No Amplifiers here.


PowerDAC-SX

The main purpose of the PowerDAC-SX is converting a digital input signal (Toslink) DIRECTLY into a corresponding analogue signal, powerful enough to drive most speakers. In short a high power MULTIBIT D/A converter.

The PowerDAC-SX has NO (pre) amplifiers in the signal path AT ALL. I might put a big sticker on the PowerDAC-SX with the text:
[NO AMPLIFIERS INSIDE!] to make absolutely clear there are NO amplifiers inside the box.

The PowerDAC-SX has NO PWM circuits inside, might as well put that on the sticker as well: [NO PWM CIRCUITS INSIDE!]


By completely removing ALL analogue components from the signal path, ALL related degrading and distortion is completely eliminated. The best way to fix a problem is using a radical solution like simply throwing out ALL problematic circuits instead of attempting to fix what cannot be fixed.



So there is NO pre-amplifier and NO amplifier inside the PowerDAC-SX. This is a bit strange at first as we grew up with the myth that we always need an amplifier to drive a speaker. The PowerDAC-SX is a whole different beast that is no longer chained to conventional analogue circuits and related unavoidable degrading.

So Toslink IN -> speaker OUT, almost completely digital and lossless DC-coupled signal path.


How does the PowerDAC-SX generate the analogue output signal for the speaker without amplifier?

We start with a LINEAR (not switched!) power supply. The voltage can be set SLOWLY in order to change the output VOLUME of the PowerDAC-SX.

HIGHER supply voltage = MORE output power, LOWER supply voltage is LESS output power.

The PowerDAC-SX runs on REGULATED power supplies that only need to supply little power when generating low level signals. So we have a clean, virtually ripple and noise free regulated power supply to work with. Most power amps do not have a regulated power supply for the output stage and the global feedback loop ALSO has to deal with power supply rail noise! Most Power amps require a bias current that constantly consumes power even when no signal is generated. The bias current is a desperate attempt to fix problems of imperfect (non-linear) components. Higher bias current = LOWER distortion, in class A we use a hefty bias current and waste a lot of power in order to “fix” semiconductor or tube properties that cannot be fully fixed.


The Power supply could also be replaced by a 12V battery as there is no switching involved. It is just a DC power supply that can be set to the desired voltage (similar to a linear lab power supply) in order to change output POWER of the PowerDAC-SX.



Next we have to convert this DC voltage into an analogue signal WITHOUT using any analogue components or circuits in the signal path and WITHOUT any bias current.

The PowerDAC-SX is a MULTIBIT converter it is NOT a PWM converter! PWM converters translate sample numbers into a DIGITAL ON/OFF signal with varying pulse width. The pulse width corresponds with the audio sample (number). Often analogue circuits like comparators are still required in order to create the PWM signal. The PWM switching frequency has to be filtered out and there are no perfect filters. The PWM switching frequency will interfere with the sample rate unless both are perfectly synchronised and equal. The slightest pulse / voltage deviations (nanosecond / micovolt) can already introduce large errors and high distortion. PWM itself also introduces PWM related distortion. So keep in mind that for this reason PowerDAC-SX does NOT use PWM anywhere in its circuits!


Each bit from the source (up to 24 bits per channel) represent a value. The PowerDAC-SX uses 18 bits and this corresponds to numbers 0 … 262144. Why just 18 bits and not 24 bits? The human auditory system can resolve up to 18 bits max. under IDEAL conditions (dead quiet room, perfectly healthy auditory system). In practice few if any audiophiles have an anechoic chamber (dead quiet room without any reflections). This represents such massive dynamic range (lowest bit just barely audible in a dead quiet room) that the peak value (262144) would cause permanent hearing damage after a while. In short, 18 bits is more than sufficient for digital audio PLAYBACK.

Recordings have limited resolution where the lowest bits (24 bit system) just contain random (thermal) noise. The only practical use for 24 bit systems is for digital editing, creating sufficient headroom when mixing multiple signals. For audio playback 24 bits is overkill and usually makes matters slightly worse because more bits also translates to more bit switching noise as every toggled bit produces certain amount of noise. After exhaustive testing, 18 bits simply offered the best PRACTICAL results for digital audio PLAYBACK.


POWERDAC-SX CONVERSION METHOD

The 18 bits for each channel that represent audio sample values will ALL change at the SAME moment according to the sample rate. With 44.1 KHz sample rate, the sample values will change 44,100 times per second. With 192 KHz sample rate the samples will change 192,000 times a second. This has absolutely NOTHING to do with PWM / switch-mode, it is simply the rate of the samples being generated by the digital audio source.

The MULTIBIT conversion method used in all MULTIBIT D/A converters and also in the PowerDAC-SX will use a bit switch for every bit and NO PWM. We have 18 bits for the left channel and 18 bits for the right channel, so 36 bit switches that ALL change state 44,100 … 192,000 times per second SIMULTANEOUSLY.

Different bit switch COMBINATION -> different output voltage. As long as the bit switches don’t change (during an entire sample) the output voltage stays CONSTANT.


Each bit drives a corresponding electronic bit switch. The MOSFET bit switches are either fully ON or completely OFF. Each switch drives a section of a power resistor array that will attenuate the SUPPLY VOLTAGE according to the SAMPLE value (0 … 262144). In other words it DIRECTLY translates a SAMPLE NUMBER into a corresponding VOLTAGE at the output of the power resistor array. One could view this system as a very fast (up to 192,000 voltage changes / second) power potentiometer that attenuates the supply voltage accoding to the sample value.

The resistor array is constructed in such way that the power resistor array OUTPUT IMPEDANCE stays CONSTANT regardless of the generated voltage.

On max. volume setting the number 26,144 corresponds to 32 VOLTS. This means that the lowest bit corresponds to 32 / 2ˆ18 = 122 microvolts at the speaker terminal output. All steps between 0 and 26144 are accurate down to the microvolt range offering an extreme accuracy / LINEARITY AT THE SPEAKER TERMINALS that is almost impossible to obtain with any analogue signal path. Because the PowerDAC-SX does not amplify and generates the DESIRED voltage directly it cannot clip. The circuit is also fully DC-coupled so there isn’t any non-linear component in the signal path that could possibly cause any degrading.


So the PowerDAC-SX takes a sample number (0 … 262144) and translates it into a corresponding VOLTAGE using at least 18 bit-switches to set the output voltage at the speaker terminals with extreme accuracy and linearity. The PowerDAC-SX D/A converter is a fractal 16 MULTIBIT converter, fractal 16 means that the 4 highest bits are translated to 16 smaller sub bits. This offers higher accuracy, lower BIT switching noise and lower output impedance. The X in the name refers to the MATRIX circuit that emulates 4 identical D/A converters using just a single one. The MATRIX circuit doubles the output voltage (bridge circuit emulation) and it ensures that the converter starts at ZERO volts for superb low level performance and cool operation (zero stand-by power, proportional power consumption depending on generated output voltage).


The PowerDC-SX does not amplify, so noise can’t get amplified either. It offers the maximum obtainable signal to noise ratio, a noise floor so low that it can no longer be heard, a noise floor so low that it seems there is no noise floor.


The PowerDAC-SX is based on one of the most advanced MULTIBIT power converters. One converter (MOSFET bit switches + resistor array) is used in 4 different ways, simulating a BRIDGE multibit converter.
This will double the voltage at the output from 16V DC supply voltage (max.) to 32V peak to peak voltage BETWEEN the speaker terminals. The signal polarity is NOT handled by the D/A converter but by a separate polarity control circuit that SWAPS the speaker terminals (using MOSFET switches) every time the signal polarity changes from plus to minus or from minus to plus. As max. recording frequency is limited to say 20 KHz, the maximum polarity change rate would be 20000 polarity changes / second for the highest trebles.

Unlike conventional low power multibit D/A converters,
the PowerDAC-SX converter always starts at zero volts. With zero volts on the power resistor array, power dissipation (no signal) will be ZERO.
When generating a signal more or less voltage is dissipated in the resistor array and the speaker. So power consumption is proportional to the audio signal being generated. This way the PowerDAC-SX runs cool.
Another advantage is that only the lowest bits will be used / switched for low level signal generation and the higher bits won’t have any effect on low level signal accuracy. With most low power multibit DACs the highest bits have major impact on accuracy and switching noise. So this common issue with multibit converters is elegantly solved in the PowerDAC-SX.

The PowerDAC-SX contains NO LOCAL and NO GLOBAL feedback loops and has / needs NO bandwidth limitation. This translates to the highest practical playback phase accuracy as the PowerDAC-SX offers a massive bandwidth that cannot be obtained with conventional audio amplifiers. Why do we need highest possible PLAYBACK bandwidth? because we are working with highly complex audio spectra and the slightest alteration of this highly complex spectrum will translate to a more fuzzy and distorted (dynamic distortion) signal. Conventional amplifiers with very large bandwidth begin to offer advantages despite the fact that the highest audio frequency is only around 20 KHz. So even when using conventional amplifiers it makes sense (and is audible) that amplifiers with much larger bandwidth offer a cleaner and more transparent music reproduction. The PowerDAC-SX has theoretical bandwidth of around 100 MHz (in practice limited by wiring), the remaining practical bandwidth is still so high that it is very easy to obtain a very high level of transparency when processing complex music spectra.

The PowerDAC-SX transfer characteristic is fully linear, starting at zero volts, therefore it requires NO bias current and the resistor array can be driven starting at zero volts.


PowerDAC-SX STEERING METHOD

Drivers within a speaker respond to the magnetic field (permanent magnet) the number of turns on the voice coil and the CURRENT running through the voice coil. So in order to have the speaker cone follow the signal accurately we have to drive it with a constant CURRENT. When the voltage is kept constant, the slightest variation in speaker impedance will lead to incorrect current, incorrect cone positioning and distortion / coloration.

Most tube amps offer quasi constant current steering (CURRENT IS KEPT MORE OR LESS CONSTANT) and are quite COMPATIBLE with driver properties, this is possibly why tube amps are often preferred by audiophiles. The issue with quasi constant current steering is at speaker resonance, Here the speaker impedance rises, in order to maintain constant current, more power is dumped on the speaker and this makes the resonance much worse. As resonance mainly occurs with bass, we often notice that tube amps have difficulties with bass performance.


Semiconductor amps offer constant voltage steering (VOLTAGE IS KEPT CONSTANT NO MATTER WHAT). This steering method is LESS compatible with driver properties. When speaker impedance dips, much more power is dumped on the speaker. This not only causes clearly audible distortion, it can also overload / destroy the amplifier.


Amplifiers -always- have local feedback (feedback loop at component level like transistor, MOSFET or tube) and most also have global feedback (feedback loop from speaker output back to the amplifier input, large loop with many components within that loop).

The feedback loops can turn an amplifier into an oscillator at higher frequencies where feed back slowly changes in feed forward as a result of given circuit phase shift. So in order to keep the feedback loops more or less stable, the amplifier bandwidth has to be LIMITED. Depending on bandwidth, phase distortion is introduced, this often results in blurred sound reproduction as the audio spectra is not reproduced accurately enough. If for some reason the feedback loop locks out for a split second we get distortion. Depending on amplifier open loop properties we could end up with temporary high distortion. Usually this temporary disruption of the feedback loop and resulting distortion / coloration increases with increased output power.

In practice we have to figure out the best MATCH between both, amplifier (with its unique properties) and speaker (with its unique dynamo properties). If we are lucky we MIGHT find the golden combination of amplifier and speaker.


Speakers are not just passive devices like resistors for example, as soon as the speaker is producing music and the voice coil is vibrating in a strong magnetic field (permanent magnet) we have a dynamo that generates energy. The louder the speaker plays, the more power the speaker dynamo will generate. This dynamo voltage is INJECTED into the feedback loop(s) and often the feedback loop(s) will be disrupted by this dynamo voltage. This causes incorrect operation of the feedback loop and related much higher distortion, at this point data sheet specs no longer apply as these are based on a lab test setup with passive load that won’t act as a dynamo.

Steering methods are often incompatible with the speaker and / or cause unwanted distortion (enhanced resonance, overload, distortion). The performance of ALL amplifiers can be degraded by the speaker dynamo voltage as this can disrupt feedback loops.


The PowerDAC-SX offers quasi constant power steering (OUTPUT POWER IS KEPT MORE OR LESS CONSTANT). Because the output power is kept constant, overload conditions cannot occur. PowerDAC-SX can therefore drive any load from dead short to open circuit without issues. At speaker resonance the output power is kept constant so it won’t boost resonance like a tube amp would. This way the bass is kept perfectly under control.

The PowerDAC-SX has no analogue circuits / components and NO local and NO global feedback loops that could get disrupted. The speaker dynamo voltage therefore CANNOT disrupt the PowerDAC-SX, the result is VERY low distortion and NO coloration regardless of volume setting.


SUMMARY:

The PowerDAC-SX has NO (pre) amps in the signal path, so all related distortion and degrading are COMPLETELY eliminated.
The PowerDAC-SX does not amplify so it cannot clip
The PowerDAC-SX has NO PWM circuits, just LINEAR power supplies and bit switches that create a MULTIBIT power D/A converter.
The PowerDAC-SX starts at zero volts and has a separate polarity switch circuit. This translates to power consumption proportional to the signal being generated and zero power consumption when no signal is generated.
The polarity switch circuit also ensures that only the lowest bits are used for low level signal generation, this means that the highest bits no longer have any effect on low level accuracy and that switching noise is at absolute minimum.
The PowerDAC-SX steering method prevents overload, resonance, distortion and coloration issues. This translates to transparent music reproduction regardless of volume setting.
The PowerDAC-SX has NO local and NO global feedback loops, so no problems with speaker dynamo voltage.
The PowerDAC-SX provides a single box digital audio set with minimum wiring (Only Toslink, mains and speaker wiring required). It offers an almost direct conversion from sample value (number) to corresponding voltage on the speaker output.