Circuits that are forgotten but should be recalled

What are legacy ICs? Let’s look at some of the original components and discuss why you should reuse them.

As electronics mature, they also begin to show signs of aging. As innovation wanes, it takes longer and longer to achieve breakthroughs, and people are less willing to take risks. Decades ago, some integrated circuit (IC) companies were willing to bring some unusual and ingenious components to market. For example, in the 1970s, Signetics was brave enough to create a novel integrated circuit, Hans Camenzind's "555 Timer." This integrated circuit became the earliest legacy integrated circuit.

What about other circuits? This article describes those circuit components and why they should be reused.

AD639 Sine Converter

Long ago, Barrie Gilbert did some research on bipolar junction transistor (BJT) differential amplifier circuits, which have a hyperbolic tangent (tanh) transfer function:

Where I0 represents the emitter source current and thermal voltage. VT is 26 mV. In order to reduce nonlinearity, an external emitter resistor RE is connected in series with the emitter in the amplifier. During the research, Gilbert fully applied and embodied the motto in engineering: "If you can't change it, use it."

The hyperbolic tangent function is vaguely related to trigonometric functions. Alan Grebene of Exar used a single differential amplifier in the XR2206 function generator (FG) integrated circuit to convert a triangle wave into a sine wave. The results were not satisfactory, but acceptable for a first generation of work. Gilbert made more refinements to the basic idea. He proposed multiple hyperbolic tangent functions with the output of a differential amplifier offset by a fixed voltage between the inputs. This approach expanded the functionality (and input range) and also led to other innovations, such as those used in the AD639 sine converter.

This 16-pin IC is a trigonometric wonder. It was so powerful that it was destined to become a legacy IC. Unfortunately, Analog Devices (ADI) withdrew the AD639 from the market without a replacement. I don't know why, not even Gilbert himself. The AD639 seemed destined to become a legend. It synthesized all the basic trigonometric functions (sine, cosine, tangent, cotangent, secant, cosecant) and their inverses.

The sine function accuracy is 0.02%, which is better than the sine output of most function generators, and the total harmonic distortion (THD) is also better than many audio amplifiers. The IC has two function generators, and also has compensation circuits, as well as a multiplier and a divider. It is priced for the niche market, so it cannot enter FG instruments and other sine wave equipment that requires accuracy or low THD. It is nominally 1.5MHz.

Perhaps the only problem is that because the AD639 was so attractive, Analog Devices (ADI) put a high price tag on it, preventing it from entering the market as a commercial component. Perhaps Rochester Electronics, as the dominant supplier in the "tail" market, can resurrect it and capture the profits it would otherwise have generated. If the components that didn't make it the first time can be used in new designs, there is no reason for Rochester Electronics to limit itself to being a replacement component supplier for obsolete equipment.

CA3096 Bipolar Junction Transistor (BJT) Array

This is a highly versatile basic unit similar to the AD639 sine converter - a transistor array. RCA has developed a BJY array containing a row of CA3000 series circuits. For NPN type BJTs, some components have fTs exceeding 1GHz, which is very suitable for today's new designs.

RCA reorganized and eventually became Intersil, but lost its old large-scale wafer fabs. Tektronix designed the longitudinal amplifier in its 2205 oscilloscope based on the CA3046 (or its equivalent National LM3046), which was attractive for implementing fast two-quadrant or four-quadrant multipliers.

Intersil has a lot of legacy, but the supply of these parts is limited and dwindling. These parts should be returned to some existing process. It is not a major development project, but these parts will be very useful. Intersil does have a replacement, the HFA3000 series SOIC parts with multi-GHz fTs, but with correspondingly lower breakdown voltages (see Figure 1).

The original CA3000 series was adapted to ±12V supply voltages, but the HFA series is designed for ±5V supply voltages. The integrated circuits can withstand voltages up to about 10V. A further improvement in the HFA series is the PNP-type BJT, which is dielectrically insulated rather than a lateral transistor like the CA3096 (Figure 2).

The CA3096 is a versatile component that has three NPN BJTs and two PNP BJTs. One drawback is that the fT of the lateral PNPs is only about 6 MHz (it is difficult to make the thin substrate required for lateral BJTs). However, for many circuits, this specification is not a major obstacle.

For example, a feedback amplifier has a quasi-static gain of 3 and a bandwidth of more than 50 MHz (see Figure 3). It has two forward channels. The slow channel goes through the PNP current mirror and the fast channel goes through Q2 in the differential amplifier input stage. It uses all five array BJTs. The only other semiconductor component is the avalanche diode Z1.

You don't want to make this circuit design a new product because of the erratic supply of components. Also, HFA components don't have a voltage range. If there was an IC family with performance comparable to the CA3000, but with dielectrically isolated PNPs, it would be a welcome addition to the IC heritage.

Caption 3

This feedback amplifier has a quasi-static gain of 3 and a bandwidth of over 50 MHz. It uses all five array BJTs of the CA3096. The only other semiconductor component is the avalanche diode Z1.

MC14500B Industrial Control Unit

The 16-pin MC14500B from Motorola is a single-bit, 1 MHz complementary metal oxide semiconductor (CMOS) processor. It has three single-bit registers (flops) and an arithmetic logic unit that can execute 16 instructions. Newer microcontrollers have replaced the MC14500B, but that's not the point. The MC14500B is a large general-purpose logic block that requires only an external counter for the program counter (PC) and a program memory driven by the PC.

The data memory is also an input/output (I/O) memory. Four bits of the memory output drive the opcode input on the MC14500B; the other bits are used for 8-bit bidirectional latch I/O addressing (MC14599B) and 8-input multiplexer or data selector (MC14512).

The single-bit accumulator is also called the result register (RR). Instructions include: load RR, load RR's complement, AND RR data, complement data and AND, OR, complement data and OR, not NOR (same), store and store valid RR output to the complement pulse write line, transfer input data to the input register or output register, skip the next instruction if RR=0, pulse flag O output or F output. The other two instructions are the unconditional transfer instruction (JMP) and the return instruction (RTN), which also output flag pulses. The unconditional transfer instruction (JMP) can be used to load an address to the PC. The RTN instruction outputs an RTN flag and skips the next instruction.

A built-in oscillator generates the clock that drives the PC. The rising edge of the clock increments the PC, and instructions are fetched while it is high. When the clock is in a low period, the instruction is decoded and executed.

What is the advantage of the MC14500B today, as it uses bit-serial processing and is input/output (I/O) intensive? It requires the use of additional counters, program and data memory, and it will remain an obsolete component because it cannot compete with lower-cost 8- to 16-pin programmable flash ICs that are easy to use and have more powerful features. Although the MC14500B is interesting, it requires too many bit conversions to be put back into production. Although the coefficients of the MC14500B are inspiring, this component will remain forgotten.

MC14549 and MC14559 Successive Approximation Register

These successive approximation registers (SARs) were once a component in Motorola's 4000 series complementary metal oxide semiconductor (CMOS) digital integrated circuits. SARs have 8 bits per IC and can be cascaded to get more bits. They are used to build successive approximation analog-to-digital converters (ADCs). Inside the SAR is a shift register and a parallel load register.

Despite its simplicity, SAR is a very useful digital function. The SA algorithm starts with the median of an interval and searches the interval using an asymptotic Boolean comparison. If the voltage is higher, the most significant bit is set and the next bit is tested until all bits have been determined. A complete conversion of n bits takes n clock cycles, regardless of the digitized value.

By adding a comparator and one or two SAR ICs to drive additional DACs, a simple analog-to-digital converter (ADC) can be added to the system along with the remaining subcomponents. Although this level of integration is semi-discrete today, it is still feasible for many applications with multiple DACs and multiple comparators (and the requirement is to use a simple ADC).

Successive approximation registers can also be used for automatic ranging, with fewer steps on average than sequential ranging. Similarly, a variable gain amplifier (VGA) gain can be set using an SA search over a large gain range. The bit weights may no longer be binary, but may be decimal or a 1-2-5 sequence. But if it is monotonic, then this scheme will work.

MC4530 Dual 5-Input Majority Gate

Integrated and marketed as a rather odd logic function, this is the dual 5-input majority gate sold by Motorola. If 3 or more of the 5 inputs are valid, then the output is valid. This component may be of interest to those who wish to find novel uses for existing logic components. There is an XOR (Non-Identity) gate at the output from one of the W inputs to set the priority of the output.

What is it used for? This component is used in some unusual applications, but it can induce some creative thinking. In a redundant system, if five or fewer subsystems present a state, the device will make a decision. For example, if the vital signs monitors in the intensive care unit of a hospital show that three or more of the five patients have problems, an emergency triage state is entered.

By adding a comparator and one or two SAR ICs to drive additional DACs, a simple analog-to-digital converter (ADC) can be added to the system along with the remaining subcomponents.

By connecting a high input to a low input, two of the three control computers can determine the output (as in an aircraft). Alternatively, if multiple banks of capacitors are charged asynchronously, and m of the n banks have outputs indicating full charge, then there is enough charge to enable the ignition device. This functionality is egalitarian; any m of the n can trigger an event. By using inverting input logic, a statically stable multi-legged robot can trigger a fault condition if fewer than m of its n legs are on the ground.

By cascading the output of one majority gate to the input of another majority gate, you can implement a voting hierarchy. The first five produce one vote among the next five. Although this is a more practical use, it is still an uncommon logic function. No wonder this circuit has been forgotten.