Choosing an Amplifier

Uncommon or Moderate Difficulty
2U Difficulty

Introduction

When designing an RF transmitter or receiver, choosing an amplifier (or amplifiers!) is one of the most critical tasks for ensuring that it functions optimally. If it’s your first time, though, it can be difficult to parse through all of the information provided on datasheets to try and find an optimal part, especially if you don’t really know what your looking for. This article is going to focus on RF amplifiers, and how you can ensure that you’re choosing a good one for your transmitter or receiver.

To help you choose an amplifier, I’m going to go step-by-step through several of the more important characteristics you’ll need to know. This is by no means an exhaustive list, but in broad strokes you can go through this list one by one, in order, to narrow down your search.

  1. Frequency Range
  2. Characteristic Impedance
  3. Gain
  4. Is it a LNA or a power amplifier?
    1. Noise Figure
    2. P1dB
    3. Psat
  5. Power Added Efficiency & Heat dissipation
  6. Maximum Input Power
  7. Temperature Range/Stability
  8. Input/Output VSWR
  9. Operating Bias Voltage
  10. Operating Current
  11. Form Factor/Case Suitability

Prerequisite Reading

  1. Overview of Key Terms – Antennas
  2. Noise Figure and the Friis Formula for Noise
  3. Noise Floor

Important Information

It’s important to note a few assumptions before getting into the weeds. In this article I’m focusing on amplifiers for the most common types of transmitters/receivers used on CubeSats today, single center frequency, with constant envelope modulation schemes. If you’re planning on building a multi-frequency system or one with a modulation scheme that isn’t constant envelope (like On-Off Keying or Amplitude Modulation), you’ll probably need to do a bit of extra legwork in order to make sure the parts you’re choosing will satisfy your requirements. That being said many of these terms and factors of merit are still relevant, so read on, just be aware that this article isn’t going to go as in-depth as you may need it to.

When explaining a few of these items on the list above, I’ll be referencing a Mini-Circuits component datasheet for ease of understanding, referenced in the “References” section at the bottom. It’s important to note that not all of these parameters will always be available in every part’s datasheet. Sometimes you’ll have to calculate them yourself based on other information that’s in the datasheet, and sometimes manufacturer’s just don’t include what you’re looking for. Here I want to emphasize, as with all things in the CubeSat world, the importance of testing. Always make sure you have time to test the device you’re building thoroughly, as the real world doesn’t always jive perfectly with what you find in datasheets!

1. Frequency Range

The frequency range is the most important and easiest way to narrow down your search for an amplifier, and it’s as simple as it sounds. Determine the operating frequency of your transmitter/receiver, and make sure that it’s within the operating frequency range of the radio you’re trying to build. On the example datasheet shown in Figure 1, note the frequency range circled in green and marked with a “1”.

Figure 1: The introductory section of an amplifier datasheet

2. Characteristic Impedance

The characteristic impedance (or Z0) of an amplifier is the impedance which it’s designed to be fed by and output to. If you don’t remember what an impedance is, check this article for a refresher. Almost all high frequency electronic components these days are designed to have 50 ohms as their characteristic impedance. 75 ohms was once commonly used for things like broadcast television sets, but it’s becoming rarer. Because of their ubiquity and low cost, I’d recommended that you stick with 50 ohm components unless you have a very pressing reason to switch. The characteristic impedance is circled in Figure 1 and labeled with the number “2”.

3. Gain

An amplifier’s gain describes the ratio of the signal power output from the amplifier to the input power while the power is within the amplifier’s linear region (more on that in section 4).

G = \frac{P_o}{P_i}

The equation above is for linear gain G, where output power is P_o and input power is P_i. You can use any units for power, as long as input and output power use the same units. Gain, being a ratio, is a unitless quantity.

As with most quantities in communications, you’ll often see gain listed as a logarithmic quantity with units of decibels (dB). You can convert from linear gain to logarithmic gain using the equation

G_{dB} = 10*log_{10}(G_{linear})

Correspondingly, you can convert from logarithmic gain to linear gain using

G_{linear} = 10^{\frac{G_{dB}}{10}}

The gain is labelled in Figure 2 with the number “3”.

Figure 2: Electrical specifications section of an amplifier datasheet

ADDENDUM ON DATASHEETS

In the specifications section (Figure 2), note that there are several columns. This section is a quick explanation on each.

Condition (GHz) – The frequency the parameter is measured at (in this case, in GHz).

Vd=5.0V/Vd=3.0V – Indicates that values were measured using a 5 Volt or 3 Volt bias voltage

Min./Typ./Max. – Minimum/Typical/Maximum measured values within acceptable production variation range. You can normally design using the typical value, but beware that each part may have some variance within the minimum/maximum range.

Units – The units of the measured value.

4. Use-Case Dependent Quantities

Amplifiers are used in many places in a CubeSat communication system. Two of the most common uses are Low-Noise Amplifiers (LNAs) in receivers and Power Amplifiers (PAs) in transmitters. The actual construction of amplifier circuits is beyond the scope of what we cover on Crashcube, but essentially you’ll need to know two things. LNAs are designed, as you might expect based on their name, to have a low noise figure. PAs, on the other hand, are designed to have their linear regions extend up to higher output power values.

4-1. Noise Figure

All amplifiers are inherently imperfect, and each will introduce some amount of noise into the signal it’s amplifying. The noise figure of an amplifier is a quantity that describes how much noise is in the signal at the output of the amplifier relative to the noise at the input. This is quantified as a signal to noise ratio (SNR). The noise figure of an amplifier is defined as the ratio of the input to output signal-to-noise ratios (a ratio of ratios, I know that’s a little bit confusing). Similarly to the way gain can be expressed as a logarithmic quantity, noise figure is also often expressed in decibels.

NF = \frac{SNR_i}{SNR_o}
NF_{dB} = 10*log_{10}(\frac{SNR_i}{SNR_o}) = 10*log_{10}(SNR_i) - 10*log_{10}(SNR_i)

If you look closely, you’ll notice that noise figure has the input and output values inverted compared to the equation for gain. This is because we want to have positive values for noise figure (due to both common convention and to make some of the equations we use it in easier to manage), and because an amplifier will ALWAYS have a lower SNR at its output compared to its input.

Why is this, though? It’s important to remember that, while we distinguish between the signal and the noise when doing analysis, to an amplifier they both appear together as a single input current and voltage. There’s no way for it to distinguish or separate the two. This means that when a signal (with some amount of noise) goes into an amplifier, it amplifies both signal and noise equally. Since in reality, all amplifiers will also add some additional amount of noise to the signal they amplify, this means that the noise value at the output will ALWAYS be higher than at the input.

The noise figure of an amplifier is more important in the case of LNAs than it is for PAs. This is because typically an LNA is being used in a receiver to amplify a very weak signal that is closer to the noise floor. As an example, let’s say your input power is -90 dBm and you have a noise floor power of -110 dBm. If you have an amplifier with a noise figure of 12 dB, your output SNR after amplification is only 8 dB. Depending on what modulation and error correction methods you’re using on your satellite link, this may not be enough to decode it properly! If you use an LNA with a lower noise figure, such as 3 dB, it may be possible to decode that same signal properly since its output SNR would then be much higher at 17 dB.

For a PA, this is less important because the input power to the amplifier is often far higher than the noise floor. As an example, a typical power amplifier might have 30 dB of gain and be fed with a signal that’s been generated by a transceiver at 0 dBm. If the noise floor power for this example system is -80 dBm before amplification, an amplifier with a noise figure of 12 dB would increase the signal power to 30 dBm and the noise power to -38 dBm. This increase will have very little impact on the quality of the transmitted signal. The SNR in this case would be reduced from 80 dB at the input to 68 dB at the output.

Once the signal travels all the way across space to a receiver on the ground, both the signal and the noise will be greatly reduced in power due to geometric spreading out of the power density. Let’s assume that the geometric spreading losses are 120 dB and that the receiver has a similar noise floor power to the one in the first half of this example, -110 dBm. Our hypothetical receiver now has a signal input power of -90 dBm, above the noise power floor, but the noise that was amplified by the PA at the beginning is now at -158 dBm, far below the level that is inherently within the receiver from external noise sources.

Table 1 below shows numerically what was described above.

 Input to PAOutput of PASpreading LossInput of Receiver
Signal Power0 dBm30 dBm-120 dB-90 dBm
Noise Power-80 dBm-38 dBm-120 dB-158 dBm
SNR80 dB68 dB68 dB
Table 1: Example of how noise power from a PA gets reduced below receiver noise floor
4-2. P1dB Compression Point

Amplifiers are generally linear devices, but only up to a point. By a linear device, I mean that the gain of an amplifier is not affected by the input power of the signal. For a 20 dB amplifier whose linear range encompasses both, for example, a -50 dBm input signal and a -20 dBm input signal should both be amplified equally (to -30 dBm and 0 dBm output, respectively).

The linear range of an amplifier is defined as the range of output power values where the amplifier will operate as a linear device. Typically, an amplifier will operate linearly for all power values up to the point where it becomes compressed (i.e. there’s no relevant lower bound other than the noise floor). The P1dB compression point is used as the upper bounding point of the linear region, and refers to the output power when the amplifier loses 1 dB of linearity (also sometimes denoted as OP1dB).

Figure 3 below illustrates the linear region and where it breaks down near the P1dB compression point, becoming non-linear. You’ll notice that on this plot there are a few other points labelled, the second-order and third-order intercept points. These (particularly the third-order intercept, IP3) are used for measuring and quantifying the linearity of devices. They are more complicated to explain conceptually than the P1dB compression point, so I won’t do so in this article, however I highly recommend this video from Rohde & Schwarz [3] that I think explains them very well.

Plot of power output linearity (Y-axis) vs. input power (X-axis) [2]

The P1dB compression point is more relevant to power amplifiers used in transmitters than it is for the amplifiers used in receivers. This is because in almost all cases you’ll want to have a high transmit power to make it easier to close your link budget. It’s important to choose an amplifier whose P1dB compression point is at or above your desired output power so it won’t be stressed trying to amplify an already too-strong signal. Stressing parts more than they need to be is a quick and easy way to cause them to fail, and for something as critical as your transmitter, having an amplifier fail could mean your whole satellite fails!

4-3. Psat

Sometimes datasheets for amplifiers will also include a value P_{sat} that refers to the saturated power point. This is the absolute maximum output power level of an amplifier, where the slope of the input-output curve becomes horizontal. It is always above (at a higher power level than) the P1dB compression point.

While it’s generally preferable to operate an amplifier below the P1dB point, if using a modulation with a constant power envelope (such as BPSK or QPSK, among others), it is possible to operate closer to the saturation region with careful design. Generally though, it’s better to avoid it when designing smallsat systems. The extra margin of error you get from being a bit more conservative is almost always worth it when it comes time to test your devices.

5. Power Added Efficiency (PAE) and Heat Dissipation

Power Added Efficiency refers to how efficient the amplifier is while taking into account the gain of the amplifier, and can be described by the below equation, where P^{RF}_{out} is the output RF power, P^{RF}_{in} is the input RF power, and P^{DC}_{in} is the DC power consumption/dissipation. It’s important here to note that P^{RF}_{out} - P^{RF}_{in} will NOT equal the gain of the amplifier! This is because you need to use linear power units (such as milliwatts, Watts, etc.) instead of logarithmic ones for this equation to be valid.

PAE = 100 \cdot \frac{P^{RF}_{out} - P^{RF}_{in}}{P^{DC}_{in}}

In the datasheet in Figure 2 I’ve circled the absolute maximum power dissipation and labelled it with a “5”, which you can use as a conservative estimate for the DC power consumption when calculating PAE. PAE is important primarily because a cubesat will have a limited ability to generate power using solar cells and it’s important to use that power efficiently, but it is also important for another reason. Every milliwatt of heat that’s generated by using power has to be managed by your cubesat’s thermal control system. It could be a good or bad thing to generate more heat, depending on your design, but either way it’s an important systems engineering consideration.

Generally, PAE and heat dissipation are more relevant to the choice of PAs for transmitters versus LNAs for receivers. It is important in both cases to be efficient, but LNAs will often be consuming much less power overall, so an inefficient ampifier using 100 mW of DC power will be much less of a concern than one using 5 W of DC power.

As a good rule of thumb, a power amplifier with reasonable PAE for a cubesat transmitter will be somewhere in the 20% to 40% range. Amplifiers will also generally be more efficient at lower frequencies vs. higher ones. You may be able to find an inexpensive amplifier in the VHF band with a 40% PAE relatively easily, but amplifiers for S-band above 25% may be much more expensive or difficult to locate.

6. Maximum Input Power

In Figure 2, the maximum input power is circled and labelled with a “6”. Maximum Input Power (CW) refers to the maximum power input that an amplifier can sustain without damage. This is a value that only indicates whether or not an amplifier will be damaged by an input signal at that level. It does not indicate whether or not an input signal will be effectively amplified without distortion or intermodulation problems, so using this value together with the gain can’t be used as a shortcut to properly determining whether or not your signal will be compressed.

The (CW) in the label refers to power using a Continuous Wave modulation scheme. This means that the power is valid for modulation schemes have a consistent amplitude, such as phase modulation and frequency modulation. Modulation schemes that vary the amplitude of the signal are referred to as pulsed wave or pulsed modulation schemes (like on-off keying or the amplitude modulation used for AM radio). Most modulation schemes used for satellite communication are continuous wave, though, since they tend to be more spectrally efficient. Pulsed modulation likely isn’t something you’ll need to worry too much about unless you’re tracking aircraft using ADS-B [4].

7. Temperature Range

As mentioned in Section 5, thermal considerations are important for cubesat systems engineering, and therefore must be taken into account when choosing parts. Both of the temperatures labelled as “7” in Figure 2 refer to the temperature at the mounting points or leads of the package, not the temperature of the package itself. There are two temperatures listed.

The Operating Temperature is the range of temperatures that the amplifier can safely operate at while powered on and in use.

The Storage Temperature is the range of temperatures that the amplifier can be exposed to while powered off with no current flowing through it.

8. Input/Output VSWR

The VSWR, or Voltage Standing Wave Ratio, is an important characteristic for determining how much of the power of a signal is transmitted through an amplifier, and how much is reflected at its input and output ports. In figure 2, it is labelled with an “8”. For a more detailed explanation on VSWR, see this article.

Essentially, for more optimal matching and therefore power transfer, we want the VSWR to be as close to 1:1 as possible. A value of 1.5:1 corresponds to 96% of the power being transferred to the amplifier, and 4% being reflected. A value of 2:1 corresponds to approximately 90% throughput, 10% reflection. A value of 3:1 corresponds to approximately 75% throughput, 25% reflection. To calculate this value, you can use this handy online calculator from All About Circuits, and then convert the Mismatch Loss from dB to a linear value [5].

This is an especially important value if you have multiple amplifiers and are using them together. If the reflected portion of the signal bounces between the output of the first amplifier and the input of the second, you can quickly end up with an unstable system that leads to overload and failure. This is why engineers will often put a small value attenuator (1 or 2 dB) of the same characteristic impedance as the amplifiers between them. While it does slightly decrease efficiency, it’s an easy way to impedance match the two amplifiers better over a wide range of frequencies and often doesn’t waste much power or decrease link budget significantly, due to the way that system noise temperature is cascaded [a more detailed article on this subject is in progress].

9. Operating Bias Voltage

The operating bias voltage is the DC voltage that will be applied to the amplifier in order to power it. Most of the time, an amplifier will expect a very particular voltage, and so the supplied bias voltage must be regulated to that value.

Power subsystems can be designed in several ways, either providing regulated or unregulated voltages to the various units/boards that make up the subsystem. If you have certain regulated voltages available, it is often preferable to choose an amplifier that uses one of those voltages, as otherwise you’ll have to convert the voltage a second time. Each voltage conversion stage introduces inefficiencies, and if you are using a switched mode converter like a buck, boost, or buck-boost, you can also introduce noise into your signal that will have to be filtered out.

The operating bias voltage is labelled in Figure 2 with a “9”.

10. Operating Current

Operating current is the current that the amplifier draws when operated at a bias voltage. This is an important consideration, since the power subsystem will have to be able to supply an appropriate amount of current, and the PCB you design for the radio will likewise need to handle it appropriately. In most cases, you won’t require more than 1 or 2 Amps for power amplifiers and even less for LNAs. This is marked in Figure 2 with a “10”.

11. Form Factor/Case Suitability

The form factor of the amplifier tells you what size it is, how many electrical connections it requires, and what the layout and dimensions of those connection points to a PCB are, also called a footprint.

The device footprint is labelled as “11” in Figure 1 at the top of the page. Many components will use industry standard packaging, and be described simply as such. This component, for example uses the SOT-343 package. The benefit of using these standardized packages is that many devices that have a similar size and number of electrical contacts, even if they have different functions, can use the same footprint. This makes life much easier for PCB designers using the components, as they don’t have to draw a custom footprint for each and every part.

More information on the footprint of most standard packages can be found online. The SOT-343 package has information available from Infineon Technologies, for instance [6].

Conclusion

I hope this article was helpful for any aspiring cubesat communication engineers. If you find any errors or important omissions in the information presented here, we encourage you to submit feedback via the comments and we will endeavour to correct anything that needs correction or clarification. Happy orbiting!

References

  1. Mini-Circuits PSA4-5043+ Low Noise Amplifier Datasheet. Retrieved on Feb 3, 2020. URL: https://www.minicircuits.com/pdfs/PSA4-5043+.pdf
  2. RF Wireless World, “IIP3 vs. OIP3”. https://www.rfwireless-world.com/Terminology/IIP3-vs-OIP3.html
  3. Rohde & Schwarz, “Understanding Third Order Intercept”. https://youtu.be/m-2H8ddSwTI
  4. https://www.mathworks.com/help//comm/ug/airplane-tracking-using-ads-b-signals.html
  5. All About Circuits, “VSWR/Return Loss Calculator”. https://www.allaboutcircuits.com/tools/vswr-return-loss-calculator/
  6. Infineon Technologies, “SOT343 Package Overview”. https://www.infineon.com/dgdl/SOT343-Package_Overview.pdf?fileId=5546d462580663ef015806a5338d04ef

Was this helpful?

0 / 0

Ian Bennett
Latest posts by Ian Bennett (see all)
Leave a Reply 0

Your email address will not be published. Required fields are marked *