|Mark's Project Pages/Audio Projects/GainClone/Control Section|
The control section of these amplifiers needed careful consideration if they were to successfully integrate with my remote-controlled preamp. They need to respond to the 12V trigger signals generated by the preamp, but I also wanted a front panel standby switch for the occasions when these amplifiers aren't being used with the preamp.
Important Safety Warning/Disclaimer:
Sorry to start this page with such a notice, but this is a serious business. The basic problem is that I didn't have space in the box for a small 50Hz transformer to power the control electronics, but in the interests of the environment and my electricity bill, I wanted to ensure the amplifiers had a minimal power consumption during standby. If you want to copy this control system, then please consider using a small mains transformer to power it.
So, what are the requirements of the control system? Fortunately, they're very simple. Because the power amps turn on and shut down cleanly with no pops and clicks, there is no need for an output muting relay. Because the LM3875 is comprehensibly protected internally, there's no need for temperature sensing or similar. Basically, we just need to control the power to the main transformer.
The "user interface" is very simple - just a single front-panel push-switch containing a red LED which lights when in standby, and a blue LED that lights when the amplifier is on. The push-switch needs to have a toggle action, switching the unit on and off as necessary.
In addition, there is a 9-way D-type connector on the rear panel which connects to the preamp. The 12V trigger signals from the preamp are present on this connector, and a small rear-panel switch is used to select between "main" and "surround" trigger signals. Each of these signals goes high when the appropriate power amplifier is required, and low when the amplifier can be switched off.
The two inputs, the switch and the trigger inputs need to be independent of each other - the 12V trigger should not stop the switch from functioning. This sounds like an ideal application for a microcontroller such as a PIC16F84, but too be honest, this is overkill. I've managed to implement all of these functions using just two HC TTL IC's.
I'll build up this slowly, and explain it in detail in the hope that this might help people who are interested in designing similar systems. Despite the ease and economy of microcontrollers like PICs, this sort of logic design is still useful for simple systems like this, and the principles described here were used in the NICAM tuner and power supply projects.
Starting with the front panel switch, this shows how to get a toggle action from a momentary push-switch using a D-type latch. Normally, the inverter has logic "1" on its input, thanks to the 47K resistor. This means that the clock input is low. When the switch is pressed, a logic "0" is applied to the input of the inverter, hence the clock rises to logic "1". This clocks the D-type register, and as the data input is derived from the /Q output, the output toggles. If the output was "0" initially, this means that /Q was at logic "1". Note how the red LED would be lit during this state. But on receipt of the clock pulse, the logic "1" seen at the Data input would be transferred to the Q output. This standard divide-by-two circuit is seen in lots of applications.
To make this a reliable and practical circuit, the switch contacts need to be "de-bounced". The addition of the 10uF capacitor achieves this. When the switch is pressed, the capacitor is rapidly discharged via the 1K resistor. This resistor protects the switch contacts by limiting the current that flows through the contacts to a peak of around 5mA. Once the switch has been released, the capacitor charges via the 47K resistor, which takes around half a second in this case. Any contact-bounce that occurs during this time is effectively filtered out by this long time constant. Key to the operation of this is the Schmitt trigger input stage on the inverter - do not use a standard inverter here!
They're not shown on the diagram, but the D-type latch has Set and Clear inputs. These can override the operation described above - not surprisingly, asserting the Set input will make the Q output go high, and it will stay high until the next event. Asserting Clear will make Q go low. So, if these inputs can somehow be connected to the 12V trigger input signal, this will give us the required operation...
And here's how it's done. The 12V trigger signal is passed via an opto-coupler. This is an essential safety requirement in this case because the "earth" is actually live, but even if this wasn't the case, it's good practice because it removes the risk of earth loops forming.
With a 12V trigger voltage present, the LED in the opto-coupler lights, turning on the internal transistor. This means that there is a logic "0" at the input to the first inverter, and logic "1" is present on the output. The second inverter inverts this. The outputs of the two inverters could be applied directly to the D-type latch shown above, but this would stop the push-button switch working because the /SET and /CLEAR inputs override the D and CLOCK inputs. Hence the RC network...
The 47K resistors keep the /CLEAR and /SET inputs high, which means the front panel switch can work normally. Remember, these are active-low inputs, so they need to be driven low to clear or set the latch. So when the 12V trigger input goes high, the transistor turns on, meaning that the output of the first inverter goes high. At the same time, the output of the second inverter goes low. The 10n capacitor connected to this inverter is uncharged, so as the second inverter output goes low, the /SET signal also goes low. This sends the "set" command to the latch, and the power amp switches on. After a short period of time, the 10n capacitor becomes fully charged, and the /SET input becomes high again. This has no effect on the state of the latch, but it does mean the front panel switch is able to work again.
When the 12V trigger signal disappears, the process happens to the /CLEAR input. So, what about the /SET input in this instance? The 10n capacitor had charged to 5V before, so when the inverter output goes high, the voltage at the /SET input tries to rise to +10V. Clearly the latch won't like this, so the diode conducts to trap this.
You'll notice the 47uF capacitor. This performs a similar "debounce" function as before. Without it, the amplifier would switch on and off as rapidly as the input signal dictates. So a 10Hz square wave would cause the relay to buzz away obediently, destroying the contacts and possibly the amplifier in the process. With the addition of the capacitor, it will just turn on. It will power up instantly as the transistor rapidly discharges the capacitor via the 1K resistor, but there's a time constant of around 2 seconds keeping it switched on, defining the minimum amount of time the amplifier will be power up for. I thought that this was worth adding as my 12V trigger signals are generated by the PIC in my preamp, and I'd hate a software problem to be responsible for destroying my power amplifiers!
This is how the Q output of the D-type latch actually switches the amplifier on and off. On the face of it, a simple relay drive circuit, using a BBC337 to turn on the relay coil. But note that the relay coil is supplied from an unregulated supply rail. With that in mind, I added the BC548 and the 68 ohm resistor to turn the drive circuit into a two-transistor current source.
The remaining components ensure the circuit behaves properly when initially powered up. Reset circuitry is essential on any circuit containing logic! The 47uF capacitor is initially discharged when the circuit is first connected to the mains supply. This means that a logic "1" is applied to the input of the inverter, and a logic "0" is present on the output. This inverter is connected directly to the /CLEAR input of the D-type inverter, meaning that the D-type will be reset. The diode is there to ensure that this does not affect the operation of the 12V trigger circuitry described above.
Note also the connection to the base of the BC548. During the reset period, which lasts for around 1.5 seconds, the BC548 is conducting, meaning that the relay coil can not be energised, irrespective of the state of the D-type latch. The diode ensures that the reset circuit does not affect the current-source operation, although to be honest, this could probably be omitted.
The final diode, in parallel with the 100K resistor, is essential to protect the inverter (as is the 1K resistor). When the circuit is powered off, the 47uF capacitor has 5V across its terminals. As the supply rail collapses to 0V, the negative terminal of the capacitor will go negative by 5V. When this happens, the diode conducts, ensuring that the capacitor discharges itself into the supply rails, thus protecting the inverter and as a bonus ensuring that the reset circuit quickly gets properly reset.
Finally, the power supply. As mentioned above, this circuit has to run directly from the mains supply as there isn't space in the enclosure for a small standby transformer. But should you want to build a similar version of this, then don't use this scheme - use a case that is large enough to fit a separate small (say, 3VA) transformer.
This supply might not look like much, but it has been thoroughly tested using a variac and safety isolation transformer. All the components are generously over-rated, and should be reliable in service. This circuit works here because of the tiny current consumption of the circuit - this objective was kept in mind while designing the logic and also selecting the mains relay.
The mains is half-wave rectified by the high-voltage diode. This halves the average voltage seen by the circuit. Next the 3K9 resistors drop this down - these resistors actually dissipate less than a watt, but in the worst-case scenario of the input diode or thhe two zener diodes short-circuiting, they will still be comfortably below their power ratings. Sharing the load between two resistors is good for long-term reliability, as they will be subject to less voltage stress, especially in the event of surges.
The smoothing capacitor is also massively over-rated. The two zener diodes will limit the highest voltage to around 44V, so I could fit a 50 or 63 volt device there. But should one of the zener diodes fail open-circuit (unlikely, I know), this capacitor will be subject to higher voltages, and we don't want any fireworks!
You'll see that the unregulated voltage falls to around 22V when the relay is on. The value of the series resistors were chosen to give around this voltage - more than enough for a 24V device. During standby, this unregulated voltage rises to the limit set by the zeners - around 42V with my prototype. This is why I thought the current source relay driver would be a good idea. The 1K5 resistor drops enough voltage to ensure that the input to the 78L05 regulator never exceeds its rated voltage.
So, a neat and efficient circuit that comfortably fits into the available space, and achieves the required functionality. When I unpack my notes, I'll be able to include more detail, but during standby, it consumes a tiny amount of power - comfortably below the recommended 2W that TV set manufacturers aim for.
On to the next section - mechanical details...
©2004 Mark Hennessy
Contact Me | Site Map | Disclaimer