On Open Loop Closed Loop Class D Amplifier
Time:2022-11-18
Views:1473
PSRR: About Open Loop Closed Loop Class D Amplifier
In the past, power supply rejection ratio (PSRR) has become an excellent measurement method to measure the performance of amplifier in suppressing power supply output noise. However, due to the emergence of more and more Class D amplifiers and their efficiency advantages, it is far from enough to rely solely on PSRR as an indicator of power supply noise suppression performance. This is more obvious than the PSRR specification of the open-loop closed-loop digital input I2S amplifier. In many cases, the PSRR specification is the same, but when listening to amplifiers that are lower than the ideal power supply, there will be obvious differences in audio performance. This paper gives an overview of the traditional PSRR measurement method, explains why it cannot completely capture the power rejection performance of Class D amplifier in the bridge load (BTL) structure, and introduces an alternative method to measure the power noise impact in Class D amplifier.
To understand why the PSRR measurement method cannot fully capture the power rejection performance, we need to review the era when Class AB amplifiers dominated consumer audio electronic devices. As is the case today, Class AB amplifiers are generally configured in a single ended (SE) or BTL output structure. In fact, it is very common for SE AB amplifier to have split rail power supply (i.e. ± 12V), because the power supply is mainly based on transformer, and the cost of adding the second rail is not too high. BTL structure is more used in audio systems without split rail power supply. Regardless of SE or BTL structure, Class AB amplifier itself has good PSRR, because of its basic architecture and output level that is usually much lower than the power rail voltage.
As far as Class AB amplifier is concerned, PSRR measurement method can relatively well show the power supply noise suppression performance of amplifier, while as far as SE structure is concerned, it requires particularly accurate power supply noise suppression performance of amplifier (we will discuss it in detail later). If we push the time forward, we will find that Class D amplifier was very popular in the market at that time. They have changed the market shape with extremely efficient operation, thus realizing considerable innovation in industrial design, especially in smaller size. However, compared with Class AB amplifiers, their architectures are fundamentally different, and their output structures are almost only BTL.
In the BTL structure, the Class D amplifier has two output stages, which are composed of four FETS (also called full bridge). The SE Class D amplifier has only one output stage, which is composed of two FETS (also called half bridge). Compared with SE structure, BTL output structure has many advantages, including 4 times of output power given in the case of power rail, better bass response, and excellent on/off clicking and popping performance. Some disadvantages of BTL architecture are that you need twice as many FET transistors. This means a larger silicon chip size and higher related costs, and the cost of reconstruction filters (LC filters) will double. In today‘s market, although SE and BTLD amplifiers can be seen, most of them are BTL.
In the D-type BTL structure, the traditional PSRR measurement method is powerless. In order to better understand the reason, we need to understand the working principle of Class D amplifier and how PSRR is measured. Class D amplifier is a switching amplifier, whose output performs rail to rail switching at extremely high frequencies (usually 250kHz or higher). The audio signal is used for Pulse Width Modulation (PWM), the switching frequency (square wave). Then, the reconstruction filter (LC filter) is used to extract the audio signal from the carrier frequency. These switching architectures are extremely efficient (the same structure is used in some switching power supplies), but they are more sensitive to power noise than traditional Class AB amplifiers. After careful consideration, it is not difficult to find that the output of the amplifier is essentially a power rail (pulse width modulation), so all power supply noise is directly transmitted to the amplifier output.
Power supply rejection ratio (PSRR) is a measure of the performance of amplifier in suppressing power supply noise (ripple). It is an important parameter when selecting audio amplifiers, because audio amplifiers with low PSRR generally require higher cost power supplies and/or large decoupling capacitors. In the consumer electronics market, power cost, size and weight are important design considerations, especially when product size is shrinking, prices are falling rapidly, and portable design is becoming increasingly common.
In the traditional PSRR measurement method, the power supply voltage of the amplifier consists of a DC voltage and an AC ripple signal (Vripple). The audio input is AC grounded, so there is no audio signal when measuring. All supply voltage decoupling capacitors are removed so that Vripple is not artificially attenuated (see Figure 1). The output signal is then measured and the PSRR is calculated using Equation 1:
However, this traditional PSRR measurement method makes the power supply noise obviously exist at the output, and the reconstruction filter exists before and after. The PSRR measurement method does not give us any indication. The reason why the PSRR measurement method fails is that the input AC is grounded during measurement. In reality, the amplifier will play music, which is where things start to get interesting.
Yle="PADDING BOTTOM: 0px; WIDOWS: 2; TEXT-TRANSFORM: none; TEXT-INDENT: 0px; MARGIN: 0px 0px 20px; PADDING LEFT: 0px; PADDING RIGHT: 0px; FONT: 14px/25px Tahoma, arial; WHITE-SPACE: normal; ORPHANS: 2; LETTER-SPACING: normal; COLOR: rgb (0,0,0); WORD-SPACING: 0px; PADDING-TOP: 0px; - webkit text size adjust: auto; - webkit text stroke width: 0px">When playing audio, The power supply noise is mixed/modulated with the incoming audio signal, and the resulting distortion is transmitted to the audio frequency band in varying degrees. The inherent cancellation effect of BTL structure can no longer eliminate noise. The industry has given this phenomenon a very vivid name: Intermodulation Distortion (IMD). IMD is the result of mixing two or more different signal frequencies together, which generally does not form some additional signals for the frequency of any harmonic frequency (integer multiple).
Before discussing how to remedy some of the shortcomings of the PSRR measurement method, let‘s first discuss the feedback function. If you have been drinking coffee and following the discussion in this article, you will not be surprised by some power noise problems of Class D amplifier itself. If it is not a feedback function, it is a serious problem. (In high-end audio applications, open-loop amplifiers sound good, but that‘s another case. They generally have very stable, high-performance power supplies and extremely high cost targets.) To compensate for power supply noise sensitivity, the designer will design a system with a highly stable power supply (which will increase the cost), or use a Class D amplifier with feedback function (also called a closed-loop amplifier).
Today, most analog input Class D amplifiers in the consumer electronics market are closed loop. However, the digital input I2S amplifier is another case. The I2S amplifier is directly connected to the audio processor or audio source through a digital bus. By eliminating unnecessary digital to analog conversion, not only the cost can be reduced but also the performance can be improved. Unfortunately, there are not many closed-loop I2S amplifiers on the market today, because it is very difficult to build a feedback loop that samples the PWM output and adds it to the input I2S digital audio stream. In an analog feedback system, you can add analog outputs to analog inputs, making implementation easier. However, with the development of I2S market, most I2S amplifiers should follow the same development path as analog input amplifiers and adopt feedback architecture.
Obviously, PSRR is not an effective method to measure the power supply rejection performance of BTLD class amplifiers. So, what should we do next? Or return to the vivid voice term intermodulation. We need to measure the intermodulation distortion generated when playing audio and its corresponding THD+N variable curve. Before doing so, let‘s move back to the SE architecture. In the SE architecture, no matter whether it is a Class AB, Class D or Class Z amplifier, you will not get the offset effect of the BTL architecture, because one end of the speaker is connected to the amplifier and the other end is grounded. Therefore, in the SE architecture, the traditional PSRR measurement method has a better indication of power supply noise suppression, regardless of whether it is Class AB or Class D amplifiers.
Now, let‘s go into the laboratory and get some data. The following is a series of measurement methods, in which we analyzed and compared the power supply ripple IMD in an open loop and a closed loop I2S amplifier. A 1kHz digital tone is injected into the input of the amplifier, and a 100Hz, 500mVpp ripple signal is injected into the power supply. The IMD is observed by using a differential output FFT with an audio accuracy built-in FFT function.
The experimental results show that there is almost no sideband when measuring the IMD of a closed-loop I2S amplifier at 1 kHz input signal. This feedback loop is excellent at suppressing intermodulation distortion.
Another experiment showed the same IMD measurement method, but this time it was for an I2S open-loop amplifier. The 900 Hz and 1.1 kHz sidebands are very obvious because there is no feedback to suppress IMD.
But in terms of audio quality, IMD is not a simple measurement method that can give you many qualitative methods. One option is to conduct the same experiment, but now it is to measure the THD+N variable curve, which is exactly what we will do in the next two measurement methods. THD+N is measured with a 1kHz digital audio signal and 500mVpp power supply ripple. The ripple frequency of the power supply varies from 50Hz to 1kHz.
In Figure 2, observe the THD+N scanning of the open loop part under different power supply ripple frequencies. The red line indicates the amplifier performance without ripple in the power supply, which represents the ideal condition. Other curves represent ripple frequencies varying from 50Hz to 1kHz. Note that as the ripple frequency increases, the frequency bandwidth affected by distortion also increases. Please note that open-loop performance is better in a stable power supply environment, but it will increase costs and be at a disadvantage in today‘s highly competitive world of consumer electronics.
Observe the same THD+N scan as shown in Figure 3, but now for the closed-loop amplifier. Feedback will suppress intermodulation distortion, so you don‘t see any ripple noise impact on audio performance.
conclusion
In this paper, we reviewed the traditional methods of measuring PSRR, and explained that "?
The reason why it cannot measure the ripple effect of power supply in BTLD class amplifier. The inherent cancellation effect of the BTL output structure, coupled with the lack of audio signals during the measurement, produces a false reading. This is a serious disadvantage of the specification, because the power supply noise rejection performance is extremely important when selecting a Class D amplifier, especially when observing the performance difference between digital input (I2S) closed-loop and open-loop amplifiers. To obtain more accurate power noise suppression images, you need to inject a 1kHz audio signal into the input and noise into the power supply to study the performance of IMD and THD+N. Finally, we introduce how the closed loop Class-D amplifier can compensate the power supply noise, while the open loop amplifier cannot. In the highly competitive consumer electronics market, cost is the most critical, and closed-loop architecture can reduce system cost is a very important design consideration.
|
Disclaimer: This article is transferred from other platforms and does not represent the views and positions of this site. If there is any infringement or objection, please contact us to delete it. thank you! |
Business consulting
13823761625
Mail me