Transformer control for the modular high end audio amplifier

Project presentation


The transformer control module presented here is part of my modular audio amplifier system. Large toroid transformers require some kind of soft start circuit in order to mitigate their inherently large inrush current. Basically there are two methods powering up large toroid transformers: First is to start the transformers with resistors or NTCs in series and after some time bridge the inrush limiting elements with relay contacts. This is very simple and works well up to a certain transformer size. I use this method for my transformer control presented here. The second option would be a very sophisticated control like described by Emeko. I bought such a module for powering up the transformer of the high power amplifier module and this works extremely well. The chassis of the modular amplifier could accept a transformer with roughly up to 1kVA, but a 400VA transformer is more than sufficient, hence the simple soft start should be sufficient. Here is the ultimate guide to soft start circuits for further reading.

Design features

  • Zero power consumption in off state
  • Power on and off control by a single DPST momentary-on push button
  • One contact of the power button carries mains voltage in off condition and once the amplifier is running, the mains voltage is removed entirely from the button, avoiding any disturbance of the amplifier internals by mains voltage traversing the chassis
  • 12V regulated DC output to power other control modules like the loudspeaker protection
  • Instant AC fail signal (open collector) to other modules like the loudspeaker protection module for instant loudspeaker mute on AC loss events
  • Fully compatible with 110V to 120V and 230V to 240V mains voltage by voltage selector switch (to be operated inside the chassis only). All transformers connected should have two 110V to 120V primary windings. The control modules mains voltage operated relays also adapt to different mains voltage.
  • Supports one power transformer and one small auxiliary transformer
  • Dedicated self resetting fuse for each primary winding of the power transformer
  • 5x20mm glass tube fuse as main fuse for the whole amplifier
  • EMI filter to keep HF junk on the mains voltage out of the amplifier


Circuit description

J1 is the AC inlet, which is EMI filtered. The safety earth is routed to one terminal. This can be used to connect the screen winding of the power transformer to safety earth. The connection to the chassis is established by the EMI filter, which needs to be bolted to the chassis in a reliable way. Safety earth is also connected to three of the mounting screw locations, but this is not a safety earth connection that should be relied on. F1 and F2 are the main fuses for the whole amplifier. They are not meant for overload protection, but should be rated below the mains filter nominal current. Those fuses should never blow except in case of a short circuit of the AC mains. MOV1 provides some protection against mains voltage surges. This is actually a new requirement for mains operated equipment. Read here more about this. In case of a mains voltage surge, the fuses F1 / F2 will likely blow regardless whether the amplifier is powered on or off. Switch S1 is for mains voltage selection. All subsequent transformers need two primary windings with 110/120V each in order to take advantage of this feature. While most parts of the world have 230/240V mains voltage, I might move some day and I want to take my audio equipment with me in any case. Transformers with dual primary windings are not that much more expensive, but much offer a lot more flexibility. K1 and K2 are normal off and ensure that the amplifier consumes zero power in its off state.

This is the power on sequence: The dual pole single throw normal open switch PS1/PS2 is pressed by the operator. PS1 energizes the coil of relay K2 in 110/120V configuration and also K1 in 230/240V configuration. Both K1 and K2 have 120V windings. In 110/120V case, only one winding primary winding is energized, while in 230/240V case, both windings are. Now T1 powers up (either using one or both primary windings) and the 12V DC power supply comes up. As soon as the 12V DC supply is up, K3 now closes its contacts in parallel to PS1 and in case of the 110/120V case, also powers the coil of K2, which then energizes the second primary winding of the transformers. This is the moment the operator can release the power on switch. I expect this to take only a few mains voltage cycles. The power transformer is energized through ICL1 and ICL2, which are NTC inrush current limiters. The power transformer is overload protected by self resetting fuses F3 and F4. The auxiliary transformer is not protected by dedicated fuses, as well as T1, which powers the 12V DC supply. C5 charges up on the 12V supply and after some time, defined by R7 and C5, tilts the Schmitt trigger made up of Q7 and Q8. After the time needed to tilt the Schmitt trigger, the coils of K4 and K5 become energized. Also, K6 could become energized, but this is prevented by the open connection of power switch PS2, which has been released by the operator in the meantime. In case the operator holds the power switch pressed (for a few seconds), the amplifier would immediately enter shutdown mode again. The normal closed contacts of K4 disconnect half of the power switch, namely PS1, which is used for power on only. This ensures that the wires running to the power switch no longer carry mains voltage, reducing the risk that the interior of the chassis is polluted by noisy mains voltage and operating the amplifier with the lid open becomes a bit less dangerous. The normal open contacts of relay K5 close and short the inrush current limiters ICL1 and ICL2, which have run hot in the meantime and are no longer needed as the power transformer is running. This allows ICL1 and ICL2 to cool down again.

The 12V DC power supply is powerful enough to supply other control circuitry of the amplifier. The 12V DC power is provided on two pairs of terminals. One pair of terminals is for the power switch illumination and the other pair for other modules like the loudspeaker protection for example.

The option to connect an external DC supply at terminal DC_IN is ill-advised and stems from attempts to improve the control sequence. GND_3 must be connected to safety earth in any case.

On the AC_LOSS terminal, the module reports mains AC loss. Q2 is an open collector output where Q2 becomes conductive once a failure of the AC is detected. This allows other modules like the loudspeaker protection to take action like to instantly disconnect the loudspeakers in case of an AC loss event.

The power off sequence is initiated by the operator pressing the same power switch again. Since PS1 is deactivated now, this cannot trigger a power up sequence anymore. PS2 can now energize the coil of K6, which only became possible after the amplifier was fully powered up and Q6 enables to close this circuit. The normal closed contacts of K6 break the circuit of K1 and K2 coils and the mains power is removed from the whole amplifier. K3, which keeps the whole circuitry alive, is supposed to release first on power loss since this is controlled by the UVLO circuit formed by D6 and Q3. In case the power button is pressed too long, the control will enter a hick-up mode trying to revert to the fully powered up state. This is very unpleasant, but I haven't figured out yet how to avoid it. The simple solution is to press the power button for a reasonable duration only, which is up to one second roughly.

In case the power supply of the control is below roughly 11.7V for too long, Q9 and Q10 would shut down the amplifier. This feature is not really required, but stems from a problem with ill-defined states of the first revision of the control unit. This problem was solved on the AC mains side of the control, but I kept the shutdown on under-voltage nonetheless.

The whole operation is fairly complex, which is due to all the features that this control module offers. It would be much easier to operate the control unit from an always on supply and not doing so was a deliberate design decision. Honestly I didn't imagine this would complicate everything that much. Fully disconnecting the amplifier without any stand-by power drain is probably worth it.


All BC transistors can be substituted by any small signal transistor. Obsolescence of components goes faster than I can change my component library and schematics and so those two transistors are my default low voltage small signal transistors.


I built an earlier revision of the control module and the schematic presented here already addresses the issues I found with the earlier revision. I have resettled the components to the second revision of the PCB and tested the module. It mostly works now as expected except the hick-up issue on power down in case the power button is pressed too long.

The root cause for failure of the first revision was lack of circuit simulation. Usually I simulate everything extensively to make sure all works well, but in this case, I only simulated some parts of the circuit. Transformers and relays are pretty inconvenient to simulate and this is why I didn't simulate the complete circuit. A proper simulation would have saved a lot of effort and cost.

Repeating mistakes is much more fun than to learn from them and so the second revision also has some quirks as well.