Integrated Analog: What’s Missing?

Analog functionality has been the stepchild of the electronic world ever since the digital revolution. The digital domain is getting all the attention, the mind share, and the development dollar share ever since Gordon Moore came up with his famous law.

Analog can't brag of putting twice as much functionality per surface area every 18 months. The smallest features are reserved for the high-performance processors from Intel, AMD, and the ARM licensees. The analog world is still largely plowing into the micron domain, while digital is at the deep sub-micron level.

Yet, there have been several revolutions in analog functionality since the advent of the digital age.

Here’s one example at work: RF devices, even high-performance devices, are now routinely integrated onto digital CMOS chips. Nobody thinks twice about a WiFi USB dongle, yet all the functions except for the antenna are now integrated on a single chip. Have you opened a DTV dongle? Most have two or even a single chip, handling all of the signal processing from the antenna to the USB interface.

Of course, the amount of actual analog electronics on these chips is reduced to a bare minimum: a low noise amplifier (sometimes), a mixer, a filter or two, and there goes the analog-to-digital converter(s) to put everything into 1s and 0s.

Another example is the replacement of much analog functionality with digital signal processors.

All high-end studio audio equipment is now digital. At the other end, even in extremely price-sensitive applications, digital is also replacing analog. A little while ago I was looking at the specs for a low-power FM transmitter of the type used to send music from your phone to a car radio. Even on these extremely price-sensitive applications, the old-fashioned analog multiplex generator is now digital. No analog filters, just a coupling capacitor at the input, another one at the antenna output, and there you go: The chip contains VCO, synthesizer, modulator, filters, and an I2C or SPI interface.

In my mundane comparison of day-to-day applications, I have also been able to replace much conventional analog circuitry with digital functionality. Many microcontrollers now come with what would have been considered high-performance analog-to-digital converters just a few years ago, and fast enough processors to do much of the previously analog dirty work in the digital domain.

While programmable analog chips from Cypress and Anadigm provide functionality that would be difficult or expensive to emulate completely digitally, the increased performance is not always needed, and the bulk of the market can be served with more conventional digital +A/D +D/A approaches.

Of course, there are many potential pitfalls when replacing analog electronics with digital, and anyone thinking that it’s just easier is probably in for some surprises.

However, while there are specific issues, the overall difficulty and, more importantly, the development cost may not be very different, once we factor in the greater ease with which changes can be made further down the development cycle or, should I say it, in production. Then, production costs themselves are likely to be lower for products using digital technology as opposed to analog, because in most digital circuits fewer parts are subject to tight tolerances, reducing your BOM costs.

What has been your experience with programmable analog? We invite your comments and questions. Please join the conversation.

3 comments on “Integrated Analog: What’s Missing?

  1. antedeluvian
    February 5, 2014


    Great to see you blogging again!

    I am too far back along the trailing edge to be aware of the fancy new functions that you mention. However I am intrigued by the insertion of digital isolators to delta-sigma isolators to achive analog isolation. I actually mention this in the third part of my blog on Analog Isolation.

  2. Didier9
    February 9, 2014

    Hi AD,

    Sorry it took so long to get back to your comment, my blog was published the day I left for Europe. It was a crazy short trip and I had no time for Internet or email… Good to be back home and good to be blogging again!

    I have used digital isolators from Linear Tech and Analog Devices in an “analog replacement” function, sort of. Here was the application:

    In a high power switch mode power supply, I had the need to monitor voltages, currents and temperature of components on both sides of the isolation barrier. I had the main processor on one side, monitoring the points that were galvanically connected to it's side of the bus, and a small processor on the other side connected to the main processor via serial UART and digital isolators. The secondary processor was monitoring the points that were on its side, converted the voltages to binary and shifted them to the main processor over serial, which handled the user interface and “normal” (read: slow, like over temperature) protection functions. This was not very fast but for the most part I did not need it to be very fast. One channel had to have a fast response time to overvoltage so the secondary processor could also drive a single opto-coupler indicating a need for immediate shut down without having to wait for the round robin serial loop. Running the ADC at full speed, the response time of that circuit was a fraction of mS, which was well sufficient.

    The advantage of that solution (over installing digital isolators in a delta-sigma converter output) is that it gave me a back channel (main->secondary processor) at no additional cost which later allowed me to implement such functionality that was not originally required in the design.

    There were actually two similar designs in that project and I switched from the Linear Tech part to the Analog part for the second project to take advantage of the smaller size. I did not need the increased isolation performance of the Linear part.



  3. Steve Taranovich
    May 6, 2018

    Hello @Removed User—sorry, but this is the second time your comment was 'This is nice article'—–I appreciate that, but I would like to know why you think this is a nice article from 2014. You have no biography, so I do not know who you are. 

    I am very concerned with 'spammers' and since I do not know who you are, then I may have to make you a 'Removed User' again—I do apologize if you are a serious reader, but you need to contribute more in your comments so I can be sure you are a serious reader who can add to our audience's knowledge and experience.


    Please reply with more details as to who you are

Leave a Reply

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