Advertisement

Blog

The Allure and Risk of Second Sources

One inescapable reality of analog (aka linear) ICs is the huge number of products in a given class. Consider that most basic building block, the op amp: there are literally thousands of distinct versions available from dozens of vendors. Although it can be overwhelming at first to identify the “best” one to use in a given application, vendor selection guides are a major help in getting started.

Admittedly, the selection challenge is not as onerous at it may seem at first. The broad category of op amps divides into subcategories such as power, high-speed, low-noise, and precision, to cite a few of the many “slices” of the op amp world. Still, understanding the tradeoffs, however subtle, that each op amp offers compared to its fairly close siblings can be a challenge.

Among the many viable devices, there are also second sources – usually claimed to be drop-in, identical replacements for a specific device; there are also alternative sources which are similar in many but not all specifications. In principle, it’s nice to be able to put a second source on the BOM, since it provides some pricing competition opportunities as well as supply-chain and availability peace of mind. These alternative sources require a lot more work to make sure they meet the specs you need, and that their differences are of little consequence in your application – once you identify those differences.

However, even with a specific op-amp group, there are many subtle variations and tradeoffs in top- and second-tier parameters such as input offset voltage, input bias current, noise across various bands…it’s a long list. A second-source may be a “just fine” alternative, or maybe not. After all, this backup part may have the same design (either as a copycat design or via a legitimate licensing agreement) but still is fabbed, tested, or qualified differently. There’s a lot of “magic” in all ICs, especially in analog/linear ones.

Then there are what are known as improved second sources, or ISSs. This is where a vendor releases a part which is identical in most aspects but improves on one or more parameters. Problem is that you may specify the ISS version on the BOM, and count on that improved specification for things to work just right. However, some well-intentioned purchasing agent along the way orders the non-ISS version, and inexplicably, things don’t work as expected. Or perhaps the person kitting and assembling the prototype orders or pulls that non-improved part, and no one catches that difference.

Sometimes, however, the problem is a circuit designer who is setting the stage for trouble. Many years ago, I worked with a brilliant circuit engineer whose schematics were usually perfect from day one. In other words – and this was before the widespread use of reference designs – he would pore over data sheets and application notes, and a schematic that was right, at least on paper. Of course, there were often real-world subtleties that affected his designs, but that wasn’t his fault – it was usually an error or omission by the IC vendor.

But he could also be a little too smart for himself. He once specified an amplifier from a top-tier vendor, because his detailed reading of their op-amp schematic (yes, it was on the data sheet) showed a protection diode on the output. Normally, this diode is not invoked unless there’s an overvoltage situation or problem with the load. He decided to take advantage of this “free” diode and not add an external one, even though the protection it offered was not called out in the data sheet.

Long story short: the purchasing team had a problem getting this particular IC from the vendor but was relieved to see that true second sources were available. They went to the component engineer on the staff, who compared the data sheets and OK’ed using a guaranteed second source from another qualified vendor.

Of course, that’s when problem of failing op amps started to pop up. It took a lot of digging and recalling “why did I do it that way?” to uncover the root problem: this legitimate second source did not have the diode. The first vendor couldn’t be faulted, as they had never called out the diode or its function in the data sheet; perhaps it was just a test of a design topology or fab-process step that they weren’t ready to formalize, or something else. Whatever the reason, no other sources had it in their versions, and that was the problem. Luckily, adding an external diode solved the problem (whew, at least that was easy fix) but the time and frustration of tracking down the cause took a serious toll on the team and delayed other projects.

The lessons here are clear: be grateful there are second and alternate sources for many analog ICs; if it is not explicitly called out on the data sheet, don’t try to take advantage of an IC-design’s internal characteristics; and even if it’s a second source, dig deep and ask the right questions of the design team. Sometimes, being too clever can turn around and bite you.

Have you ever had frustrating problems with what appeared to be true, legitimate second sources? And what’s been your experience with alternate sources?

Related Content

Debugging Tactics: 4 Ways to Increase Integration

Simple Product, but Subtle Troubleshooting

Perfect Sensor, Imperfect Test

3 comments on “The Allure and Risk of Second Sources

  1. John Dunn
    June 27, 2018

    I had a 2N6578 power transistor second source issue with Motorola versus RCA way back when, but also a problem with just the one source, Motorola, gradually altering their devices over a time span of several years as they sought to improve production yields.

    Please see:  http://licn.typepad.com/my_weblog/2011/09/semiconductor-evolution-john-dunn-consultant-ambertec-pe-pc.html 

     

     

  2. Steve Taranovich
    June 28, 2018

    Thanks for sharing that John experience John—we all need to be careful with so-called 'second source' ICs because they are not all true second sources.

     

    Back in the early 80s, I had designed the ubiquitous 555 timer into a circuit design. Later, purchasing asked me to provide a second source. 

    I had selected a Signetics NE555 (https://spectrum.ieee.org/tech-history/silicon-revolution/chip-hall-of-fame-signetics-ne555 ) as the original part and chose an RCA 555 timer as the second source. After reviewing the data sheets, I was convinced that they were true replacements for each other.

    Hans Camenzind had designed the NE555 in the early 70s. He attended the Analog Aficionados dinners hosted by Jim Williams in Silicon Valley—but I digress.

    When I tested each timer in my circuit, I found that the RCA 555 timer did not make my circuit perform as designed. After some further investigation, I compared schematics, and 'lo and behold', the RCA circuit had a very seemingly minor design difference that probably worked in 99.9 percent of designs, but not in my design.

    I don't remeber which 555 I had chosen next as my second source, but before testing it, I compared schematics and they were identical (Be careful here because identical schematics can have gotchas too—that;s another story for next time—probably John Dunn has experience there as well)

    I tested the new second source and it worked in my design (Be careful with second sources when the design or data sheet gets a new revision—that can also bite you)

  3. nathandavidson
    July 2, 2018

    I think that in every scenario, it's important to have a backup plan, and I think that's the idea with this second source right? At the end of the day, we just want to make sure that the network or system is set up in a way that prevents any sort of loss of capacity or any downtime is reduced if there are errors and issues that pop up and while such instalments may complicate the system, they are there for that reason! I'm sure that we will be able to streamline the framework better as we work through the options! 

Leave a Reply