1 Introduction
Based on the 0.35 micron N-well process, the full-time division of the power supply is designed. At the same time, in order to reduce the disordering of the residual voltage, T / H (Tracking-Maintenance) demodulation technology is used. When the circuit is operating at 150 kHz in the chopping frequency, the input equivalent noise reaches 31.12 nV / Hz. In the op amp design of the D audio amplifier, the signal's low harmonic distortion and the noise design format the design. For audio signals in the range of 20 to 20 kHz, the distortion of op amp is mainly caused by voltage disorders and low frequency 1 / F noise. The CMOS process relatively high 1 / f noise and voltage disorders make this problem are particularly serious. When the offset voltage of the circuit is required to be lower than 1mV and the input equivalent noise is less than 100 nV / Hz. Ordinary CMOS is difficult to meet demand. Common static disorders zero technology, such as Trimming, although it can eliminate the impact of voltage disorders, but cannot reduce 1 / F noise. The best way to solve this problem is to use dynamic offset-cancellation techniques such as automatic steady and chopping techniques. Auto Zero Tiechnique is sampled by samples low frequency noise and offset, and then subtracts them from the instantaneous value of the operational amplifier, achieving reduction in offset and noise. Since automatic steady zero technology is used as the principle of capacitive sampling, it is easily folded into the baseband frequency in the circuit operation, and the wider bandwidth of the op amp, the more noise on the sampling capacitor, usually 70 nV / HZ. Chopper Technique is the principle of modulation and demodulation, movement of low frequency noise and offset to the high frequency portion, using low-pass filter filtration, due to no thermal noise, the noise voltage is automatically stable The zero technology is lower. However, the influence of chopper switch charge injection and charge feedthrough effects can still produce a residual voltage disorder of 100UV. Moreover, the use of chopper switches, the thermal noise level of the device will increase. 2 The principle of working principle of chopper op amp is shown in Figure 1, where Vin is an input audio signal, a chopping switch modulation of the frequency of fch, amplitude is 1, according to Nyquist sampling principle, in order to Avoid the aliasing of the letter input signal, the FCH must be much greater than 2 times the signal bandwidth. Figure 1 After the principle of chopping is modulated, the signal is moved to the width harmonic frequency of the chopped square wave. This signal is enlarged by an operation amplifier that is gain AV. At the same time, the input noise and input offset voltage are also amplified. The output signal is 1, the frequency is 1, the frequency is the chopper switch modulation of the FCH, the output signal is: As can be seen from the formula (1), after the second chopping, the input audio signal is demodulated to the low frequency segment, and the voltage offset and low frequency noise are only moved to the chopped square wave after only one modulation. On the frequency harmonics, after low-pass filtering, the high frequency component in the output signal is filtered out, and the low frequency component is reduced to an audio signal, thereby achieving precise amplification of the audio signal. Fourier analysis of the output signal is obtained, and the final input noise spectrum density (PSD) of the op amp is obtained is: where the coefficient k is related to the noise parameters of the process. 3 The design of the op amp circuit This article is designed for the CMOS full circuit structure. Composed of chopper switches, main delivery circuits, output stage, and common-mode feedback circuits. The operating voltage range of the circuit is 2.5V ~ 5.5V. The circuit structure of the chopper operation amplifier is shown in Figure 2. Figure 2 The circuit structure of the chopping op amp input chopping switch completes the modulation of the audio signal, and the chopping switch is introduced into the residual voltage offset in the upper edge and the lower edge of the clock, and Fig. 3 is the waveform of the residual offset voltage at zero input. Figure 3 (a) Residual offset voltage (B) Chopper signal Figure 4T / H demodulation and control timing by analyzing the CMOS switch characteristic analysis can result in an equivalent input residual offset voltage is Vos, RMTS = 2VSPIKETFCH, where T is MOS switch The time constant can see three methods from this formula: 1. Reduce the chopping frequency: 2. Remember the input resistance; 3. Reduce the charge injection effect of the chopper switch. Since the corner frequency of MOS tube 1 / F noise is generally above several tens of kHz, reduce the chopping frequency of 1 / F noise, and the input resistance is only related to the internal resistance of the signal, it is difficult in the design. Reduce the input resistance, so only the charge injection effect of the switch can be considered. For this purpose, the chopping switch is used to use a complementary clock structure, and the minimum line width is used in size. On the one hand, the transmission on the conduction resistance can provide a larger voltage swing; on the other hand, the charge injection and feedback are reduced. The impact reduces residual voltage disorders. Taking into account the 1 / f noise characteristics of the PMOS tube than the NMOS tube, the input tubes MP1 and MP2 use large-area PMOS tubes, which can reduce the voltage disorders caused by the mismatch of the device, but also reduce transistor 1 / F noise. The corner frequency, improve the noise characteristics of op amp. To minimize residual voltage disorders, the output of Fold-Cascode op amps uses T / H demodulation technology, circuit structure, and timing as shown in Figure 4. The working principle of the circuit: When the tracking signal is closed, K5 ~ K8 is turned off, the output signal is held on the capacitances C1 and C2, when the circuit is output, K1 ~ K4 is broken, K5 ~ K8 closed, C1 and C2 The voltage value is loaded to the load capacitor C3 for sum. Since the voltage on C2 is superimposed to the load capacitor, the reverse direction is reversed, so the residual voltage disorder of the amplifier can be effectively canceled. Since the demodulator uses high-resistance points. Therefore, the smaller area of NMOS tube switches can be used, and the effect of reducing the output pole point can be reduced. The main transportation uses a full-differential folding Cascode structure, in the structure of Class-D, due to frequent operation of the output power MOSFET large current, the generated electromagnetic interference (EMI) will form a strong ripple on the power source, in actual application It is found that when the chip operates under the 5V supply voltage, the power fluctuations caused by EMI can reach ± 2V, and the full differential structure can improve the power supply suppression ratio and common mode suppression ratio of op amps, and the impact of power supply noise and common mode noise. Moreover, the mirror pole is avoided, so that a stable characteristic can still be exhibited for larger bandwidth. In order to provide a higher gain and voltage output swing, add the common transport output stage after Fold-Cascode. Using secondary op amp. Analyze the frequency stability of the op amp. It is temporarily regardless of the influence of the chopping switch, it can be inferred that at least three LHP poles are present, which are the main pole WP1 introduced by the Miller compensation capacitance, and the output filter capacitor generated WPOUT. The non-polar WP3 introduced for the first non-primary pole, and the non-polar WP3 introduced by Folded-Cascode (source of Mn1), and the relationship between the three is wp1
