With the speed that technology changes and advances, products can't be expected to remain on the market forever. But what happens when a product's retirement interferes with your project?
For this week’s entry, I had been planning a tutorial and project build based on an IMU. For those of you who may be relatively new to electronics, IMU stands for Inertial Measurement Unit, and they are used to measure specific force, angular rate and, depending on the chip, also magnetic field. It accomplishes this by using accelerometers, gyroscopes and magnetometers. For years, you would have needed three separate chips on your board. As technology continues to improve and shrink, it is now generally more cost-efficient for manufacturers to combine all three sensors into a single chip, with footprints in the area of 3.5 x 3 x 1mm, as is the case with the LSM9DS1.
Thanks to their diminutive size, IMUs can be found everywhere you look - aircrafts and UAVs, gaming controllers, phones, space heaters, satellites, watches, VR goggles. Why, without IMUs, we never would have gotten to watch a chimpanzee riding on a Segway.
So I worked on my project and accompanying tutorial, and all was going well, until I learned that the IMU I was using has been EOL-ed. Again, for those who might not know, EOL stands for End Of Life, and it’s the term used when a manufacturer terminates production of a component.
So what does this mean? Well, it will mean different things for different people. For most of us, it’s really no big deal. We may not even realize that a product has been retired until we pull that one board or chip out of our collection, start to build a project around it, and only when we go to re-examine its specs online do we learn of its demise. We can most likely continue happily with our project, and it will work perfectly for us for years.
Slightly worse is when you’re planning to repeat your build, whether it’s to place multiples of your amazing project around your home, share them with friends or colleagues or, worst of all, if you’re planning on custom-made gifts for everyone in your family. Learning too late that you can only get enough chips to make gifts for eleven of the fifteen members of your family can create a real pre-holiday Sophie’s Choice.
The third and definitely worst-case scenario is when you’re prototyping a design, trying to bring a product to market. I can clearly remember a couple of times during my tenure in SparkFun tech support when customers had been using certain chips for a board they were planning to send to production. One in particular had spent about a year on the design, revisions and firmware. He felt he finally had it exactly how he wanted it, only to learn that the manufacturer was retiring the chip around which his product was based. This can be an annoying setback to a medium or large company, but for any entrepreneur or small business, it can be devastating. To spend a year developing a product, counting on that product’s future revenue to recoup your development costs, only to have it evaporate shortly before you go to production can be catastrophic.
Now this is in no way intended to scare anyone. The rapid progress and growth we’ve witness in the tech industry, epsecially in the last decade or so, is, and should remain, absolutely thrilling! Look back at Jeff Han’s TED Talk from 2006, entitled - get this - “The radical promise of the multi-touch interface” (emphasis mine). When we listen to the crowd's awed reaction when he places two fingers on the screen and zooms in on a picture, and we realize that this radical technology is now as ubiquitous as indoor plumbing, we can’t help but remain excited for what we might be able to do a decade from now. However, if you are planning large runs, or a long-term design based on a particular chip or sensor, it’s always good to be aware of where the component might be in its life cycle, and what similar components might be available. You can always check the news coming from companies like Invensense, Bosch and ST Microelectronics.
As for me, this was a minor annoyance at best. I had spent a couple weeks designing a project, coding it, designing and printing a custom 3D enclosure and documenting it all for a build tutorial, but now there’s no sense in me sharing instructions for something that others won’t be able to build. I’ll find a suitable replacement component, do a little re-design, tweak the code and in a few weeks, release the new and improved version of what I had planned to release today. Had I just been building this project for myself, there would have been no effect on me at all, but there is an important lesson here for me, and for all of us who tinker, hack, make or manufacture.
With the speed at which technology is improving, product design is changed fairly regularly. Major components, like the Atmel 8-bit AVR series, or the ARM® Cortex®-M0+ chips, should be around for years to come. But with smaller components like sensors, the turnover seems to be a bit quicker. Still, at the very minimum, you can expect any sensor to remain on the market for at least three years before an EOL announcement is made, and even after that, availability may last for another year so. So don’t be dismayed or timid about using something in your project, especially if, as is the case for most of us, it is a one-off. Products change and grow, and we all need to be flexible and creative enough to be able to have our projects and products grow with them.
I don't have much to add to the electronics part of the discussion, but Banana Republic and Brooks Brothers are my go-to sources for cardigans (definitely wait for a good sale!).
Signed, your former SparkFun hipster :)
Thanks for the tip, Shawn! I used to go to L.L. Bean, but the last time I looked they seemed to only have the really pricey ones.
For instance, due to limited range of motion in my shoulders (thanks indirectly to my physical handicap), I can NOT wear a “pull-over” sweater, and cardigans (button-up) sweaters have become scarce due mainly to the fashion dictators.
I hear you. Try buying a station wagon (I don't want a crossover, and hatchbacks are too small) on today's US car market without paying for a premium brand like Volvo (though I do like the look of their wagons). The used market of these beasts is getting slim. Yet, last time I was in Europe (Sweden 9-10 years ago... damn, has it been that long?) most of the rentals were wagons (and standard transmission).
Gosh -- it's been about 14 years since I've been to Europe! Many are surprised that when I buy a new vehicle I'm adamant about getting a manual transmission, but it's actually due to my physical handicap -- the extra excercise from the clutch is just enough to keep my left leg as strong as my right leg. Before I got my first manual transmission (in 1980), my left leg had always been a lot weaker than my right leg. They're getting harder and harder to find!
You're never sure it is a successful design until you get EOL notice on at least one critical component.
In this case there is a pin compatible replacement with lower power consumption: https://www.invensense.com/products/motion-tracking/9-axis/icm-20948/
The biggest EOL problems I've had are with low-cost through-hole power FETs. They seem to be dropping like flies. I was changing the FETs for my course every year! Some of the packages had very short leads and would not stay firmly in the breadboards.
I finally decided to have the students solder SOT23 FETs to a breakout board, so that we could use the same part numbers each year and so that the header pins were long enough to stay in the breadboards.
What happens when there is no replacement product?? The Simblee has been EOL, NOTHING else can do what it did. I'm pissed!!
Thanks for the tip, Shawn!
Lots to say on this topic...
To begin, it's not just a problem in electronics. For instance, due to limited range of motion in my shoulders (thanks indirectly to my physical handicap), I can NOT wear a "pull-over" sweater, and cardigans (button-up) sweaters have become scarce due mainly to the fashion dictators.
The EOL problem is not just a problem in the hobbyist and small company world. It's a significant part of why the Space Shuttle was retired -- they were having trouble getting spare parts for the electronics.
I've been on the "other side" of this, working for a company that made chips. One of the things that most companies try hard to do is to make the "new version" as compatible with the "old version" as (and when) possible. They try to keep the pin assignments as close as possible to the old, and the interface "backwards compatible". Ever wonder why, say, the low-order bits on a 12-bit sensor are located so far away from the upper 8 bits when you look at the "register map"? It may be because they wanted to keep the map from the 8-bit part, so that if you're transitioning to the 12-bit part you can just let those low-order 4 bits "fall on the floor", using exactly the same routines to do the job. It can be worth the effort for even the hobbyist to do a little "homework" and find out how close the new part is to the old.
When I left the first company I worked for after getting my Engineering degree, one of the big topics of discussion was a part that had been around well past it's "expiration date". The part had originally been done using Rubylith technology. The mask shop had said that the latest set of reticles (from which the actual etching masks were made) would be the last because the Rubyliths were worn out. The big discussion was whether to make a new set of Rubyliths (the machinery to do this had been surplused several years before, though the place that bought it still had it operational, and we still had the magnetic tapes with the files on it) or to put the design into (then) modern CAD systems, doing a die shrink at the same time. BTW, the part had been flagged as "obsolete" several years before, and we gone from putting "watermark" of "not for new designs" in the datasheet to big bold letters "OBSOLETE" so that you were not able to read the datasheet, but the volumes kept going up. (There were some other parts that we actually sent customers to other manufacturers for, but we were the last manufacturer of this part.) Unfortunately, I never did hear the outcome.
I've been "doing" electronics for more than half a century, and recall seeing pictures of IMUs back then that were the size of a dorm fridge, and likely took at least a couple of good strong men to lift. They've sure come a long ways!