Today's Audio Final Amps

are normally still equipped with traditional toroid transformer with rectifier and capacitor battery. This is a generator for a 100 Hz hum in the supply voltage of the power amplifier. Considering also the CE test and the suitability this is not a brilliant solution.

Better Final Stages

are already equipped with switched mode power supplies. They serve, however, to often as replacement for toroid transformer and thus just for show. The rectification of the mains voltage is carried out via rectifiers and capacitor battery with the known disadvantages. A power conditioner is not realized here.

One Manufacturer

is known to us which makes the effort to use a real PFC power supply in his power amplifiers. It is the company named Halcro. The solution is, however, some thing like a "One-cylinder".

At the behold BPA768 four "one-cylinder" are nested resulting in a "four-cylinder". There are consequently four storage inductors, four switching transistors, four diodes and four separate storage capacitors operating independently from each other and then finally feed their power to a common bus bar ( like a crankshaft at a four-cylinder car engine). As this is well known, four-cylinder engines are running much more smoothly.

There is still something special in the power supply: The switching frequency is strictly synchronous with the audio sampling frequency. By that there are absolutely no beats and comparable conditions between power supply and analogue assemblies.

The switching frequency is exactly 96kHz, as exact as the real time clock included in the pre-amp APU768 or even better. This applies also if a digital source has only 44,1kHz or works inaccurately. A PLL system especially  established in the power supply provides that. No other final stage amplifier so fare is equipped like that.

The D/A-Converter is also a "four-cylinder" as described in case of PFC switching power supply. This has not only the same positive backgrounds as described for the power supply but is also used to process cleanly the high sampling frequency of 768kHz. Four D/A-Converters of type AD1853 from Analog Devices are used. These components are stereo D/A-Converters.

We use each both channels together as a mono block and interleave them chronological into each other. 16 peaces of 24Bit D/A-Transducers are therefore for the recovery of the valuable analogue audio signal. The signal is then led on strictly symmetrical amplifiers for amplification and filtering. Used are the lowest-noise operational amplifiers today available, the AD797 of Analog Devices.

Today's analogue final stage amplifiers

operate with conventional difference amplifier to reach the required voltage gain. The numerous disadvantages shall not be listed here. This disadvantages justify, that some manufacturers use tubes in this place. We are, however, not tube specialist and don't want to become it either.

The Right Solution is Called:

discreet transimpedance amplifier. It has to be discreet since there are no monolithic available for the required voltages and especially not for an optimal control of a analogue final power stage. The used transimpedance amplifiers are built up completely symmetrically, i.e. mirror symmetrical to the mass potential. The required know how comes not least from our high-frequency engineering.

Conventional Final Stages

have an analogue idle current adjustment which must compensate the temperature fluctuations and the spread of transistors. It achieves that more or less good. The negative consequences resulting from it are well described in the literature and otherwise. Many compromises have to be made. Added to that are the problems of overload protection and short-circuit rating. They further intensify the analogue.

Completely Different and New Approach:

At behold BPA768-400 there are simply no "annoying" source-resistances at the power transistors and also no feedback to protection circuits. There are no analog protective circuits. The solution is effected by digital control via a signal processor. The construction is relatively simply to be understood. The current is not measured via a source resistance but via a resistance installed in the drain voltage feed.

Idle Current Control by Processor

There the current has no negative influence on the audio signal at all. On this high tension potential the measured current than is analogue to digital converted and passed over  to the processor via optical coupling devices. This is carried out for the positive as well as the negative voltage feed. The processor has an algorithm enableling the correct  idle current to flow on at any time even at full modulation. No analogue final stage can operate like that.

The So-Called "A-Operation"

doesn't exist in reality as the current passing to the loudspeaker is always missing in the respective other final stage half. The "A-final-stage" do not exist for other reasons too but it would lead much too far to explain this here. The processor in the behold BPA768 passes the calculated information again via optical coupling device to a D/A-Converter "sitting" on the audio signal itself. ( May be taken literally and imaginable)

This D/A-Converter

adjusts in real time the operating points of the complementary transistors within 10.4µsec and thus faster than any audio signal. It naturally regulates also the effects arising unavoidable with the modulation of the transistors via their characteristic. The physical laws can not be avoided but used skilfully and without disadvantages.

At the behold BPA768 we call that: Digital real time idle current tracing.

A positive waste product of this method is that the idle current can be set at any time to the requirement of the audio signal. It is therefore not fixed and adjustable.

The idle current can be adapted during the operation and also the switched mode power supply can adjust the supply voltage processor-controlled due to the fact that the loudness of the analogue signal is known. Based on that it seems obvious to adjust everything automatically for the chosen volume. At low volume no high operating voltage is required.

On the other hand a higher idle current is meaningful to keep low the zero crossing distortions and vice-versa. With that at the same time the dissipation power can be optimised resulting in lowering the temperature and consequently the cost of electricity . It furthermore extends the lifetime. The behold BPA768 provides that for the first time ever.

The used signal processor naturally has also the task of the overload protection. The processor calculates in real time the present dissipation power in the transistors. By this method a much better use of these transistors is possible. The calculation processes the information of the supplied operating voltage und currents considering simultaneously the transferred output power.

Calculating difference from both yields the dissipation. This is carried out of course separated for all four half bridges and their positive and negative half waves. Resulting those eight processes of calculation are running in parallel. In case the highest permitted value at one of the eight branches is exceeded, this block is switched off within a few µsec.