STMicroelectronics Shortage

[robotix3]

Due to the strike (https://www.eenewseurope.com/news/strike-calls-continue-stmicroelectronics-france) in France and COVID, there is a global shortage of STM micros that is much worse than the average component. How is everyone dealing with this right now? We've reached out to every broker we can find and no one seems to have anything that is close in terms of pin-count and memory requirements to our original MCU.

I've only used Atmega and ST (but mostly ST) micros, what would be the easiest manufacturer to port our existing code to?

[tunk]

I don't have a clue, but I'm left with the impresession
that ST has several microcontroller families. Which one
have you been using?

[robotix3]

We're using the STM32G030K8T6, there are a lot of crosses in other families but they are also suffering from shortages.

[robotix3]

Just interesting that there seems to be no talk of this on social media, being that basically every electronics company on earth is affected by this.

[station240]

https://twitter.com/OmzloElec/status/1331021524174852098

I want 200 STM32F030C6T6 now 
24 Nov 2020

I've got a project of mine I finally got around to making, uses the STM32F042F4P6.
Sure it's available from somewhere, but I don't want to make 10 of something, have them sell out instantly and risk not being able to make them in bulk.
Nor do I want to order 100x, and not have them sell.

[Kjelt]

Hurry up while stock lasts

[exe]

How many do you need? https://octopart.com/search?q=STM32G030K8T6&currency=EUR&specs=0

[VK3DRB]

De ja vu.

During the GFC, Atmel's workers making switch detection IC's in France went on a long strike, resulting in Atmel shutting the plant down. Two week's notice to engineers like myself who had just developed a product using their touch switch IC's. I had to painfully use a Microchip 16F1947 MCU to do adaptive matrix touch switch detection and it cost the project several weeks, headaches and sleepless nights. During this time, Atmel refused to answer any emails about when they would recommence manufacture and they offered offered zero support. They could not give a rats arse. I avoided Atmel like the plague until Microchip bought out Atmel, where maybe there could be some trust re-established. Microchip had a strong history of being a reliable supplier.

The French go on strike a the drop of a hat. Their militant unions often put the French public to ransom. At the bottom of the sewer is the gilets jaune who are basically anarchist criminals. In Australia our unions have not put the public to ransom since the 1970's. We have moved on from those dark days when the once militant Transport Workers Union put the public to ransom with their petrol strikes, train strikes, tram strikes and airline strikes. We barely ever hear of anyone going on strikes in Australia. Maybe STM should consider relocating their plant to Australia. They would be welcomed here.

[JPortici]

one of our contractors gave us a headup on ST increasing lead times... two months ago!
they are beginning to give ever increasing lead times on older parts (like the F1) so that you move on to more recent, (possibly) pin to pin compatible chips.
Other manufacturers as well, like ADI, and they said this is all due to COVID because either the big customers (automotive) have their production lines on hold, or other big customers (medical) need specific parts and they are focusing production on those

[VK3DRB]

The problem STM will have is people will avoid using their micros in new designs if things stay bad for too long, and move to more reliable suppliers. It sounds like STM has no alternative source. Remember the RAM shortage of the late 1990's where prices skyrocketed. This was partly due to the one Sumitomo chip plastics plant (Indonesia) going up in smoke. The industry was at fault for having a single source. The only winners were brokers who hogged the memory chips. Having a single source is risky, and particularly foolish when the parts are made in a country run by a tyrannical dictator. The major banks of the world have hidden backup locations in difference cities, so if the central processing bureau is hacked by "state actors" or bombed, the other can continue operations effective almost immediately. Seems it is not the same with key electronics parts where there are no alternatives.

[exe]

Has anyone tried EFM32? https://www.silabs.com/mcu/32-bit . I set on stm32 because I'm familiar with ChibiOS, but may be it's time to move on or try something new.

[dietert1]

The idea that ARM MCUs are becoming difficult to buy is strange. And if so, it appears a problem of the supply chain rather than of production of a certain brand.

And the approach of implementing one design with no variation possible may not be enough in a crisis. At least one should be able to make the product with another member of the same MCU familiy, like in the example above F6P6 instead of F4P6. Better you can have two designs with one MCU from STM and one from Microchip. I think it is great that we can work with a common code base, at least the application part.

Regards, Dieter

[Kjelt]

Porting code for a serious product say 100-200kloc from uC manufacturer A to manufacturer B even when the core is the same and the compiler is the same can take a team of three SW devs about 6 to 9 months of time depending on the amount of testing you require.
My previous company has done this multiple times because of new peripheral features for future features and/or lower price/quantity, but there can be lots of surprises. You would like to already have automated endproduct system testing in place for product verification.

[dietert1]

A pain if you have to do it in a case of emergency. If you plan it from the beginning it may even help to better structure the firmware into system parts that depend on the hardware platform and application parts. Yes, a redundant effort will cause extra cost. But that depends. In an industrial supply chain you may win a contract if you can prove that you have a plan B.  In former times they would call it "second source".
I remember very well the discussions in March when it turned out that China won't be able to supply protective medical equipment for the whole world, because they needed it themselves. Here in Germany people became nervous and started telling lies about the necessity of masks etc. Later they said they will enforce national supplies in critical sectors of public health. Don't know whether it is ever going to happen. Nobody wants to be second source.

Regards, Dieter

[S. Petrukhin]

We're using the STM32G030K8T6, there are a lot of crosses in other families but they are also suffering from shortages.

Are you know that the Chinese comrades have a clone of STM32 called GD32. I tested GD32F103C8 - didn't notice any problems. 

[VK3DRB]

We're using the STM32G030K8T6, there are a lot of crosses in other families but they are also suffering from shortages.

Are you know that the Chinese comrades hav a clone of STM32 called GD32. I tested GD32F103C8 - didn't notice any problems. 

We should not be supporting piracy and IP violation, irrespective it is is made by comrades in China or comrades in the (former) Soviet Union. The Soviets pirated a lot of copyrighted chips made in the USA. The GD32 might work "the same" as the STM32 but no-one in their right mind would use it for a commercial, industrial or medical product. You'd get sued if you substituted one of these for the real thing and something went wrong, eg: early life failure, corner case issues, temperature issues etc. Besides, you have no come-back when something goes wrong and no support.

[ANTALIFE]

We're using the STM32G030K8T6, there are a lot of crosses in other families but they are also suffering from shortages.

Are you know that the Chinese comrades hav a clone of STM32 called GD32. I tested GD32F103C8 - didn't notice any problems. 

We should not be supporting piracy and IP violation, irrespective it is is made by comrades in China or comrades in the (former) Soviet Union. The Soviets pirated a lot of copyrighted chips made in the USA. The GD32 might work "the same" as the STM32 but no-one in their right mind would use it for a commercial, industrial or medical product. You'd get sued if you substituted one of these for the real thing and something went wrong, eg: early life failure, corner case issues, temperature issues etc. Besides, you have no come-back when something goes wrong and no support.

Full of bullshit as expected from you.

Where is the BS in that comment?

Also the author of this blog contacted ST about GD, and yea it looks like GD is not doing things by the book:

Those GD32xxx devices are effectively a kind of clones of our STM32, but it's a pure piracy: no agreements of any kind between ST and Giga Device, no license... nothing. ST legal people are in charge of this problem.

https://olimex.wordpress.com/2015/11/09/chinese-clones-attack-stm32-microcontrollers/

[BravoV]

Unethical, maybe, as unethical as AMD vs Intel, but nothing illegal.

They're too young to know and aware of that.

[Siwastaja]

We're using the STM32G030K8T6, there are a lot of crosses in other families but they are also suffering from shortages.

Are you know that the Chinese comrades hav a clone of STM32 called GD32. I tested GD32F103C8 - didn't notice any problems. 

We should not be supporting piracy and IP violation

Making compatible products (i.e. reproducing same public external interfaces: like having a steering wheel) has absolutely nothing to do with piracy or IP violation.

Similarly, I could call you a murderer. Which you hopefully are not. Like Gigadevices are not pirates or IP violators.

There is also no proper ground over which you could get sued by using Gigadevices products. On the other hand, I can sue you over a murder. It's simple: get a lawyer to apply the paperwork. It's just not going to fly very well, unless there is a political reason for it to work.

AMD vs. Intel is a great analogy, really (there was Cyrix and others, as well). Try to understand it. And in the end, both have benefited from it; AMD has reproduced the Intel's interface, and later, Intel has reproduced the AMD's interface (AMD64) when it worked out better.

[VK3DRB]

Remember the FTDI FT-232 chip debacle. Knock-offs were made in China at a fraction of the original cost. When FTDI did a strategic driver update for their chips, the fake chips ceased to work. That was smart but with one BIG exception. Through the supply chain, innocent purchasers of equipment using those knock-offs got their equipment bricked, which FTDI presumably failed to anticipate along with the backlash that followed. Piracy damages industry, stifle innovation and rips off the innovative inventors of technology. Ethical engineers will use the original devices that are still under patent protection and shun the "copy watch" brigade. Same with music and videos - unless it is public domain, piracy is theft. Even the egg-throwing badly-behaved Justin Bieber is entitled to the rewards of creating so-called intellectual or artistic property.

IMO, if functionality within the STN chip is protected by patents, and if the GD chip violates these patents, we should not be using them, even if STM supply is crippled due to a strike. But if no patents are violated, use the GD chip by all means, but of course at your risk.

https://olimex.wordpress.com/2015/11/09/chinese-clones-attack-stm32-microcontrollers/

[chickenHeadKnob]

AMD vs. Intel is a great analogy, really (there was Cyrix and others, as well). Try to understand it. And in the end, both have benefited from it; AMD has reproduced the Intel's interface, and later, Intel has reproduced the AMD's interface (AMD64) when it worked out better.

Not a good analogy. Intel and AMD had a very profound and comprehensive second source agreement for a wide variety of Intel designs, eproms, 8086, 80286, '386, 8051 and more. Intel provided AMD with full mask tapeouts. In the 1970's the idea was that customers, (especially the military) of chip company’s  would be very reluctant to adopt a design unless it could be source from multiple fabs.
 
 Intel was not the behemoth it is today. It started out making small DRAMS to replace core memory in mainframes. The first DRAM was  intel's 1103. Then IBM adopted the 8088 and Intel became a processor focused company. By the mid 1980's Intel was fed up with handing masks to AMD and wanted out of the contract pre-maturely. AMD was still able to make 386's from intel masks but that was the last in the second source agreement. Lawsuits were launched and I think they were settled out of court.

 As far as I know GigaDev never had such an intimate business arrangement with ST.

[dgtl]

Yeah... F03x and F07x in smaller packages were gone from all suppliers. Had to redesign multiple boards to larger package to get manufacturing going again. Not nice at all.

[robotix3]

We were finally able to secure enough of an ST MCU in the same family for the next few months.

Still likely switching over to NXP since someone was saying they are a more reliable supply. (Let me know if you don't agree with this) After a few months we have no idea what our volumes will be so if ST MCUs are still at a 30 WK LT then we are in a bad place.

I've already started porting our code to NXP and so far it is a good experience, toolchain is very similar to ST.

EDIT: Also, anyone going to short STM? It's at an all time high and I'm betting a ton of companies will switch over to someone else because of this.

[Kjelt]

ST vs GD : ST pays hefty fees for use of the Arm Cortex core design, does GD too or did they copy paste that as well 

[tonyh88]

Interesting info about this shortage. Looks like not all the stm32 are affected equally by this shortage or maybe demand is lower for some families.