The AD741— a Precision 741


Neither the 741 nor the 101A were designed as true high precision amplifiers. In the years following the development of the 741 and 101A op amps, other IC manufacturers looked into refining the performance of these popular products for the precision analog marketplace. In 1971 ADI took a key step towards this, and acquired Nova Devices of Wilmington, MA. This added both technology and design capability to the ADI portfolio, which was to be immediately useful for the manufacture of linear ICs.

One of the first ADI ICs to be produced was an enhanced design 741 type op amp, the AD741 (Reference 9: Data sheet for AD741J, K, L, S Lowest Cost High Accuracy IC Op Amps, Analog Devices, Inc., January 1973). A schematic of this circuit is shown in Figure 18 below.
The AD741 monolithic IC op amp
Figure 18: The AD741 monolithic IC op amp

Although this circuit looks deceptively like the μA741 of Fig. 17, it should be understood that there are many subtleties affecting IC performance that don't necessarily appear on the data sheet. In this case, some key differences include a thermally balanced layout, evident from the use of a cross-quad input stage (denoted by the Q1-Q4 cross markings), and the quad operation of the current mirror load transistors, Q5-Q6. In addition, a better output stage was used, with higher efficiency transistors.

The premium version of this design was the AD741L, which achieved an offset of 500μV(max), a drift of 5μV/°C(max), a bias current of 50nA(max), and a minimum gain of 50,000 (94dB) into a 2kΩ load. ADI also produced an improved 301A type amplifier, the AD301AL, with DC specifications similar to those of the AD741L above.

In 1973, the AD741L sold for $6.00 in 100 piece lots, while the AD741J could be purchased for just $1.25 in the same quantities. With the establishment of general purpose IC op amps a fact, other designers began to focus on greater precision. This was sought through the reduction of various errors; lower bias currents, lower offset voltage, higher gain, etc. A couple of the development paths that follow the general thread of higher IC op amp precision will now be discussed (For coherence, this super-beta precision op amp (and other) threads will be presented in a continuous fashion. In actual time of course, each thread paralleled many other concurrent IC op amp developments).

Super-Beta IC Op Amps— LM108 to OP97

After his release of the 101 op amp, Bob Widlar began to explore the super-beta bipolar transistor technique (To avoid confusion, the term "super-beta" should really be "super-HFE"— but "super-beta" has stuck). A super-beta transistor is one subjected to extra diffusion steps, so as to raise the forward gain from a typical 200, to several thousand or more. Used in the input stage of an IC op amp, a pair of super-beta transistors can potentially reduce input currents by a factor of 10-20 times.

However, the use of super-beta transistors isn't exactly straightforward, because the super-beta process reduces the breakdown voltage to 5V or less. This factor requires extra circuitry around the super-beta devices, to buffer high voltages normal to an op amp.

The first IC to use super-beta transistors was the LM102 voltage follower of 1967, by Bob Widlar (Reference 10: Robert J. Widlar, "The LM110 An Improved IC Voltage Follower," National Semiconductor LB-11, March 1970 (The use of super-beta transistors in follower-connected IC op amps LM102 and LM110)), followed by the upgraded LM110 in 1970. Widlar also published a more general description of super-beta transistor operation, in Reference 11 (Reference 11: R. J. Widlar, "Super Gain Transistors for ICs," IEEE Journal of Solid-State Circuits, Vol. SC-4, August 1969 pp. 249-251 (General principles of super-beta transistors in IC op amps)).

The 102 and 110 voltage follower ICs were somewhat specialized parts. Internally configured as unity-gain buffers, there was no user configuration needed (or possible). Nevertheless, the use of the super-beta devices at the input established their viability, at least in one context of application.

Meanwhile, in these very early years of the technology, Bob Widlar wasn't the only designer working on super-beta concepts as applied to op amps. At Motorola Semiconductor, Solomon, Davis and Lee developed the MC1556 op amp, reporting on it in early 1969 (Reference 12: Jim Solomon, William Davis, P. Lee, "A Self-Compensated Monolithic Operational Amplifier With Low Input Current and High Slew Rate," ISSCC Digest of Technical Papers, February 1969, pp.14-15 (The use of super-beta transistors in the MC1556 IC op amp))(Rather ironically, the 1556 op amp may not have gotten all the credit due, as perhaps the earliest use of super-beta devices, within a general purpose op amp. A second irony is that the paper itself is better known (and often quoted) for the establishment of input stage gm reduction as a means of raising slew rate).

This two-stage op amp design used a combination super-beta NPN pair, combined with a PNP pair as the input stage. With a quoted super-beta transistor gain of 4,000, the design had a 2nA input current. It was also known for an appreciably higher slew rate than the 741 or other devices available at the time.

In late 1969, Bob Widlar contributed another IC op amp design, the LM108 (Reference 13: Bob Widlar, "IC Op Amp Beats FETs on Input Current," EEE, December 1969 (The super-beta input LM108 IC op amp)). The LM108 was the first of what turns out to be a long line of precision IC op amps with low input currents, by virtue of a super-beta input transistor front end. A simplified schematic of the LM108 is shown in Figure 19, below.

In this circuit the super-beta NPN devices are indicated by a wider base in the symbol, and the remaining transistors are high voltage types. Q1-Q2 make up the super-beta input differential pair, and are cascoded by Q5-Q6. The diode drops biasing this cascode are arranged so that Q1-Q2 see a 0V VCE. The second stage on the 108 is a PNP differential pair, Q9-Q10, with a balanced load, Q21-Q22. The output voltage is developed at the emitter of Q14, and buffered by a class AB output stage.
The LM108 super-beta input monolithic IC op amp
Figure 19: The LM108 super-beta input monolithic IC op amp
The 108 design achieved a notably low bias current, typically under 1nA at room temperature. Offset voltage was typically 700μV and 2mV(max), and gain was 300,000 (or 110dB). It had very wide input and output ranges, typically ±14V operating from ±15V supplies, and it consumed 300μA of quiescent current. Further, it could operate down to supplies of ±2V, making it useful on 5V rails. A point worth noting here is that the 108 differed from Widlar's previous LM101/101A designs with a load rating of 10kΩ (whereas the 101/101A could drive 2kΩ at rated gain). This was obviously a byproduct of the low power nature of the 108 design.

The basic LM108 design was later upgraded by National, to the LM108A. This was a 500μV(max) offset voltage version of the part. An internally compensated version was also offered, the LM112.
Later on in time, many other companies brought out their own competitive versions of the 108 and 112 op amps, with similar sounding names, and some with much improved performance. In the 1970's, ADI was one such company, offering the AD108 and AD108A, with specifications like the originals.

In 1969 Marv Rudin and Garth Wilson formed Precision Monolithics Incorporated (PMI), a brand new company with a charter of precision linear ICs. PMI introduced their counter to the 108A, the OP08, in 1976. This wasn't simply a second source to the 108A, but a revised and upgraded design by George Erdi and Larry Farnsley. Erdi was known as the father of the Fairchild μA725 (and the SSS725, at PMI). Erdi came to PMI in 1969, from Fairchild, where he had already established some key op amp design concepts (see narrative on 725 to OP07).

The new OP08 design added a thermally balanced layout, to reduce offset voltage and to increase gain. This was reflected in an offset voltage of 150μV(max) for the best grade, a minimum gain spec of 50,000 (94dB) into a 2kΩ load (other specs were comparable to the 108A). At the same time the PMI OP12 was introduced. This was a device similar to the OP08, but with internal compensation, and one which competed with the 112.
The OP97/297/497 super-beta input monolithic IC op amp
Figure 20: The OP97/297/497 super-beta input monolithic IC op amp

Another IC company to introduce 108/112 style designs was Linear Technology Corporation (LTC). Formed in 1981 by former National and Precision Monolithics engineers, Linear Technology introduced their own super-beta op amps, the LT1008 and LT1012 in 1983 (Reference 14: George Erdi, Jim Williams, “Precision Op Amp Serves Host of Needs,” Electronic Design, September 1, 1983 (The LT1008 and LT1012 super-beta input IC op amps)). Designed by a team headed by former PMI op amp designer George Erdi, the LT1008 featured trimmed offset voltage of 120μV(max), a drift of 1.5μV/°C(max), and a minimum gain of 120,000 (~102dB) driving 2kΩ. A notable feature of these amplifiers versus the earlier 108A types was the use of input bias current cancellation, allowing an LT1008 bias current as low as ±100pA(max).

Precision Monolithics followed up on the OP08 and OP12 designs with the PM1008 and PM1012, released in 1987. These were designed by Peter Gaussen of the Twickenham UK design center. The PM1008 had specs comparable to the LT1008, and the PM1012, to the LT1012. Also in 1987, the performance bar was raised a bit higher by the introduction of the even more tightly specified PMI OP97, an internally compensated super-beta input op amp functionally like the 112 or 1012. A simplified schematic of the OP97 family is shown above in Figure 20.

The OP97 best grade (A, E) offers an offset voltage of 25μV(max), a drift of 0.6μV/°C(max), a bias current of ±100pA(max), and a minimum gain of 200,000 (106dB) driving a 2kΩ load. It is notable that the OP97 was marketed as a "low power OP07," which of course it isn't, technically speaking (Former PMI and ADI op amp product line director Jerry Zis relates that while the OP97 may have been marketed with a focus on the OP07 users looking for a lower power device, this niche was nevertheless a real need. Of course, it also helps to have great specs, plus a family of dual and quad devices, which the standard OP07 never did have— but which the OP97/297/497 family eventually provided). The OP97 uses a two-stage topology, the OP07 a three-stage. Not at all the same inside— but to many users, lower power with precision can be very important, rendering the internal differences moot.

The OP97 is still available even today, as are the other dual and quad family members, the OP297 (dual) and OP497 (quad) devices. The latter devices were designed by Derek Bowers, adding laser trimming (as opposed to the use of zener-zap trim on earlier family devices), and were released in 1990 and 1991.

Share this article :

0 comments:

Post a Comment

Please wait for approval of your comment .......