Many modern circuits tend to work from a single supply voltage of 3V. But often they need a virtual earth at half the supply voltage for efficient operation. The splitter shown in the diagram bisects the supply voltage with a high-resistance potential divider, R1-R2, and buffers the resulting 1.5 V line with an op amp. Since the op amp used is not a fast type, the output is decoupled by capacitive divider C2-C3. This ensures that the impedance of the virtual earth point remains low over a wide frequency band. Because the potential at the junction C2-C3-R3 is fed back to the inverting input of IC1, the circuit becomes a standard voltage follower.

Resistor R3 ensures that the regulation remains stable. The circuit can regulate ±2mA without any difficulties. Because of the low current drawn by IC1, and the high resistance of R1 and R2, the overall current drain is low. In the absence of a load, it was 13µA in the prototype, of which 1.5µA flows through R1-R2. Finally, since IC1 can operate from a voltage as low as 1.6V, the splitter will remain fully operational when the battery nears the end of its charge or life.

The circuit in the diagram enables a negative voltage to be derived without the use of integrated circuits. Instead, it uses five n-p-n transistors that are driven by a 1 kHz (approx) TTL clock. When the clock input is high, transistors T1 and T2 link capacitor C1 to the supply voltage, UIN, which typically is 5 V. During this process, transistor T5 conducts so that T3 and T4 are off. When the clock input is low, T5 is cut off, whereupon transistors T3 and T4 are switched on via pull-up resistor R6 and either R4 or R5.

This results in the charge on C1 being shared between this capacitor and C2 Since the +ve terminal of C2 is at ground potential, its –ve terminal must become negative w.r.t. earth. The high level at the clock input must be of the same order as the positive input voltage, UIN, otherwise T1 cannot be switched on. The clock frequency should be around 1 kHz to ensure a duty cycle ratio of 1:1. Altering the ratio results in a different level of negative output voltage, but this is always smaller than that with a ratio of 1:1.

The title of this article naturally raises the question of why we think that the generous selection of fully integrated voltage regulators needs to be extended with a version constructed using discrete components. In other words, what does this circuit offer that the well-known ‘three-leggers’ don’t have? To start with, we can point out that this circuit is refreshingly simple for a discrete version. Three semiconductors, three resistors, a capacitor and a diode are all it needs. Of course, that’s still more components than an integrated regulator, so what exactly are the advantages of this circuit?

They are to be found in three areas: voltage range, bandwidth and current rating. The last of these is the primary strength of this circuit, since the maximum current depends only on the specifications of the output transistor. With the BD680, as used here, a current of 4 A can be delivered at a collect-emitter voltage of 10 V with adequate cooling (Rth = 3.12 K/W). The peak current is even 6 A. Try matching that with an integrated voltage regulator! The maximum input voltage is 30 V with the illustrated version of the circuit (UDSmax of T1), but this can easily be increased by using special high-voltage transistors.

The same applies to the bandwidth, which can be extended as desired, without any modifications to the circuit, by using high-speed transistors. Generally speaking, wide bandwidth is also not one of the strong points of integrated voltage regulators. As noted, the circuit is basically very simple. A zener diode (D1) fed with a constant current of around 1mA by a JFET current source (T1) provides the reference potential. C1 is connected in parallel with D1 to provide well-behaved startup behaviour (soft start). This capacitor also provides additional buffering and decouples noise and other disturbances. The startup time is around three seconds.

The only additional item that is needed for the voltage regulator is an output buffer for the reference potential. This takes the form of a sort of super-Darlington using T2 and T3. This works very well, but has the disadvantage that the output voltage is a bit lower (one diode drop) than the Zener voltage. P1 can be added to correct this, but this does reduce the regulation of the circuit. If the voltage difference is not important, it is thus better to replace P1 with a wire jumper. The main specifications of the voltage regulator are listed in Table 1.

This article should satisfy those who might want to build a low power FM transmitter. It is designed to use an input from another sound source (such as a guitar or microphone), and transmits on the commercial FM band – it is actually quite powerful, so make sure that you don’t use it to transmit anything sensitive – it could easily be picked up from several hundred metres away. The FM band is 88 to 108MHz, and although it is getting fairly crowded nearly everywhere, you should still be able to find a blank spot on the dial.

NOTE: A few people have had trouble with this circuit. The biggest problem is not knowing if it is even oscillating, since the frequency is outside the range of most simple oscilloscopes. See Project 74 for a simple RF probe that will (or should) tell you that you have a useful signal at the antenna. If so, then you know it oscillates, and just have to find out at what frequency. This may require the use of an RF frequency counter if you just cannot locate the FM band.

Description

The circuit of the transmitter is shown in Figure 1, and as you can see it is quite simple. The first stage is the oscillator, and is tuned with the variable capacitor. Select an unused frequency, and carefully adjust C3 until the background noise stops (you have to disable the FM receiver’s mute circuit to hear this).

Because the trimmer cap is very sensitive, make the final frequency adjustment on the receiver. When assembling the circuit, make sure the rotor of C3 is connected to the +9V supply. This ensures that there will be minimal frequency disturbance when the screwdriver touches the adjustment shaft. You can use a small piece of non copper-clad circuit board to make a screwdriver – this will not alter the frequency.

The frequency stability is improved considerably by adding a capacitor from the base of Q1 to ground. This ensures that the transistor operates in true common base at RF. A value of 1nF (ceramic) as shown is suitable, and will also limit the HF response to 15 kHz – this is a benefit for a simple circuit like this, and even commercial FM is usually limited to a 15kHz bandwidth.

Capacitors
All capacitors must be ceramic (with the exception of C1, see below), with C2 and C6 preferably being N750 (Negative temperature coefficient, 750 parts per million per degree Celsius). The others should be NPO types, since temperature correction is not needed (nor is it desirable). If you cannot get N750 caps, don’t worry too much, the frequency stability of the circuit is not that good anyway (as with all simple transmitters).

How It Works
Q1 is the oscillator, and is a conventional Colpitts design. L1 and C3 (in parallel with C2) tunes the circuit to the desired frequency, and the output (from the emitter of Q1) is fed to the buffer and amplifier Q2. This isolates the antenna from the oscillator giving much better frequency stability, as well as providing considerable extra gain. L2 and C6 form a tuned collector load, and C7 helps to further isolate the circuit from the antenna, as well as preventing any possibility of short circuits should the antenna contact the grounded metal case that would normally be used for the complete transmitter.

The audio signal applied to the base of Q1 causes the frequency to change, as the transistor’s collector current is modulated by the audio. This provides the frequency modulation (FM) that can be received on any standard FM band receiver. The audio input must be kept to a maximum of about 100mV, although this will vary somewhat from one unit to the next. Higher levels will cause the deviation (the maximum frequency shift) to exceed the limits in the receiver – usually ±75kHz.

With the value shown for C1, this limits the lower frequency response to about 50Hz (based only on R1, which is somewhat pessimistic) – if you need to go lower than this, then use a 1uF cap instead, which will allow a response down to at least 15Hz. C1 may be polyester or mylar, or a 1uF electrolytic may be used, either bipolar or polarised. If polarised, the positive terminal must connect to the 10k resistor.

Inductors
The inductors are nominally 10 turns (actually 9.5) of 1mm diameter enamelled copper wire. They are close wound on a 3mm diameter former, which is removed after the coils are wound. Carefully scrape away the enamel where the coil ends will go through the board – all the enamel must be removed to ensure good contact. Figure 2 shows a detail drawing of a coil. The coils should be mounted about 2mm above the board.

For those still stuck in the dark ages with imperial measurements (grin), 1mm is about 0.04″ (0.0394″) or 5/127 inch (chuckle) – you will have to work out what gauge that is, depending on which wire gauge system you use (there are several). You can see the benefits of metric already, can’t you? To work out the other measurements, 1″ = 25.4mm

NOTE: The inductors are critical, and must be wound exactly as described, or the frequency will be wrong.

The nominal (and very approximate) inductance for the coils is about 130nH.This is calculated according to the formula …

L = N² * r² / (228r + 254l)

… where L = inductance in microhenries (uH), N = number of turns, r = average coil radius (2.0mm for the coil as shown), and l = coil length. All dimensions are in millimetres.

Pre-Emphasis

It is normal with FM transmission that “pre-emphasis” is used, and there is a corresponding amount of de-emphasis at the receiver. There are two standards (of course) – most of the world uses a 50us time constant, and the US uses 75us. These time constants represent a frequency of 3183Hz and 2122Hz respectively. This is the 3dB point of a simple filter that boosts the high frequencies on transmission and cuts the same highs again on reception, restoring the frequency response to normal, and reducing noise.

The simple transmitter above does not have this built in, so it can be added to the microphone preamp or line stage buffer circuit. These are both shown in Figure 3, and are of much higher quality than the standard offerings in most other designs.

Rather than a simple single transistor amp, using a TL061 opamp gives much better distortion figures, and a more predictable output impedance to the transmitter. If you want to use a dynamic microphone, leave out R1 (5.6k) since this is only needed to power an electret mic insert. The gain control (for either circuit) can be an internal preset, or a normal pot to allow adjustment to the maximum level without distortion with different signal sources. The 100nF bypass capacitors must be ceramic types, because of the frequency. Note that although a TL072 might work, they are not designed to operate at the low supply voltage used. The TL061 is specifically designed for low power operation.

The mic preamp has a maximum gain of 22, giving a microphone sensitivity of around 5mV. The line preamp has a gain of unity, so maximum input sensitivity is 100mV. Select the appropriate capacitor value for pre-emphasis as shown in Figure 3 depending on where you live. The pre-emphasis is not especially accurate, but will be quite good enough for the sorts of uses that a low power FM transmitter will be put to. Needless to say, this does not include “bugging” of rooms, as this is illegal almost everywhere.

I would advise that the preamp be in its own small sub-enclosure to prevent RF from entering the opamp input. This does not need to be anything fancy, and you could even just wrap some insulation around the preamp then just wrap the entire preamp unit in aluminium foil. Remember to make a good earth connection to the foil, or the shielding will serve no purpose.

Some older Strand dimmer units used a zero to -10V control signal, and the standard analogue control voltage is zero to +10V. This project allows the easy conversion from one standard to another. This is a very simple project, but may turn out to be a lifesaver for small theatre groups and the like. It has come to my attention that there are still a great many old Strand dimmers very much in use. The problem is that they are just too reliable, and won’t go away … but, they use a zero to -10V control signal, so are incompatible with the dimmer unit in these project pages, and with any new commercial analogue control console.

In addition, there are no doubt quite a few old lighting consoles that use this standard, which means that they can’t drive modern dimmer packs. As it turns out, a simple opamp inverter will convert either standard to the other. This is shown in Figure 1.

There is really nothing to it. Use as many circuits as needed, and a simple power supply (such as that in Project 05) will drive as many of these inverters as are likely to be required in any lighting setup. The above circuit has two channels, and may be simply repeated as many times as you need to get the required number of channels. The 100 ohm resistors on each output are there to prevent the opamps from oscillating when supplying a capacitive load (such as a coax cable).

With an input of zero volts, the output will also be at zero volts. As the input increases (or decreases in the case of the -10V control) the output will change by exactly the same value, but in the opposite direction. Wiring is not critical, the 1458 opamps specified are very cheap (but more than capable of doing the job), and they can be built very simply on Veroboard or similar. Supplies should be bypassed to common (ground) with 10uF electrolytic caps.

Firstly, I’d like to stress that the intended use of this circuit is only one of many possible applications. Apart from the obvious usage as a headphone amplifier, the circuit can be used for a range of applications where a wide bandwidth low power amplifier is needed. Some of the options include …

• Reverb drive amplifier – ideal for low and medium impedance reverb tanks
• High current line driver – suitable for very long balanced lines
• Low power speaker amplifier – better performance than small integrated amps
• … and of course, a headphone amp.

In short, the amp can be used anywhere that you need an opamp with more output current than normally available. Since most are rated for around ±20-50mA, general purpose opamps are not suitable for driving long cables or anywhere else that a relatively high output current is needed.

As a headphone amplifier, this design is very similar to others on the ESP site, but the main difference is that this one (and P70) has been built and fully tested. The design is fairly standard, and every variation was checked out before arriving at the final circuit. A photo of the prototype is shown below, and at only 64 x 38mm (2.5 x 1.5 inches) it is very small – naturally, the heatsink is not included in the dimensions.

The amplifier is capable of delivering around 1.5W into 8 ohm headphones, and 2.2W into 32 ohms – this is vastly more than will ever be needed in practice. The use of a 120 Ohm output resistor is recommended, as this is supposed to be the standard source impedance for headphones. Unfortunately, many users have found that their ‘phones perform better when driven from a low impedance source.

The circuit is based on an opamp, with its output current boosted by a pair of transistors. Distortion is well below my measurement threshold at all levels below clipping into any impedance. Noise is virtually non-existent – even with a compression driver held to my ear, I could barely hear any, and I couldn’t hear any with headphones.

WARNING
Headphones are rated in dB SPL at 1mW, and this amplifier (like many other similar headphone amps) is capable of producing extreme SPLs. The levels obtainable are sufficient to cause almost instantaneous permanent hearing damage! Never operate the amp at very high levels, and never switch the amplifier on with signal while wearing you headphones.

Always start with the volume control at minimum, and gradually increase the level until it is comfortable, but not too loud. Because of the very low distortion, it is easy to increase the level too far without noticing. Your ears are precious – safeguard them at all times.

Note the warning above – this is serious. Most headphones are capable of at least 94dB SPL at 1 mW, with some as high as 107dB SPL. Even 10mW is enough to create sound levels capable of causing hearing damage, so you must be very careful to avoid damaging levels.

Continuous dB SPL	Maximum Exposure Time
85	8 hours
88	4 hours
91	2 hours
94	1 hour
97	30 minutes
100	15 minutes
103	7.5 minutes
106	< 4 minutes
109	< 2minutes
112	~ 1 minute
115	~ 30 seconds

Table 1 – Maximum Exposure to SPL

Note that the exposure time is for any 24 hour period, and is halved for each 3dB SPL above 85dB. The above shows the accepted standards for recommended permissible exposure time for continuous time weighted average noise, according to NIOSH (National Institute for Occupational Safety and Health) and CDC (Centers for Disease Control). Although these standards are US based, they apply pretty much equally in most countries – hearing loss does not respect national boundaries.

Description

The amplifier itself is fairly conventional, and is very similar to another shown on this site (see Project 24). This amplifier does not include the active volume control, because in general it is far easier to get a good log pot (or simply ‘fake’ the pot’s law as described in Project 01). Likewise, it does not include the cross-feed described in Project 109. If this is desired, it is very easy to implement on a small piece of tag board, or even ‘sky hook’ the few components off the bypass switch. Full details of how to do this will be included in the construction guide when PCBs are available.

The output transistors are biased using only resistors, rather than constant current sources. Extensive testing showed that using current sources made no discernible difference to performance, but increased the complexity and PCB size. Using separate caps for each biasing diode does make a difference though – and although it is relatively minor, the use of the two caps is justified IMHO.

The bias diodes should be 1N4148 or similar – power diodes are not recommended, as their forward voltage is too low. This may result in distortion around the crossover region, where one transistor turns off and the other on. As shown, crossover distortion is absolutely unmeasurable with the equipment I have available.

Above is the schematic of one channel. Resistors and caps use the suffix ‘R’ for the right channel. The second half of the dual opamp powers the right channel. Note that the volume control shown is optional, and is not on the PCB. If needed, it may be mounted in a convenient location and the output connected to the inputs of the board as shown. D1 and D2 (L and R) are 1N4148 or similar.

One of the reasons the amp is so quiet is that the entire board runs from a regulated supply, so hum (in particular) is eliminated. Although an unregulated supply can be used, this is not recommended. The supply should be separate from that used for your preamp, because of the relatively high current drawn by the amplifier (at least with low impedance ‘phones). A P05 preamp supply can be used, and will ensure optimum performance.

The prototype amplifier has flat frequency response from 10Hz to over 100kHz. Distortion is below my measurement threshold with any level or load impedance, and output impedance is almost immeasurably low. Your headphones may be designed to operate from a 120? source impedance (many are), so this may be added if it improves sound quality. Adding any series resistance will reduce the available power, but it is already far greater than you can use. Without series resistance, the minimum power into various load impedances is given below (based on ±15V supplies).

Impedance	Power (Direct)	120 Ohm Feed
8 Ohms	1.5 W	35 mW
32 Ohms	2.2 W	99 mW
65 Ohms	1.1 W	136 mW
120 Ohms	595 mW	149 mW
300 Ohms	238 mW	121 mW
600 Ohms	119 mW	82 mW

Table 2 – Output Power Vs. Impedance

This is not especially comprehensive, but will cover the majority of headphones in common use. In all cases, the available power is more than needed … not so you can damage your hearing, but to allow adequate headroom for transients.

Construction

While it may be possible to build it using Veroboard or similar, there is a high risk that it will oscillate because of the very wide bandwidth of the amplifier. A capacitor may be added in parallel with R4 (L and R) to reduce the bandwidth if stability problems are encountered. Although I used an NE5532 opamp for the prototype, the circuit will also work with a TL072, but at reduced power. You may also substitute an OPA2134 or your favorite device, taking note of the following …
The standard pinout for a dual opamp is shown on the left. If the opamps are installed backwards, they will almost certainly fail, so be careful.

The suggested NE5532 opamp was used for the prototype, and performance is exemplary. Devices such as the TL072 will be quite satisfactory for most work, but if you prefer to use ultra low noise or wide bandwidth devices, that choice is yours.

Construction is fairly critical. Because of the wide bandwidth of the NE5532 and many other audio grade opamps, the amplifier may oscillate (the prototype initially had an oscillation at almost 500kHz), so care is needed to ensure there is adequate separation between inputs and outputs. Even a small capacitive coupling between the two may be enough to cause problems.

As shown in the photo, this amplifier needs a heatsink. While it can operate without one at low power using high impedance headphones, you need to plan for all possibilities (after all, you may purchase low impedance ‘phones sometime in the future). The heatsink does not need to be massive, and the one shown above is fine for normal listening levels. An aluminium bracket may be used to attach to the chassis – I recommend 3mm material. Note that the heatsink should always be earthed (grounded).

The output transistors must be insulated from the heatsink. Sil-Pads™ are quite suitable because of the relatively low dissipation, but greased mica or Kapton can be used if you prefer. If you use the suggested 3mm aluminium, you can drill and tap threads into the heatsink, removing the need for nuts.

Testing

Connect to a suitable power supply – remember that the supply earth (ground) must be connected! When powering up for the first time, use 56 ohm “safety” resistors in series with each supply to limit the current in case you have made a mistake in the wiring. These will reduce the supply voltage considerably because of the bias current of the output transistors.

If the voltage at the amplifier supply pins is greater than ±6V and the output voltage is close to zero, then the amplifier is probably working fine. If you have an oscilloscope, check for oscillation at the outputs … at all volume control settings. Do this without connecting your headphones – if the amp oscillates, it may damage them.

Once you are sure that all is well, you may remove the safety resistors and permanently wire the amplifier into your chassis.

Crowbar circuits are so-called because their operation is the equivalent of dropping a crowbar (large steel digging implement) across the terminals. It is only ever used as a last resort, and can only be used where the attached circuit is properly fused or incorporates other protective measures.

A crowbar circuit is potentially destructive – if the circuitry only has a minor fault, it will be a major fault by the time a crowbar has done its job. It is not uncommon for the crowbar circuit to be destroyed as well – the purpose is to protect the device(s) attached to the circuit – in this case, a loudspeaker.

Description

There’s really nothing to it. A resistor / capacitor circuit isolates the trigger circuit from normal AC signals. Should there be enough DC to activate the DIAC trigger, the cap is discharged into the gate of the TRIAC, which instantly turns on … hard. A TRIAC has two basic states, on and off. The in-between state exists, but is so fast that it can be ignored for all intents and purposes.

The BR100 DIAC (or the equivalent DB3 from ST Microelectronics) is rated for a breakdown voltage of between 28 and 36V – these are not precision devices. Needless to say, using the circuit with supply voltages less than around 40V is not recommended, as you will have a false sense of security. The supply voltage must be higher than the breakdown voltage of the DIAC, or it cannot conduct. Zeners cannot be used as a substitute for lower voltages – a DIAC has a negative impedance characteristic, so when it conducts, it will dump almost the full charge in C1 into the gate of the TRIAC. This is essential to make sure the TRIAC is switched into conduction.

The TRIAC is a common type, and may be substituted if you know the specifications. It’s rated at 12A, but the peak current (non-repetitive) is 95A, and it only needs to sustain that until the fuse (or an output transistor) blows. A heatsink is preferred, but there is a good chance that the TRIAC will blow up if it has to protect your speakers, so it may not matter too much. The 0.47 ohm resistor is simply to ensure that the short circuit isn’t absolute. This will limit the current a little, and increases the chance that the TRIAC will survive (albeit marginally). Feel free to use a BT139 if it makes you feel better – these are rated at 16A continuous, and 140A non-repetitive peak current.

The peak short circuit current will typically be about 90A for a ±60V supply, allowing ~0.2 ohms for wiring resistance and the intrinsic internal resistance of the TRIAC, plus the equivalent series resistance of the filter capacitors. That’s a seriously high current, and it will do an injury to anything that’s part of the discharge path. Such high currents are not advised for filter caps either, but being non-repetitive they will almost certainly survive.

Construction & Use

Apart from the obvious requirement that you don’t make any mistakes, construction is not critical. Wiring needs to be of a reasonable gauge, and should be tied down with cable ties or similar. C1 must be polyester. While a non-polarised electrolytic would seem to be acceptable, the circuit will operate if the capacitor should dry out over the years. This means it will lose capacitance, and at some point, the crowbar may operate on normal programme material. This would not be good, as it will blow up your amplifier!

Make sure that all connections are secure and well soldered. Remember that this is the last chance for your speakers, so it needs to be able to remain inactive for years and years – hopefully it will never happen. The circuit doesn’t have to be mounted in the amplifier chassis – it can be installed in your speaker cabinet. Nothing gets hot unless it operates, at which point no-one really cares – it just has to save the speakers from destruction once to have been worthwhile.

Remember that the crowbar circuit absolutely must never be allowed to operate with any normal signal. A perfectly good amplifier that triggers the circuit because of a high-level bass signal (for example) will very likely be seriously damaged if the crowbar activates. To verify that no signal can trigger it, you may want to (temporarily) use a small lamp in place of R2, and drive the amp to maximum power with bass-heavy material.

A speaker does not need to be connected. If the lamp flashes, your amp would have been damaged. If this occurs, you may want to increase the value of C1. Note that bipolar electrolytics should never be used for C1, because they can dry out and lose capacitance as they age. This could cause the circuit to false-trigger.

This little project came about as a result of a design job for a client. One of the items needed was a mic preamp, and the project didn’t warrant a design such as the P66 preamp, since it is intended for basic PA only. Since mic preamps are needed by people for all manner of projects, this little board may be just what’s needed for interfacing a balanced microphone with PC sound cards or other gear. Unlike most of my boards, this one is double-sided. I normally avoid double-sided PCBs for projects because rework by those inexperienced in working with them will almost certainly damage the board beyond repair. I consider this not to be an issue with this preamp, because it is so simple. It is extremely difficult to make a mistake because of the simplicity.

As you can see, the board uses a PCB mounted XLR connector and pot, so is a complete mic preamp, ready to go. Feel free to ignore the terminals marked SW1 (centred between the two electrolytic supply caps), as they are specific to my client’s needs and are not useful for most applications. The original use was to use them for a push-button switch that activated an audio switch via a PIC micro-controller. They are not shown on the schematic.

The DC, GND and output terminals may be hard wired to the board, you may use PCB pins or a 10-way IDC (Insulation Displacement Connector) and ribbon cable. Power can be anything between +/-9V and +/-18V with an NE5532 opamp. The mic input is electronically balanced, and noise is quite low if you use the suggested opamp. Gain range is from about 12dB to 37dB as shown. It can be increased by reducing the value of R6, but this should not be necessary. Because anti-log pots are not available, the gain control is not especially linear, but unfortunately in this respect there is almost no alternative and the same problem occurs with all mic preamps using a similar variable gain control system.

The circuit is quite conventional, and if 1% metal film resistors are used throughout it will have at least 40dB of common mode rejection with worst-case values. The input capacitors give a low frequency rolloff of -3dB at about 104Hz. If better low frequency response is required, these caps may be increased to 4.7uF or 10uF bipolar electrolytics. These will give response to well below 10Hz if you think you’ll ever need to go that low.

The project PCB measures 77 x 24mm, and the mounting centers for the pot and XLR connector are spaced at 57mm. If preferred, a traditional chassis mounted female XLR can be used, and wired to the board with heavy tinned copper wire. The PCB pads for the connector are in the correct order for a female chassis mount socket mounted with the “Push” tab at the top.

This is yet another project born of necessity. It’s a simple circuit, but does exactly what it’s designed to do – dim LED lights or control the speed of 12V DC motors. The circuit uses PWM to regulate the effective or average current through the LED array, 12V incandescent lamp (such as a car headlight bulb) or DC motor. The only difference between the two modes of operation is the addition of a power diode for motor speed control, although a small diode should be used for dimmers too, in case long leads are used which will create an inductive back EMF when the MOSFET switches off.

The photo shows what a completed board looks like. Dimensions are 53 x 37mm, so it’s possible to install it into quite small spaces. The parts used are readily available, and many subsitiutions are available for both the MOSFET and power diode (the latter is only needed for motor speed control). The opamps should not be substituted, because the ones used were chosen for low power and their ability to swing the output to the negative supply rail.

Note that if used as a motor speed controller, there is no feedback, so motor speed will change with load. For many applications where DC motors are used, constant speed regardless of load is not needed or desirable, but it is up to you to decide if this will suit your needs.

Description

First, a description of PWM is warranted. As the pot is rotated clockwise, the input voltage changes linearly with rotation. At first, the voltage is such that the comparator output is just narrow spikes, which turn the MOSFET on for a very short period. Average current is low, so connected LEDs will be quite dim, or a motor will run (relatively) slowly. As the input voltage coming from the pot increases, the MOSFET is on for longer and longer, so increasing power to the load.

Figure 1 shows how the PWM principle works. The red trace is the triangle wave reference voltage, and the green trace is the voltage from the pot. When the input voltage is greater than the reference voltage, the MOSFET turns on, and current flows in the load. Because the frequency is relatively high (about 600Hz), we don’t see any flicker from the LEDs, but the tone is audible from a motor that’s PWM controlled. The PWM signal is shown in blue. The average current through the load is determined by the ratio of on-time to off-time, and when both are equal, the average current is exactly half of that which would be drawn with DC.

The circuit is shown in Figure 2. U1 is the oscillator, and generates a triangular waveform. R4 and R5 simply set a half voltage reference, so the opamps can function around a 6V centre voltage. U2A is an amplifier, and its output is a 10V peak to peak triangle wave that is used by the comparator based on U2B. This circuit compares the voltage from the pot with the triangle wave. If the input voltage is at zero, the comparator’s output remains low, and the MOSFET is off. This is the zero setting.

In reality, the reference triangle waveform is from a minimum of about 1.5V to a maximum of 9.5V, so there is a small section at each end of the pot’s rotation where nothing happens. This is normal and practical, since we want a well defined off and maximum setting. Because of this range, for lighting applications, an industry standard 0-10V DC control signal can be used to set the light level. C-BUS (as well as many other home automation systems) can provide 0-10V modules that can control the dimmer.

While a 1N4004 diode is shown for D2, this is only suitable if the unit is used as a dimmer. For motor speed control, a high-current fast recovery diode is needed, such as a HFA15TB60PBF ultra-fast HEXFRED diode. There are many possibilities for the diode, so you can use whatever is readily available that has suitable ratings. The diode should be rated for at least half the full load current of the motor, and the HFA15TB60PBF suggested is good for 15A continuous, so is fine with motors drawing up to 30A.

Construction

While it’s certainly possible to build the dimmer on veroboard or similar, it’s rather fiddly to make and mistakes are easily made. Also, be aware that because of the current the circuit can handle, you will need to use thick wires to reinforce some of the thin tracks. This is even necessary for the PCB version. Naturally, I recommend the PCB, and this is available from ESP. The board is small – 53 x 37mm, and it carries everything, including the screw terminals. The PCB is double-sided with plated-through holes, and has solder masks on both sides.

The MOSFET will need a heatsink unless you are using the dimmer for light loads only. It is necessary to insulate the MOSFET from the heatsink in most cases, since the case of the transistor is the drain (PWM output). For use at high current and possible high temperatures, the heatsink may need to be larger than expected. Although the MOSFET should normally only dissipate about 2W or so at 10A, it will dissipate a lot more if it’s allowed to get hot. Switching MOSFETs will cheerfully go into thermal runaway and self destruct if they have inadequate heatsinking. You may also use an IGBT (insulated gate bipolar transistor) – most should have the same pinouts, and they do not suffer from the same thermal runaway problem as MOSFETs.

As noted above, there are many different MOSFETs (or IGBTs) and fast diodes that are usable. The IRF540 MOSFET is a good choice, and being rated 27A it has a generous safety margin. There are many others that are equally suitable – in fact any switching MOSFET rated at 10A or more, and with a maximum voltage of more than 20V is quite ok.

Testing

Connect to a suitable 12V power supply. When powering up for the first time, use a 100 ohm “safety” resisor in series with the positive supply to limit the current if you have made a mistake in the wiring. The total current drain is about 2.5mA with the pot fully off, rising to 12.5mA when fully on. Most of this current is in the LED, which is also fed from the PWM supply so you can see that everything is working without having to connect a load.

Make sure that the pot is fully anti-clockwise (minimum), and apply power. You should measure no more than 0.25V across the safety resistor, rising to 1.25V with the pot at maximum. If satisfactory, remove the safety resistor and install a load. High intensity LED strip lights can draw up to ~1.5A each, and this dimmer should be able to drive up to 10 of them, depending on the capabilities of the power supply and the size of the heatsink for the MOSFET.

As readers will know, there are already several power amplifier projects, two using IC power amps (aka power opamps). Both have been popular, and this project is not designed to replace either of them. However, it is significantly smaller than the others, so it makes building a multiple amp unit somewhat easier because the space demand is much lower. It’s quite simple to include 4 amps (two boards) into a small space, but be aware that good heatsinking is essential if you expect to run these amps at significant power levels.

The TDA7293 IC uses a MOSFET power stage, where the others featured use bipolar transistors. The main benefit of the MOSFET stage is that it doesn’t need such radical protection circuitry as a bipolar stage, so unpleasant protection circuit artefacts are eliminated. There are no apparent downsides to the TDA7293, although it was found that one batch required a much higher voltage on the Standby and Mute pins than specified, or the amps would not work. This is not a limitation, since both are tied to the positive supply rail and are therefore disabled.

This particular project has been planned for a long time, but for some reason I never got around to completing the board or the project description. This is now rectified, and it’s ready to “rock and roll”. The board is very small – only 77 x 31mm, so getting it into tight spaces is easy … provided adequate heatsinking is available of course.

Description

The TDA7293 has a bewildering number of options, even allowing you to add a second power stage (in another IC) in parallel with the main one. This improves power into low impedance loads, but is a rather expensive way to get a relatively small power increase. It also features muting and standby functions, although I’ve elected not to use these.

The schematic is shown in Figure 1, and is based on the PCB version. All unnecessary functions have been disabled, so it functions as a perfectly normal power amplifier. While the board is designed to take two TDA7293 ICs, it can naturally be operated with only one, and the PCB is small enough so that this is not an inconvenience. A LED is included to indicate that power is available, and because of the low current this will typically be a high brightness type.

The IC has been shown in the same format that’s shown in the data sheet, but has been cleaned up for publication here. Since there are two amps on the board, there are two of most of the things shown, other than the power supply bypass caps and LED “Power Good” indicator. These ICs are extremely reliable (as are most power amp ICs), and to reduce the PCB size as much as possible, fuse clips and fuses have not been included. Instead, there are fusible tracks on the board that will fail if there is a catastrophic fault. While this is not an extremely reliable fuse, the purpose is to prevent power transformer failure, not to protect the amplifiers or PCB.

I normally use a gain of 23 (27dB) for all amplifiers, and the TDA7293 is specified for a minimum gain of 26dB, below which it may oscillate. Although this is only a small margin, tests so far indicate that the amp is completely stable. If you wish, you may increase the gain to 28 (29dB) to give a bit more safety margin. To do this, just change the input and feedback resistors (R3A/B and R4A/B) from 22k to 27k.

The circuit is conventional, and is very simple because all additional internal functions are unused. The LED is optional, and if you don’t think you’ll need it, it may be omitted, along with series resistor R3. All connections can be made with plugs and sockets, or hard wired. In most cases, I expect that hard wiring will be the most common, as the connectors are a pain to wire, and add unnecessary cost as well as reduce reliability.

The TDA7293 specifications might lead you to believe that it can use supply voltages of up to ±50V. With zero input signal (and therefore no output) it might, but I don’t recommend anything greater than ±35V if 4 ohm loads are expected, although ±42V will be fine if you can provide good heatsinking. In general, the lower supply voltage is more than acceptable for 99% of all applications, and higher voltages should not be used unless there is no choice. Naturally, if you can afford to lose a few ICs to experiments, then go for the 42V supplies (obtained from a 30+30V transformer).

Construction

Because of the pin spacings, these ICs are extremely awkward to use without a PCB. Consequently, I recommend that you use the ESP board because it makes building the amplifier very simple. The PCBs are double sided with plated-through holes, so are very unforgiving of mistakes unless you have a good solder sucker. The best way to remove parts from a double sided board is to cut the pins off the component, then remove each pin fragment individually. This is obviously not something you’d wish to do if a power amp IC were installed incorrectly, since it will be unusable afterwards.

The diagram above shows the pinouts for the TDA7293V (the “V” means vertical mounting). Soldering the ICs must be left until last. Mount the ICs on your heatsink temporarily, and slide the PCB over the pins. Make sure that all pins go through their holes, and that there is no strain on the ICs that may try to left the edge off the heatsink. When ICs and PCB are straight and aligned, carefully solder at least 4 pins on each IC to hold them in place. The remaining pins can then be soldered. Remember, if you mess up the alignment at this point in construction, it can be extremely difficult to fix, so take your time to ensure there are no mistakes.

This amplifier must not be connected to a preamp that does not have an output coupling capacitor. Even though there is a cap in the feedback circuit, it can still pass DC because there is no input cap on the PCB. I normally include an input cap, but the goal of this board was to allow it to fit into the smallest space possible, and the available board space is not enough to include another capacitor. A volume control (typically 10k log/ audio taper) may be connected in the input circuit if desired.

Note that the metal tab of the TDA7293 is connected to the -Ve supply, so must be insulated from the heatsink. The more care you take with the mounting arrangement, the better. While you can use a screw through an insulating bush and a piece of mica to insulate the tab, a better alternative is to use a clamping bar of some kind. How you go about this depends a lot on your home workshop tools and abilities, but one arrangement I’ve found highly satisfactory is a suitable length of 6.25mm square solid steel bar. This is very strong, and allows good pressure on the mica (or Kapton) for maximum heat transfer. Naturally, heatsink compound is absolutely essential.

Do not be tempted to use silicone insulation washers unless you are using the amp at very low supply voltages (no more than ±25V). Its thermal transfer characteristics are not good enough to allow the amp to produce more than about 10 – 20W of music, and even that can be taxing for silicone washers. The amp will shut down if it overheats, but that curtails one’s listening enjoyment until it cools down again.

Power Supply

A suitable power supply is shown below, and is completely unremarkable in all respects. The transformer may be a conventional (E-I) laminated type or a toroid. The latter has the advantage of lower leakage flux, so will tend to inject less noise into the chassis and wiring. Conventional transformers are usually perfectly alright though, provided you take care with the mounting location.

The bridge rectifier should be a 35A 400V type, as they are cheap, readily available and extremely rugged. Electrolytic capacitors should be rated at 50V. The cap connected across the transformer secondary (C4) should be rated at 275V AC (X Class), although a 630V DC cap will also work. This capacitor reduces “conducted emissions”, namely the switching transients created by the diodes that are coupled through the transformer onto the mains supply. The power supply will work without this cap, and will most likely pass CE and C-Tick tests as well, but for the small added cost you have a bit of extra peace of mind as regards mains noise.

The supply shown includes a “loop breaker”, which is intended to prevent earth/ ground loops to prevent hum when systems are interconnected. Please be aware that it may not be legal to install this circuit in some countries. The diodes must be high current types – preferably rated at no less than 3A (1N5401 or similar). The loop breaker works by allowing you to have the chassis earthed as required in most countries, but lets the internal electronics “float”, isolated from the mains earth by the 10 ohm resistor. RF noise is bypassed by the 100nF cap, and if a primary to secondary fault develops in the transformer, the fault current will be bypassed to earth via the diodes. If the fault persists and the internal fuse (or main power circuit breaker) hasn’t opened, one or both diodes will fail. Semiconductor devices fail short-circuit, so fault current is connected directly to safety earth.

Be very careful when first applying mains power to the supply. Check all wiring thoroughly, verify that all mains connections are protected from accidental contact. If available, use a Variac, otherwise use a standard 100W incandescent lamp in series with the mains. This will limit the current to a safe value if there is a major fault.

When the loop breaker is used, all input and output connectors must be insulated from the chassis, or the loop breaker is bypassed and will do nothing useful. The body of a level pot (if used) can be connected to chassis, because the pot internals are insulated from the body, mounting thread and shaft.

Note that the DC ground for the amplifiers must come from the physical centre tap between the two filter caps. This should be a very solid connection (heavy gauge wire or a copper plate), with the transformer centre tap connected to one side, and the amplifier earth connections from the other. DC must be taken from the capacitors – never from the bridge rectifier.

The order of the fuse and power switch is arbitrary – they can be in any order, and in many cases the order is determined by the physical wiring of the IEC connector if a fused type is used. With a fused IEC connector, the fuse is before the switch and it cannot be removed while the mains lead is inserted.

I have shown a 2A slow-blow fuse, but this depends on the size and type of transformer and your mains supply voltage. Some manufacturers give a recommended fuse rating, others don’t. The fuse shown is suitable for a 150VA transformer at 230V AC, and is deliberately oversized to ensure that it will not be subject to nuisance blowing due to transformer inrush current. A 2A fuse will fail almost instantly if there is a major fault.

Make sure that the mains earth (ground) is securely connected to guarantee a low resistance connection that cannot loosen or come free under any circumstances. The accepted method varies from one country to the next, and the earth connection must be made to the standards that apply in your country.

WARNING: This power supply circuit requires experience with mains wiring. Do not attempt construction unless experienced, capable and suitably qualified if this is a requirement where you live. Death or serious injury may result from incorrect wiring.

Testing

Never attempt to operate the amplifier without the TDA7293 ICs attached to a heatsink!

Connect to a suitable power supply – remember that the supply earth (ground) must be connected! When powering up for the first time, use 100 ohm 5W “safety” resistors in series with each supply to limit the current if you have made a mistake in the wiring. If available, use a variable bench supply – you don’t need much current to test operation, and around 500mA is more than enough. If using a current limited bench supply, the safety resistors can be omitted. Do not connect a speaker to the amplifier at this stage!

If using a normal power supply for the amp tests, apply power (±35V via the safety resistors) and verify that the current is no more than 60mA or so – about 6V across each 100 ohm resistor. No load current can vary, so don’t panic if you measure a little more or less. Verify that the DC voltage at both outputs is less than 100mV. Using another 100 ohm resistor in series with a small speaker, or an oscilloscope, apply a sinewave signal at about 400Hz to the input and watch (or listen) for signal. The signal level needs to be adjusted to ensure the amp isn’t clipping, and the waveform should be clean, with no evidence of parasitic oscillation or audible distortion.

If everything tests out as described, wire the amplifier directly to the power supply and finish off any internal wiring in the amp. Once complete, it’s ready to use.