The distributed amplifier is an unconventional technique that allows an amplifier designer to escape the tradeoff between gain and bandwidth.
With conventional amplifiers, if the gain of one stage is not enough, the designer has to cascade stages. The midband gain of the resulting two-stage amplifier is calculated by simply multiplying the midband gains of each of the stages. However, the bandwidth (3 dB cutoff frequency) of the two-stage amplifier is lower than the bandwidth of each of the stages by itself. In most situations the resulting rise time, tr, is closely approximated by
- tr = (tr12 + tr22)1/2
where tr1 is the risetime of the first amplifier and tr2 that of second amplifier.
For example, cascading two amplifiers having a gain of 10 and a rise time of 3 ns, and a gain of 12 and a rise time of 4 ns, respectively, will result in a mid-band gain of 120 and a rise time of 5 ns.
Consider a designer who is working with a technology that produces amplifier stages like the first amplifier in the example above. If he needs a total gain of 100 with a rise time of 3 ns, he/she is constrained by the gain-bandwidth trade-off and is unable to meet both goals simultaneously.
In a distributed amplifier, several stages are connected together to form what in effect is a "transmission line with gain". The gain is the sum (not the product) of the gains of the stages, whereas the bandwidth of a distributed amplifier is the bandwidth of each of the stages.
Thus, it is possible to construct an amplifier with a gain of 100 and a rise time of 3 ns by using ten instances of the ×10, 3 ns amplifier from the earlier example connected to form a distributed amplifier.
The key difference between a distributed conventional cascaded-stage amplifier is that in the former, the input of each stage is the original signal, not the output of a previous stage, thus eliminating the cumulative degradation of rise time that occurs in cascaded stages.
One of the most important challenges when building distributed amplifiers is avoiding reflections in the signal path. For example, when the input signal reaches the input of one stage, parasitic capacitance of that stage must not cause an impedance discontinuity in the signal path, which would cause reflection.
Since eliminating the parasitic capacitance is not possible, the approach is usually to reduce the capacitance of the transmission line around each amplifier input (thereby increasing its impedance) so that the amplifier's parasitic capacitance can substitute for the capacitance of that region of the transmission line.
The design of distributed amplifiers is closely related to the design of lumped-element delay lines made from L-C sections. This, in turn, is based on the notion that a transmission line can be modeled as a series of L-C sections.
In a properly designed and constructed distributed amplifier, the input signal passes by (and is applied to) the inputs of each of the stages, and is finally absorbed by a terminator, which has the characteristic impedance of the input transmission line of distributed amplifier. In practice, there are several limits that apply to the distributed amplifier technique. One of these is the real component of the input impedance of many amplifying devices. The real component necessarily causes energy loss in the transmission line, which limits the number of stages.
The idea of a distributed amplifier goes back to British Patent 460,562 by W.S. Percival in 1936.
In 1948, Ginzton, Hewlett, Jasberg and Noe published a paper on distributed amplifiers in the Proceedings of the IRE, first using the term "distributed amplifier". Around the same time, Hewlett met Logan Belleville of Tektronix in a Portland restaurant and described the concept on a napkin.
In the fall of 1948, Howard Vollum and Dick Rhiger built a 6 ns rise time distributed amplifier under a US government contract (for radar applications). The prototype was attached externally to an early 511 oscilloscope.
Vollum, Belleville and Rhiger went on to design the 50 MHz 517 oscilloscope incorporating a distributed vertical amplifier.
The 580 series (1959) were the last Tektronix scopes to use distributed amplifiers.
These Tektronix instruments contain distributed amplifiers:
- Distributed Amplifier @ Wikipedia
- W. S. Percival, British Patent Specification No. 460,562, Improvements In and Relating to Thermionic Valve Circuits, Filed 24 Jul 1936, granted 2 Jan 1937
- E. L. Ginzton, W. R. Hewlett, J. H. Jasberg, J. D. Noe, “Distributed Amplification”, Proceedings of the IRE, pp 956-969, August 1948.
- John Kobbe and Bill Polits, US Patent 2,930,986, Distributed Amplifier. Filed 29 Feb 1956, granted 29 Mar 1960.
- John Addis, Good Engineering and Fast Vertical Amplifiers, in Jim Williams (Ed.), Analog Circuit Design: Art, Science and Personalities (1991), p.110
- G.Nikandish, R.Staszewski and A.Zhu, The (R)evolution of Distributed Amplifiers: From Vacuum Tubes to Modern CMOS and GaN ICs. IEEE Microwave Magazine Vol. 19 Issue 4, June 2018, p.66+