Leading-edge technology in electronics: differential amplifiers in oscilloscopes

　　Some semiconductor companies have recently begun offering fully differential amplifiers, but they have been used in cutting-edge electronic devices for decades. These differential amplifiers are differential at both the input and output, doubling the output range. Their input and output ports are closed paths with no common ground node. Isolation from ground improves waveform quality. Grounding is important only for static analysis and common-mode range, since the integrity of the input and output circuit loops themselves is maintained.

　Leading technology in electronic equipment

　　While the electronics industry doesn’t have major events like the Indy 500 or Grand Prix to test new concept cars, the development of electronic products has its own testing grounds. For example, during World War II, MIT’s Radiation Laboratory achieved some outstanding results in the development of radar. Even to this day, people still admire the “Rad Lab” literature of that era, which contains theoretical writings that were adopted for some electronic products. An earlier example is the theory proposed by Vladimir Zworykin when he developed the television at RCA. This was a true technological breakthrough, completely different from the incremental improvements of today’s products.

　　In the 1950s, Tektronix had a group of very innovative engineers, including Howard Vollum, Jack Murdock, Cliff Moulton, John Kobbe (who invented the JK trigger), and Bill Polits. They not only developed cathode ray oscilloscopes that could be used in laboratories, but also created the most inspiring ideal technology development environment. In the oscilloscope market driven by technological innovation, it once won more than 70% of the market share, and the root of its success lies in advanced engineering technology. Because the founder himself is an engineer with great inventiveness, Tektronix is ​​a typical engineering-oriented company that encourages creative innovation.

　　Tek and HP (now Keysight) are great examples of test and measurement (T&M) instrument companies, and high-performance measurement instruments are the "track" of the electronics industry. After all, in order to measure the performance characteristics of the circuits being developed, the measurement instrument circuits themselves must be good enough. As a result, many interesting circuits are found in T&M equipment, and the differential amplifier is one of them.

　　Differential Amplifier in an Oscilloscope

　　Oscilloscopes have used fully differential amplifiers for decades (why did it take so long for them to become commercial ICs?), and they are usually found in the vertical amplifier, where the probe voltage is amplified by a precise gain and the amplified waveform is then applied to the vertical deflection plates of the CRT. Except for the first stage, which is usually driven by a ground-referenced probe, they are all fully differential amplifiers.

　　To demonstrate, let's look at the vertical amplifier of a Tek T935A 35 MHz oscilloscope, which, although now obsolete, was state-of-the-art and inexpensive in the 1970s.

　　Figure 1: Input buffer amplifier stage.

　　The input buffer amplifier stage in Figure 1 was scanned from a manual. (By the way, the circuit diagrams in the old Tek "instruction manuals" were works of art, far superior to today's CAD drawings. Such is the price of technological progress!)

　　The first stage consists of JFETs Q4222A and B. The probe waveform is input to the gate of Q4222B. It forms a x1 buffer amplifier with the other JFET below it, with a near zero offset voltage between input and output. This is accomplished by using matched JFETs, and treating the lower JFET as a current source. Its gate is connected to the -8V supply, and the voltage VGS across R4225 (the 20Ω resistor at the source) corresponds to the drain current flowing in the JFET above it. The JFETs are matched, and at the same current, the upper JFET has the same VGS. Therefore, the lower end voltage of the corresponding 20Ω resistor R4224 is the same as the input gate voltage, and biases the JFET at the zero TC operating point. The current in the upper JFET increases slightly with the base current of Q4232, but it is small and the matching is adequate.

　　This amplifier drives a fully differential amplifier in the second stage, consisting of Q4232 and Q4234. Only the upper BJT (Q4232) needs to be driven by the amplified waveform, while the lower input is dynamically (AC) grounded at the base of Q4234 to the oscilloscope probe circuit ground, completing the input circuit return. Since the vertical amplifier (like all amplifiers) has an input offset error, the unused input can be used for input offset error adjustment, which is called DC balance in oscilloscope terminology. Balance means that the oscilloscope amplifier is highly differential, and it is necessary to operate both sides of the amplifier under the same static (DC) conditions.

　　The output of the second stage is also differential. This stage is just an emitter follower with no voltage gain, but it needs to provide a high input impedance to the JFET buffer and drive the third stage with a low impedance. In other words, it provides a voltage source for the next stage. However, at its differential output, the input waveform is not yet differentially balanced because there is no gain interaction between the emitter followers and no splitting of the input waveform occurs between them. The second stage is differential only when there are 2 inputs and 2 outputs. In the absence of voltage gain, the input differential voltage is equal to the output differential voltage.

　　The next three stages of the delay line are shown in Figure 2, which is a continuation of the same amplifier.

　　Figure 2: The last three stages of the delay line.

　　Q4258 and Q4268 form a fully differential amplifier stage, sharing emitter resistor R4254, a 63.4Ω 1% resistor. Resistors R4257 and R4267 are connected to the -8V supply and are much larger than R4254, approximating the current source of the BJT emitter.

　　The waveform at the base of the upper Q4258 BJT is split by the emitter circuit and shared with the lower Q4268 BJT (roughly half each), so a balanced waveform appears at the load resistor, with equal amplitude and opposite polarity. If R4254 or RE is split into two series resistors of value RE/2, then their midpoint will be the "virtual ground" null node of the balanced input differential amplifier. At this stage, half the amplitude of the input waveform (applied only to the upper BJT) will appear at the null node.

　　The next stage (Q4274, Q4284) is the second half of a complementary cascode amplifier - the common base. It is fully differential, as is the final common collector (Q4276, Q4286).

　　Stage Gain

　　To calculate the differential voltage gain of the complementary cascode stage, note that the emitter incremental (or small signal) resistance of Q4274 and Q4284 (shunting resistors R4271 and R4281, both 825Ω) is much smaller, so most of the ΔIC (i.e., incremental current from Q4258 and Q4268) flows through Q4274 and Q4284, developing a voltage across load resistors R4273 and R4283 (both 499Ω). The purpose of the 825Ω resistors is to provide emitter bias current for the common-base stage. The stage gain is primarily determined by the collector load resistance and emitter resistor R4254:

　　Where the upper and lower voltages are represented by the subscripts u and l. The difference between them is the input and output differential voltage. Both the upper and lower sides of the amplifier affect the overall gain, so x2 is added in front of the BJT gain of Av. Because RE (R4254) is very close to the dynamic emitter resistance re of the BJT, a better gain approximation is to increase RE by 2 x re in the denominator of the gain equation, where:

　　Av ≅ -12.9, the emitter current of each BJT is 3.72mA. The loading of the input impedance of the next stage by the load resistor and the current loss α of the two cascode BJTs are neglected. Do you think the amplifier designer wanted a gain of -10?

　　Conclusion

　　Fully differential monolithic amplifiers have been available for many years, such as the ADI AD8138, for driving high-resolution ADCs and other high-performance (high-speed and high-precision) amplifier applications. In fact, their predecessors have been used in oscilloscopes for decades. Are there other good monolithic amplifier products in measurement instrumentation circuits that semiconductor companies might be able to discover?