Unexpected Amplifier Warm Up

Question

The following diagram shows the signal processing stage of an audio system that I've recently implemented on a perfboard.

The images of the circuit board are given below:

The circuit starts with a buffer which filters out possible DC component inside the input signal. Afterwards, a second-order Linkwitz-Riley crossover network comes into play which divides the low and medium-high frequency signals at approximately 500 Hz. Then, those two signal branches are fed into the amplifiers which are based upon bridge-configurated TDA2822Ms. Finally, the outputs drive two loudspeakers (a woofer and a tweeter).

When I energise the circuit and feed an audio signal, both amplifiers heat up aggressively whilst those two potentiometers aren't even halfway through. Additionally, even though I decrease the supply voltage down to +/-3V, those amplifiers still get very hot. Moreover, as I turn the pots, I can't get sounds out of the speakers at all (I've tested the circuit in a lab. There, I used a waveform generator and as I changed the frequency of it, I've easily noticed the change in pitch on the loudspekers, but the sound intensity was so low that my ear has had to be so close to the cone of the speaker.)

Of course, in general, amplifiers are prone to high amounts of power dissipation which in turn cause high temperature rise in them. However, this is an intriguing case for me as this is not the first time that I am building up an audio system which utilises TDA2822Ms. My desktop loudspeaker system uses one of those and although it is supplied with 10 V, when I increase the position of its potentiometer, it works fine without lack of sound and excess heat generation.

What could be the problem with the circuitry?

Answer 1

the main problem is inefficiency induced heat rise. The schematic shows 4 to 8 ohms making the impedance mismatch uncertain. The datasheet shows 32 Ohm is optimal for 9V and lower impedance speakers demand lower supply voltage.

this is simply due to impedance matching and not following the datasheet on the OP's part and being unclear about which speaker was used and what power was being drawn to cause the apparent problem.

Ohm's Law will explain.

• the output voltage drop from each supply rail is given at only 5 mA max as 1.5V & 1V for each driver in differential output. thus 18V-5V= 13Vpp but more likely 12Vpp max
• 6V/4 Ohms is 1.5A pp.
• Ohms Law means the 5V drop x1.5A is 7.5Watts peak maximum at full power into a chip that is happy with dissipating 0.5W

do you see the problem yet?

• The speaker choice is mismatched to the voltage. either riase to 32 Ohms or Lower supply to 5V then 2Vac/4 Ohms is 0.5App and this is also 1W . using more supply like 6V is possible then 6-2.5V = 3.5V

Other risks are Oscillations result when parasitic shunt capacitance with series inductance couples to non-inverting inputs with positive feedback (PFB) gain >1 then...

That results in full power dissipation with inadequate heatsinks resulting in high temperatures.

Examine your ground noise, long wires that parallel to PFB high impedance inputs. Gnd is supposed to be 0V. Or is it xx mVac? Use wide busbar or braid for gnds. Same for V+'S and decoupling caps must be low ESR.

The RC snubbers seem to be insufficient load to reduce to RF output >> 20 kHz Try 2 Ohms and compute C accordingly. Try NFB , same and wire neatly to avoid long wires in high current loops that are not using twisted pairs to cancel inductive EMI at RF. Compute 10 nH / cm impedance at measure RF oscillations for output stage and PFB inputs. Use AWG 30 Magnet wire twisted pairs for critical I/O's.

• always use short gnd clips for 10:1 probes if noise > 10 MHz.

IF THERE ARE NO OSCILLATIONS > 20kHz then it is a simple Thermal problem with high thermal resistance. e.g. > 50'C / Watt

Suggestion

With 10V AND 4 Ohms load, you can imagine 50% of 25 Watts being shared somewhere. To reduce heat rise to +25'C max that means you need to sink 1 watt/1'C rise which is a large heatsink. Since Class A is low efficiency.

The datasheet shows a 10 cm^2 heatsink ground plane and only 1.5W Max in free air and 4W max with large heatsink.

Stick a heatsink or Alum plate 3x3cm on IC with 3M thermal tape or equiv.