Fullscreen HTML5Med Res (??MB)HTML4

Mini-D Stereo 10W/Channel Class-D Audio Amplifier This little chip can deliver a whopping 30 watts! With no heatsink! Main Features • • • • • • • • • • • • • • • • • • Stereo or mono Class-D amplifier on a single, small PCB No heatsink required Low EMI DC power supply, wide operating voltage range Drives one or two 4-8Ω speakers Selectable gain On-board volume control RCA input sockets Shutdown mode Output short-circuit protection DC offset protection Over-temperature shutdown with auto resume Selectable output power limit with soft clipping Low quiescent current Reversed supply polarity protection Input signal overload protection Power and fault indicator LEDs Under-voltage and over-voltage lock-out 74  Silicon Chip This tiny Class-D amplifier module can work in two modes. In stereo it can deliver more than 10W per channel or you can connect its output channels in parallel to deliver more than 25W into a single speaker. It is up to 91% efficient, with selectable gain, volume control and other features such as a low-power shutdown mode and over-temperature, over-current, short circuit and speaker protection. By NICHOLAS VINEN H OW CAN A CHIP this small deliver so much power? And how can it deliver so much power without needing a big heatsink? The answer to both questions is Class-D operation. It’s a switching amplifier and its efficiency can be over 90%. High efficiency is also good if you want to run it from a battery since it will last longer. And if running from mains, you don’t need a bulky power supply; a 1A plugpack should be more than adequate. We published our first switching amplifier design, the CLASSiC-D, in November & December 2012. It’s a powerful beast, able to deliver up to 250W into a 4-ohm load or 500W into an 8-ohm load (bridged) with low distortion. Lots have been built since its publication. But while you may want the high efficiency of Class-D, the CLASSiC-D is simply too big and expensive for many applications where you only need a few watts of audio, perhaps running off a small battery – for busking, for siliconchip.com.au GVDD PVCCL BSPL PVCCL PBTL Select OUTPL FB Gate Drive OUTPL OUTPL FB LINP Gain Control PGND PWM Logic PLIMIT GVDD LINN PVCCL BSNL PVCCL OUTNL FB OUTNL FB FAULT Gate Drive OUTNL SD GAIN0 TTL Buffer SC Detect Gain Control GAIN1 PLIMIT Reference PLIMIT Ramp Generator Biases and References Startup Protection Logic AVDD AVCC PGND DC Detect Thermal Detect GVDD PVCCL BSNR UVLO/OVLO LDO Regulator PVCCL GVDD Gate Drive GVDD OUTNR OUTNN FB OUTNR FB RINN Gain Control PLIMIT PGND PWM Logic GVDD RINP PVCCL BSPR OUTNP FB PVCCL Gate Drive PBTL TTL Buffer OUTPR PBTL Select OUTPR FB AGND PGND Fig.1: block diagram of the TPA3113D2 Class-D audio amplifier IC. The left & right channel differential inputs are buffered and fed to Schmitt trigger stages where they are compared against a ramp (triangle) signal. The resulting PWM signals are then fed to PWM logic blocks which then drive two bridge-mode stereo switching amplifiers. example. Or say you want to build a pair of self-powered computer speakers. Whatever the reason, a few watts can go a long way. That’s where Mini-D amplifier module comes into its own. It’s based on the Texas Instruments TPA3113D2 which contains two complete bridgemode stereo switching amplifiers. It’s so efficient that it doesn’t need a heatsink for normal program material; the PCB itself dissipates the heat. Only a simple output filter is required to minimise the amount of RF interference generated by its switchmode operation. This consists of just four ferrite beads and four ceramic capacitors, or eight components for the two channels. All the components are surface-mount types, selected so that they are straightforward to solder. Because the Mini-D module’s outputs are bridged, it has good power delivery even with moderate supply siliconchip.com.au rails. With a 12V supply, it can deliver at least 5W per channel into 8-ohm speakers or 2 x 10W into 4-ohm loads. More power is available with higher supply voltages. Unusually, the Mini-D can also operate in mono mode, with the outputs paralleled. This doubles its current capability, allowing more power into low-impedance loads, eg, 25W or more into 4Ω. By the way, we’ve said this in the past but it bears repeating: while the output transistors in Class-D amplifiers spend most of their time either on or off, they aren’t really ‘digital’ amplifiers. While there may be some digital circuitry involved, they still work on the principle of analog negative feedback to generate the correct output waveform for a given input signal. Class-D amplifier operation We won’t go into the full theory of how a Class-D amplifier works but let’s look at the functional block diagram of the TPA3113D2 IC (Fig.1) which is the heart of the circuit. The two inputs are differential. Looking at the left channel, the signals are fed to LINP (in-phase) and LINN (ground/out-of-phase) at top left. The feedbacks from the switching outputs, OUTPL FB (positive) and OUTNL FB (negative), pass through low-pass RC filters internal to the IC and these four signals all go into a differential amplifier which performs this analog computation: (LINP - LINN) x GAIN - (OUTPL - OUTNL) The GAIN setting is determined by the state of two digital inputs, GAIN0 and GAIN1, which control the resistances in this part of the circuit to select an effective gain of 20dB, 26dB, 32dB or 36dB. The output of this differential amplifier then passes through another RC low-pass filter, to further attenuate September 2014  75 Parts List 1 double-sided PCB, code 01110141, 46 x 85mm 4 HI1812V101R-10 ferrite beads, SMD 4532/1812 (FB1-FB4) (element14 2292377) 2 PCB-mount switched RCA sockets, white & red (CON1CON2) OR 2 2-way pin headers plus shielded cable, header plugs and chassis-mount RCA sockets 3 2-way mini terminal blocks, 5.08mm spacing (CON3-CON5) 1 3-way pin header, 2.54mm pitch (CON6) 3 shorting blocks 1 10kΩ dual gang 9mm log potentiometer (VR1) OR 2 10kΩ mini horizontal trimpots (VR2-VR3) OR 1 20mm length tinned copper wire or two component lead off-cuts 3 2-way pin headers, 2.54mm pitch (LK4-LK6) 3 tapped spacers with M3 x 6mm machine screws (optional, for mounting) Semiconductors 1 TPA3113D2PWP Class-D Audio Amplifier IC, HTSSOP-28 (element14 1762987) 1 IRFML8244 N-channel Mosfet, SOT-23 (Q1) (element14 1857298) 1 BSS84 P-channel Mosfet, SOT23 (Q2) (element14 1431318) the switching artefacts in the signals, and then into a differential buffer. During normal operation, with the output correctly tracking the input (after gain is taken into consideration), the output of these amplifiers will be virtually nil, ie, the two differential lines will be at the same potential. Any deviation from this state means that the amplifier output must swing one way or the other. The buffered signal passes through the PLIMIT block which allows an external voltage to limit the maximum output swing, for speaker overload protection. The signals then pass into a pair of Schmitt-trigger comparators where they are compared against a ramp (triangle) signal, generated by an internal oscillator. This is a common method for pro76  Silicon Chip 5 5.6V zener diodes, SOT-23 (ZD1ZD5) (element14 1431238) 2 BAT54A dual Schottky diodes, SOT-23 (D1,D2) (element14 2114869) 1 high-brightness green LED, SMD 3216/1206 (LED1) (element14 2217905)* 1 high-brightness red LED, SMD 3216/1206 (LED2) (element14 1226389)* Capacitors (all SMD 3216/1206** unless stated) 2 100µF 25V low-ESR radial electrolytics 7 4.7µF 25V X7R ceramic (element14 1828835) 6 220nF 50V X7R ceramic (element14 1327724) 8 1nF 50V NP0/C0G ceramic (element14 2280692) 4 330pF 50V NP0/C0G ceramic (element14 3606090) Resistors (all SMD 3216/1206** 1%) 9 100kΩ (element14 1811974) 2 10kΩ (element14 1811973) 2 100Ω (element14 1632521) 5 10Ω (element14 1591420) 2 4.7Ω (element14 2142059) 2 0Ω (element14 1632520) (LK1-3) * or use 2 x 2-pin headers with offboard LEDs ** SMD 2012/0805 size parts can also be used ducing PWM (pulse width modulation), typically used in motor control circuits. The main difference here is that the operating frequency is much higher; around 310kHz. This is necessary to allow accurate reproduction of audio signals up to 20kHz. PWM output These signals then pass through the PWM logic to the Mosfet gate drivers and then the totem-pole output stages, consisting of N-channel Mosfet pairs. This chip uses a ‘centre-aligned’ or ‘dual-ramp’ PWM, a different modulation scheme to that used in many other Class-D amplifiers. This is shown in Fig.2 and is possible because the TPA3113D2 always operates in bridged mode. In the quiescent condition, both outputs are driven in-phase with a 50% duty cycle (top of Fig.2) and this results in no current flowing in the speaker(s) or filter at all. To drive the output positive, the duty cycle of the positive output is increased while the negative output duty cycle decreases (middle of Fig.2). This is done by shifting both the leading and trailing edges of both waveforms. Since none of these edges line up, this spreads RF emissions out, making them easier to filter. To drive the output negative, the reverse condition occurs (bottom of Fig.2). Since the output transistors are Nchannel Mosfets, a supply above the positive rail is required for the upper gate drive. This is generated by four 220nF capacitors between the OUTPL & BSPL terminals, OUTNL & BSNL etc. When the respective output is low, its capacitor charges through an internal diode from GVDD (~7V) and when the output goes high, the capacitor charge maintains the associated boost pin 7V above that output, sufficient to keep the upper Mosfet conducting. The block diagram also shows the protection circuitry, including shortcircuit detection, output DC offset detection, high temperature detection and under/over-voltage lock-out. Should any of these fault conditions occur, the output drivers are all switched off. The over-temperature cut-out kicks in when the die temperature hits 150°C and operation resumes once it has dropped by around 15°C. When the chip is running in mono mode, as set by the PBTL input pin, the PWM logic is modified slightly so that OUTPL and OUTNL carry an identical signal. At the same time, OUTPR and OUTNR are both driven with the same out-of-phase PWM signal, allowing the pairs of outputs to be paralleled. Speaker wires Because ‘centre-aligned’ PWM is used in this chip, only a very simple output filter is required to minimise the amount of RF interference generated. This consists of just four ferrite beads and four ceramic capacitors. The data sheet states that the ferrite bead output filter is sufficient for twisted speaker wires up to 1.2m long. We imagine that standard figure-8 speaker wires should also be OK, given that the conductors are in close proximity. If you want to use longer speaker leads or are particularly concerned about radio interference, you can add siliconchip.com.au No output (quiescent) 33µH OUTP OUTP L1 OUTN 33µH OUTP-OUTN Speaker Current C1 1µF OUTN L2 0V Positive Output OUTP C2 1µF Cutoff Frequency = 27kHz, Speaker Impedance = 8Ω OUTN 15µH OUTP OUTP-OUTN Speaker Current L1 C1 2.2µF 0V 15µH Negative Output OUTN OUTP L2 C2 2.2µF OUTN Cutoff Frequency = 27kHz, Speaker Impedance = 4Ω OUTP-OUTN Speaker Current 0V Fig.2: the quiescent (top), positive output (middle) and negative output (bottom) signal waveforms for the TPA­ 3113D2 Class-D audio amplifier IC. an external LC output filter. This could be wired externally to the board, ie, between the output terminals and speakers. Note that you would need to keep the components relatively close and run some connections to a PCB ground point. One disadvantage of this approach is that the filter component values must be selected based on the speaker impedance. Also, the inductors must handle the peak load current (up to 4A in some cases) without saturating. The recommended filters for 8-ohm and 4-ohm loads are shown in Fig.3. Note that an LC filter may also give improved efficiency. Speaker impedance For supply voltages up to 15V, the unit can drive speakers with nominal impedances from 4-8Ω. Above 15V, however, it isn’t recommended to drive 4Ω speakers. Plenty of power for 4Ω loads is already available at supply voltages below 15V anyway. To drive 4Ω speakers from a supply above 15V, it’s necessary to run the Mini-D in mono mode; more on that later. To drive two speakers in this mode, you will need to build two boards but in exchange for that, you siliconchip.com.au Fig.3: an external LC filter can be added if long speaker leads are to be used, with the filter component values selected according to the speaker impedance. These two diagrams show the recommended values for 8-ohm and 4-ohm loads. get more power and higher efficiency. Circuit description The full circuit is shown in Fig.4. All the real work is done by IC1. The left & right channel input signals are applied to RCA connectors CON1 and CON2. Alternatively, pin headers may be fitted in their place for connection to chassis sockets or another board. From this point on, we shall refer to the operation of one channel only. The signal first passes through a low-pass RF-rejecting filter, comprising a 100Ω series resistor and 1nF ceramic capacitor. Both the signal and ground pins are then AC-coupled to the volume control potentiometer (VR1) via 4.7µF ceramic capacitors. The signal ground is also connected to power supply ground via a 4.7Ω resistor, taking advantage of the differential inputs provided by the IC. This 4.7Ω resistor reduces the chance of hum being injected into the signal due to the common input grounds. The volume control potentiometer is either a dual-gang log pot (VR1) or two horizontal trimpots (VR2 & VR3), the latter used for a pre-set volume level. If you don’t need volume control at all, simply link out VR2 and VR3. Regardless, the wiper of each pot goes to the non-inverting input for each channel (pins 3 & 12) while the bottom (ACgrounded) end goes to the inverting inputs (pins 4 & 11). The TPA3113D2 can handle a strictly limited voltage range at each input pin of -0.3V to 6.3V so we have added protection components to limit these voltages when the power is off or in case a high level signal is applied (which is common when plugging and unplugging RCA leads). These parts consist of 5.6V zener diodes (ZD1-ZD4) and parallel Schottky diodes (D1 & D2) between each input and ground. The zener diodes take care of clamping positive signal swings while the Schottky diodes clamp negative excursions more effectively. The outputs of IC1 pass through the recommended output filter, consisting of four large ferrite beads (FB1-FB4; HI1812V101R-10) and four 1nF C0G ceramic capacitors. C0G capacitors have a very low temperature coefficient (±30ppm) but also low ESR (equivalent series resistance) and ESL (equivalent series inductance); just what we need to suppress sharp voltage spikes. September 2014  77 Speaker impedance: 6-8Ω; 4-8Ω in mono mode or in stereo with up to 16V supply Power LED1 (green) can either be an on-board SMD LED or it can be mounted off-board via pin header CON7. Supply current for LED1 and LED2 is around 1-2mA, so high-brightness types should be used. Continuous output power: 2 x 5W or 1 x 10W (12V, 8Ω) Power supply Specifications Supply voltage: 8-25V DC Quiescent current: typically <40mA active, <2mA shutdown Peak output power: 2 x >15W or 1 x >30W (thermally limited) THD+N: typically <0.1%; see Figs.6 & 7 Signal-to-noise ratio: 100dB Frequency response: 20Hz-20kHz ±1dB; see Fig.9 Efficiency: up to 82% (stereo), 91% (mono) Gain: 20dB, 26dB, 32dB or 36dB Under-voltage lockout: ~7.5V Output offset voltage: typically within ±1.5mV Power supply rejection ratio: typically -70dB Switching frequency: ~310kHz We have also added snubbers, consisting of 330pF C0G ceramic capacitors in series with 10Ω resistors, from each output to ground. They are actually wired to the boost supply pins but these are AC-coupled to the outputs via much larger 220nF capacitors so the effect is the same. These reduce radiated EMI further by limiting the output voltage slew rates. We have used a 1:1 voltage divider between GVDD (pin 9; ~7V) and ground, with a 4.7µF filter capacitor, to set PLIMIT (pin 10) at 3.5V. This limits the output amplitude to about ±11V (22V peak-to-peak). Thus it will only limit the output power with a DC supply over 20V. If you are trying to get the maximum possible power from the chip at 24V, you could reduce the upper divider resistor to 47kΩ but in most cases it won’t make much difference; the ‘soft clipping’ provided by this limiter may have some benefits in reduced treble artefacts if you are going to drive the amplifier that hard anyway. Other features 100kΩ pull-ups on GAIN0 and GAIN1 allow links LK4 and LK5 to define these input states. A table in the circuit diagram shows the possible settings. With a gain of 20dB (10x), input sensitivity is 425mV RMS for a 12V supply and 850mV RMS for a 24V supply. With the gain set to 36dB (63x), input sensitivity is 67mV RMS for a 12V supply and 135mV RMS for a 24V supply. The unit can handle signals up to 78  Silicon Chip at least 3V RMS. For line-level signal sources such as CD players, 20dB of gain should be plenty, so most constructors should stick with that. The FAULT output (pin 2) is connected to pin 1 on CON6, which can go to a microcontroller pin (but with some provisos, see below). It goes low if the IC detects that an output is short-circuited or there is a DC offset fault. The FAULT signal also switches P-channel Mosfet Q2 via a resistive divider (which ensures that Q2’s gate is not over-driven). If there is a fault, Q2 switches LED2 (red) on. This can either be an SMD LED mounted on the board or an external LED wired up via pin header CON8. The shut-down input (pin 1) is also connected to CON6 (at pin 2) and is pulled up by a 100kΩ resistor so that the amplifier will power up automatically. If pulled to ground, the amplifier shuts down and only draws about 250µA. However, that doesn’t include the current for LED1 and the various pull-ups, which increase total shutdown current to around 2mA. If a shorting block is placed on LK6 and an output short circuit is detected, once the short has cleared, the amplifier will automatically resume operation. Otherwise, short circuit faults are ‘latched’ and the unit remains off (with LED2 lit) until the power is turned off and back on again. Over-temperature faults are automatically cleared and LED2 will not light if IC1 overheats; rather, output will simply cease and then resume once it has cooled. The 8-25V DC supply (from a battery, plugpack or power ‘brick’) comes in via terminal block CON3, with Mosfet Q1 providing reverse polarity protection. If the supply polarity is correct, Q1’s gate is pulled positive via the 100kΩ resistor. This switches Q1 on, so current from the circuit can flow back to the supply ground. However, if the supply polarity is wrong, Q1’s gate will be pulled negative relative to its source and Q1 will remain off, so no ground current can flow and the circuit is protected. Q1’s drain-source voltage is rated at 25V, so as long as the DC supply is within the specified range, this will be sufficient to block the supply voltage. Zener diode ZD5 limits Q1’s gate voltage to a safe level when the supply voltage is above 20V. There’s little else to the power supply other than the bypass capacitors, which consist of one 100µF electrolytic, one 220nF X7R ceramic and one 1nF C0G ceramic for each pair of power VCC pins, ie, PVccL (pins 27 & 28) and PVccR (pins 15 & 16). The analog supply, AVcc, is at pin 7 of IC1 and has a 10Ω/4.7µF RC low-pass filter to remove switching noise. IC1’s internal Mosfet gate supply regulator has a 4.7µF output filter capacitor at pin 9 (GVDD). Mono (parallel) mode To operate in mono mode, LK1 and LK2 are fitted and LK3 is left out. The speaker in then connected between CON4 and CON5 as shown on the circuit diagram. LK1, LK2 & LK3 are 0Ω surface-mount resistors. In this case, you can also omit FB2, FB3 and the two associated 1nF capacitors. Plus you can omit CON1 and its associated components as the mono signal is fed into the right input (CON2). Note though that you will only get more power in mono mode (also known as PBTL or Parallel Bridge-Tied Load mode) with a low-impedance speaker, eg, 4Ω. This is because with higher speaker impedances, you will run into clipping before the maximum siliconchip.com.au siliconchip.com.au September 2014  79 2 2 1 10k LOG 32dB 36dB 26dB K1 VR1a 10k D2 BAT54A VR3 10k K1 A A G 5.6V 5.6V ZD3 A K ZD1 A K 5.6V 5.6V ZD4 A K ZD2 A K X7R 4.7 µF GAIN 0 LK4 100k 10Ω GAIN 1 LK5 100k LK6 AUTO RESET IN IN MONO LK2 OUT LK1 OUT MODE STEREO OUT IN LK3 STEREO/MONO MODE LINKING K2 K2 100k 100k ‘MINI–D’ 10W X 2 CLASS D AMPLIFIER OUT LOG VR1b 10k K λ A D1 BAT54A VR2 10k 10k FAULT LED2 D S Vcc GND X7R 4.7 µF 100k X7R 4.7 µF 100k X7R A ZD1- ZD4 K 100k 10 11 12 9 6 5 4 3 14 7 1 2 C0G 1nF A CATHODE BAND 220nF * EITHER DUAL LOG POT VR1 OR TRIMPOTS VR2 & VR3 ARE FITTED; THE LATTER FOR PRESET VOLUME STEREO LK3 100k FAULT LOW ESR 100 µF 25V SHUTDOWN CON6 DIAG 27 15 BSPL 16 BSNR OUTNR OUTPR BSPR BSNL OUTNL OUTPL PVccR IC1 TPA3113D2 28 LEDS K 8 0 K1 24 K2 BAT54A 19 AGND PAD PGND PGND PLIMIT RINN RINP GVDD GAIN1 GAIN0 LINN LINP PBTL AVcc SD FAULT PVccL A 21 20 18 17 22 23 25 26 X7R 330pF C0G FB4 FB3 330pF C0G 330pF C0G FB2 FB1 330pF C0G LOW ESR 100 µF 25V A K 10Ω 1nF C0G 1nF C0G 10Ω 10Ω 1nF C0G 1nF C0G 10Ω ZD5 5.6V G 100k G S D IRFML8244, BSS84 PAD UNDER 28 S D 1 – – + 1 CON5 – + RIGHT SPEAKER 2 DC 14 SPEAKER CONNECTION IN MONO MODE CON4 2 1 LEFT SPEAKER Q1 IRFML8244 CON3 1 POWER IN 2 + 8–25V TPA3113D2 NOTE: FB1–FB4 ARE HI1812V101R-10 SMD FERRITE ‘BEADS’ (1812 SIZE, 8A RATED) X7R 220nF LK2 MONO X7R 220nF X7R 220nF LK1 MONO X7R 220nF 220nF C0G 1nF Vcc Fig.4: the complete circuit diagram of the Mini-D 10W x 2 Class-D Amplifier. The audio input signals are fed in via CON1 & CON2, filtered and fed to the LINP (pin 3) and RINP (pin 12) inputs of IC1 via dual-gang volume control VR1. The amplified outputs appear at pins 25 & 23 (left channel) and pins 18 & 20 (right channel) of IC1 and these drive the left and right speakers via ferrite bead/capacitor filter networks. Links LK1-LK3 select either mono or stereo operation (see table), while links LK4 & LK5 set the gain. Mosfet Q1 provides reverse polarity protection, while Mosfet Q2 drives the fault indicator LED (LED2). SC 20 1 4 IN OUT OUT IN LK5 IN OUT 4.7 µF X7R 4.7 µF X7R 4.7 µF X7R 4.7 µF X7R K Q2 BSS84 GAIN 20dB POWER λ LED1 LK4 IN 4.7Ω C0G 1nF (MONO INPUT) RIGHT INPUT CON2 100Ω C0G 1nF 100Ω 4.7Ω CON1 LEFT INPUT CON8 K A (LED2 WHEN OFF PCB) CON7 A K 1 (LED1 WHEN OFF PCB) A 100k 220nF 4.7 µF 0 + LK3 STEREO 100 µF (LK1) 10Ω FB2 FB3 220nF 220nF 1nF 1nF 330pF 1nF 220nF Lout CON4 1nF 330pF 10Ω FB1 1nF CON5 4.7 µF 100k 330pF 10Ω (LK2) D2 ZD4 100k IC1 TPA3113D2 LK5 LED1 LK4 LED2 1nF 4.7 µF 10Ω Q1 10Ω 330pF FB4 + VR1 220nF 220nF 100k 100k (VR3) ZD5 – 100 µF LK6 + CON2 100Ω 4.7 µF 10kΩ + 10kΩ LOG POWER CON3 100k G CON6 S 100k A 4.7 µF A 1nF F ZD2 ZD1 1nF 100Ω 4.7 µF 10k 10k 4.7Ω Rin 100k (VR2) – Right in 100k – CON1 ZD3 Lin D1 4.7Ω 4.7 µF Q2 Left in Rout INSTALL DOTTED LINKS AT VR2 & VR3 FOR FIXED VOLUME ONLY – SEE TEXT NOTE: INSTALL LK3 (0 Ω) FOR STEREO. OMIT LK3 & INSTALL LK1 & LK2 FOR MONO Fig.5: follow this parts layout diagram to build the Mini-D amplifier. You can either install potentiometer VR1 or trimpots VR2 & VR3 for volume control (see text). Alternatively, leave all these parts out if no volume control is required and link out VR2 & VR3 as indicated. output current becomes the limiting factor. With a 4Ω speaker at 15V in mono mode, output power is up to 30W, which is pretty good! Even if you don’t need the extra power, it’s preferable to use the module in mono mode as it improves efficiency. 24V battery operation Since the maximum recommended operating supply voltage for IC1 is 26V and there are a number of 25V-rated components in the circuit, we don’t recommend running directly from a 24V battery. In theory, if you increased the voltage ratings of the 25V capacitors and Mosfet Q1, you might get away with it as the absolute maximum specified for IC1 is 30V. But it’s outside the recommended operating voltage range so we don’t suggest doing that. A better option is to use a 24V lowdropout pre-regulator, eg, by placing a 12V zener diode in series with the ground pin of an LM2940CT-12 regulator to ‘jack it up’ to 24V. You will need appropriate input and output filter capacitors. The LM2940 is only rated at 1A but is unlikely to run into current limiting during normal operation. It may need a small heatsink though, as it could dissipate up to 5W. PCB layout Being a switching amplifier, instantaneous currents can be high and the voltage rise/fall times are very short, so the the design of the PCB has been quite rigorous. We also wanted to keep switching noise away from the analog circuitry. Bypass capacitors need to be near IC pins and the output filter must be kept tight for maximum EMI suppression. There are also thermal considerations, given that the amplifier 80  Silicon Chip IC uses the board as a heatsink. We’ve placed ground planes on both the top and bottom of the board immediately under IC1 and fanned them out to the full width of the board. There are 15 vias placed directly under the IC, on and around its thermal pad, both to reduce ground impedance for better performance and to help conduct heat from the IC to the bottom side of the board where it can be effectively radiated away. The 1nF and 220nF bypass capacitors are immediately adjacent to the IC, with the 1nF C0G types the closest, as they have the best high-frequency performance. The placement of the 100µF electrolytics is less critical. Note that there is provision to use 22µF 25V SMD multi-layer ceramic capacitors (1812 size) instead but the cheaper electros do the job well. The IC’s pin layout is well-optimised, with the main power supply and all output related pins on one side, which we have orientated towards the right side of the PCB. Thus the filter components are placed immediately between the IC and CON4/CON5 at right. The analog ground pin (pin 8) is on the left side of the IC and this is the only point at which the power ground meets the signal (analog) ground. Construction Fig.5 shows the assembly details. Apart from some of the components being relatively close together, the only tricky thing about building this board is soldering IC1 (a magnifying lamp will come in handy here). We used hot-air reflow as this (or oven reflow) is best for ICs with thermal pads (like the TPA­3113D2). The equipment is surprisingly cheap; we paid around $60 for an Atten 858D+ hot-air soldering station while hot-air reflow wands can be had for as little as $25. But you can do it with a regular soldering iron too. For hot-air, the trick is to use a very thin layer of fresh solder paste (kept in the fridge!). Spread this sparingly on the pads, drop the IC on top, heat it (gently at first) until all the pins reflow and then for a few seconds longer and Bob’s your uncle. If all you have is a regular iron, apply some no-clean flux paste to the thermal pad on the board and also the pad on the bottom of the IC. Then melt a small amount of solder to both; just enough to tin them. Start with the PCB pad so you can get an idea of the correct amount. If you add too much, add a bit more flux and then remove the excess with some solder wick (harder to do with the IC!). Having tinned both, place some fresh flux paste on all the IC pads on the PCB, including the thermal pad, then pop the IC down in place, checking its orientation. Next, move it slightly out of the way, tin one small corner pad and then slide the IC into place while heating that pad. Now check that the IC lines up with all its pads. If it’s misaligned, reheat and gently nudge it into place. Try to avoid getting solder on any other pads Use a magnifying glass (or magnifying lamp) to check carefully that all the pins are sitting properly over their pads, then tack down the diagonally opposite pin. Re-check the alignment, then solder the rest of the pins, making sure not to disturb either of those first two solder joints. Having soldered the pins, it will then be necessary to flip the board over siliconchip.com.au and apply enough solder to the pad on the bottom to transfer heat through the vias. Heat this solder until the flux between the IC and board vapourises, indicating that the thermal pad has reflowed. This will take a good few seconds but don’t overdo it as you could cook the chip. Regardless of which method you used to solder the IC, check carefully for bridged pins (again, use a magnifying glass) and clean up any that look dodgy up with some flux paste and a clean piece of solder wick. The bridges should clear easily; press the wick down onto the board but not over the IC pins as they are small and easily damaged. As a final measure, it’s a good idea to clean the flux residue off the board using a specialised flux cleaner (or in a pinch, an alcohol or acetone) and then carefully check all the soldering, again with a magnifying glass. Check that all the bridges are gone and that the solder has flowed cleanly onto all the pins and pads. Remaining parts There are nine SOT-23 package transistors, diodes and zener diodes to solder. These are quite easy as the pins are well spaced but don’t get the various device types mixed up. Start with Q1 and Q2, then solder D1 and D2 and finally the five identical zener diodes. The easiest method is to put a bit of solder on the central pad and slide the device into place while heating that pad. Then solder the other two pads (a dab of flux paste makes it easier) and refresh the first. Now move onto the SMD passives, starting with the resistors and then the capacitors and ferrite beads. Use siliconchip.com.au Mini-D THD+N vs Power 08/12/14 13:27:27 Filter: AP AUX-0025 + 20Hz-80kHz bandpass Total Harmonic Distortion + Noise (%) 5 2 1 0.5 4Ω+47µH <at> 15V 0.2 0.1 8Ω+47µH <at> 18V 4Ω+47µH <at> 12V 0.05 8Ω+47µH <at> 12V 0.02 0.01 0.1 8Ω+47µH <at> 14.4V Solid = Stereo mode Dashed = Mono mode 0.2 0.5 1 2 5 10 20 Power ( W atts) Fig.6: distortion versus power for a range of load impedances and supply voltages. Performance is generally better for 8Ω loads but power delivery is higher into 4Ω. Note the test load series inductance, to simulate loudspeakers. 10 Mini-D THD+N vs Frequency 08/12/14 13:28:51 Filter: AP AUX-0025 + 20Hz-80kHz bandpass 5 Total Harmonic Distortion + Noise (%) Left: the completed PCB assembly. Don’t be intimidated by the SMD parts; they’re quite easy to install if you follow the instructions in the text but you do need a good magnifying glass (or magnifing lamp), tweezers and a soldering iron with a small chisel tip. 10 2 1 0.5 0.2 4Ω+47µH <at> 12V, 1W 0.1 0.05 8Ω+47µH <at> 12V, 1W 0.02 0.01 20 50 100 200 500 1k 2k 5k 10k 20k Frequency (Hz) Fig.7: distortion versus frequency at 1W. As usual, the distortion rises with frequency but it also rises at the low end due to coupling capacitor-induced distortion. It’s below 0.1% between 40Hz and about 1.5kHz. a similar method as for the SOT-23s. The main thing to check for with these parts is that the solder has flowed onto the pad and not just the end of the component. As before, adding flux smoothes solder flow. Note that the resistors will have printed values on them but the capacitors and ferrite beads will not, so check the packaging before fitting them. Remember to fit either 0Ω resistor LK3 only (stereo mode) or LK1 and LK2 only (mono mode). If using the SMD LEDs, they can September 2014  81 V+ VCC CON3 + S1 K K A A − (OPTIONAL CLAMP DIODES) MICROCONTROLLER 1 1k 2 470nF 1 FAULT POWER 10Ω SHUTDOWN 2 100 µF 25V MINI-D 3 MINI-D FAULT SHUTDOWN 3 GND CON6 GND CON6 B: START-UP DELAY & SHUTDOWN WITH A SWITCH A: CONNECTING A MICROCONTROLLER TO THE MINI-D Fig.8(a): the shutdown pin (pin 2) of CON6 can be pulled low under no-signal conditions (eg, using a microcontroller) to reduce power consumption. The RC filter shown provides slew rate limiting, while external clamp diodes may also be required with some micros (see text). Fig.8(b) at right shows how to add a capacitor (eg, 100μF) to give a switch-on delay, while a DPDT power switch (S1) can be used to eliminate switch-off clicks or pops. go in next but first you will have to check their polarity. Unfortunately, markings are inconsistent so use a DMM in diode test mode and try connecting the probes both ways around. When the LED lights, the red probe is to the anode and this goes towards the bottom of the PCB (marked with “A”). We used a green LED for LED1 and a red LED for LED2. Through-hole components That’s it for the SMDs so once you’re confident that they’ve all been +3 soldered correctly, there are just a few through-hole parts left. If you aren’t using an on-board volume control, solder wire links in place of VR2 and VR3 where shown. Also, if using off-board LEDs, fit 2-way pin headers CON7 and CON8 in place of the LEDs. Next, move on to links LK4-LK6, CON6 and the inputs (if you aren’t fitting RCA sockets). That done, dovetail two screw terminal blocks together and solder them in place for CON4 & CON5 (wire entry holes facing outwards). CON3 can then go in. Mini-D Frequency Response 08/12/14 12:39:04 Filter: AP AUX-0025 + 80kHz lowpass +2.5 +2 Amplitude Variation (dBr) +1.5 +1 +0.5 +0.0 -0.5 -1 -1.5 Set-up & testing Initially, fit LK4 and LK5 (note that they go in vertically) and LK6. Turn the volume pot(s) to minimum, then apply DC power to CON3 (say, 12V) and measure the current. It should be just under 40mA (but possibly as high as 55mA) and LED1 should be on while LED2 should be off. If anything is wrong, switch off immediately and check for faults. Also double-check that you have connected the supply wires with the correct polarity. Assuming that all is OK, switch off and connect a signal source such as a CD player, MP3 player, oscillator or mobile phone. Connect the speaker(s), then switch back on and slowly turn the volume up. It’s now just a matter of making sure it sounds right. If you get to maximum volume and it’s still too quiet, switch off and increase the gain by changing LK4 and/or LK5 but remember to turn the volume down before re-applying power. Shutdown control -2 -2.5 -3 If you are using onboard volume control pot VR1, fit it now (or trimpots VR2 & VR3). RCA sockets CON1 & CON2 can then go in, followed by the electrolytic capacitors (take care with their orientation). 20 50 100 200 500 1k 2k 5k 10k 20k Frequency (Hz) Fig.9: the amplifier’s response is effectively flat in the audible frequency range. There is a low-frequency roll-off due to the high-pass filter formed by the input coupling capacitors and volume pot, while the rise at the high end can be attributed at least partially to the inductance of our test load. 82  Silicon Chip To reduce power consumption when power is applied but no signal is present, you can pull the shutdown input (pin 2 of CON6, pin 1 of IC1) low to enter a power-saving state. However, there are a couple of provisos. First, the data sheet specifies that this pin should be slew rate limited to 10V/ms unless the source impedance siliconchip.com.au SIGNAL HOUND USB-based spectrum analyzers and RF recorders. The TPA3113D2 IC (circled) should be installed first, following the procedure described in the text. The photo above right shows the corresponding heatsink area for this IC on the back of the PCB. It’s connected to a thermal pad on the top of the board by 15 vias. This larger-thanlife-size view shows the heatsink pad on the underside of the TPA3113D2 Class-D audio amplifier IC. is at least 100kΩ but it doesn’t say why. Confusingly, they also show sample circuits where a “control system” (eg, a microcontroller) drives the shutdown pin via just a 1kΩ series resistor, which is unlikely to limit the slew rate to their specification. We would be tempted to try that but not knowing the reason for the limitation, a safer approach would be to add an RC filter, as shown in Fig.8(a). The same comments apply if you’re going to use a switch, relay, transistor or something else to pull down the shutdown pin. If connecting a micro in this manner, note that the on-board pull-up resistor could pull its control pin above the micro’s supply voltage. Normally, the microcontroller pin will have a clamp diode to its positive supply rail to limit the voltage on that pin to a safe level. However, some micros lack a positive clamp diode (eg, 5V-tolerant pins on a 3.3V micro) and in that case, you will need to add an external clamp diode (or a low-voltage zener to ground) to protect the micro – see Fig.8(a). The situation is the same if connecting the FAULT signal to a micro. Powering up & down We didn’t hear any clicks or pops or run into other issues when powering the Mini-D up or down normally but there are a couple of issues noted in the data sheet which constructors should be aware of. If the signal source is powered up at the same time as the Mini-D and there are large initial transients on those signals, that could trigger the siliconchip.com.au SA44B: $1,320 inc GST • • • • • Up to 4.4GHz Preamp for improved sensitivity and reduced LO leakage. Thermometer for temperature correction and improved accuracy AM/FM/SSB/CW demod USB 2.0 interface SA12B: $2,948 inc GST • • • Up to 12.4GHz plus all the advanced features of the SA44B AM/FM/SSB/CW demod USB 2.0 interface The BB60C supercedes the BB60A, with new specifications: • • • • Another view of the completed PCB assembly. Links LK1-LK3 have been configured for stereo operation; ie, LK1 (0Ω) in, LK2 & LK3 out. DC offset protection in the Mini-D and once that’s activated, its outputs will remain disabled until the power is switched off and on again. So in that case you need to hold shutdown low until the audio signals stabilise. This can be achieved with a capacitor between the shutdown pin and ground. A 22µF capacitor will give a switch-on delay of around 100-200ms, a 100µF 500-1000ms and so on. Or if a micro is connected to shutdown, it can do the same job. The data sheet also states that pulling shutdown low before power is removed will minimise clicks or pops. While not strictly necessary, this can be achieved using a DPDT power switch; see Fig.8(b). This will bring shutdown low almost immediately while the supply capacitors take some SC time to discharge. • The BB60C streams 140 MB/sec of digitized RF to your PC utilizing USB 3.0. An instantaneous bandwidth of 27 MHz. Sweep speeds of 24 GHz/sec. The BB60C also adds new functionality in the form of configurable I/Q. Streaming bandwidths which will be retroactively available on the BB60A. Vendor and Third-Party Software Available. Ideal tool for lab and test bench use, engineering students, ham radio enthusiasts and hobbyists. Tracking generators also available. Silvertone Electronics 1/8 Fitzhardinge St Wagga Wagga NSW 2650 Ph: (02) 6931 8252 contact<at>silvertone.com.au September 2014  83