Neither had I until I came across an unusual amplifier from Maxim, that slipped below Electronics Weekly’s radar when it first came out.
Class-D audio amplifiers (switching audio amplifiers) have done a lot to increase efficiency and cut heat compared with traditional analogue Class-AB amplifiers, without sacrificing much audio quality when it is done correctly.
However, as power output drops, switching losses start to dominate, which is why amplifiers aimed at 32Ω headphones are still analogue.
While they are analogue, the latest headphone amplifiers are analogue with a twist: the amplifier is Class-AB, but they get their power from a two-level voltage source which will be at the higher level for sonic peaks and the lower level for quieter passages.
This sort of power-saving operation is called Class-G – and has even been used in some large amplifiers. Some of Bob Carver’s 1980s ‘Cube’ amplifiers used Class-G operation as a heat-reduction strategy to deliver over 200W from cases far smaller than many thought possible at the time.
In a rather left-field move, Maxim has combined Class-D switching amplification with Class-G-style rail switching and come up with Class-DG operation, implemented in its MAX98308 3.3W (from 5V into 8Ω+68µH) mono audio amplifier.
To get the power, it has differential outputs to double potential output swing, and a charge pump that can generate a negative rail when more than 0-5V of output excursion is needed.
Multi-level output modulation is employed in to draw maximum power from the lower impedance positive battery rail, rather than the higher impedance charge-pump-generated negative rail.
This is accomplished by generating PWM signals that swing from ground to the positive rail, or from ground to the negative rail, at either end of the bridged load, rather than continually swinging from positive to negative rails.
“When the negative rail is not needed, the output is drawn entirely from the standard supply. This scheme results in high efficiency over a wide output power range,” said Maxim. Better than 80% efficiency is claimed from 350mW to 2.2W output.
Fixed gains of 8.5, 11.5, 14.5, 17.5, and 20.5dB set by a select input, or there is a MAX98307 version which has adjustable gain set by external resistors.
1.77W is available from a 3.6V lithium ion battery.
Both the 3.3W and 1.77W figures are at 10% THD+noise. For 1% THD+noise, 1.54W is available from 3.6V, and 2.85W from 5V.
At 500mW from 3.6V, efficiency is 84%. Quiescent current is 1.85mA at 3.6V.
Active RF emissions limiting is built in, as well as click and pop suppression, thermal protection, over-current protection, and a low current shut-down mode.
This is a very small chip: a 1.7×1.3mm 12bump wafer-level package (WLP) needing only 4.7 and 10µF capacitors (the external resistor version is in a larger 16pin TQFN).
In a slightly larger (2.1×2.1mm, 16bump WLP), Maxim also makes a 3.6V class-D audio amplifier with a more conventional built-in full-time inductive boost converter (needing 2.2µH and 22µF) to provide the audio section with added headroom.
Called MAX98502, from 3.6V it can deliver 2.2W (10% THD+N) or 1.7W (1%) – and these figures roughly double if the load is changed to 4Ω+33µH.
Quiescent current is 1.7mA at 3.6V, and it will run from 2.5 to 5.5V.
The inductive version is slightly more efficient, however, it will not deliver 3.3W into 8Ω at any voltage, although it will deliver over 4W into 4Ω.
For those interested in higher power audio, many semiconductor firms make single-chip Class-D amplifiers. For example, Maxim’s MAX98400A delivers 2x20W into 8Ω loads or 1x40W into 4Ω. There is a reference design (No 4320) for a 2.1 channel (two 2in high frequency speakers and one 5in woofer) Bluetooth docking station on its website complete with enclosure plans.