3W RMS Stereo Amplifier Is Very Small
I recently came across these little 3W class-D amplifier modules that had sprung up all over eBay. At $1 a piece delivered, I figured that there was little to lose in ordering one and testing it out.
The board measures 2cm x 2cm and contains the PAM8403 chip and the circuit as show below, lifted directly from the datasheet. As you can see, there isn’t much to it.
Performance is claimed to be 2.5V to 5.5V supply voltage with output of 3W into 4 ohms at 5V with 10% total harmonic distortion, and 0.15% THD at 0.5W. Efficiency of 83% when driving 4 ohms, and 87% for 8 ohms.
I made a quick lash up using a 5V 2A power supply and a couple of reasonably quality bookshelf speakers, and listening to some MP3 music being played by a computer was very pleased with the result: “Perfectly listenable to” was my assessment. Others agreed and so I decided that this would be the perfect amplifier for listening to music in my being-converted-to-solar-power office and radio shack. See below for the result.
1. Class D
The high efficiency comes about from the so-called class-D operation of the amplifier. In a class-D driver circuit, as I understand it, one output terminal is connected hard to the supply rail and the other is connected hard to ground, and alternated via a MOSFET H-bridge arrangement. The MOSFETS, being very low resistance when turned either fully on or fully off, dissipate relatively little power.
The amplitude of the input signal is sampled to create a variable width pulse train (260kHz in this case) that drives the MOSFETS, the width of the pulse train being proportional to the input amplitude. This diagram from Wikipedia explains it better:
2. Power Output and RFI
The datasheet specifies the power output as being ’3 watts’ but it is not mentioned whether this is RMS or peak, nor per channel or for both combined, so a measurment was in order: with a 1kHz sine input and driving a 4 ohm load, the scope showed around 3.6V RMS across the speaker which, by my calculation, is 3.2 watts RMS. It’s a little suprising then that the datasheet doesn’t trumpet ’3W RMS per channel’, given that it’s quite impressive for such a small device.
The waveform on the speaker leads shows the 260kHz pulses. Persumably, the density of the pulses corresponds to the sine amplitude, but this isn’t really obvious. Clearly the speakers are being relied on to do the low pass filtering, and the cables are likely to be radiating significant RFI.
With the amplifier and speakers set up about two feet away from the HF receiver, sure enough, the interference was prominent every 260kHz across the HF bands up to 15 metres where I couldn’t detect it. The datasheet mentions the filter capacitor on the power input pin should be sufficient to prevent RFI from the power leads, and for speaker leads of less than 20cm length, no further suppression should be required. For runs greater than 20cm, it suggests a ferrite bead and capacitor as a low pass filter at the chip’s speaker outputs.
I reason that I can live with the RFI as I am not going to be listening to music and at the same time operating HF, and so haven’t tried the ferrite low pass, but persumably these devices are being used in consumer appliciances such as LCD monitors that have FCC (or equilivent) approval, and so the RFI must be managable.
3. Finished Product
The components wired up and being tested prior to assembly. The case is from a cheap USB hub, the innards being used in another project, and the existing holes being suitable for mounting the various switches, knobs and jacks etc, and the remaining holes allowing for air flow. Copious amounts of hot-melt glue were used to fix the parts inside the case.
The larger board is a switch-mode power-supply module (available on eBay for around $2) set to provide the 5V for the amplifier module. Input voltage can be anything from around 7V to 30V or more. I’ll be running it from a 12V solar-charged battery.
The finished product with power switch, LED, volume pot and input jack on the front panel. On the rear are the power input jack, and exit points for two short leads for connection to the speakers. As you can see, it’s not very big.
All in all, and listening to some relaxing Pink Floyd as I write, I am very happy with the little amp, so happy in fact, that I have just lashed out a whopping $2.36 bulk buying five more modules. Ideas for novel applications are welcome. Rock on.