1. EE434 Jason Adams Mike Dierickx
Class D Audio Amp
EE434
Jason Adams
Mike Dierickx

2. Basic Operation
The Class D Audio Amplifier uses Pulse Width Modulation to change an analog audio signal into a digital continuous-time signal.
This digital signal is fed through a set of switching power MOS devices to create high efficiency, as in all switching mode power devices.

3. Basic Design Requirements
Uses 3.5 Volts of supply and receives an input signal that swings from -5 to +5 Volts
Has a left and right channel
Has digital volume control
Can be fit into a die of 900 um square

4. Design Flow – Attenuator
Since the signal swing of the input is greater than the swing of our allowed supply lines, the input phase needs to be attenuating
Also, the output impedance of our previous stage is unknown, but can be assumed high
Therefore, an operational amplifier-based attenuator was chosen

5. Attenuator – Cont.
It was decided that this would be the logical place to add our volume control as well, and a scheme using eight possible levels at 3 dB increments (factors of 2) was implemented
This is done by changing the value of the feedback resistor in the attenuating op amp

6. Potentiometer Realization
Only one transistor is allowed to conduct at a time, so the op amp in the circuit can see anywhere from 64 kOhms down to 500 Ohms of total resistance in its feedback loop.
Note the transistors have widths of 6 um or 12 um. This is so that their conducting resistance is negligible compared to the other series resistances.

7. Volume Control - Decoder
Rather than have the volume controlled by 8 pins, a 3 to 8 decoder is used.
This decoder was made using Synopsys, the original and program-optimized code are included as items 2 and 3 in the handouts.

8. Final Tuning
The input resistance on the op amp was selected as 200 kOhms, giving a maximum gain of 64/200, or .32.
With an input signal swing of 10 Volts, this means the maximum swing after this phase is 3.2 Volts, so clipping is impossible if the input is in the specified range.

9. Testing
The testing was done using simply a 1 Volt DC signal, so that changes in the gain would be very visible. The results are item 4 in the handout.
The eight different levels at 3 dB increments are clearly identifiable.

10. Results

11. Triangle Wave Generator
The triangle wave serves as the carrier in the class D audio amp. The circuit to generate this signal is item 5 of the handout.
This follows a basic schematic from the Sedra and Smith textbook, consisting of a bistable multivibrator on the left and an integrator on the right

12. Triangle Wave Results
Using the equations in the textbook, we were able to create a generator to operate at 175 kHz, which represents significant oversampling. The simulation results showed the actual frequency to be closer to 155 kHz, still highly oversampled. The results are item 6 in the handout.

13. Triangle Results

14. Modulator
The modulator is simply a comparator, a very simple part of the circuitry which uses only an op amp.
The output is high (+1.75 V) when the modulating triangle wave is above the incoming attenuated audio signal. This is why 3.2 V p-p for the tri wave was important

15. Modulator Simulation

16. Modulator Simulation

17. Final Stage – Switching Power
The final stage is the power transistors that make the amplifier a switching power device, and a low-pass filter that extracts the signal from the carrier
This stage was decided to be OFF-CHIP. Using the whole die area for one NMOS transistor (w = 900u with 250 multiplier) produced only about 37% power across the speaker

18. Final Stage – LP Filter
The LP filter is off-chip in virtually every class D amplifier, since the use of inductors and the size of the capacitors makes integration nearly impossible
The purpose of the LP filter is NOT to improve the audio quality, but rather to eliminate the energy in the higher harmonics to protect the speaker

19. Bridge Simulation

20. Attenuator Layout

21. Attenuator PLS

22. Triangle Generator Layout

23. Triangle Generator PLS

24. High-level Schematic

25. High-level PLS

26. Conclusions
In spite of the difficulties, we both feel the project was very educational and really helped to bring together the material we had been covering in the lab.

27. Conclusions
However, throughout the process there were a few elements that continued to annoy us:
Trouble with on-chip current
Computer lock-ups and loss of data
Op Amp troubles

28. Conclusions However, there were many things that went well:
Once we had a functional op amp, we found out all out designs were correct
All things considered, the layout went VERY well, and was probably completed in under 25 man hours.

29. THE END