Anxieties Of Hardware Bringup During Parts Shortage

The module on a green PCB, connected to the Pixhawk controller, powering the servo rail

[Dirksavage88] tells us a story about developing a simple BEC in times of chip shortage. He needed a small 5V/3A regulator board for a servo rail on his drone, and decided to use one of the new integrated-inductor modules from Texas Instruments. Hardly requiring any external parts, such modules are exceptionally nice to use for all your power rail needs, albeit at a slightly increased cost – the downside is that, as the parts shortage hit, most of them have been out of stock. Originally priced at about $7 USD, the asking price for these specific modules, LMZM33603, has climbed as high as $800. Somehow, he obtained a few of these modules nevertheless, and went on designing a board.

It can be daunting to test your very first PCBs when the silicon you’re putting on it is effectively irreplaceable for your purposes. TI is known for their wacky footprints, and this module is no exception – the solder paste application took a bit of time, and seeing small solder balls around the module after reflow didn’t exactly reassure him. Thankfully, when he powered it all up, the module worked wonders, and took its rightfully earned spot in his drone’s servo turret. He says we can expect the next revision of his design in 2024, or whenever it is that the reported 100 week lead time is due. In case some of us could use them, Eagle files are available on GitHub!

Quite a few of us are lucky enough to have enough crucial parts for what we need, but most of us got a good few projects shelved until better times – take this WiFi-enabled wall charger project, for instance. Even bigger projects are suffering, from SmoothieBoard to Raspberry Pi. Just a year ago, we had our readers share their chip shortage stories.

22 thoughts on “Anxieties Of Hardware Bringup During Parts Shortage

  1. “or whenever it is that the reported 100 week lead time is due.”

    I suspect the majority of chips/modules like this will go straight from 100 weeks lead time to EOL with nothing in between.
    Nobody’s actually building any meaningful fab capacity at these nodes afaik, so don’t expect shortages to get any better until the new 28nm fabs start coming online and manufacturers shift some of their new products to these nodes to free up capacity in older fabs.

    1. That would be my concern also.. even maybe parts on long term availability promises… they could be like “Well, crap happened, go fish.”

      Maybe a bonus for us hacky types though, is that it keeps the surplus industry on life support for a bit longer, both by things they can cash in on now, and two years hence when warehouses of “orphaned” parts start letting go where you can get cool stuff for cheap to play around with, but nobody gonna plan a product with it, with only a few thou available.

    2. Already happened with a couple of Altera FPGA’s I use in some long-tail industrial products. I have a handful of boxes, designed a decade ago, where we still randomly build a half-dozen a year to support long-term customers.

      The legacy Cyclone FPGA’s slowly got more expensive over the years, but the quantities never justified a board spin.

      Recently, they went into long, long backorder mode and now I have been informed that they will not be produced again.

      I totally understand this, parts get to end of life and fab capacity is limited and the world has to move on… but damn, I am getting tired of juggling this whole 2020’s thing.

  2. I really do not understand why he designs in that product.
    If space or weight was at a premium then I could understand, but on that PCB there is plenty of space, so why design in an USD 6 chip when a 40ct chip with a 30ct inductor is also just fine?

    My best guess is this is just a test pcb for this chip, he had one laying around and only found out it’s unobtaneum after the pcb’s were ordered and he wanted to get more.

    The chip shortage is a real pain though. I’ve been thinking about designing a BLDC controller around the STSPIN, but even the “normal” STM32 chips are out of stock, and even if I could get a batch, supply is too uncertain to design in exotic parts.
    The only option you have is to first buy enough stock for a year (or two) of production and only finish and order PCB’s once you’re sure you have all the parts, because as soon as your current stock runs out you’re likely to have to redesign a part of the pcb for the parts you can then find.

    And of course lot’s of companies work this way at the moment, and all this hoarding adds an enormous amount of stress to the supply chain.

    Recently I saw a documentary about ASML, and one of their representatives stated that over 90% of the products they ever made is still in production, and 30+ year old wafer steppers are still chugging along.

    I bought a new PC in December 2021 which was a month or so after the price for the Ryzen 5600G had come down to something reasonable. Most of the AMD processors were simply out of stock everywhere but in the last few months stock has increased and a lot of models are available again. for a while the budget processors of AMD were also available, but at the moment it seems like they’re out of stock again. AMD clearly gives priority to the higher end of their processors.

    Reasonably priced AMD video cards have also become available again. In December you pretty much had a choice of buying a pc with integrated graphics, or pay EUR 1500 for a video card, (if you could find one at all). The bitcoin (and similar) mining craze had a big finger in the cause of that shortage though.

  3. Ow! Solder an irreplaceable chip on a PCB and only afterwards break off from the panel.

    That sounds like a great way to cause solder failures or even chip failures. Breaking a PCB off the panel will inevitably flex the PCB – and all the joints and everything attached to the board.

    1. @04:11 you see he’s further weakening the V-grooving with a knife, which is a quite sensible precaution if you do not have one of the dedicated (but extremely expensive) cutoff wheels for V-grooves.

      There are also guidelines for the distance of parts to V-grooves and this looks like it’s well on the safe side.
      Main concern is not the solder though, but brittle parts such as ceramic capacitors and resistors breaking.

      De-paneling after soldering is quite normal.

      1. This method of depaneling will be throwing bit of fibreglass into the air where he can breath it in. It’s a health hazard, silicosis or the lung fibrosis caused by this is irreversible.

        He needs a couple of tool upgrades, a temperature controller iron, 0.6mm solder or something that doesn’t look like fencing wire and definitely a way to de-panel that’s safe.

        Obviously he has been using these tools for a long time as he seems to do fairly well with them but at the same time, when a failure happens you could blame anything – long process elimination.

  4. Smoothie had to completely redesign the board because of this. Smoothie v2 was designed (two years work) around the LPC43XX, and after Covid, that chip became fully unavailable, and we had to switch to STM32H7 (much more powerful, but a full redesign was needed).

    Kliment (who you might know from writing pronterface, and the 4pi board, another early 32bit reprap system) actually did the work of redesigning, completely saved the project.

    We have been fighting for months to get all the parts to make the boards, and we just got the last one of them a few days back, and are about to start actual production.

        1. Fair enough, be worth putting some of this on another Kickstarter update, as it sounds like things are really ticking along now on the prime, and thinking about how to let folks pay for shipping of the prime that is ready (when it is) even if they originally intended to wait for the whole lot etc..

          1. The way Kickstarter works, we are supposed to do a survey (it’s integrated into the site) as soon as we have a type of boards ready to ship (but not before).
            Getting very close for prime.
            There’s an update in the works also.

          2. « Even bigger projects are suffering, from SmoothieBoard to Raspberry Pi. »

            Reading Smoothie categorized as a “big project” really warmed my heart. I still think of it like that nest of wires on a breadboard in my garage a decade ago, but sometimes people talk about it like it’s a huge thing, and it feels really weird. Helps with not feeling like an impostor, and feels really good for all the people over the years who’ve worked so hard to make it what it is.

          3. No this comment system is wonderful at being wonky…

            I’d agree though Smoothie has to be considered a “big project” now – at least to most everyone here, as its the brains of so many things!

  5. I have problems with all his extra engineering when suitable fixes are available, cheap, and common. One such would be to use an existing ESC with bec not attached to a motor. Not enough channels? “Y” harness.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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