|
In principle, the output stage of most switching amplifiers has the same working principle.
PWM = Pulse Width Modulation is simply a matter of variable open/close time of the transistors.
Simplified, you can compare the transistors in the output stage to two switches. Pls. see fig 1.
 Fig. 1
The longer switch 1 is "ON" (and switch 2 is "OFF") the longer the "positive" excursion of the loudspeaker will be and the longer switch 2 is "ON" (and switch 1 is "OFF") the longer the "negative" excursion will be.
And if the two switches are on for the same period of time within one switching cycle there will be no excursion of the driver (no sound). Pls see fig 2.
 Fig. 2
Most PWM amplifiers today are self oscillating which means that the amplitude of the ANALOG input signal decides the switch frequency. This means that the switch frequency is high (> 300k Hz) when the audio input signal level is low, and that the switch frequency is low (= moving towards the audible frequency range) when the audio input level is high = when the output from the amplifier is high.
The advantage of this topology is that it is cheap and simple to design since it uses well proven feedback from the analog output to the analog input partly to create the oscillation and partly to create a low THD amplifier.
The disadvantages are that it requires an analog input signal (= it is really an analog amplifier!). So, the PCM (Pulse Code Modulated) signal from your CD player is converted to analog and then the analog signal is converted to the PWM signal.
Also, the feedback loop typically isn't linear which is why distortion often increases towards higher frequencies.
Finally the fact that the switch frequency moves closer to the audible frequency range (typically it is limited not to go lower than 80 - 100 kHz) when you turn up the volume results in risk for creating an offset working point for the tweeter. Even though you cannot hear it directly, harmonics of a 80 or 100k Hz switch frequency can bias the tweeter.
The Lyngdorf true digital amplifiers use a fixed switch frequency (@400k Hz). Furthermore, we convert the PCM signal directly to the PWM signal for the output stage in a PCM-to-PWM EquibitTM modulator.
Below in fig 3, is a sine wave, in terms of a PCM signal, in the upper panel of the figure. In a PCM signal, each discrete sample represents a specific amplitude. The corresponding pulse-width-modulated signal at the same sampling rate (frequency) is shown in the lower panel.
The magnitude of each PWM sample is described in terms of the pulse width, as opposed to the pulse height in a PCM signal. So, the 24-bit PCM digital audio signal is fed to the modulator where the audio data is up-sampled 4 times. The EquibitTM modulator then translates the up-sampled signal to a PWM signal having the same switching frequency.
 Fig. 3
This means that we have an unbroken signal path without sound deteriorating conversions.
The very unique thing feature of the Equibit technology is that the PCM to PWM conversion is made without using feedback. Which actually is a necessity since you cannot make a feedback loop taking the analog signal at the speaker terminals and feed it back to the digital PCM signal! That is just not feasible!
So, the Lyngdorf true digital amplifiers are open loop amplifiers - no feedback used at all.
It is quite obvious that developing such an amplifier is a much more complex process. It is simply more expensive since it requires very stable and ripple free power supplies and other special solutions such as the Equibit for the PCM to PWM conversion and extremely linear design of both power supply, output stage and reconstruction filter.
The advantage of this meticulous design is that e.g. the very linear and low distortion simply results in a more musical sounding amplifier. If you consider an acoustic instrument it gives a fundamental tone and a lot of harmonic overtones. For e.g. pianos and violins there is considerable energy in the overtones compared to the fundamental tone. If the distortion versus frequency of the amplifier is not flat (which it rarely the case for a typical switching amplifier) you will add more or less distortion from the fundamental to the natural harmonics and actually destroy the balance of the natural harmonics. However, when the distortion (which, as already mentioned, is very low) is the same at all frequencies you can preserve the natural balance of the music you listen to. We have conducted experiments with this, and actually test persons would prefer higher but linear distortion compared to lower but nonlinear distortion over frequency. So, linear distortion is key to musicality.
The advantage of the fixed switch frequency is first and foremost that it is so far away from the audible frequencies. This gives the possibility of constructing a more efficient reconstruction filter. It is almost self explanatory that a passive filter (in principle it actually only consists of a coil and a capacitor, see fig 4) only supposed to filter away a very narrow frequency band can be made more efficiently than a filter that is constructed to filter away a broad frequency range.
 Fig. 4
The reconstruction filter of a switched amplifier is often a point that is overlooked - partly because the ideal components are expensive and take up board space and partly because the filter is regarded to be way out of the audible frequency range. But even filters placed octaves above audible frequencies affect the linearity within the audible frequencies: In the Millennium amplifier we have used no compromise Jensen capacitors resulting in a unsurpassed low distortion and linearity in the filter.
|
