Advertisement

Blog

Integration: Old but New

By its very nature, digital has always led analog in the level of integration possible at any point in time. Jack Kilby’s first device, after all, was a digital circuit.

It is clear that this trend towards higher levels of integration is continuous. I believe, however, we are in a period in this evolution of analog integration that moves us into a new era.

Integration started out quite simply: the op-amp. Bob Widlar developed the first integrated op-amp, the uA702 at Fairchild in 1963. It was not very popular due to various quirks. In 1965, he developed the uA709, which integrated, I believe, 14 transistors and 15 resistors. The ’709 was very successful and the race was on.

The main drawback of the ’709 was that it had to be compensated (stabilized) with an external capacitor. In 1968, Dave Fullagar, also at Fairchild, developed the internally compensated uA741 op amp. The ’741 had a few more transistors and resistors than the ’709, but the key difference was that it also integrated a capacitor that compensated the device.

Analog circuits became more complex over the following years, but analog ICs were all of the “building block” type. Some ICs did manage to combine a couple of similar building blocks. The next step was to integrate dissimilar functions in one IC.

Perhaps the most famous of the early ICs to integrate dissimilar functions is the MAX232. The idea for this part came from Charlie Allen of Maxim, whom Fullagar once described as “the best applications engineer I have ever encountered.” Charlie noticed that a lot of equipment was operating completely from 5V supplies, except for the RS-232 drivers, which required ±12V supplies. If an RS-232 chip could be made to run from 5V, then the equipment makers could eliminate these two supply rails. In 1987, Dave Bingham designed the MAX232, integrating charge pumps, a type of DC-DC converter, to generate the larger voltages needed for the RS-232 drivers.

Over time, this integration of dissimilar functions has increased. Equipment designers have benefited, as their designs require fewer circuits, can be developed more quickly, and are smaller and less expensive.

So why do I say we are entering a new era in analog integration? As the number of dissimilar functions integrated in a single IC increases, we are now seeing circuits that integrate so much that they take up a major portion of the end equipment.

While earlier analog ICs were developed to provide certain functions that could be used in a number of end equipment types, these more highly integrated components are targeted to specific single applications.

Much of the system architecture is now in the chip, and the chip manufacturers now have to be experts in the end equipment and work very closely with the makers of the equipment.

This is fundamentally different than the days where all analog ICs were building blocks. As with integration in general, this trend will continue toward higher and higher levels of integration.

8 comments on “Integration: Old but New

  1. TomMurphy
    January 11, 2013

    Certainly, integrated circuits moved us past the days of stand alone “building blocks,” but I'm not really sure I see how we've reached a turning point to a “new age.”  Can you explain what was the tipping point that you think brought us into a new realm of analog design, please?

  2. TheMeasurementBlues
    January 11, 2013

    The MAX232 is high on my list of great parts. I used many of them on RS-232 ports for printers and modems.

  3. TheMeasurementBlues
    January 11, 2013

    Mark, have you ever read “Makers of the MicroChip? Read my review.

  4. Mark Fortunato
    January 11, 2013

    Tom, this is certainly no hard corner we are turning.  The key difference is that the level of integration in analog/mixed-signal is getting so great that much of the architecture of the end euipment is going into the chip.  This is fundamentally different that just providing building block chips that can go nearly anywhere with the system architecture being completely in how the many chips are combined at the board level.

  5. patrick_m
    January 14, 2013

    Granted, there will always be the 'purist' who'd go the discrete route for corner-case designs that need a very specific optimization, but so much is now integrated at such high performance, that the benefits of discrete are going asymptotic really quickly. Increasingly, much more is to be gained in many cases by focusing on the overall system and how it meets the final objectives at a broader level. Issues such as customer satisfaction, UI, software come to mind.

    But this, will always be a good discussion as technologies improve, and the users/designers make clearer their needs.

  6. Brad Albing
    January 14, 2013

    That is a sweet part. Used a bunch myself. Solved a lot of problems in one fell swoop.

  7. Brad Albing
    January 14, 2013

    I had almost forgotten about the '709. Probably best to on retrospect. It was a tough one to use. Better than op-amps made with 12AX7s of course, but still….

    The LM301 was a little better. All things considered, I'm glad we're not still back there.

  8. MClayton
    January 17, 2013

    While Kilby and Noyce were focused on Patents, the Military was focused on what they called Batch Electronics.   And yes, some were simply looking to replace Von Braun's relay racks with anything else.  But while the thin film gurus worked on many digital replacement, the earliest IC R/D folks focused on analog solutions for the military and aerospace.  Batch Electronics Conferences were popular in early 1960's, as IC industry struggled with its identity, Solid Circuits a la TI, Molecular Electronics a al Westinghouse (which were mostly analog at first), and finally IC name from Fairchild.  Perhaps some of you can comment on the early 1960's, with no implanters, no plasma tools, only Bipolar Devices and some fancy isolation, metallization, and gettering processes to stabilize those early analog parts.  RF, IF, amps, and attempts at single chip miilitary radios????

Leave a Reply