Perspectives on Preamplifiers

Highly signal tolerant preamplifiers, the kind that will stand up to being used in co-located environments, depend on two approaches. The first and most common is to select a robust front end transistor, like those typically used in CATV amplifiers, and secondly the use of a multipole preselecting filter. Unfortunately such a filter must be properly terminated in Zo at each end if the integrity of the passband response is to be maintained. The main problem with trying to use a multi resonator preselector with the typical preamplifier is that the best noise performance for the preamplifier often occurs at an impedance other than Zo, as such there is a built-in conflict with this approach.

The reason for this is that when one matches a transistor at room temperature (typically about 290 degrees Kelvin) to Zo in any traditional manner, one defines the input noise temperature at the device temperature, i.e. 3dB noise figure. This subject is covered, with mathematical justification, in a publication by William A. Rheinfelder titled "Design of Low Noise Transistor Input Circuits", Hayden Book Company,1964, LOCCN 64-13750. If this is the case, how does one get noise performance at a temperature lower than ambient out of an amplifying device? The answer, as Rheinfelder illustrates, is to mismatch the device. He includes graphs of mismatch ratio vs noise performances below the ambient temperature of the device. Indeed, unless one makes special arrangements in the design, most preamplifiers are not Zo or 50 ohms at the input. The actual match can be very poor with a VSWR in excess of 5:1 as measured on a vector network analyzer or reflectometer.

So how does one get around this conundrum? The simplest way is to lower the ambient temperature directly to cryogenic levels and match the amplifying device directly to Zo. All physical laws are satisfied with this approach. This has been done in radio astronomy applications with MASER, parametric amplifiers and FETs for many years. Unfortunately cryogenic cooling does not work with bipolar transistors, as these cease to function at very low temperatures. On the other hand it does work well with GaAs FETs as these devices, unlike their bipolar cousins, show enhanced performance at low temperatures. There are also other techniques involving circulators/isolators, as well as Kurokawa topologies, some of which still involve cooling the terminations for best performance. Needless to say, cryogenic solutions are not very practical for amateur radio applications.

At the time Rheinfelder published his work as cited above, the modification of input impedance for achieving simultaneous gain and noise match, (the condition necessary to make a preamplifier work with a multi resonator preselector) via feedback was not well explored for VHF/UHF devices. Since then, however, two forms of negative feedback have emerged that satisfy the requirement. One uses reactance feedback (specifically the addition of small inductance to the emitter or source terminal of the device) and the other is transformer feedback in what is commonly known as the "Norton amplifier" by D. Norton, Microwave Journal, May, 1976. Both forms of negative feedback use non-dissipative circuit elements in the feedback path, that is, reactances and transformers as opposed to resistors. These techniques are applicable to single stage amplifiers.  There is also a technique known as "second stage reflection" where in the second stage is set up to reflect enough energy, out of phase, back through the first stage to cancel the first stage input reflection. Techniques like this are being used in some modern multi-stage integrated circuit designs.

It should be noted that Valley and Wallman explored the application of inductive reactance in low frequency vacuum tube amplifiers and published their results in the "MIT Radiation Labs" Vol. 18 in 1946. The technique did not gain wide acceptance, as most designers felt that adding inductance to the cathode would invite instability. Early attempts at H,Y and Z parameter characterization also proved troublesome at VHF/UHF frequencies due to stability problems in the measurement apparatus. It was not until stable and full S parameter characterizations of semiconductor devices were developed in the late 60s, along with computer analysis programs such as Compact, that designers concluded that the techniques were workable.

The inductive reactance feedback works when the gain term of the semiconductor device shows a complex form ( Beta = beta - j beta' ) where the complex component of Beta is large enough that when multiplied by an inductive reactance looks like a virtually noiseless resistor. This is a quite simplified explanation and not intended to be a full discussion of the theory.

Considering all this, how good must the return loss be in order to be good enough? Practical experience dictates that return loss in excess of 20dB (VSWR <1.22) shows very little degradation in the passband shape. Below 15dB (VSWR >1.43) we begin to see noticeable effects and less than 10dB (VSWR >1.93) shape distortion is severe.

Examples are shown in the “Preamplifiers” page. The 1296 MHz amplifier and the 144 MHz amplifier use inductive feedback, and the 5700 MHz amplifier, being multi stage, uses both inductive feedback as well the second stage reflection method. The 432 MHz amplifier features a modern IC that meets all requirements. Today, there are numerous devices that are fully S parameter characterized and there are also a number of free "student versions" of analysis programs available that will handle one or two active devices. I would encourage anyone interested in these methods to avail themselves of the software and data and experiment and learn about the design.

Finally, all results cited in the “Preamplifiers” page are the result of actual physical measurements made in the laboratory and not data sheet derived or simulated.