Ask Hackaday: How Is The Chip Shortage Affecting You?

Some friends of mine are designing a new board around the STM32F103 microcontroller, the commodity ARM chip that you’ll find in numerous projects and on plenty of development boards. When the time came to order the parts for the prototype, they were surprised to find that the usual stockholders don’t have any of these chips in stock, and more surprisingly, even the Chinese pin-compatible clones couldn’t be found. The astute among you may by now have guessed that the culprit behind such a commodity part’s curious lack of availability lies in the global semiconductor shortage.

A perfect storm of political unintended consequences, climate-related crises throttling Taiwanese chip foundries and shutting down those in the USA, and faulty pandemic recovery planning, has left the chipmakers unable to keep up with the demand from industries on the rebound from their COVID-induced slump. Particularly mentioned in this context is the automotive industry, which has seen plants closing for lack of chips and even models ditching digital dashboards for their analogue predecessors.

Chips on order everywhere on the Mouser website.
Chips on order everywhere on the Mouser website.

The fall-out from all this drama in the world’s car factories has filtered down through all levels that depend upon semiconductors; as the carmakers bag every scrap of chip fab capacity that they can, so in turn have other chip customers scrambled to keep their own supply lines in place. A quick scan for microcontrollers through distributors like Mouser or Digi-Key finds pages and pages of lines on back-order or out of stock, with those lines still available being largely either for niche applications, unusual package options, or from extremely outdated product lines. The chances of scoring your chosen chip seem remote and most designers would probably baulk at trying to redesign around an ancient 8-bit part from the 1990s, so what’s to be done?

Such things typically involve commercially sensitive information so we understand not all readers will be able to respond, but we’d like to ask the question: how has the semiconductor shortage affected you? We’ve heard tales of unusual choices being made to ship a product with any microcontroller that works, of hugely overpowered chips replacing commodity devices, and even of specialist systems-on-chip being drafted in to fill the gap. In a few years maybe we’ll feature a teardown whose author wonders why a Bluetooth SoC is present without using the radio functions and with a 50R resistor replacing the antenna, and we’ll recognise it as a desperate measure from an engineer caught up in 2021’s chip shortage.

So tell us your tales from the coalface in the comments below. Are you that desperate engineer scouring the distributors’ stock lists for any microcontroller you can find, or has your chosen device remained in production? Whatever your experience we’d like to know what the real state of the semiconductor market is, so over to you!

133 thoughts on “Ask Hackaday: How Is The Chip Shortage Affecting You?

  1. The Smoothieboard v2 Kickstarter is redesigning the entire board from the original LPC4330 to the newer STM32H7, entirely because if we were to wait on the LPC4330 to be available again, it might be up to a year before we can ship boards.

    Kliment (the author of Pronterface) has offered to do the board redesign (thanks SO MUCH to him), and we hope to have a prototype run in a bit more than a month or so.

    Jim Morris (who wrote the original LPC4330-targetted Smoothie v2 firmware), is also nearly done porting that firmware to STM32H7.

    It’s crazy to have to completely redesign the board, but it’s much better than having to wait a year for the current chip to be available.

    We were nearly ready to start producing the LCP4330-based board at the beginning of this year, and it’s as we were working on setting up production for the Kickstarter backers’ boards, that we realized how many of the chips were now unavailable (or for some of them, had price markups of 1000-5000% sometimes ).

    Also, the STM32H7 is a bit more expensive, but *much* more modern than the LPC4330, so all the backers of the Smootheboard v2 Kickstarter are getting a massive hardware upgrade, completely for free.

    ( You can see a comparison table of the two chips here: https://docs.google.com/spreadsheets/d/1DDMflx5Cr8gOOoUEgHUeac6ta_YS8VEiQUdXbmI4kdU/edit?usp=sharing )

    We also have to change the stepper drivers, and a few other components, all because of shortages.

    This has been so stressful, and a crazy situation to handle, but we are seeing the light at the end of the tunnel.

    We need to put out an update to the Kickstarter about this, but things keep changing so fast and often, it keeps getting delayed. Should be soon now though.

    Crazy times.

      1. Ok, so that’s where we actually are now a step ahead of the problem:

        We’ve bought enough STM32H7s, a month ago, to be able to ship all the Smoothieboard v2-mini and v2-prime,

        (as well as enough chips to produce several runs of prototypes for the new STM32H7 version of the design)

        which (-mini and -prime) are the two designs we’ll be shipping first (v2-pro, which there are a lot less of to deliver, isn’t even fully finalized design-wise, and will be fully designed and tested *after* the other two boards ship).

        So *hopefully* this means we are out of the woods in terms of stock, and will be able to produce and deliver what we have to, moving forward.

        1. Thinking about it a bit further, our order might actually be part of why the supply situation at mouser is so bad, :) I know we ordered *most* of the stock they had at the time we put our order through. Which is a bit scary, underlines how little there was to go around.

        2. The manufactures I know in China are not counting on things getting better before 2023. Theit cash economy has the same effect as a disrupted Just-In-Time economy. There is no such thing as “Net 30” or 60 or 90. In normal times a 30% down will get things made. More likely 70%. Now they have spot markets and can’t predict prices. It really screws them up. Plus the Bank of China is a social control tool controlled by Emperor Xi. To top it off, their population is dropping.

          1. Keep dreaming. They start 3rd highly skilled, west educated, zero BS generation. 10-15 years ago was last bell for invasion (economic or military, whatever works). Do they already get the blade alloy tech for jet turbines you ask? No, at least officially. But when do, hold to your hats.

          1. We have secured enough of the microcontrollers, and *some* drivers (after changing drivers to one that was more easily available).

            We’re still looking for some more drivers, and have plans/good expectations we’ll manage on that side of things.

  2. Resistors, capacitors, microcontrollers, and voltage regulators have been a problem. I stocked up on everything but it won’t be enough for the year. The voltage regulator issue caused me to re-evaluate and add to a board re-spin I am doing. I don’t have large volumes, but when parts are not available I have to get replacements from somewhere – I had designed with what I thought were generic parts with many commonly available alternatives, but those are gone too, so now I’m going to an even more popular pinout to get the most replacement options.

  3. I’m running a Kickstarter that ended in April. I had to change the MCU and I/O expander sources and models at least 5 times before I found one with enough stock to get me what I need, and even then I only have about 30 spares now for a production run of over 500.

    I’ve also pretty much put most of my other Tindie projects on hold that don’t run on DIP MCUs (seems to be the only thing not impacted).

  4. I’m having problems with capacitors and gate drive ICs. Some package options are available but they tend to be the ‘retarded’ options like SOIC-16 wide which takes up extra board space. Many parts also only available from US distributors.

    1. I’m no designer or engineer, just do hobby stuff but IMOP it’s definitely a bad deal when such mundane items as an IR2110 or 2104 along with a huge list of other components cannot be obtained without waiting 2 months or more to show up in your mailbox.

  5. I need 200 of

    TPS4H160AQPWPRQ1

    To complete an on hold order at the contract manufacturer.
    They have everything but that part.

    Let me know if you have any that you can sell me.

    Thanks,
    Wolf

      1. I remember Sparkfun had a writeup a while ago when they had this problem with the ATmega328 and they ended up getting counterfeit chips that were useless, so it’s probably going to happen on a bigger scale this year.

      2. I’d say “temptation” is not even close!

        If counterfeits get sold in ordinary times when chips are not scarce, to think counterfeits would not be prevalent when there is…!

        1. But counterfeits need to be manufactured as well. If the squeeze is on all manufacturing, counterfeits may be affected nearly as much. The only area that may not be affected is relabelling of existing parts.

      3. I got my first batch of counterfeit chips a month ago. The part was laser engraved (which is a bad sign) vs the real one that was advertized with printed part numbering. It was blank part as ESD diode test shown. i.e. no silicon inside. Aliexpress refunded me without fuzz. I looked around and saw a handful sellers showing the picture of the fake engraved version, so buyers beware! I really could have used one for a recent project. The same project used a $0.40 module fom China and thankfully it was good.

        I am good for parts as I am used to unreliable shippping delays from China and minimum quantities. Got a small stock pile of cheap uC, LDO when they were on sale and junk piles I can scrap for passives. I have enough of projects with hardware done but the software part was too hard.

  6. I sell a display replacement kit for Pure evoke flow radios (on Tindie) which has an adaptor board which used a stm32f103c8t6. These were assembled by JLCPCB but when the price went from 2USD to over 19USD, making it unaffordable.

    However the GD32E230 is pin compatible (but not binary compatible). I had to port my code from OpenCM3 to their SDK but it wasn’t that bad. The GD32 parts are around 2 USD and actually have better performance than the ST parts.

    David

  7. Same same – I designed a board around the TPL7407LAPWR and foolishly didn’t buy the chips when I selected them and now they are out until July.

    I ended up re-spinning the board to use the venerable ULN2003’s that I cannibalized from a 15 year old project.

  8. I design electronics for experiments that fly to space. This requires short runs of PCBs on a very fixed schedule – if you’re not ready, the rocket’s not going to wait around for you.

    This spring we’re working on flight boards for a mission launching this fall, and about a third of our BOM was out of stock at the normal suppliers; not only MPUs and sensors, but bread and butter components like SMD resistors and caps. This often happened between confirming stock availability, sending off the PCB, and getting the PCB back – in one instance I saw 30,000 parts disappear overnight. After subbing everything we could we still had to scramble to find a number of parts. Octopart was a lifesaver – it showed us the remaining stock (often in the single digits) at distributors literally around the world. We got most of what we needed from those (with foreign payment and MOQ drama). The last few missing parts were obtained by networking with friends asking if they had any spare backstock in their labs. We barely got everything we needed, and that was to assemble only 10 PCBs.

    I’ve been doing this a long time and have never seen anything like this in my life. I also predict that we’re about to see significant shortages of consumer electronics – pretty soon this is going to affect everyone, not just those of us on the front lines.

  9. I am mostly doing hobby projects, so most of my boards are one-offs. At this point I have started scraping the bottom of my parts bin for leftover microcontrollers from old projects, and just design around whatever I can find. On the STM32 front it seems that the highest pressure is on lower-performance LQFP-48 F-series devices. For many L-series parts there are at least some left. High-performance devices such as F7 series parts also seem to be impacted less. The crazy demand for the popular F103C8T6 has meant that many F-series parts in TQFP-48 are out of stock since they are pin-compatible. I think if I would need a smaller F-series part now, I would probably try to find one in a TQFP-64 or one of the lead-less packages.

    1. Same, but my company has to make 200 pcbs with stm32f103, now that they are unobtainium, we had either to do it with atmega32e5 (trying to cram our firmware from 62kb/10kb used memory to 32kb/4kb would not be pretty) so we decided for completely new version with esp32.

  10. I once caused a shortage by designing a board with 20 SMD Posistors on it and ordering a run of 500 boards. The same board also made it difficult to find a particular analogue switch for a few months. I literally went from supplier to supplier buying up all the stock I could find and boom! Sadly there is often less stock in the pipeline that we think.

  11. Have had full and partial reels of R, C, L, transistors, MCUs (mostly TI and NXP), and some other stuff sitting on my shelves for 1 to 12 years. It is all almost gone. The last three full reels (LDOs) are going to be dropped off today. What stock remains is by design. So a bit of a win for us small independent contractors that have been wondering how to dispose of excess stock.

    Most of the supply chain will recover in 10 to 20 months, but there will be some level of permanent damage to the affected industries.

  12. The shortage has really killed my vibe. Prior to the shortage I procured parts by the 100’s and 1000’s and have built up a local parts library of 100,000 parts or so (R, L, C, Micros, ASICs, amps, fets, etc). I feel justified in my techno-gluttony. With this library I was able to consign parts to manufacturers when no replacement was available. Unfortunately some of my designs have parts that are not in my library in enough quantities and are not available at the usual suspects. All of my BoMs need substitutions or consigns which really sucks. Hopefully the industry can pull through the backlog before my stash dries up.

  13. I have a stock of around 100 STM8s (S003F3 in LQFP32, IIRC) for a project that never made it to completion; I needed them for the four pin-accessible CAPCOM channels and external clock that TIM1 allows.

    Maybe it’s time to sell them.

  14. For ham radio builders, the hardest shortage to deal with has been the lack of the Si5351A synthesizer chip. It’s at the heart of a lot of recent radio designs. (Oddly, the Si5351A breakout boards are still available for now despite the nearly total unavailability of bare chips.) Shortages of STM32-based microcontroller boards like the Blue Pill have also hurt.

    1. But the Blue Pill almost never has a real STM32 – there are like 6 or 7 identified counterfeits. Finding a BP with a real STM32 in it is like a needle in a haystack.

      1. True now. But they had real STM32 processors when they first appeared; the clone processors came later. The Black Pill boards mostly still contain actual STM processors for now; I suspect that will also change over time. And there now seem to be shortages of the Blue Pill boards with clone processors as well.

        There is also a Chinese clone of the Si5351A now, the MS5351. It seems to work with existing code, so they copied the entire programming architecture of the Si5351A. Only the 10 pin TSSOP version appears to have been cloned so far, not the versions in larger packages or the Si5351B or Si5351C. Fortunately that’s the one in all the radio designs; it’s the cheapest, and radios rarely need more than three oscillator outputs.

        I haven’t yet seen any reports on how the spectral purity of the outputs of the clone compare to the real Silicon Labs chips — which in a sense we will never see again because they’re in the part of SiLabs that was recently sold to Skyworks, and they may be rebranded by the time they come back in stock.

      2. Just because there are plenty of counterfeits *now* doesn’t mean bluepills were “almost always” counterfeit. Don’t confuse proportion and history, they don’t have the same units.

  15. I’m a grad student designing readout electronics for prototype MEMS gyros, we had a few board designs sent to production and a Digikey BOM sent to a local distributor (Digikey charges pretty significant shipping rates + customs to my country) early May, with most things having at least a thousand units in stock. Since then, my country has gone into full lockdown and we came back this week to find that our order had not been placed, and half our items were now out of stock with 40 week lead times. At least we could find compatible parts without a respin (mostly analog building blocks) but it was definitely a close call.

  16. I’ve started brushing up on my soft(ware) skills, using my stock of ESP32 boards and making them talk to cloud servers of my own making and to each other through the nets.

    The shortage will be over at some point, and I’ll know a little more about interconnecting my projects by then :)

  17. The biggest problem for me in the Linux board business is Ethernet chips from TI and Realtek. Even if they can be found at less than their weight in gold, they must be allocated to a run of a particular board, meaning the others just stay out of stock.

  18. For me it’s the weird rolling connector shortages.

    I end up having to interface with modules from a lot of manufacturers, and each of them uses a different connector.

    Just last week I finally shipped a small run of product that had been sitting around for 6 weeks waiting for one Hirose part that, until recently, had been such a commodity item that I didn’t bother stocking it.

  19. I work at a contract manufacturer. The pandemic has revealed good and bad design engineers.

    Some of our customers have design considerations… like second sources, the component company’s history and structure (ST vs Mirochip), shipping methods, countries of origin, etc. Those companies are doing fine. Most have fewer than 3 engineers, many just one.

    Customers with lazy design engineers are hurting. These are the kinds of engineers who put in ZERO effort researching single source components or bothering to create a flexible design. They’ll use parts like the ESP32, ‘neopixels’, or really anything from LCSC. These engineers just throw it over the wall and blame their co-workers or Trump when they go line-down. Most of these clowns are really just IT guys, invested time into programming and now identify as hardware engineers.

    1. You know you sound like a snob, right?

      IT people learning new skills and taking on new challenges is a great thing, not something to skoff at and look down upon.

      You’re the clown.

      — The IT people.

        1. > Don’t get so offended. You almost prove his point.

          That’s like … my whole joke …
          I guess tone isn’t easy to detect in these sorts of circumstances.

    2. To extrapolate from present producibility of a product to past goodness, badness, or laziness of design engineers seems like a bit of a stretch. I’d argue that a more nuanced (and generous to the design engineers) characterization of the situation is that the parts shortage has revealed differences between designs that met producibility requirements vs those that failed to meet (or that never had) producibility requirements.

    3. It is rarely a design engineer problem, it is a bean counter problem.

      Beancounters are known for trying to push the need for sending out orders to the last possible second, crying sacrilege if you dare to suggest they get stock before the hour its’ needed.

  20. I’m hobbyist and presently I am not doing big production run, I haven’t any any issues with my current stockpile. If I ran out of ATTiny85 SOIC-8 I can switch to PDIP I also have on hand.

    I’ve intentionally held off making new projects that would require MPU that I don’t have on hand.

  21. As a hobbyist, I’ve got enough ‘stuff’ to keep me busy for a long time. I was planning on upgrading my R&D computer to a AMD 5900X processor, but unavailability and inflated prices when available is keeping me away.

  22. Hah, as if the chip shortages weren’t bad enough.. I just got an email from one of our boardstuffers today that the board house is out of raw stock of a copper clad core used on one of our most popular items. Closest product is ~.18mm thicker and we use two per stack. Now we have to figure out if we need to retest EMC and if it’s still going to fit in it’s enclosure.

    Good times… good times.

    1. The company I work for has had the same problem! We’ve never even put any thought into the brand of FR4 core material that we use, and now we’re all experts… at least until the next shortage of something we’ve never had a shortage of before.

  23. Over the last few months, we’ve seen a LOT of desperate buyers looking for parts from sellers who bought too much inventory. We’ve helped save a few production runs from very long supply-related delays, which is great!

    U-Blox, TI CAN/automotive parts and STM micros are especially rare now, but they do sometimes show up in the marketplace. For example, 16K of the STM32F423ZHJ6 were just added. https://xscomponents.com/product/stm32f423zhj6/

    Sellers are often willing to negotiate too. Good luck out there.

  24. We’ve redesigned parts around old kinda lousy chips we produce, because they’re made on a process node that has excess capacity right now as it’s an old fab.
    The hate coming from the automotive tier 1’s shrieking at us for parts, and saying they’ll go elsewhere and never use us again, and we’re all “where, exactly, will you go that’s better?” This is messing up a lot of long term working relationships.

  25. I went to a job interview about two months ago and although I’m under an NDA, I can describe some of what I learned.

    The overall project has to do with delivering a quantum computer under contract to the federal government. I was offering my services for building a support organization post-implementation. I ended up declining to proceed with the process due to weirdness I felt around the CEO.

    From the first conversation with the recruiter, I detected a subtle undercurrent of a suggestion that the CEO was a time traveler and the architectural foundation of their project was secreted to them from the future. I’m looking for tangibles regarding compensation and they kept promising that this technology rollout wasn’t about profiting in the current day, but that everyone involved would see wealth beyond imagination provided by the developments that would be enabled by this quantum computer project. She (the CEO) pointed at people like Musk and Bezos and said they were wealthy not because of proprietary ideas, but because they applied those ideas in lucrative ways before others came to recognize the opportunity. This project would usher in tenfold Bezos-scale opportunities for all involved. Her certainty in this premise was fueled by an implied first-hand experience of what was to come.

    After the second-round interview attended by the CEO, I became paranoid. It was oddly syncratic that my previous engagement had its contract abruptly cancelled and then immediately afterward I had received the call from the recruiter. I get a lot of recruiter solicitations, but was inclined to pick this one up because I suddenly was looking for my next gig. How did they know I needed a job? Did they pull levers to make me need a job?

    Anyway, they told me their timeline (pun intended) for this device is delayed due to the chip shortage. I asked them if they were concerned they could default on delivery of the contract. The CEO said she knew how to manage schedule expectations of demanding customers and likened it to a bronze shortage that had delayed some other project she had worked on in the past called the Antikythera mechanism.

    1. sounds like a terrible working environment even if stacks of cash are floating around. not worth it at any cost. eccentric billionaires are nothing new though so I don’t doubt what you experienced. that level of secrecy is often secured by blood however.

        1. I don’t doubt that he was told that story, not that it actually happened. look at the death of howard hughes, watched ice station zebra on loop and turned into a hairy vegetable doing it. elon musk is a likely unmedicated bipolar and he could end up the same way. money tends to isolate the super rich, then they surround themselves with syncophants who cater to their every whim.

          1. Come on with the armchair over-the-internet half-assed random medical diagnosis … you *know* that’s a terrible idea.

            Especially considering Musk just recently started being open about his Aspergers.
            Which is a much better fit for his behavior than “unmedicated bipolar” … My brother has Aspergers and is pretty much a clone of Musk (without the billions, unfortunately), quirks and all.

    2. I guess they didn’t pay enough attention when visiting the future… Could’ve gotten their hands on some (even mediocre) Covid vaccine to get more than enough money to stock up on any required chips before the shortage. Guess they didn’t learn enough from that Antikythera mishap…

    3. Time travelers who are caught off guard by a chip shortage, may have other undisclosed issues. Good call on passing this one up. You must have a good sense of what’s coming…

    4. Turns out there are more than a couple of female CEO’s of a quantum computing companies. Go figure. Reminds me of – The Inventor: Out for Blood in Silicon Valley.

      I suppose when you are selling snake oil, it doesn’t hurt to be a time traveling genius. The “shortage of bronze” in (one assumes) 100 BC must have been a real downer.

    5. Sounds like Meredith from the uBeam unfame. Unless the chip shortage triggered a chain reaction that caused Jobs-impersonating female CEOs to multiply.
      Sadly, since it sounds like a government contract, no one will be able to publish failures like this and the guilty will lead a lifetime of deception.

  26. These shortages world even occur without crises or disasters. The large number of products and variants makes it impossible for manufacturers and distributors to estimate future demands, parts that cannot be sold in million quantities only get one production run per year at best. Long gone are the days when there actually were “standard” integrated circuits, made by more than one manufacturer and 100% interchangeable. The stuff you can buy at Digikey and others should be seen as excess stock of custom silicon, without guaranteed future availability.

  27. Just a hobbyist with lots of projects half-built. I think the only way it’ll affect me is that I was thinking of building a rover with the kids this summer. Not the end of the world if we have to put it off another year.

    1. Me too! I didn’t realize that they might be correlated though. Do you think this relationship is co-incidental, or are you thinking this this is causal?

      My parts bin is experiencing chip shortages, and though I hadn’t connected them before, ever since Renesas, or perhaps even before, my purchasing agent was operating on a-little-too-late as opposed to the jit and doomsday hoarding options when it came to chip sourcing. The only exception being the first week of January, but I think that was more closely related to football than the new fiscal year. However, we have been able to respin with nuts and pretzels. I find the pretzels to be slightly underpowered, but extensive tests show the UI hasn’t significantly suffered, but we needed to offset with bigger traces to the drink fridge. The nuts (especially pecans) however, are significantly overdriven, but I have able to balance that with additional storage capacity.

      A recent order however had an odd mix of counterfeit variety packs with some even swapping out popcorn in the same chip package. I have been able to mostly avoid them, but once they build up enough, I find an implementation of several bga packages in a multi-layer approach has demonstrated that they aren’t entirely useless. Just a lot more work for the same performance. Nothing that couldn’t be replaced with a 5 minutes to 6 (N2) but a good thought experiment none the less.

  28. I’ve got a healthy stock of mcu’s (mostly 8, 16 and 32 bit PIC mcus), arduino boards etc for my hobbies. I’ve ordered DACs, bus drivers, logic level shifters etc past few month without issues.
    If I run out, I can easily recycle some of my old projects or scavange from old PCBs or use vintage micro processors. I just bought 10 20MHz Z80s and VDPs with some sram. Enough to keep me busy for a couple of years at least.

    I did see the PiStorm project having problems getting the clpd or fpga to build boards. I think the makers of that board are redesigning it using a more readily available chip but who knows how long that stock will last.

    As the parts for medical devices have also been affected, it’s a really worrying situation. Hopefully these will get precedence over more commodity based manufacturers.

    1. Had some custom silicon get delayed an extra month because of this – got out of TSMC about on schedule but then stuck waiting for packaging (in Belgium I think). Fortunately have them now, out being assembled along with the rest of the parts we bough 6 months early (just to avoid any squeeze on them).

  29. nobody cares, but I have had to reflow my GPU at least 3 or 4 times because it started failing just as everything started going up in price, and i’m not patient enough to reball it entirely and a BGA is not worth the cost for something that’s push ing 8 years old.

    Can’t find a GPU anywhere, and I have been too slow to get in on the LTT VAG drop every single effing time.

    So frustrating to not be able to purchase for a machine that is literally required for me to work.

    also, screw your stupid tariffs, if you want to screw over China, you should have supported global foundries (and other homegrown chip fabs) rather than trying to put in too little too late after the issues have already arisen. I literally blame our government for not fostering competition and sitting on their laurels and waiting until literally millions of us are screwed. Good luck trying to create an entire supply chain out of thin air.

    1. I buy a new cellphone on average about once per quarter these days, it’s honestly easier to just have my old phone shredded before departure and to buy whatever the new flagship is when I arrive in another country than to deal with possible confidentiality and trade secret issues.
      The IT department has some software they use to set them up with my contacts and everything, but phones aren’t my specialization.
      It would be nice for them to be refurbished instead of destroyed, but company policy doesn’t let us do that.

  30. I’ve largely shelved all projects using an STM32 MCU because the supply chain is just that horrid.
    Thankfully, I’ve not had trouble getting boring ass ATMega328s, yet.

    I had to scrounge a particular IP67 pushbutton switch from 6 different vendors. My contract assembler laughed at me when I handed them 1020 parts in 6 different bags for that part.

    I’ve been sitting on a thousand MX150L connectors for a couple of months. When I saw them come in stock, I bought them. I have a backorder in on another connector I need – originally 45 weeks. Connector availability is preventing manufacturing product. This isn’t a big surprise as Molex is a supplier for both Ford and GM (and others I’m sure?) of automotive connectors… Which just happen to be what I’m ordering. Sigh.

    Infineon and ST integrated power switches have been scarce. I bought Rochester out of BTN8980s because BTN8982 and BTN8962 are sky high, when they’re in stock. Thankfully been able to get the VN5 and VN7 parts I’ve needed, but it’s been dicey.

    I haven’t been paying that much attention to prices on jellybean passives because I’ve just been happy to still be able to get them.

    The only vendor that has been able to consistently overdeliver has been Espressif. I’ve gotten a few thousand ESP32-WROVER-E 128/64 modules this year and they all arrived on time or earlier than promised.

    I’ve never seen anything like this. The supply chain is totally hammered. I’ve been buying 6-12 months of need if I see parts available versus ordering 2-3 weeks in advance of production usually.

  31. I’m a hobbyist ask nothing critical, though my Glasgow Interface Explorer is severely delayed because of it: they’re hoping the early birds will get theirs by year end, while mine was supposed to ship in the next week.

    The last few weeks of Desk of Ladyada have also been shortage related: they’re making more mechanical/input device new products because they can buy them.

    I’ve also heard some popular imus and display bridges are nearly impossible to source.

    The dark side of “lean” manufacturing: fat aids survival for a reason.

    I wonder how much of the popularity of the stm32f103 is due to things like the blue pill (I imagine some), how much is due to having reasonable clones commercially available (probably some), how much is just brand recognition based on that, and how much is other reasons: it’s hardly the only low-ish cost Cortex-M micro (in normal times).

    1. Mostly it’s that the Blue Pill boards are really really cheap (well, were) for their level of capability; significantly cheaper than the Atmel-based M0 boards, for example. I’m guessing at first that the STM microcontroller on it was very inexpensive because there was mass demand for it in some other application such as automobiles. Then the clone chips happened and the boards got even cheaper.

      Those development boards aren’t what’s causing the shortages in any case. Popular though they may be, I’m sure they’re a tiny fraction of total demand for those chips.

      1. Oh sure, I wasn’t blaming the blue pill directly for the shortage. I just see f103 chips in actual products more often than I’d expect given the sheer number of chip vendors out there (and given that the f103 used to cost more than e.g. the f405 which is a much nicer chip)

  32. Funny you should ask. I had not really pondered it much But as of late I have been getting a lot of counterfeit stuff on eBay, more than usual, and some things have been like vaporware. I bought a bunch of A/D’s and the first guy canceled my order because they failed “quality control”. Like they even have that in China. The next batch the guy got back to me like 2 days later and told me I could have a refund or pay quite a bit more for them. I told him I had already paid and I would bitch to eBay. He canceled me and I did bitch about the shakedown. He literally has 12000 items listed so he makes them a lot of money so I am not holding my breath. The next batch was about 2X the price of the first two and he canceled my order as the next guy at about 3X the original price. I got a note from the guy who tried to gouge me in the meantime and he said he had already shipped. WTF? So it will be interesting to see if I get anything from him or not. On the fake or wrong parts front, I wrote to one guy and flat out asked if the parts were real and what he was advertising and he said things are as advertised. Oh, if not it will be another eBay refund and more low grade parts for something else.

    So yea I have been feeling the squeeze, and I have seen prices really shoot up. Things I was paying like $3 for 10 of are now going for $2 a pop. Ouch.

  33. I’ve been getting many of my ongoing designs finished. Fortunately I purchase the parts for a limited run of whatever I’m making in advance.

    With luck my backlog will ride out the production hiccups.

    For customer’s PC server components, well… the wait’s just longer for new replacement hardware unfortunately.

  34. I have redesigned almost all our boards to work with STM32L* (they were on stock when STM32F* were not) series. They are build inside 1kW+ units, lol. I am also doing the SW as well.

    + I made several assembly variants for different crystal, power chips and pressure sensors (there are no left, all used for medical right now) footprints. I am in contact with our assemble service and purchase department all the time.

    More or less 4 years of work re-designed in like 5 months. Alone. With everyone really unpatient, expecting the “stupid PCB” to work at the first try. I feel totally burned out right now, how long it is going to take?

    1. How feasible, practical, and plausible would it be to modularize the microcontroller onto a universal-footprint subassembly that only does the digital logic?
      I’m thinking maybe an inch square or similar, with castellated edges for direct board-board soldering and holes for headers so that it’s easier to debug and has more flexibility – Kind of like the ESP8266 modules
      Then a shortage of the MCU would only mean that module needs redesign instead of a much larger PCB.

      1. Could be a solution as well… but the costs would be higher as you are producing 2 PCBs. Not only the PCB purchase costs, but the whole “product management” behind it…

        Luckily STM32 are quite pin compatible (if you take care!). I spend some time with Cube MX and datasheets and I am able to fit several variants of STM32F10*, STM32F4* and STM32L4* on the same board (with help of 0R resistors bridging traces the in the right places.

        Of course, 3 years ago when I was designing it, I did not take care about that… There were like 250k+ STM32F103 on stock. Wild times, huh…

  35. I work for a medium-sized company where we make our own chips.
    We are currently on rations for wafers, for one product line maybe 200 wafers a month, in dribs and drabs.
    Once you have set up with one foundry it’s difficult to copy the design elsewhere, and besides the other foundries have similar problems with capacity.

    * Not my usual handle* for obvs reasons…

  36. Pent-up demand combined with transportation capacity logjams combined with manufacturers not wanting to increase production requiring expensive equipment only to be caught with excess capacity when demand returns to normal mean this shortage is going to last a long time.

    1. This is really my fear as well….
      There will be a lack of incentive to expand capacity as increased demand and prices is basically a license to print money for fabs.
      Why not just keep everything the same, not need to invest anything more, and then not even have to worry about the risk of reduced demand on your newly built capacity when this all starts to normalize?

      If all things stay the same, this will be a long correction back to normal as people and organizations will do what any human does in a shortage; buy everything now, further perpetuating shortages and supply chain issues!

      1. We will probably not return to “normal”. But a rebalancing was in order, that “normal” was not sustainable, and now we are seeing how marginal it was. I think most supply chains and part catalogs need to be about 10 times simpler. Even before the shortages, which meant we could deal with changes, and find substitutes, it was a mess, parts arbitrarily going in and out of stock with 30 week lead times all over the place. Very few companies were managing it well, there was a lot of wasted time and energy everywhere.

  37. It’s nice to read about all the projects that can just do simple redesigns to get production running again.
    In our company, thats just not going to happen, unless there is REALLY no other way out. Reasons are pretty simple, we work in a highly regulated industry where all products will have to be certified and approved. Recertification is VERY expensive and time consuming if you have to do these certifications for a global market with a lot of local regulations that need to be approved.

    On one product we were kinda lucky, for now. Intel had announced that all the MAX-II CPLDs are EOL, so we got ready to push out one LTB order to cover the production for a few years. We were already starting the evaluation for a new chip and redesign of the product too. But then Intel realized that demand for these older CPLDs is still high enough to keep the fab running for a few years longer. So our redesign got put on hold to free ressources to solve other supply chain problems.

  38. I just finished revision 2 of my Riskeyboard 70 (https://gfycat.com/alienatedflatcanvasback) analog hall effect keyboard PCB with 3D printed switches and stabilizers but I can’t order it! Everywhere is out of STM32F401C*U6 and STM32F411C*U6 microcontrollers. The entire world also appears to be out of the OH49E-S hall effect sensors I was using so basically my keyboard is going to be a no-go for a while, sadly.

    In the mean time I’m trying to find alternate hall effect sensors and MCUs that have at least 6 analog pins + 8 GPIO that supports USB HID (and an ADC clock speed that will let me read 70 sensors in under a millisecond). I’ll probably work out a way to do a hand wire build because there’s still plenty of the through-hole 49e hall effect senors available. Sigh.

  39. We were using the MKL05Z32VFM4 on nearly all of our products. Then they suddenly dried-up. We bought-out all (~100k) of the inventory of a couple of similar variants (Sorry if you were using a MKL26Z128CAL4R)

    Now I’ve been spending my time these past few weeks porting 20+ projects to the other chips. But at least we can build product for a little while longer…

    I would love to show a photo of the tiny adapter board that my colleague designed to mount a 36-WLCSP 2.4×2.4mm (Wafer-Level Chip-Scale Package) on our existing HVQFN-33 5x5mm footprint.

  40. “How Is The Chip Shortage Affecting You?”

    It’s bad. One example is a medium sized telecommunications network I’m supposed to build in Asia where wel need to start shipping to toward the end of 2021. Unfortunately the whole thing is on hold because I can’t get firm-fixed pricing and delivery on the equipment I need. For example shortest lead times on microwave radios is six months IF I order TODAY with 100% payment in-advance or 20% cash down and 80% secured by an at-sight irrevocable and confirmed Letter of Credit (which the beneficiary will immediately discount). If I try to place equipment orders later this year, the manufacturers are all say “all bets are off”. If this situation continues, I’m going to have to shut the business down.

  41. Has anyone developed a pattern of what sorts of chips are harder to find than others? Is it more older parts, or certain packages? Parts popular in certain industries? Certain temperature ratings?
    Have you found a way to do “defensive design”? Obviously there is the order-parts-before-designing-the-PCB, but are there other recommended practices?
    Scanning around DigiKey yesterday, I noticed strange things like discrete semiconductors that aren’t available on full reels, even though there are 100,000+ of cut-tape inventory.

Leave a Reply to Foldi-OneCancel 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.