, if we halve the load impedance we only double the power. At the same time as halving the load resistance, we double the current which quadruples the losses. The only practical option is to increase the supply rails and that’s exactly what this DC-DC converter is designed to do. It can deliver supply rail voltages up to ±70V DC, so that you can run really high power amplifier systems (up to 180W per stereo channel). Fig.2: an amplifier powered from a 14.4V rail can typically deliver a maximum output of about 11V p-p or about 3.9V RMS (note: the scope shows a slightly low RMS figure). This means that the maximum power which can be delivered into a 4Ω load from a single-ended configuration is about 3.8W or about 15W from a bridged configuration. 34  Silicon Chip rectifier and filter capacitor stages to develop the plus and minus DC output rails. Mosfets Q3-Q5 drive the top half of the step-up transform­er, while Q8-Q10 drive the bottom half. These in turn are driven by a switchmode circuit which has feedback applied from the DC output. This feedback circuit acts to reduce the width of the pulses applied to the Mosfets if the DC voltage rises above a preset value. Conversely, the pulse width from the driver circuit increas­es if the output voltage falls below the preset value. Note that the two Mosfet driver circuits are switched in antiphase, so that when one half of the winding is conducting, the other is off. The resulting primary drive is stepped-up in the secondary windings. Apart from the voltage feedback which maintains a constant output voltage regardless of load, the switch­ mode driver circuit also detects overcurrent conditions via resistor Rsc. If overcur­rent occurs, the pulse width drive to the Mosfet gates is re­duced. Note that the voltage across Rsc is amplified by over-current amplifier IC3. Circuit details Fig.3 shows the final circuit for the 600W DC-DC Converter. It’s based on a dedicated switchmode IC, the TL494 (IC1). This device contains all the necessary circuitry to gener­ate complementary square wave outputs at pins 9 and 10 and these drive the gates of the Mosfets via buffer stages. The device also contains control circuitry to provide output voltage regulation and low voltage dropout. Fig.4 shows the internal circuitry of the TL494. It is a fixed frequency pulse width modulation (PWM) controller contain­ing a sawtooth oscillator, two error amplifiers and a PWM com­ parator. It also includes a dead­time control comparator, a 5V reference and output control options for push-pull or single ended operation. Fig.3 (left): the final circuit is based on a TL494 dedicated switchmode IC (IC1). It generates complementary PWM signals at pins 9 & 10 and these drive the parallel Mosfet switching devices via buffer stages. IC3 monitors the voltage across RSC to provide current overload protection. October 1996  35 Fig.4: this block diagram shows the internal circuitry of the TL494 PWM controller. It includes a sawtooth oscillator, a PWM comparator, a dead-time control comparator, two error amplifiers and a 5V reference. Emitter followers Q1 & Q2 provide the complementary PWM output signals at pins 9 & 10. The PWM comparator generates the variable width output pulses by comparing the sawtooth oscillator waveforms with the outputs of the two error amplifiers. By virtue of the diode gating system, the error amplifier with the highest output vol­ tage sets the pulse width. Dead-time comparator The dead-time comparator ensures that there is a brief delay before one output goes high after the other has gone low. This means that the outputs at pins 9 and 10 are both low for a short time at the transition points. This so-called “dead-time” is essential since without it the Mosfets driving one half of the step-up transformer would still be switching off while the Mosfets driving the other half were switching on. As a result, the Mosfets would be destroyed as they would effectively create a short circuit across the 12V supply. Fig.5 shows the pin 9 and pin 10 output signals at the maximum duty cycle. Note that each output is high for only 44.7% of the time, indicating that there is 5.3% dead-time. One of the error amplifiers in IC1 is used to provide the under-voltage cutout feature. This is achieved by connecting its pin 2 (inverting) input to the +12V rail via a voltage divider consisting of two 10kΩ resistors. The non-inverting input at pin 1 connects to SPECIFICATIONS Supply voltage ......................................................................... 10-14.8VDC Maximum output power .............................................................600W RMS Maximum input current ....................................................... 63A continuous Standby current ................................................300mA (mainly fan current) Output voltage ....................................................................±70V maximum Efficiency at full load ..........................................................................>80% Overcurrent cutout .......................................................... 80A peak approx. Over-temperature cutout .....................................................................80°C Under-voltage cutout ............................................................................ 10V 36  Silicon Chip IC1’s internal reference at pin 14 via a 4.7kΩ resistor. When the voltage at pin 2 drops below 5V (ie, when the battery voltage drops below 10V), the output of the error ampli­fier goes high and the PWM outputs at pins 9 & 10 go low, thus shutting the circuit down. The over-temperature cutout operates in a similar manner. The sensing device is thermal cutout device TH1 and this is mounted on the main heatsink along with the Mosfet output transistors. As shown on Fig.3, it is connected in series between the voltage divider on pin 2 and the positive supply rail. If the heatsink temperature reaches 80°C, TH1 opens and so the circuit shuts down by switching the PWM outputs low as be­fore. Note the 1MΩ resistor between the non-inverting input at pin 1 and the error amplifier output a pin 3. This provides a small amount of hysteresis so that this particular error amplifi­er operates as a comparator. The second error amplifier in IC1 is used to control the output voltage of the converter and provide current limit protec­tion. This amplifier has its inputs at pins 15 and 16. Let’s consider the voltage regulation role first. In this case, the feedback voltage is derived from the positive side of the bridge rectifier and is attenuated using a voltage divider consisting of VR1, a 47kΩ resistor and a 10kΩ resistor to ground. The resulting voltage is then fed via D7 to pin 16 of IC1 and compared to the internal 5V reference which is applied to pin 15 via a 4.7kΩ resistor. Normally, the attenuated feedback voltage should be close to 5V. If this voltage rises (due to an increase in the output voltage), the output of the error amplifier also rises and this reduces the output pulse width. Conversely, if the output falls, the error amplifier output also falls and the pulse width in­creases. The gain of the error amplifier at low frequencies is set by the 1MΩ feedback resistor between pins 3 & 15 and by the 4.7kΩ resistor to pin 14 (VREF). These set the gain to about 213. At higher frequencies, the gain is set to about 9.5 by virtue of the 47kΩ resistor and 0.1µF capacitor in series across the 1MΩ resis­tor. This reduction in gain at the higher frequencies prevents the amplifier responding to hash on the supply rails. The 27kΩ resistor and .001µF capacitor at pins 6 and 5 respectively set the internal oscillator to about 44kHz. This is divided using an internal flipflop to give the resulting complementary output signals at pins 9 & 10, which means that the resultant switching speed of the Mosfets is 22kHz. Pin 4 of IC1 is the dead-time control input. When this input is at the same level as VREF, the output transistors are off. As pin 4 drops to 0V, the dead-time decreases to a minimum. At switch on, the 10µF capacitor between VREF (pin 14) and pin 4 is discharged. This prevents the output transistors in IC1 from switching on. The 10µF capacitor then charges via the 47kΩ resistor and so the duty cycle of the output transistors slowly increases until full control is gained by the error ampli­fier. This effectively provides a “soft start” for the converter. Resistor R1 has been included to provide more dead-time if necessary. It prevents the 10µF capacitor from fully charging to 5V and this increases the minimum dead-time. R1 (1MΩ) is only necessary in those rare circumstances when current limit­ing occurs at full load. This is indicated by a buzzing sound from the transformer. Current limiting The current limiting circuit is based on op amp IC3. This is wired as a non-inverting amplifier with a gain of 101 and is used to monitor the voltage Fig.5: these waveforms show the complementary pulse signals from the TL494 PWM controller at the maximum duty cycle. Note that one output always switches low before the other switches high and that each output is high for only 44.7% of the time, indicating a 5.3% dead-time. Fig.6: these waveforms show the converter performance when there are transient load changes from no-load to almost full load. The con­verter is supplying the power rails to an amplifier which is driving a 4-ohm load at 317W when the signal is on (this corresponds to more than a 500W load on the converter when efficien­cy is taken into account). The middle trace shows the 100Hz tone burst input signal, the top trace is the positive supply rail for the amplifier (20V/div) and the lower trace is the negative supply rail (20V/div). Note the small voltage droop and minimal overshoot when the load is removed. developed across resistor RSC. The output of IC3 in turn drives the pin 16 input of the second error amplifier in IC1 via diode D8. RSC is actually a short length of wire with a value of about 0.7mΩ. It is connected between the commoned Mosfet sources and ground, which means that all the transformer primary current flows through it. October 1996  37 Despite the heavy-duty nature of the circuit, the 600W DC-DC Converter is easy to build since virtually all the parts are installed on a single large PC board. A large heatsink and a fan at one end help keep things cool. As long as the current through RSC remains below 79A, the output of IC3 will have no affect on the operation of the error amplifier. However, if the current attempts to rise above 79A, the output of IC3 will rise above 5.6V and so the voltage applied to pin 16 of IC1 will rise above 5V. As a result, the output of the error amplifier rises and this reduces the output voltage and thus the current. Complementary outputs The complementary PWM outputs at pins 9 & 10 of IC1 come from internal emitter follower transistors. These each drive external 10kΩ load resistors. They also each drive three paral­leled CMOS non-inverting buffer stages (IC2a-c and IC2d-f). These in turn drive transistors Q1 and Q2 on one side of the circuit and Q6 and Q7 on the other side. Thus, when pin 10 goes high, Q1 turns on and drives the paralleled gates of Mosfets Q3-Q5 via a 4.7Ω resistor. Note that each Mosfet gate is connected via a 10Ω “stopper” resistor to minimise any parasitic oscillations which may otherwise occur while the paralleled Mosfets are switching on 38  Silicon Chip and off. When pin 10 subsequently goes low, Q2 switches on and quickly discharges the gate capacitance of Mosfets Q3Q5, thus switching them off. Pin 9 then switches high at the end of the dead-time period and Q6 switches on Q8-Q10 to drive the other half of the transformer primary. Q1, Q2, Q6 & Q7 have been included to ensure that the Mosfets are switched on and off as quickly as possible. This minimises the time that they spend in the linear region where they dissipate high power. Zener diodes ZD2 and ZD3 ensure that the Mosfets are pro­tected against switching spikes generated by the transformer. If the voltage between the drain and gate of any Mosfet rises beyond the zener breakdown voltage plus the gate threshold voltage, that Mosfet switches on to suppress the voltage. Diodes D1 and D2 prev­ent the gate signals from shorting to the drains via the zener diodes. Note the 1Ω resistors connected between the cathodes of ZD2 & ZD3 and the drains of the Mosfets. These prevent large currents from flowing in the PC board tracks. The high-current paths between the drains of the Mosfets and the transformer primary are run using heavy-duty wiring. Note also the six 10µF capacitors between the centre-tap of the transformer primary and the commoned Mosfet sources. These capacitors are there to cancel out the inductance of the leads which carry the heavy currents to the transformer. The transformer, T1, is a relatively small ferrite-cored unit designed to be driven at high frequencies. The primary winding is made up of flat copper sheet with two turns on each side of the centre-tap. The secondary uses conventional enamelled copper wire with the number of turns set to provide the required output voltage. In summary, the power Mosfets in each phase of the circuit alternately switch each side of the transformer primary to ground, so that the transformer is driven in push-pull mode. When Q3-Q5 are on, the 12V supply is across the top half of the prim­ary winding, and when Q8-Q10 are on the supply is across the bottom half. This alternating voltage is stepped up by the transformer secondary and applied to bridge rectifier D3-D6. This produces positive and negative supply rails with respect to the secondary centre tap. These rails are then filtered using four 2200µF capacitors. PARTS LIST 1 PC board, code 05308961, 310 x 214mm 1 2-unit rack case (without rack front panel) 1 front panel label 1 fan heatsink, 214mm long x 69mm wide with fins on one side cut off 1 12V DC fan, 80 x 80 x 24mm 2 Clipsal BP165C18 brass link bars 1 63A (A3 type) cartridge fuse (F1) 1 Neosid 17-745-22 iron powdered ring core (L1) 1 Philips ETD49 transformer assembly with 3F3 cores (T1) (2 cores 4312 020 38041, former 4322 021 33882, 2 clips 4322 021 33922) 3 5mm LED bezels 1 5mm red LED (LED1) 2 5mm green LEDs (LED2, LED3) 6 PC stakes 2 2AG fuse clips 1 1A 2AG fuse (F2) 4 TOP3 insulating washers 4 TO-220 insulating washers 10 insulating bushes 2 6-10mm cable glands 1 80°C cutout switch (TH1) 1 100kΩ horizontal trimpot 1 2-metre length of red 4GA cable (length dependent on installa­tion) 1 2-metre length of black 4GA cable (length dependent on in­stallation) 1 6-metre length of 3.5 sq. mm multi-strand wire (length dependent on installation) 1 55mm length of 3.5 sq. mm multi-strand wire (Rsc) Inductors L1a and L1b limit the peak transient currents in the diodes. Note that L1a and L1b are wound as a compensated choke on a common ferrite core, so that the flux generated by L1a’s winding is cancelled by the flux generated by L1b. This prevents the core from saturating. LEDs 2 and 3 connect across the positive and negative output rails respectively, to indicate that these rails are present. The 6.8kΩ resistors limit the LED current. Voltage regulation is achieved by sampling the positive supply rail and 1 1.5-metre length of 3.3 sq. mm black multi-strand wire (for T1) 1 400mm length of 3.3 sq. mm red multi-strand wire (for T1) 1 1-metre length of 1.78mm dia. solid core insulated wire 1 1.2-metre length of blue hookup wire 1 400mm length of red hookup wire 1 400mm length of green hookup wire 1 2-metre length of red hookup wire for ignition connection (length dependent on installation. 1 1.2-metre length of 1.5mm dia. ENCU (for L1) 1 6-metre length of 1.25mm dia. ENCU (for T1 secondary) 1 150mm length of 0.8mm tinned copper wire 4 large eyelets for 8mm dia. wire with 12mm hole 6 eyelets for 3mm dia. cable and 3mm screws 3 eyelets for 4mm dia. cable and 4mm screws 4 1/8th inch x 9mm long cheesehead screws 10 3mm x 15mm screws 24 3mm x 6mm screws 3 3mm x 9mm screws 13 3mm nuts 6 3mm star washers 4 9mm tapped standoffs 7 15mm tapped standoffs 3 4mm dia. x 15mm screws plus nuts & star washers 2 8mm dia. x 15mm bolts, nuts & washers 1 12mm dia. x 15mm bolt & nut 1 copper strip, 75 x 18 x 0.6mm feeding this back to pin 16 of IC1 via a voltage divider network. The internal error amplifier on this pin then controls the PWM comparator to provide voltage regulation, as described previously. Trimpot VR1 allows the output voltage to be set to the desired value. Power supply The 12V supply from the car battery connects via heavy duty cable and fuse F1 to the centre tap of T1. Because of the high currents involved, there is no on/off switch. 1 copper strip, 295 x 41 x 0.315mm 10 small cable ties Semiconductors 1 TL494 switchmode controller (IC1) 1 4050 CMOS buffer (IC2) 1 LM358 dual op amp (IC3) 2 BC338 NPN transistors (Q1,Q6) 2 BC328 PNP transistors (Q2,Q7) 6 BUK436-100A Mosfets (Q3-Q5, Q8-Q10) 4 1N914, 1N4148 signal diodes (D1,D2,D7,D8) 4 MUR1560 15A 600V fast recovery diodes (D3-D6) 1 16V 1W zener diode (ZD1) 2 47V 400mW zener diode (ZD2,ZD3) Capacitors 4 2200µF 100VW electrolytic (Philips 2222 050 19222) 1 100µF 16VW PC electrolytic 2 10µF 16VW PC electrolytic 6 10µF 100VW MKT polyester (Philips 2222 373 21106) 2 0.47µF MKT polyester 4 0.1µF MKT polyester 1 .0056µF MKT polyester 1 .001µF MKT polyester Resistors (0.25W 1%) 2 1MΩ 3 4.7kΩ 1 470kΩ 1 2.2kΩ 2 47kΩ 7 10Ω 1 27kΩ 2 4.7Ω 6 10kΩ 6 1Ω 4 6.8kΩ 0.5W Miscellaneous Solder, insulating tape, heat­shrink tubing, battery termi­nals Power for the rest of the circuit is supplied via the igni­tion switch (or a separate switch could be used). LED1 indicates the presence of the 12V rail and is supplied via a 2.2kΩ resis­tor. In addition, a 12V fan is wired directly across the supply and this runs continuously whenever power is applied. Finally, a 10Ω resistor and 16V zener diode (ZD1) provide protection against transient voltages for the low current circuitry. That’s all we have space for this month. Next month, we shall give the SC full construction details. October 1996  39