Control

Power switch - click to see more! There is a surprising amount of control electronics in this amplifier. Firstly, the unit has a tactile front panel switch which also contains the blue power LED. There's also an D-type connector mounted on the rear panel that accepts the 12V trigger signals from the preamp.

The power is carefully managed to avoid switch-on surges that can cause fuses to blow. The LM4780 mute inputs have to be controlled to prevent pops and bangs from the speakers. The temperature of both heatsinks is monitored and the amplifier is muted if the temperature rises excessively - should the temperature fail to fall to a normal level within a suitable time, the amplifier is shut down. Finally, should a DC fault be detected in any of the channels, the amplifier is switched off.

Most audio power amplifiers use output relays which, in conjunction with a suitable delay circuit, prevent switch-on pops and bangs. Additionally, in the event of a DC fault these relays can be used to isolate the speakers from the faulty amplifier. However, properly implemented National Semiconductor "Overture" IC's are free from switch on/off noises so these relays are unnecessary. So in the event of a DC fault, the power is switched off and the energy contained in the smoothing capacitors could end up being dumped into the connected loudspeakers. But this is better than no protection at all, and as the smoothing capacitors are relatively small, the potential for damage is much reduced.

While initially planning to implement these functions using discrete circuits, I decided that employing a software-based solution would be much more dependable. The various time delays could be readily altered if necessary, and would not be reliant on electrolytic capacitors that are prone to aging. New functionality could be added, and the component count would be reduced.

The mains transformer is switched off during standby, so some means of powering the control circuitry is required. There wasn't space in the enclosure for a small standby transformer, so I employed a capacitive dropper - a well-known technique that's used in lots of mains-powered applications. However, adopting this strategy requires a knowledge of safe working with live mains, and I do not recommend you try this unless you understand this risks. The earth of the PIC circuitry is at half-mains, and connecting earthed test equipment could cause serious risks!

Important Safety Warning/Disclaimer

This control circuit is powered directly from the mains electricity, and in the interests of safety, you should not attempt to copy this unless you are fully aware of the risks involved, and the necessary precautions required during PCB layout, component selection and construction.

The author disclaims all responsibility for any loss or damage, and reading further implies agreement with this disclaimer. You have been warned!

If you want to copy this control system, then please consider using a small mains transformer to power it instead of the capacitive dropper scheme detailed here.

Power control

The control circuitry is split between two circuit boards - the first one manages the supply of power to the mains transformer and also generates a regulated 5V rail to power the PIC. Additionally, this board generates the active-low /MAINS-FAIL signal.

The torroidal transformer requires an inrush-limiting function to prevent fuses blowing and lights dimming at power-up. This is usually catered for by connecting a power resistor in series with the transformer for the initial power-on, then shorting out that resistor with a relay after a short delay. This is most commonly done using a circuit like this:

Series soft-start scheme (7K)

As you can see, this makes for an incredibly simple control circuit, that can readily be "retro-fitted" to an existing amplifier if required. The transistor turns the relay on after a short delay, bypassing the soft-start resistor. The power for the relay coil comes from a convenient part of the power supply.

There are a few issues with this control scheme. If for some reason the relay fails to turn on, the equipment remains powered via the soft-start resistor - for something like an audio amplifier, this might go unnoticed if it's used at low levels. Meanwhile the soft-start resistor will rapidly overheat, causing damage to itself or the surrounding circuitry. This failure mode is more common than you might think - the relay contacts can go bad over time, or the control circuit, although simple, can go wrong. Also, the simple (but often used) control circuit shown above doesn't reset as quickly as is desirable, even with the extra diode to help discharge the capacitor. So, very brief power outages can cause fuses to fail.

The soft-start bypass relay is powered all the time the unit is on, which is wasteful. And the power switch and the soft-start bypass relay end up being connected in series, meaning that the mains has to pass through two sets of contacts - not good for reliability. So I decided quite early on to use a "parallel" scheme:

Parallel soft-start scheme (6K)

Controlling this layout is slightly more complicated, and unlike the series scheme above, it doesn't lend itself to being added to an existing amplifier. At switch on, the power relay is energised, supplying energy via the soft-start resistor, and after a short delay the bypass relay is switched. As above, the power for the bypass relay comes from the secondary side of the mains transformer.

However, the power relay can be switched off once the bypass relay has been energised, saving some energy. Doing this adds an important failsafe - if there is a fault in the power relay or soft-start resistor, or in the power supply of the amplifier, the bypass relay doesn't energise. If there is a fault in the bypass relay or control, the amplifier will simply be switched off when the soft-start relay is de-powered.

As the power relay only passes current for a brief period of time, you can consider alternative technology here - after all, relays are large, expensive and mechanically noisy. Two conventional relays operating together in this scheme could make quite a racket at power-up! Triac-based solid-state relays can cause problems with audio equipment because of their behaviour around the zero-crossing point, but as the relay will only be passing current for half a second or less, this isn't an issue here.

I could have purchased an off-the-shelf SSR (solid state relay), but I couldn't easily find a device that fitted into the 16mm of height that I had under my "false floor" (see later). Plus, I wouldn't have learned anything that way. Instead, I found a triac and a special opto-coupler that is designed to drive the gate of the triac - it also incorporates a zero-crossing detection circuit to ensure clean transitions. The LED in the opto-coupler requires just a few milliamps of current - easily provided by a logic circuit or a PIC...

Soft start relay

I decided to take advantage of the solid-state relay when switching off the amplifier. Of course, simply switching off the bypass relay is all that is required at shut-down, but I decided to on a cunning scheme to save wear and tear on the relay contacts. When the software receives the shut-down command, either from the front panel switch or via the 12V trigger input, the amplifier is immediately muted and the front panel LEDs are updated to indicate Standby. Then the solid-state relay is powered up. After a short delay, the bypass relay is depowered, but the amplifier remains powered via the soft-start resistor and relay. Finally, after another short delay, the solid state relay turns off. Because the opto-coupler incorporates zero-crossing detection, the transition is clean and doesn't generate any interference that could affect other pieces of equipment.

This circuit was built on Veroboard, taking great care to ensure that correct creepage distances are maintained, and using tinned copper wire to supplement the copper tracks to ensure that the potentially high currents don't damage anything. On the same board, I incorporated the standby power supply for the PIC control circuitry:

Standby PSU

The mains is reduced using a capacitive dropper. The 470nF capacitor is has an X2 rating, which means it is designed for continual use across the mains, and is designed to fail safely open-circuit in the event of a high-voltage surge. Never use standard components in these sorts of circuits!

The 100 ohm resistor is to absorb any switch-on surge (when the dropper capacitor is uncharged), and will fail open-circuit in the event of an overload. The metal-glazed 1.5M resistor is specially rated for continuous high-voltage operation, and its role in the circuit is to discharge the dropper capacitor when the mains is removed.

A full-wave rectifier produces DC, limited by the 10V Zener diode. From here, it is smoothed and regulated to 5V, ready for the PIC to use. The current demand is small - the only load is the PIC itself, the two front panel LEDs and the LEDs in the opto-couplers. For this application, a capacitive dropper is ideal, but a small mains transformer would be much safer - unfortunately I had no room in the case for one. If you are considering building a similar design, I would urge you to use a conventional transformer unless you know what you are doing!

There is also a mains failure detection circuit here. This is essential to ensure the soft-start routines are properly reset, and to avoid microprocessor lock-ups. It's a simple but very effective circuit that would work just as well with a mains transformer instead of the capacitive dropper.

Power control board (38K)

By inserting the 1n4001 diode between the diode bridge and the smoothing capacitor, we can derive the 100Hz post-rectifier waveform. This turns the first transistor on and off again 100 times a second. Every time this transistor turns on, the 470n capacitor is discharged. As the mains disappears, the capacitor quickly charges via the 68K resistor, turning on the second transistor and resetting the PIC. This happens within about 30ms, so it only needs a couple of missing cycles to respond.

Finally, here is a picture of the finished board. You can see how the limited height required careful planning. The mains connections are at the top, with connections to the transformer, the mains input, the soft-start relay and the bypass relay. The bottom wires go to the PIC control board, and carry 5 volts, the mains-fail signal and the control signal for the solid-state relay.

PIC Control

As mentioned, I initially planned to use a discrete control system, combining the temperature sensor and mute control of the Micro-Amp, and the power control of the Gainclone Monoblocks, with some minor modifications to provide a soft-start function. But then I started playing around with some ideas with a PIC, and one thing led to another...

While using a microcontroller for such a simple function seemed like overkill, space is at a premium inside the amplifier and any means of reducing component-count is welcome. And it enabled me to experiment with some interrupt-driven techniques which will form the basis of an introduction to interrupts that I'm writing for the PIC section of this site.

Without thinking too hard about it, I started experimenting with the ubiquitous PIC16F84, but this turned out to be an ideal choice here. I used most of the ports, and over half of the programme memory, perhaps surprisingly. The PIC runs at 10 MHz - this is higher than in my preamp, but chosen because the PIC trainer boards that we use at work are fitted with a 10MHz crystal. However, I was surprised when working out the interrupt-driven LED dimming routines that I could actually have made use of a faster processor!

As noted above, the PIC is powered from the mains via a capacitive dropper, which leaves the logic ground is at "half-mains" (approximately minus 150-160V w.r.t. neutral). Obviously the control circuits cannot be directly connected to the amplifier, so to allow communication with the rest of the unit, a total of 6 optocouplers are used (including the special optocoupler that forms the solid-state relay in conjunction with the triac).

The basic PIC circuit is very simple:

The power switch and Standby LED are mounted on the front panel, and the SOFT-START and /MAINS-FAIL signals travel to the power control board. Everything else is mounted on the control board:

Control board (41K)

The "live" circuitry is on the left of the board, and as much copper as possible was removed from underneath the opto-couplers, thus forming a safety isolation barrier. The connectors are as follows:

  • Front panel switch and Standby LED (bottom left, 6-way)
  • From the power control board (top left, 4-way)
  • To power-on relay (top right, 2-way)
  • To amplifier (middle right, 6-way)
  • From 12V trigger input (bottom right, 2-way)

The circuitry on the right of the board is explained next:

Bypass relay control

As previously discussed, the bypass relay connects the mains transformer directly to the mains, bypassing the soft-start triac and power resistor. The PIC energises this relay a short while after energising the soft-start relay. The power for the relay coil actually comes from the mains transformer, meaning that if there is a fault in the soft-start circuit or power supply of the amplifier, it can't switch on. This is a useful "failsafe" mode, but might cause some headscratching when fault-finding at some point in the future!

Bypass relay control (8K)

The relay is powered from the negative supply rail from the amplifier - all of the control circuitry has been located between ground and -UNREG because of the Mute requirements of the LM4780 ICs. Because the supply is unregulated, the relay is fed from a 2-transistor current source - this ensures the relay is able to pull in while the supply is low due to the action of the soft-start resistor, but when the supplies are established, it ensures that coil consumption doesn't rise - good for long-term reliability.

With no light from the opto-coupler LED, the transistor is off and no base current flows into the BC337. When the PIC lights the LED, the current source can work. The two 33K resistors are arranged so that the collector-emitter voltage of the opto-coupler transistor doesn't exceed the maximum allowed value (around 25 volts for the 4n27)

Mute control

This looks simple, but was a bit more complicated to get right. To work, a current must be drawn from the mute pins of the LM4780. This current is initially withheld by the PIC so that the switch-on is noise-free (the OPA2134 buffer causes a slight noise at power-on). Temperature sensors on each heatsink mute the amplifier in the event of overheating, and the PIC must be aware this has happened. For the "single" mode, just one of the LM4780's on the heatsink is muted, and this must not interfere with the above.

This diagram shows the circuitry, which is split across the two power amplifier modules and the control board. The three sections are joined by the "MUTE" connections (yellow wires).

The 10K resistors provide the necessary current to each LM4780 to bring it out of the mute condition. The amplifier can be muted by shorting the mute line to ground, and this is exactly what each of the thermal switches does at the appropriate temperature. The transistor, when biased on by the 100K resistor, also shorts the mute line to ground. When the PIC lights the "PLAY" LED, the transistor is turned off, and the amplifier unmutes.

The PIC knows when the amplifier is muted because the LED in the "MUTED" opto-coupler goes out. This enables the software to flash the power LED, and eventually shut down the amplifier if it doesn't cool down sufficiently quickly. Of course, when the amplifier is muted by the software (rather than the thermal switches), the PIC can't tell for sure whether the amplifier is overheated or not - this is accounted for in the software.

When operated in Single mode, the unused LM4780 is muted. Steering diodes ensure that this does not mute the remaining amplifiers.

DC Detection

As mentioned previously, the implementation of the DC failure circuitry isn't as good as it could be because the amplifier doesn't employ muting relays. But it's better than no protection, and was relatively easy to add.

The chosen scheme was designed to be very easy to implement, requiring a minimal amount of extra wires to be run around the amplifier. Like the "Mute" bus mentioned above, the "DC" connection runs across the amplifier - identified with the use of blue cables. The summing point is on the PIC control board. The 39K resistor pulls the summing point towards ground in the absence of any connected amplifier modules - this enables the PIC control board to be operated on its own without creating spurious DC fault signals - useful during testing.

The summed outputs are low-pass filtered by the 100uF capacitor - the net potential here should always be very close to 0V. Should this voltage rise to greater than 0.6V, the first BC557 is biased on, and the current flowing through the 150K causes the BC547 to conduct, lighting the LED in the opto-coupler. Similarly, should the summing point fall to -0.6V, the second BC557 conducts, again lighting the LED via the BC547. Note the high value resistors here - to maintain trigger symmetry with this simple circuit, it's important to minimise the current draw from the collectors.

DC Fault Detection (8K)

There are some compromises with this system. Firstly, the DC trigger voltage is slightly higher than is ideal - around 3V. This is because the 47K resistor for the faulty channel effectively forms a potential divider with the remaining three 47K resistors, and the 39K resistor on the PIC board. However, when DC faults occur, they are normally much greater than 3V - and to be honest 3V probably wouldn't harm most speakers.

Additionally, there is a chance that this system will fail to detect a DC fault if two faults occur at the same instant, with opposite polarities. A DC voltage of +35V on one channel coupled with -35V on another will sum to zero. But, what are the chances of this?

Of course, the other difficulty with these schemes is choosing the best value for the capacitor. There is always a conflict between reaction time, and freedom from false triggering. The chosen value offers a reasonable compromise here.

When the LED lights, the PIC immediately shuts down the amplifier and flashes an error code on the Standby LED. A power-off reset is required before it will work again. When shutting down, the PIC doesn't worry about following the power-down sequence described above - it just turns off the main relay ASAP!

12V Trigger Input

The final part of the control circuit is the easiest - the 12V trigger input.

12V Trigger Input

The rear-mounted toggle switch chooses between two of the power amplifier triggers that the preamp generates - main or surround. The yellow LED is next to the switch, and provides a useful visual confirmation of the presence of the trigger signal. The trigger input will probably be already earthed, but the 100K resistor ties it weakly to earth in case it's floating. The 1n4148 diode protects the LED in the optocoupler in the event of an AC or reverse polarity trigger signal.

Switch Details

Bulgin
     switchSince posting this project, I've had a lot of emails asking about the power switch used in this amplifier.

It's a vandal-proof stainless steel unit made by Bulgin (part number MPI002/28/BL). The action is single-pole momentary, so at the risk of stating the obvious, you will need some logic circuitry to use this as a power switch - it can't be used on its own. (Yes, I have been asked this question. On more than one occasion!)

You can choose the LED colour, the connection style (2.8mm "Faston" tags as shown, or screw terminals), and illumination style (ring or centre-dot). A word of warning: if you plan to solder to the tags, be aware that the plastic is very soft and will deform during soldering. I was lucky and "got away with it", but the switch doesn't feel quite as nice as it did when new. Try not to use it in a situation where the space behind is limited!

I bought mine from Farnell but, as usual, Rapid are slightly cheaper. The prices and part numbers were last checked Autumn 2009. Note also that the range has been expanded, and it's possible to get white, as well as dual colour LEDs in nice combinations - for example, blue and red is an option, and is what I might choose if I was building a similar project today.

Bulgin switch suppliers
Supplier Part number Price each
Farnell 430-3076 £14.85+VAT
Rapid 78-1697 £13.26+VAT
Rapid 78-1745 (with screw terminals) £11.99+VAT

More information can be found on the Bulgin website.