Increase power output in cheap PC speakers

The PC speakers used the common TDA2822 audio IC, like those found in the subsidiary channels of cheap 5.1 amplifiers. Using the original design, I measured the Pout was only 0.4W into 8 ohm @ 10% THD (see fig. 5). 

The TDA2822 was a plastic DIL with no provision for additional heat-sinking. Checking the IC's datasheet, I found that pin #4 (gnd) has the lowest Theta J-A (max. 70 deg/C). The other pins were specified at max. 100 deg/C. So, I guessed that pin #4 was also doubling as the paddle for the die. I made a piece of tin-plate approx. the size of the IC's outline. The tin-plate was soldered to pin4 (gnd). The tin-plate rested on the body of the TDA2822.

Fig.  1:Solder tin plate to IC's ground pin and then bend it over the IC body

Fig. 2: Drill a hole in PCB beside IC

Fig. 3: Place an aluminium heatsink over the tin-plate and bolt it to the PCB

Fig. 4: another view of the added heatsink

An aluminium heatsink was bolted to the tin-plate thru a hole drilled in the PCB (fig. 2). The screw at the right side of the heatsink served as a adjustable stop against the top surface of the PCB (fig. 3). Its function was to keep the heatsink level and in good contact with the tin-plate below.

After the addition of the heatsink, the TDA2822 could sustain 1.9W @ THD=10% continuously at 12V. Without a heatsink the protection circuit in the IC would have gone into thermal limiting. When Vcc is raised above 14V, the TDA2822 broke into high freq self oscillation. Perhaps, a tweak to the snubber network or PCB layout can solve the oscillation, but I didn't try it.

Fig. 5: Distortion vs. output power. The added heatsink allows the supply voltage to be increased for greater output power 

Another area worth looking into, if you want deep bass, is the output coupling cap. The measured freq resp showed the lower -3dB freq limit was ~ 100Hz. This was not a surprise considering that the original 220 uF cap has a jXc equal to 8 ohm at 100 Hz. Doubling or quadrapling this value will bring down the lower limit to 50 Hz or 25 Hz, respectively.

Fig. 6: Amplitude vs. frequency