Why does phase noise matter? Consider the quadrature direct conversion receiver in Figure 1. Designers typically focus on the phase noise of the LO to determine the overall phase noise performance for a receiver. In the case of a direct conversion receiver, the LO phase noise gets down-converted to DC. This is due to the fact that the incoming RF signal has the same carrier frequency as the LO signal. If the RF and LO frequencies are identical, then the resulting IF is centered at DC. The baseband spectrum consists of the down-converted spectrum and the parasitic phase noise spectrum. This phase noise can cause problems for the receiver if the down-converted RF signal has information in the frequency domain near DC due to the LO phase noise (one way to solve this problem is to simply go to a low-IF receiver). Unfortunately, for a direct conversion receiver, there are other sources of noise in the signal path to contend with. Specifically, the 1/f noise of the baseband circuitry causes additional problems for the receiver. This article will discuss and analyze the device 1/f noise and other noise sources in the receiver (and transmitter) to understand how amplifier additive phase noise impacts the RF signal as it traverses the receiver signal path to the receiver output.
What is Phase Noise?
Phase noise is defined as the noise arising from the rapid, short term, random phase fluctuations that occur in a signal. In the frequency domain, amplitude modulated (AM) and phase modulated (PM) noise show up symmetrically about the center, or carrier, frequency (Figure 2). Phase noise is measured as a power spectral density (PSD), or spot noise, in units of dBc / Hz (dB below the carrier) in a 1 Hz bandwidth (BW) with respect to the carrier power. The lower the phase noise, the better the performance of the system. Figure 2 shows a carrier frequency (x-axis is units of frequency) and the carrier power (y-axis is units of power in dBm) for an amplifier. The carrier frequency is a single tone shown with a gray line. Superimposed on this tone is the additive phase noise associated with the amplifier (blue curve) and its impact on the carrier frequency. The phase noise is measured in a 1 Hz bandwidth in dBc / Hz relative to the peak carrier power, and is shown as level below the carrier in dBc. The additive phase noise is referenced to the carrier frequency, and is measured at a Hz offset from the carrier in dBc / Hz relative to the carrier power. Typically, phase noise is reported in decades of frequency relative to either a 1 Hz to 10 Hz offset from the carrier frequency.
Phase Noise Sources and Typical Application Requirements
There are many contributors to additive phase noise for an amplifier. Some of these contributors are summarized below.
- Output power of the amplifier
- Carrier frequency
- Offset frequency
- Technology used internal to the amplifier
- Biasing voltage
- Noise from DC source
- Number of cascaded amplifier stages
In general, phase noise can be characterized into AM and PM categories. In Figure 2, the effect of PM is characterized at an offset from the carrier, measured in Hertz. PM effects the amplitude (y-axis) and is measured at various frequency offsets (x-axis) from the carrier frequency (grey line). Figure 3 shows a breakdown of some of the categories of these noise sources.
White Noise and Pink Noise
Of particular interest for purposes of this discussion is the circled 1/f (flicker) noise and white noise. The physics of transistor devices causes them to have both 1/f noise and white, or thermal noise. 1/f noise is also referred to as “pink” noise. This is the additive phase noise to a carrier. It is independent of power and typically dominates the noise of a device in the 1 – 100 kHz frequency offset from the carrier frequency for an amplifier. “White” PM noise is inversely proportional to the power and typically dominates in the > 100 KHz frequency offset from the carrier frequency for an amplifier.
Contributions of the Power Supply
Another potential source of additive phase noise that warrants a mention is the power supply used for the gain block. If the power supply has a noisy spectrum, this noise source can also mix with the carrier frequency to cause AM-to-PM conversion and further degrade the phase noise frequency spectrum of the amplifier. Care must be taken to ensure that this situation does occur at the board level design for the amplifier. When multiple amplifiers are used in cascade in a transceiver line-up, or when an amplifier is used to boost the LO signal in a transceiver, the additive phase noise of each amplifier must be taken into account in the noise budget.
Effects of Device Architecture
As discussed above, the frequency of reference to note here is the 10 kHz offset frequency when comparing amplifier additive phase noise performance. Table 1 shows the spot noise density that is the amplifier PSD at the 10 kHz reference frequency, which increases as the device minimum gate length reduces. This is a result of device physics. The 1/f, flicker, or “pink noise” frequency corner increases as the gate length of the device decreases. Thus, as the technology node decreases from 0.5 µm to 0.25 µm to 0.15 µm , the 1/f noise corner of the device increases in frequency. For older discrete silicon based bipolar devices, the 1/f noise corner frequency for these devices can be < 1 kHz. Close-in additive phase noise corrupts all signals that pass through an active device and is a function of the device physics and the DC biasing. This baseband noise is translated directly onto the carrier, independent of carrier frequency. Far out phase noise is also a function of DUT noise figure and is dependent upon drive level.
For modern sub-0.1 µm FET-based technologies, the 1/f noise corner frequency can be close to the MHz range. This relationship can be seen through measurements or simulations of a minimum-gate-length device by itself. The frequency noise spectrum of an amplifier can be measured simply by terminating the amplifier in 50W at the input terminal and looking at the output noise spectrum of the amplifier on a spectrum analyzer. That being said, in the above table, the additive phase noise of the amplifier is measured at an offset frequency. Since the device 1/f noise is a low frequency noise source in a device, when a 2 GHz input frequency is applied to the amplifier, one would think this noise source shouldn’t be a consideration since it is out of the frequency band of interest, but that’s actually not the case. The reason for this is that there are multiple 1/f noise sources to consider in an amplifier. The voltage and/or current bias circuits for an amplifier have 1/f noise sources that can produce additive phase noise. In fact, the additive phase noise contribution for an amplifier works the same way as it does for some of the phase noise contributions for a VCO due to the DC bias circuit. Thus, there is similarity between phase noise contributions in a VCO and additive phase noise contributions in an amplifier due to the 1/f noise in the DC bias circuit. In order to get a better idea of how this works, consider the simplified schematic of a 50W matched gain block circuit in Figure 4. In the figure, VBIAS is the bias circuit that sets up the quiescent current in transistor Q1. Since VBIAS is a circuit made up of transistors in the same process technology as transistor Q1, VBIAS has 1/f noise. Q1 acts to mix the incoming carrier frequency with the 1/f noise of VBIAS and produce one source of the additive phase noise for the amplifier.
Based on the above discussion, a comparison can be made between measurements of the additive phase noise of a 2 µm HBT amplifier and a 0.5 µm InGaAs amplifier. The 1/f noise frequency corner of a 0.5 µm InGaAs device is much higher in frequency than a 2 µm HBT device. Thus, it is expected that additive phase noise of an amplifier based on the the 0.5 µm InGaAs device will be higher in frequency close in to the carrier as compared to an amplifier based on the 2 µm HBT device.
Typical Application Requirements
Mini-Circuits conducted a survey of customers, particularly those working in sensitive defense radar applications, and feedback indicates a good benchmark number for amplifier additive phase noise is -165 dBc at 10 KHz offset. With this feedback in mind Mini-Circuits has characterized additive phase noise for a number of MMIC amplifiers that meet these stringent requirements. A summary of findings from the customer survey is shown in Figure 5 and some of Mini-Circuits’ low phase noise amplifier products are listed in Table 1.
|-150 > X > -155||Fair|
|-155 ≥ X ≥ -165||Good|
|-165 > X||Excellent|
Measuring Phase Noise
Mini-Circuits has evaluated several pieces of test equipment from different manufacturers to measure phase noise. Phase noise measurement repeatability with regards to signal carrier frequencies and applied signal power is notoriously challenging due to multiple factors, and the data is therefore only as good as the test setup. Mini-Circuits’ engineering team took great care in developing a test procedure to minimize effects that can compromise the integrity of the measurement.
Noise from Power to and from the DUT
Measurements can be dependent upon the input/output power to/from the device under test (DUT). What is needed, ideally, is a “noiseless” way to vary and measure the DUT input power and measure the resulting DUT output power, as well as the power being fed back to the test equipment’s phase detector. In order to minimize these noise issues, Mini-Circuits uses mechanical step attenuators and passive, high-directivity couplers.
Mechanical Vibration and External EMI
Phase noise test equipment and measurements are also extremely sensitive to mechanical vibration and external EMI effects. To reduce these effects, Mini-Circuits uses a test setup with an EMI shielded enclosure with vibration damping to help reduce the impact of these issues on the measurements.
AC Power Line Noise
AC power line noise can adversely affect the measurement at low offset frequencies. Mini-Circuits uses AC line filters to mitigate line noise.
Moreover, no single piece of test equipment performs optimally across all offset frequencies. Mini-Circuits has selected selected equipment that performs best in the 10 kHz to 100 kHz region in concert with the noise mitigation techniques described above. This setup has allowed Mini-Circuits to offer reliable amplifier additive phase noise measurements for amplifiers. Additive phase noise specifications for several Mini-Circuits amplifier models at different offset frequencies are shown in Table 1 below, including near-in and wideband (> 1 MHz) noise. The carrier frequency used for these measurements is either 1 or 2 GHz, as shown under “F Input.”
|DEVICE||Technology||F_Input [GHz]||PIN [dBm]||L@100 Hz [dBc/Hz]||L@1kHz [dBc/Hz]||L@10kHz [dBc/Hz]||L@100kHz [dBc/Hz]||L@1MHz [dBc/Hz]|
|PMA3-83LN+||0.25 um InGaAs||2||-6||-128||-147||-155||-161||-167|
Amplifier additive phase noise is a significant contributor to transceiver phase noise performance and designers are increasingly looking for phase noise specifications before choosing an amplifier for their systems. Mini-Circuits’ engineering team has a deep understanding of phase noise and its sources, and the company has invested in the capability to measure phase noise in amplifiers and provide reliable data to customers for a variety of applications. This article has explored the basics of phase noise in amplifiers and some of its common sources. We’ve discussed the special challenges of making reliable phase noise measurements and showcased some of the commercial amplifier products in our portfolio that exhibit phase noise performance for the most demanding applications.