The quest for the ideal rf amplifier.
The basic function of a power amplifier is always the same: to boost a low-power signal to a higher power level, to be delivered to the amplifier load. Because that role is so fundamental, it's tempting to view amplifiers as simple black-box devices, with an input, an output, and a constant amplification factor. In many instances, the black-box approach provides an adequate picture.
It fails, however, when the demands placed on an amplifier are extreme. Hardest to satisfy is the requirement for maximum capability of two or more conflicting parameters, such as the demand for broad bandwidth and high power in the same package.
Bandwidth versus powerThe demand for broadband, high-power capability has spawned a bewildering variety of amplifiers. Part of the problem is in the bandwidth limitations of power devices themselves. In any device, gain falls off at higher frequencies, largely as a function of internal parasitic capacitance. Eventually a frequency is reached where gain falls below unity, and the device stops functioning as an amplifier. To extend the bandwidth, the designer must sacrifice the size--and, with it, the power-handling capability.
Considerable effort has gone into development of output devices capable of high-frequency operation and power handling, but the conflict between these parameters has never been fully resolved. The amplifier designer faces a clear limit in the gain-bandwidth product and power capabilities of the devices, and the necessity for tradeoff and compromise in the circuit.
Along with device limitations, the designer must contend with another aspect of the broadband/high-power dilemma--the tendency for high-gain, untuned amplifiers to break into oscillation. This is of considerable importance in broadband amplifiers, where design calls for stable operation over a bandwidth often of several decades. Combined with this is the frequent requirement that the amplifier be stable under conditions of severe mismatch in the load.
Designing for broad bandwidth and stable operation within device limitations is the designer's real job--and headache. But it is exactly in this area where the designer makes a major difference in amplifier performance. Here is an opportunity to create a circuit that gets the maximum out of available devices, perhaps advancing the state of the art.
Understanding amplifier specificationsWith a basic understanding of the parameters that have to be traded off in designing an amplifier, and an understanding of what the specifications mean and how they are stated, the user can select an amplifier with confidence that it will do the job it's intended for.
How broadband amplifiers are rated.
The power ratingHigh-power rf amplifiers are usually rated for cw (continuous-wave) operation. Pulse power (see box later in this section) is not an accurate indication of an amplifier's capability, simply because of the variables involved. Rating for cw operation gives the user an understandable figure on which to base his judgment--if, that is, the methods of rating are known.
A simple statement of the power output of an amplifier, expressed in watts or dBm, is not enough to describe the amplifier's capability. Output power varies with frequency in any amplifier. The degree to which it varies allows a certain flexibility in specifying output. For this reason, the user should know how the given figure was chosen, and the tolerances within which it fits.
Stated power can represent maximum power at a specific frequency, nominal power over the amplifier bandwidth, or minimum power available at any frequency within the bandwidth. The graph illustrates the differences that can exist in power claims for a given amplifier.
Clearly, the way the designer rates his amplifier makes a difference. This depends to some extent on the intended use; certain ratings make more sense than others in particular applications. A minimum power rating guarantees availability of at least the full rated output over the entire bandwidth. This is the most desirable rating for the bulk of high-power rf applications, because it allows for headroom above a consistently predictable figure. Excess power capability is rarely a complaint.
In some special cases, it's more meaningful to specify nominal power. Frequency tuning of apparatus, for example, calls for "ballpark" power from the amplifier. In this application, the ballpark is the flatness specification, and the quality of the flatness specification is as important as, or more important than, the absolute power being delivered to the load. Knowing the nominal power rating and the flatness specification, the user can still safely assume a minimum figure for output. Amplifiers rated in this way will always deliver power greater than the rated power minus the total flatness specification.
While it's possible to calculate minimum power if both a maximum power rating and a flatness specification are given, maximum ratings tend to be deceptive figures. They sound good; they also carry the implicit admission that, over most of the bandwidth, the actual output will be less than rated output. If no flatness specification is given, minimum performance remains entirely undefined.
Amplifier Research rates most of its amplifiers by minimum power, so the user can always predict minimum performance. Ratings are also given with a flatness specification, usually ±1.5 dB; conservative ratings overall give the user an extra margin of flexibility.
One more consideration regarding output power: the drive level required to obtain rated output. This can vary widely; in the worst case, the user can find himself unable to get full use from an amplifier simply because he can't provide enough signal. All AR amplifiers are designed to require a maximum of 1 mW input for full rated output. They will withstand twenty times rated input without damage.
The amplifier bandwidthThe upper and lower frequency limits of an amplifier are defined--in Amplifier Research specifications--as the frequencies where the rated output falls below the value of the minimum power specification, or below the range of the flatness specification.
Amplifier manufacturers don't always spell out the methods they use to determine bandwidth; some amplifiers, for instance, may be rated at the "3 dB down" points. These are the upper and lower frequencies at which output falls below the rated power by more than 3 dB. Because this figure is not clearly definitive, the prospective buyer should know how the designer has arrived at his specification.
"Instant" bandwidthOne consideration which is not generally given as a part of the bandwidth specification, but which is critically important to the prospective buyer, is a factor described as "bandwidth availability." In the wide variety of rf power amplifiers on the market, there are models that have a serious limitation in certain applications--they require bandswitching or tuning at the frequency being worked, or under varying load conditions.
Such amplifiers aren't necessarily useless; they are unsatisfactory in applications calling for frequency-swept output (rf susceptibility, instrument calibration) or applications requiring constant changes in load or frequency (filter tuning, antenna trimming). In these applications, the continual need to re-tune becomes tedious and time-consuming, and may add a factor of uncertainty to the procedure.
Good designing here can provide instantly available output at any frequency in the operating spectrum--"instant" bandwidth. The stability and sweep capability implied by instant bandwidth are important factors for the prospective buyer to understand. An amplifier's stability assures the predictability necessary in research, testing, and calibration applications. Sweep capability--even if the application doesn't absolutely require it--means entirely stable, instantly available output power anywhere in the operating spectrum that the procedure calls for.
Amplifier linearityLinearity is the ability of an amplifier to deliver output power in exact proportion (the gain factor) to the input power. Linear amplifiers are required in AM applications, for example, where the linearity specification indicates how hard the amplifier can be driven before distortion appears in the output.
Gain compressionNon-linear response appears in an amplifier when the outputs are driven to a point near saturation. As this level is approached, the amplifier gain falls off, or compresses. The tracking relationship between output and input levels is a direct function of the gain factor; when the gain compresses, the amplifier's linearity is lost.
The linearity specification can be expressed as the output level at which the gain compresses by a given amount. Amplifier Research specifies linearity at the 1.0 dB gain-compression point, illustrated in the input vs output graph shown here.
Occasionally, an amplifier will be rated for both linear and non-linear operation. Non-linear amplifiers--for instance, those used in pulse mode or FM operation--are often driven at saturation to achieve the desired flatness specification. In amplifiers rated both ways, the linear output will always be some figure well below the saturated, non-linear output. If an amplifier is designed solely for non-linear use, however, the prospective buyer should know that it may not be at all suitable for linear applications. Operation of the output stages at a level below saturation will, in some cases, seriously affect flatness.
Harmonic distortionAll amplifiers create harmonic distortion (multiples of the fundamental frequency) to some degree. It shows up as a power loss in the fundamental, and is caused largely by non-linearity in individual stages. While it isn't entirely avoidable, the extent to which it occurs is still a matter of design, and therefore still in the hands of the designer. A number of techniques can be used to minimize harmonic distortion; AR has developed technology that incorporates these methods--often in combination--into high-power, broadband design.
Class A operation of small-signal stages, and--where possible--output stages, produces the lowest distortion figures. For higher-power solid-state applications where class A operation isn't feasible, class AB push-pull outputs are employed conservatively to achieve nearly comparable distortion levels. Harmonic filtering, particularly in amplifiers operating over a narrower bandwidth (less than one octave), is effective in reducing harmonic components in the output. Finally, the designer--with careful design and layout of the circuit--can shape gain and frequency in the amplifier to introduce the least amount of distortion.
Harmonic distortion is specified as the harmonic content of the overall output. The actual specification is usually a statement--in dB--of how far below the fundamental the harmonic content lies at rated output.
Intermodulation distortionIntermodulation distortion is, like harmonic distortion, always present to some degree in any amplifier; again, the culprit is non-linearity. A device operating with any degree of non-linearity and passing two or more signals acts as a mixer and introduces sum-and-difference products of the applied frequencies. Many of the same techniques described above are used to minimize intermodulation distortion, although IMD is generally not as important a consideration in broadband rf power applications as is harmonic distortion.
The expression of intermodulation distortion is somewhat more difficult than that of harmonic distortion; the variables involved--the number of tones and their relative power, frequency, and separation--preclude a simple procedure for providing a meaningful figure. To overcome this difficulty, Amplifier Research uses the intercept-point method, and the amplifiers which have an IMD rating give it at the third-order intercept point.
Amplifier protectionWhen the designer considers the ability of his circuit to withstand reflected power from a load mismatch, he makes a decision--one which drastically affects an amplifier's usefulness in many applications.
An easy way of protecting the circuit from a high VSWR is to design it to shut off under adverse conditions. This works; but it also has the effect of taking a large degree of freedom out of the hands of the user. In applications where adverse conditions are the norm--testing and aligning power filters, for instance--the load is rarely matched to the amplifier. If the amplifier shuts down immediately under adverse load conditions, the user is left with the tedious process of trial-and-error to bring the load close enough to the correct impedance to allow amplifier operation.
By combining good design techniques with conservatively rated devices, however, the designer can come up with a circuit that will continue to operate under worst-case conditions. This is more than a nicety; ability to withstand high VSWR loads means that the amplifier output stages can completely absorb reflected power, and still function. The amplifier therefore cannot be damaged by poor cable connections, faulty cables, or any of the myriad other conditions that can be expected in normal use. And because the amplifier will work into any load condition, the user has full operating capability for any conceivable application.1
Reading between the linesGiven the availability of high-quality, state-of-the-art devices, the major differences in the amplifiers on today's market boil down to a matter of design excellence. Clearly stated, conservative power and bandwidth ratings imply conservative design techniques--and the availability of headroom in the design itself. Instant bandwidth and imperviousness to high VSWR loading indicate a level of design excellence that assures the user maximum reliability and flexibility.
1Request Application Note #27
U.S. and Canada: call toll-free direct to applications engineering: 1-800-933-8181 · 160 School House Road, Souderton, PA 18964-9990 USA · TEL 215-723-8181 · FAX 215-723-5688
For engineering assistance, sales, and service throughout Europe, call EMV: Munich, 89-614-1710; London, 01908-566556; Paris 1-64-61-63-29.
© 1996 Amplifier Research