The semiconductor shortage sparked by the pandemic is showing no signs of slowing down. Although auto manufacturers were some of the first affected, the shortage has now spread and is impacting all sorts of projects, including the Smoothieboard open-source CNC controllers.
[Chris Cecil] walks through the production woes they’ve had over the last few months. It began this spring with a batch of the V1.1 boards. The prices of some of their chips started jumping, and then they were informed that the microcontroller that serves as the brains of the Smoothieboard was only available for five times the old price. In the end, they placed a smaller order, and V1.1 Smoothieboards will likely be scarce until the microcontroller’s price returns to normal.
Getting V2 of the boards into production has been even more difficult. Just weeks before the final prototype, it was discovered that the LPC4330 microcontroller the V2 was built around was also sold out worldwide. With the shortage in mind, a hole was left in the layout of the final version of V2 so that they could finish the design around whatever microcontroller they were able to get. In the end, they were able to lock down a supply of STM32H745 controllers, which are actually substantially more capable than the original device.
If you’re interested in the origins of the chip shortage, this article from January is a good place to start. This isn’t the first time parts shortages have wreaked havoc on the world of electronics—does anyone remember the global resistor shortage of ’18?
27 thoughts on “Tales From The Global Chip Shortage: Smoothieboard”
I’m nearly done writing an update to the kickstarter about this. I keep pushing it to later because of work and constant new developments, but now that HaD has made an article about this, it’s the right time to publish I guess. Will come back here with a link.
As promised, the link: https://www.kickstarter.com/projects/arthurwolf/smoothieboard-v2/posts/3293352
Also of interest, this update from Chris: https://www.robosprout.com/status-of-smoothieboard-v2-september-2021-update/
Great info from both of you, nice job laying out the shift in specifications and what it means to, very clear.
Nice to know its getting to the end of the road, excited to get a crack at this new model.
This shortage really is annoying, and far worse for the poorer student/hobbist that will only be ordering 100 of anything max, and probably less than 10… At the moment some chips I’d like to play with are only really available to me by buying a product containing them, which I’ll pass on as that is just more e-waste generated, even though the product is actually cheaper than the chip right now…
Hope the Smoothie V2 gets over the final hump ok, as I’ve been looking forward to playing with it for what feels like forever… Still should be worth it when its ready.
Look and see if what you’re looking for is available to use in any discontinued or decommissioned products, if it’s something like a microcontroller or SBC board that has been made for a specific purpose but doesn’t have a use anymore then you can pick them up pretty cheaply on AliExpress or eBay saving you money and saving the product from going to landfill. I was able to pick up a 1200W HP decommissioned server power supply for very cheap on ebay and Xilinx zynq 7000 ex Bitcoin mining control boards both very cheap to use the fpga from it (board cost £10, zynq 7010 chip costs £75) and it saves them going to waste as they were decommissioned and no longer in use. It’s very good for a student like me these disused or outdated products because of how cheap they are and it teaches you more about it than buying a board designed to be easy for hobbyists to use.
A good point, sometimes easy to apply, and might even apply to the bits I really want – but finding out will be rather hard – as I’m after not at all hobby friendly bits, but single chips like PCIe switching, PCIe – USB3 controller for a few Pi CM4 based projects. Some bits are still easy enough to find, but lots of it is not. I’ve been learning HDL for FPGA stuff, so maybe I could fake those IC’s I can’t find in a FPGA – as you say quite alot of them available, many of them large enough I would think to implement such functions, but damn if that isn’t making life hard on yourself…
Those chips probably do exist on many motherboards that are on the way to be ground up, its just finding a source of enough of the same/ footprint compatible ones that I don’t need to change the PCB for them all, and actually finding the datasheet on whatever one is on the boards you get so you can actually design the PCB at all..
LPC1768 and LPC1769 are used on a very popular aftermarket 3d printer controller board, and the manufacturer was forced to rush the release of a newer version using STM32F407 microcontrollers. Some sources even say that these LPC models were prematurely EOLed.
And to think that this SNAFU will last 12-18 months still, from some sources I’ve been reading.
Probably longer. The global supply chain is backed up a couple of years at this point. For example there is not enough money to get a container on a ship in China because there are no containers returning and no ships to put them on. Then everyone on the planet has doubled or tripled orders to fill back-orders and inventory – from cooking grills to machinery parts. I doubt if anyone is going to build container ships in time, and who would do that just for a temporary bump? The closing of ports has had a cascade of effects that will last a long time. Every 25 to 30 days spent waiting by a ship is a full round trip lost. Loose a couple of trips and how do you ever make that up with existing shipping?
Add the uncertainty from Emperor Xi’s serious demand for a return to Maoism and I guess we are looking at a global reset in manufacturing that will take many years to settle out.
I have lived thru a number of chip shortages in the 80’s and 90’s.
Initially it has been caused by lack of forward orders and a new set of products requiring parts that were not scheduled. Case in point seems that auto manufacturers failed to have forward orders due to their JIT. As soon as shortages are seen, two things happen.
Some manufacturers quickly buy up parts they need for future supply. They also start dupicating orders from multiple suppliers. Scalpers enter the market buying up items in short supply. Both of these severely extend the shortage problem exponentially! Until supply can be ramped up, things will only get worse. Once thing start to get back to normal, there will be a glut of parts as those manufacturers holding inventory built up over the shortage period will start to run down inventory, creating less demand, resulting in the glut.
So what might have been a minor hiccup turns into a disaster.
The commodity memory market have its cycles of under/over production over the year. Costly chip production capacity can only be added in long time frame while companies that chase profits in short sight quarter by quarter.
A feedback system can go into oscillation when you have too much delay at the output.
I have a feeling the auto industry may have helped start it.
I wouldn’t be surprised if they cancelled or reduced planned orders due to the vastly incorrect market predictions when covid started happening.
The auto makers did exactly that. The gave up their contracted chip manufacturing incorrectly thinking the pandemic wold reduce sales. But the cause is basically simpler. The chip makers were running at their ideal peak of 85% capacity, which gives time for maintenance and upgrades and all that. The Wuhan Flu shutdowns lasted more than long enough in China to make chip demand exceed 100% of capacity. It takes years to build a modern fab. So we are either looking at years of shortages or/and a fold back to older fab equipment and parts with larger features. Maybe even 8″ wafers. Plus new fabs outside China, if environmental laws and water supplies will allow and approvals don’t take a decade.
Typo: I think you meant “the auto industry were among the first culprits”. Turns out “lean” mostly works when everything else goes smoothly, so you just push all risk on to others.
Have you by chance heard of Taleb’s Antifragile? The part he doesn’t discuss is that whenever you can gain from disorder, you have less incentive to stop the disorder. Rich people profit from volatility.
It’s not “Weak men make bad times” like the old saying, it’s “Weak men are the victims while the strong make bad times which are good for them “
Yeah, we are all victims of The Man.
We have some products in the prototyping stage that are affected by chip supply shortages too. The solution for now, to be able to build the prototypes for firmware development and EMC pretesting, is to buy eval boards and desolder the components to put them on our board. Unfortunately we don’t have the flexibility to just switch to another component or manufacturer, as there are not many options that fit the requrements of this application.
Last week I was setting up a parts order from the BOM of a new design for a customer. It is a simple project with an STM32. The 4 chips needed are all on back-order at all the regular suppliers. No date for availability. Alternatives? Backordered.
Our alternative? Drop back to a 12 year old Samsung part that runs Linux (And we have enough old stock for 50,000 or so of the boards).
This, too, shall pass.
And once production starts meeting demand, the channels will be flooded by all the hoarders trying to offload their excess inventory.
Hoarders are selling now. Realtek Ethernet chips that were $1 are selling for $7.50 to $12 – if you can find them. WiFi chips, same. RAM and Flash climbing. If you have 100,000 chips you can profit nearly a $million without lifting a finger.
@Stephen Ogier said: “The semiconductor shortage sparked by the pandemic is showing no signs of slowing down.”
TSMC signals global chip crunch may be easing Semiconductor group says carmakers can expect upturn in supplies over coming weeks. Kathrin Hille, Financial Times – 7/15/2021, 12:50 PM:
Pffft, who knows what’s real anymore?
Good for the environment, we will recycle used parts, and use outdated ICs.
I don’t think that the switch to STM32H745 was a good idea. As we speak I am scrambling to find any H7 series MCUs anywhere for a product that we had in development for the past year. Earliest stock coming in 2022.
That’s why we ordered enough H7 chips for the Kickstarter pledges and for initial production runs in advance a few weeks back. We got this covered, for a lot of the other components too.
I’m just glad to hear that Smoothieboard V2 is still progressing forward. Based on the lack of recent updates on their website, I thought i had permanently stalled. I switched my main printer over to Duet a few years back (mostly for silentstep, but I fell in love with RRF), and packed away my V1.1, assuming I would probably not use it again. Fast forward to last month, and I recommissioned it for use on a development test bed I’m making. After loading Klipper FW, i’ve been quite pleased with it so far.
I don’t really have anything of substance to add, just glad that v2 is still moving forward.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)