Spectrum Analysis Series: 1dB Gain Compression Point Overview and Testing
[Copy link]
Semiconductor devices are very dazzling stars in the modern electronic industry. They have made great progress in recent decades. With many advantages, they have been widely used in functional circuits such as control, conversion, amplification, and calculation, and have always been favored by people. If you love it, you must accept its shortcomings. Everything has its own shortcomings, and semiconductor devices are no exception. For active devices such as RF amplifiers involved in this article, nonlinearity is one of its shortcomings.
Nonlinearity is an unavoidable topic for RF active devices such as PA and LNA. Although it is unavoidable, we still hope to keep it at a relatively low level to reduce the impact on the system. There are many parameters to measure nonlinear characteristics, among which the 1dB gain compression point is usually a must-measure item. How nonlinearity is generated, why it causes gain compression, and how to test the 1dB gain compression point will be the focus of the following introduction.
1. How does nonlinearity arise?
One of the reasons why semiconductor devices are widely used is that they can be "controlled" and used by people. For a basic transistor, taking a field-effect transistor as an example, the on and off of the transistor can be controlled by controlling the power supply of the gate, and the magnitude of the current between the drain and the source can be controlled. Transistors can be used to design RF amplifiers, and the slope of the transistor transfer characteristic curve (transconductance) determines the gain of the amplifier to a certain extent. However, the transfer characteristic curve of the transistor is not linear, which means that the gain of the amplifier is not constant.
Figure 1. Transistor DC characteristic curve and signal amplification diagram
Taking the common source amplifier as an example, the gate is used as the input port of the AC signal, and the amplified signal is output by the drain. When the DC operating point Q is designed, the RF signal will be superimposed on the operating point voltage of the gate and then amplified and output. Figure 1 shows the typical transfer and output characteristic curves of the transistor and the schematic diagram of the process of AC signal amplification.
When the RF input signal is relatively small, if the DC operating point is selected appropriately, the area mapped to the transfer characteristic curve is close to linear, and the signal can be approximately considered to be linearly amplified. As the power of the RF signal continues to increase, the area mapped to the transfer characteristic curve gradually becomes nonlinear. At this time, the amplified waveform is significantly different from the input signal, and the distortion becomes more and more obvious. This distortion is not a linear distortion such as the overall amplification or reduction of the waveform, but a nonlinear distortion.
Taking Figure 1 as an example, consider an extreme case: Assuming that the DC operating point has been selected, when the input signal increases to make the gate-source voltage Vgs lower than the turn-on threshold voltage at some times, the transistor will be turned off at these times, and of course there will be no output waveform, so the output waveform will be seriously distorted, which also means that the amplifier is already in a serious nonlinear working state!
If the DC operating point is selected appropriately, the amplifier may not have obvious nonlinear distortion within a wide input power range; however, if it is selected inappropriately, such as close to the turn-on threshold voltage, then even if the input signal is small, there may be obvious nonlinear distortion.
Speaking of this, the factors that affect the linearity of the amplifier mainly include: nonlinear transfer characteristics, the choice of DC operating point and the strength of the input signal.
The nonlinear distortion mechanism of semiconductor devices may be far more complicated than the above introduction, but the above three factors are at least part of the reasons. This also allows RF engineers who have always talked about nonlinearity to take one step closer to the goal of "knowing what it is and why it is."
Nonlinear characteristics are not completely useless. They are indeed harmful to amplifiers, but some devices use this characteristic to achieve specific functions, such as multipliers, mixers, and frequency doublers. The key depends on the specific application.
2. What is the 1dB gain compression point?
The output signal of a nonlinear circuit can usually be expanded using a Taylor series, or more specifically, a Maclaurin series:
Where, v in (t) and v out (t) are the input and output signals respectively, and c i is the constant coefficient of the series expansion.
For an amplifier, what are its output characteristics when a single-frequency signal is fed into it?
Let v in (t) = V 0 cos ω 0 t , and substitute into the above formula we can get
The individual terms in the above formula can be expanded as follows
For even powers, the product and difference only contain DC components and even harmonic components, which can be abbreviated as 2i· 0 , where i is a non-negative integer;
For odd powers, the product-sum-difference conversion results in only odd harmonic components (2i+1)· 0 , where i is a non-negative integer.
It can be seen from this that due to the nonlinear characteristics of the amplifier, when a single-frequency signal is input, its output signal will not only contain the amplified original frequency signal, but also regenerate some new frequency components. Frequency regeneration has also become a feature of nonlinear distortion.
For the frequency component 0 , the output term is
After Taylor series expansion, the higher the order, the smaller the constant coefficient c i . Therefore, for convenience, only the first three terms are considered in the above formula, and the higher-order terms are ignored.
The voltage gain of the amplifier is
The gain can be divided into two parts: the linear amplification gain, and the gain contributed by the input signal.
G = c 1 + G (V 0 )
Theoretically, the amplifier is expected to be ideally linear. No matter how large the input power is, the output power increases linearly, that is, the gain is constant. But the fact is that when the input power is large, the amplifier will enter the nonlinear working area, resulting in gain compression. Therefore, G(V 0 ) in the above formula is less than 0.
When the input power is low, the gain part related to the input signal tends to 0, and the gain at this time is close to linear gain. As the input power continues to increase, the output power increases approximately linearly. When it increases to a certain extent, the nonlinearity becomes more and more obvious, resulting in gain compression, and the output power growth rate slows down, and finally gradually tends to saturation and stability. The relationship between input and output power is shown in Figure 2.
Figure 2. Amplifier gain compression characteristics and 1dB gain compression point.
How to interpret the curve relationship in Figure 2?
First of all, it should be made clear that this is a logarithmic representation of input and output power, so the slope of the curve of an ideal linear amplifier is 1, as shown by the dotted line in the figure. The solid line represents the relationship between the actual input and output power of the amplifier. As the input power increases, the output power increases more slowly, deviating more and more from the ideal curve, and the gain gradually decreases. This is the gain compression effect.
P out (dBm) = P in ( dBm) + G (dB)
Active devices such as amplifiers usually focus on the position where the actual gain drops 1dB from the linear gain, which is called the 1dB gain compression point. The input and output powers corresponding to this point are generally marked as P 1dB,in and P 1dB,out respectively .
The higher the 1dB gain compression point, the better. When comparing two amplifiers, the one with a higher compression point has better linearity. In wireless communication systems, signals usually have a certain bandwidth. If the total power is close to the 1dB compression point, the amplifier nonlinearity becomes more obvious, and stronger harmonics and intermodulation products will be generated, thus causing interference to adjacent bands or within the band. Therefore, nonlinear distortion is a very important consideration in amplifier design.
3. How to test the 1dB gain compression point?
We have briefly introduced the basic content of gain compression caused by nonlinearity. So how do we test the 1dB gain compression point in practice?
The test method for 1dB gain compression point is more flexible. It can be automatically tested by vector network analyzer, and the compression point during one-dimensional/two-dimensional scanning can be automatically tested. It can also be manually tested based on a signal source, spectrum analyzer or even a power meter. This will be the method that will be emphasized below.
There is a certain deviation between the output power of the signal source and the set value, and the cables, attenuators and other accessories required for the test have certain losses. So do they need to be calibrated in advance?
Since P 1dB tests absolute power, the above factors still affect the test results. However, there is no need for comprehensive calibration. The P 1dB test idea introduced below is to first determine the output power setting of the signal source when the gain is compressed by 1dB, then record the set power and the power measured by the spectrum analyzer, and finally deduct the loss of cables and accessories to determine the input, output power and gain of the 1dB compression point.
It is worth mentioning that this test requires the signal source to have good linearity in output power, otherwise it will cause large test errors. Fortunately, common RF sources have good linearity in a wide power range.
Of course, in order to further improve the test accuracy, it is also possible to consider using a power meter to calibrate the signal source within a certain power range to further improve the output power linearity.
Figure 3. Typical connection diagram for 1dB gain compression point test.
During the actual test, the specific steps are as follows:
(1) Preparation before testing: Select RF cables, adapters and suitable high-power attenuators with good performance. For cables and adapters, it is especially important to ensure that the VSWR within the test frequency band is good. High-power attenuators are only required when testing PAs to protect the test equipment. For testing small signal amplifiers, the built-in attenuator of the spectrum analyzer is sufficient.
(2) Test connection: Complete the test connection according to Figure 3. If you are testing a PA, you need to introduce a suitable attenuator after it. At this time, make sure that the signal source has no RF signal output.
(3) Parameter setting: Set the frequency, power and power step of the signal source. The frequency is set according to the test frequency. It is recommended to set a low power first to ensure that the amplifier works in the approximately linear region. Set the center frequency CF and Span of the spectrum analyzer. Because the single-frequency signal is tested, it is recommended that the Span should not be too large. The setting of the internal attenuation needs to be considered comprehensively based on the input power, gain and external attenuation of the amplifier.
(4) Turn on the amplifier: Pay attention to the power-on sequence, especially for the PA, as detailed in the following notes. It is recommended to power on the gate first and then the drain.
(5) P 1dB test: Turn on the output switch of the signal source, adjust the reference level and attenuation of the spectrum analyzer, so that the CW signal spectrum is well displayed and a sufficient signal-to-noise ratio is ensured. Call up the peak marker of the spectrum analyzer and record the power P in,1 set by the signal source and the power P out,1 of the peak marker .
Continue to increase the power output by the signal source. You can use the navigation key of the signal source to gradually increase the power according to the step set in step (2) and observe the power measured by the spectrum analyzer. When the power of the signal source is adjusted to Pin ,n and satisfies the following formula
P in,n -P in,1 -(P out,n -P out,1 )=1dB
The first step of the P 1dB test is completed , but P in,n and P out,n are not yet the input and output powers of the 1dB gain compression point.
P out,n is the power measured after the amplifier output power passes through the downstream cables, attenuators and other accessories. Therefore, after compensating for these losses, P 1dB,out is obtained . P in,n is the power set by the signal source when the amplifier gain is compressed by 1dB. Remove the amplifier and directly use the spectrum analyzer to test the power on the amplifier input side, which is P 1dB,in . The amplifier gain corresponding to the 1dB compression point is
G 1dB =P 1dB,out -P 1dB,in
What should I pay attention to when testing P 1dB ?
First, the power-on sequence of the amplifier should be correct, especially the power amplifier. For GaAs and GaN amplifiers, most of them currently use depletion transistor design, which is turned on when the gate-source voltage is 0V. Therefore, the gate of this type of amplifier generally works in a negative voltage state. For this type of amplifier, for the purpose of protection, it is necessary to add a negative gate voltage first, and then add a drain voltage.
Secondly, when testing the PA, you must select a suitable attenuator based on its maximum output power to avoid over-power burning out the RF front end of the spectrum analyzer. The internal attenuator of the spectrum analyzer can usually only withstand a maximum power of 1W, so the power after passing through the external attenuator must be much less than this value.
Finally, considering the test accuracy, a suitable attenuator can be introduced before and after the amplifier to improve the input and output matching. For PA testing, since there are certain requirements for the driving power, the attenuation before the PA should not be too large to ensure that the power of the signal source after the attenuator can still drive the PA to work normally.
In addition, the spectrum analyzer should be prevented from entering the nonlinear region. You can refer to the spectrum analyzer's specification sheet to find its P 1dB . As long as the power fed into the spectrum analyzer during the test is at least 6dB lower than this value, the impact of the spectrum analyzer's own nonlinearity can be ignored.
There is also a simple judgment method. When the 1dB compression point is found, manually increase the attenuation of the spectrum analyzer. If the measured signal power remains basically unchanged, it means that the spectrum analyzer has no obvious nonlinearity. If the measured power increases when the attenuation is increased, it means that the RF front end of the spectrum analyzer has been compressed. It is necessary to further increase the attenuation until the measured power is stable, and then continue to increase the signal source output power to find the true 1dB compression point of the amplifier.
Finally, there is an open question: For PA, there may be a situation where the gain does not always show a monotonically decreasing trend when the excitation power increases. For example, as the excitation power increases, the PA gain may first increase and then decrease. So when calibrating the 1dB gain compression point, which gain (or power) should be used as a reference?
The above is what I want to share with you. I hope it will be helpful to you ~~
| 
提升卡
变色卡
千斤顶