Advertisement

Analog Angle Blog

Signal Isolation Reveals True Analog Competition

No question about it: When choosing analog components, designers have a huge array of options out there. For example, there are tens of thousands of op amps, many optimized for a specific application, and each offering a subtle combination of trade-offs in various primary, secondary, and even tertiary characteristics. At the end of the day, they are all op amps offering a basic function, despite the various implementations.

But signal isolation is a different sort of design challenge. (“Isolation” means there is no galvanic or ohmic path between input and output, yet energy representing a signal can pass through the barrier.) Isolation may be needed for a variety of reasons. These include maintaining overall signal integrity, providing common-mode rejection, ensuring user or component safety, separating “grounds” and circuit-common return paths, allowing ground-referenced digital outputs to drive floating MOSFET gates, enabling single-ended to differential signal conversion, and eliminating ground loops — and the list could go on much longer. In some cases, isolation's role is to improve performance. In others it is absolutely essential to meeting specs and having viable circuit.

But how best to achieve that isolation? That's the interesting part, since there are many fundamentally different ways to do it. The most common approaches use transformer (magnetic), optical, capacitive, and even RF-coupled techniques. Further complicating the situation, there are subsets within these topologies. For example, you can isolate a sensor signal via all-analog techniques. Otherwise you may choose to amplify and digitize it first (and you'll need isolated power, too), then just isolate the digitized serial bit stream and, if needed, reconvert to analog at the other end. Further, there are very different physical realizations ranging from monolithic ICs to hybrids and even modules.

Each of the very different approaches offers interesting and complex tradeoffs in speed, isolation voltage, bandwidth, footprint, long-term reliability, and cost. The vendor community is also quite diverse. Some sources offer only one approach and are therefore biased towards it, while others offer several types, and can give (in theory) a more even-handed assessment of the pros, cons, attributes, and trade-offs of isolation technique A versus B.

If I were interviewing a candidate for an analog circuit-design position (and thankfully, I'm not!), I would certainly ask him or her about a basic analog circuit, such as an op amp circuit with gain, of course. But for a position which requires a more experienced engineer, I'd ask about isolation, because it highlights the multiple, non-overlapping ways that you can achieve a given goal, and the trade-offs you make as you decide which approach best matches the target specifications. An engineer who understands the how and why of that, and can articulate it, is certainly a viable candidate.

What's your preferred isolation technique, and why? Do you ask analog job candidates about it?

7 comments on “Signal Isolation Reveals True Analog Competition

  1. antedeluvian
    April 2, 2014

    Bill

    What's your preferred isolation technique, and why?

    I did a series of blogs on my experience

    http://www.planetanalog.com/author.asp?section_id=3140&doc_id=562112

    http://www.planetanalog.com/author.asp?section_id=3140&doc_id=562137

    http://www.planetanalog.com/author.asp?section_id=3140&doc_id=562164

     

    But for a position which requires a more experienced engineer, I'd ask about isolation

    So, do I get the job?

  2. Bill_Jaffa
    April 2, 2014

    Yes you do, absolutely!  😉

  3. etnapowers
    April 4, 2014

    “At the end of the day, they are all op amps offering a basic function, despite the various implementations.”

     

    A designer, which has to start a new project, should have a rough idea of the requirements  that the components have to satisfy, this would save a lot of time.

     

     

  4. RedDerek
    April 9, 2014

    For my first engineering job after my degree it was not asking me about opamps, but more about phase angle and missile tracking systems; an item I clearly was not taught in college. However, it came down to how I went about trying to solve the problem was the key to getting the job. Granted, they did give some hint along the way. I got the basics down and the job.

  5. Sachin
    April 10, 2014

    It is true that as a professional designer, it is important (in fact basic) to have a clear idea of the components that will be needed to complete the design. At least know more than 95% of the minimum conditions that all the components will have to meet right from the initial napkin sessions. Then you can refine the conditions and even completely replace some components as the designer develops until you have the efficiency or precision that you need for the design.

  6. SunitaT
    April 29, 2014

    We have to make a very fast decision in order to change from analog to signal isolation. This will be a move on the right directions as the pros are greater than the cons in signal isolation. The fact that signal integrity will be maintained is a good thing meaning the quality of the signal will remain a constant factor. Also the fact that user and component safety is maintained will make a lot of end users feel safe with their components.

  7. GSKrasle
    January 16, 2018

    I would like to add a couple of thoughts to this, in addition to antedeluvian's contributions.

     

    Many projects now are communicating via RF: Zigbee, BLE, or WiFi, all of which obviously provide very effective isolation. If we have that, we don't need any more. The usual mains-isolation is no longer necessary: Ditch the Big Iron! There are several SMPS chips available to provide non-isolated power rails with good efficiency. The caveat is that you can't use USB, RS-232, or SPI unless the device is removed to a 'safe' supply. At mains-voltage, the loss in a series diode or two is pretty small, but it would be so cool to see a design with synchronous rectification….

     

    Years ago, the project books put-out by Radio Shack had a plethora of projects using a device they called a LASCR: Light Activated Silicon Controlled Rectifier. I always wondered why they were pushing such a weird little device so hard, and the best I can guess is that they were cheap and available because they were being manufactured in quantity for some other use than hobbyist projects. My guess as to what that use was is controlling the big thyristors in 'Valve Halls' of the Power Grid for interconverting HVDC and AC and AC phases. At hundreds of KV, fearsome isolation is needed, and a device like that could be attached to a fiber-optic line to safely trigger a much larger device floating at dangerous voltage. Can anyone confirm my hypothesis?

    And then there is my favourite form of isolation, the simple flying capacitor. It can be used for the power-transfer device in an SMPS (see TI's white-paper 'The forgotten converter'), but also serves to provide analog isolation, as long as voltages are within the operating range of the switches used. See the venerable LTC1043. Like switched-capacitor filters, this technology has never been deployed as widely as it should have been.

Leave a Reply

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