Why Is Analog Integration So Hard to Talk About?

Have you noticed that analog integration can be hard to talk about?

I commend all the bloggers on this site for taking a step into the void and starting the conversations here. Similar kudos to the commentors who have spoken out on the subject. But I’ve also seen outspoken engineers who jump at the chance to write up a design idea or new product suddenly grow quiet when the subject is analog integration. Why?

Is it because analog integration is new and untried? No. One could argue that analog integration has been with us since the first time two resistors were put on an IC, since the first monolithic op amp, since PWMs and ADCs first appeared on a microcontroller, since the first PMIC for a cellphone. While analog integration has been integral to analog since the beginning, perhaps it's just now reaching critical mass and affecting a cross section of end markets and product architectures in a substantial way.

Are these discussions hard because analog integration lacks importance? No. It would be hard to deny the trend toward integrating more functionality in analog ICs, and the impact that has had in end equipment ranging from smartphones to the smart grid. Analog integration will continue to enable the transformation to a more connected, more responsive world.

Is analog integration hard to talk about because it’s not well defined? Perhaps. You might describe analog integration as adding analog functions to a digital chip, or adding digital calibration to an analog chip.

My idea of analog integration could be building a complete data acquisition channel in the form of an AFE. Or it could be integrating the IC and the ecosystem — an IC with multiple analog functions, a development board, software to run it, and a simulation environment to test it. Perhaps we need to define analog integration — or classes of analog integration — to make it easier to talk about.

Is analog integration hard to talk about because it threatens the status quo? Matching the optimal amplifiers, converters, references, and more to create a high-performance circuit used to be the exclusive domain of the analog designer. Now, IC designers are performing these tasks and delivering integrated chips for a growing number of applications.

The analog engineer may not have the same level of control over the analog design when an integrated solution is used. And, it is far less likely that key components will be second sourced by multiple vendors, shifting the status quo for purchasing and manufacturing professionals as well.

As the sands underneath the analog IC industry shift toward integration, are there other reasons to watch quietly, maintaining a degree of distance and skepticism? Speak now and be heard!

7 comments on “Why Is Analog Integration So Hard to Talk About?

  1. Michael Dunn
    April 5, 2013

    I think integration is great – for specific, well-defined functions. But it's a lot harder to make more generalized chips. Think about it… Why are there thousands of different op-amp models out there?

  2. ErinM
    April 5, 2013

    Great question!  What would need to be integrated to replace those thousands of different op amp models with a smaller set that could meet the same needs?  What is driving us to make so, so many similar but slightly different amps?

  3. Brad Albing
    April 5, 2013

    Are there so many different versions of op-amps simply because manufacturers make improvements in their devices and release those improvements as new devices? Or is there a “me too” phenomena? Company Z sees Company X's product and makes a similar device?

  4. Brad Albing
    April 5, 2013

    Erin – would it be practical to have multiple op-amps in an IC where the user can pick the one they want to use and get the specs they need? The other devices in this hypothetical scenario would remain unused. i wonder if it would be economical practical.

  5. Jack Shandle
    April 12, 2013

    I tend to think of analog integration as adding analog to a digital IC. This is probably a default opinion and my guess is that I offer it simply because there typically are more digital circuits on large devices. In other words, it reflects the preponderance of logic. The other option — adding digital calibration to an analog part — is probably more interesting. After 40+ years trending toward “digital everything,” applications are now forcing us to think analog again. Digital is well suited for computers where the only human interface is a keyboard. But medical applications (and others) are providing a much wider variety of sensor inputs, for example, that make the analog interface supremely important. The trend toward wireless communication also makes it less important to do all your digital processing locally. So it seems to me that getting the analog right (doing the best possible job with those sensor inputs) might be the priority that drives the analog/digital partition.

  6. ErinM
    April 12, 2013

    Thanks for being the first to pick a definition of analog integration!   I see Bill Schweber has posted something today with a different definition, which is fine.  I don't expect everyone to have the same definition, but I am interested to learn how different people define analog integration and why.

    Love that you say the direction of adding digital to analog is interesting.  Your comments suggest to me that maybe we have become so efficient at digital processing, that the real opportunity to improve systems now lies in the analog domain.

  7. WKetel
    April 16, 2013

    Analog consideration was always there, but at the lower speeds that logic used to run at many of the analog considerations could be ignored. But as logic got faster and faster the analog properties could no longer be ignored. Reflections, ground bounce, propagation delays, and signal rise times, along with waveform distortion for a lot of different reasons. Of course, if you are meaning only the circuits intended to handle analog signals, that is a bit different. Many avoid discussing them because there are two different sets of priorities involved, and usually two different skill sets as well. And we have different vocabularies, which is actually good, since it gets really confusing to find that others hold a word to have an entirely different measning than what you take that word to mean. THAT leads to all kinds of unpleasantness.

    Besides that, it seems that the digital folks just keep wanting to go faster and faster, completly ignoring the reality that increasing effectiveness would be a better choice. The analog folks keep striving for less noise and better linearity, along with greater stability of parameters, which really are not big concerns in the digital world. So it is no wonder that folks would just as soon choose to avoid the unpleasant mixing and talking about it.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.