I think you are still missing my point.

If I am am selling 3 chips, A, B, and C in my XYZ family, with the only difference being A has 32K of memory, B has 64K, and C has 128K, the C chip could be used by customers who need any of the 3, but the A chip can satisfy only the A customers. It makes sense to me to only make the C chip.

And @MrSalts, do you really think takes longer or costs more ti make the C chip? Often, the different versions actually all have the same die but parts of the chip are disabled for the cheaper versions.

Bob

Yield might be the problem.

 

I, like many of you, have experienced the shortage in microcontroller chips. I am using a pic24 in 44-pin TSOP. I chose (before the shortage) the most capable member of the chosen family, just because I like to leave room for expansion.

So, which one becomes unavailable? You guessed it, the one I chose. Ones with smaller memory are still available, which is okay for me because I can fit in the smaller memory. But for a commercial product that needed the max memory it would be a disaster.

So why would Micrichip produce the less capable chips and not the more capable one that can replace all of the smaller ones? It makes no sense to me. They would even benefit from the higher selling price if they sold only the larger one to customers who didn’t need it. And fewer customers would be stuck with no chip they could use.

Bob

So glad I purchase in volume...

 

So glad I purchase in volume...

I've been buying parts and then designing boards. Paying for "digi-reels" in small volumes and double shipping starts to add up but customers don't blink when I quote projects.

 

One reason not mentioned much is that the new narrower line widths needed new production machinery, which, by the way, is horribly expensive and has a long lead time. So the production of the previous technology level could have gone on, but the virus plague did disrupt a whole lot of things. So a big part of the problem was poor planning and not anticipating the jump in demand. So it was a combination of a whole bunch of wrong guesses based on wrong guesses, aggravated by the jump to a newer finer-line technology that is a lot more expensive to get started with.

 

The industry was warning of shortages years ago in the second tier of semi manufacturing. The roller-coaster of semiconductor biz boom and bust has not stopped. The massive investment today is already signaling a coming bust cycle.

https://spectrum.ieee.org/global-chip-shortage-charts#toggle-gdpr

The semiconductor industry has always been cyclical, undergoing gluts and shortages as the market for personal computers, and then home electronics and smartphones, followed the ebb and flow of the global economy. But even before the COVID-19 pandemic struck in early 2020, there were signs that the “market was very, very tight,” says Russell Harrison, director of government relations at IEEE-USA. Semiconductor factories (a.k.a. fabrication facilities, or fabs) typically run at about 80 percent of their rated capacity, allowing time for maintenance, upgrades, and staffing variations. As early as the summer of 2019, the industry-wide utilization level was nearing 90 percent. That is a reflection, says Calhoun, of a growing appetite for connected home appliances and increasingly sophisticated automated driving features and digital connectivity in cars. Utilization hasn’t fallen below 90 percent since the summer of 2020, according to the SIA.

 

Yield might be the problem.

Just as a possibility. When die share the exact same processing steps but with variations in thing's like memory, it's might be reasonable to have masks that can define several die versions on the same wafer as a pattern containing the family variations when the wafer has a large number of die.

 

 

If you do everything discrete the global chip shortage doesnt bother you at all.
Is there going to be a global transistor shortage? I dont think so.

 

I think shortage is partly in the test and assembly parts, making large number of chips on a 12 inch wafer may take say 1-2 weeks but produce 100s of chips on a single wafer. Then comes the test and assembly parts which is far more labor intensive.
I think the biggest shortages are in the microcontrollers (billions are used yearly), now every thing has a mcu in it even for that matter your pen, cup, coaster, rug etc. for all that 'iot'. then the other is in the top end the high core count processors like those 16-32-64-128 - 1000 cores processors.
for the top end the crypto crash probably alleviates things somewhat.

it seemed still quite easy to buy generic op amps and power ICs (e.g. regulators, buck converters, flyback ac-dc converters, battery chargers etc) .
Adafruit once blogged about mosfet shortage, I'm not sure if it is still the case.
https://blog.adafruit.com/2021/11/1...ment-thegreatsearch-digikey-digikey-adafruit/
https://blog.adafruit.com/2022/06/1...-get-through-the-chip-shortage-makerbusiness/
New Teslas delivered with surprise missing USB ports due to parts shortages
https://electrek.co/2021/11/12/new-...ise-missing-usb-ports-due-to-parts-shortages/

the biggest laughing stock would be if the world runs out of mosfet, the basic mosfet that is.

another feature from adaftuit

oh and out of curiosity, would all the phone sellers stop selling you usb chargers saying there are no more chips? here is your phone, no charger. lol

 

The hardware portion of the shortage was caused by the use of the very latest technology, which is the new generation of finer lines, smaller everything,because that requires new production equipment, and because it must be even more precise, it is much more expensive. And because the increase in demand was not anticipated, not as much equipment was ordered. And all of this new equipment is long-lead-time stuff. So now there is not enough of the new production equipment, and also not enough workers to run all of the production.
If the designs had used the previous technology sizes there would have been plenty of production equipment to make the parts.
It is important to see that at the "bleeding edge", that blood must come from somewhere. THere are some real benefits from sticking to more mature technology!!!

 

The previous technology sizes are where the critical shortages are for manufacturing things like cars. There are are massive, expensive fab reconstructions happening in existing 'mature' facilities that will give you more product in the next six months but it will be more expensive as the only equipment you can buy used or new today are designed to also be compatible with bleeding edge upgrades even if they are being used with older technology. There has been nearly zero excess capacity in the 'mature' side for years before this critical shortage. Those with deep pocket will ride it out but expect to see many currently 'critical' fab projects pushed out due to financing pressures.

IMO due to 30+ years in the biz we are looking at a down cycle really quickly after current orders a delivered.
https://seekingalpha.com/article/44...kely-semiconductor-equipment-meltdown-in-2023

In 2021, we again are witnessing a “chip shortage.” Although in a series of four Marketplace articles I only identified microcontrollers as the chip in shortage (automotive MOSFET’s could also be a product which I'm investigating), the press continues to cover the topic.

More importantly, the reaction to this issue is mirroring what happened as a result of poor forecasting by Gartner in 1998.

According to SEMI, the industry consortium, semiconductor manufacturers worldwide will have started construction on 19 new high-volume fabs by the end of this year and break ground on another 10 in 2022 to meet accelerating demand for chips across a wide range of markets including communications, computing, healthcare, online services and automotive.

Based on significant similarities in 2000 and 2021, presented here and in more detail in my Semiconductor Deep Dive Marketplace newsletter, I anticipate semiconductor capital intensity will likely drop beginning in 2023 as it did in 2002 due to a meltdown in equipment sales.

 

Given the competency level of our current administration and given the ongoing agenda of the federal reserve bankers, in addition to their ongoing agenda, the present inflation will end with a crash. and a recession. Not intentional, but because of gross judgement errors. Like a vastly under-damped servo system.

 

Given the competency level of our current administration and given the ongoing agenda of the federal reserve bankers, in addition to their ongoing agenda, the present inflation will end with a crash. and a recession. Not intentional, but because of gross judgement errors. Like a vastly under-damped servo system.

BS, this is not political. The boom/bust cycle is as natural as breathing.

https://www.cambridge.org/core/jour...roman-empire/47837AFC17442503E6E90ED30FAD9D98

 

Adafruit made another feature on chip shortage

This time it is the STM32F405 mcu, it is so short of chips that micropython can't get their hands on it !
It is pretty extreme, STM32F405 is a popular and an 'important' chip. It is a chip that has just large enough memory and flash to run things like micropython, has pretty decent speeds 168 MHz and has the 'ART accelerator' - on chip cache, that reduces flash reads from n wait states (say 3-5 waits) to 0 waits if it is in cache plus an FPU. That acceleration is quite substantial.
And it has a good host of pheriperials 3x pretty fast ADC, DAC then other IO stuff like spi, i2c, uart, usb etc are there. pretty much packed with peripherals.
https://www.st.com/en/microcontrollers-microprocessors/stm32f405rg.html
this shortage will hit quite a lot of players, small and large ones alike.
stm32 mcus are also popular in 3d printers e.g. the more recent 'Reprap clones' are all using stm32 F103 / F4xx and NXP chips, and probably elsewhere drones, medical devices? etc.

 

Adafruit made another feature on chip shortage

This time it is the STM32F405 mcu, it is so short of chips that micropython can't get their hands on it !
It is pretty extreme, STM32F405 is a popular and an 'important' chip. It is a chip that has just large enough memory and flash to run things like micropython, has pretty decent speeds 168 MHz and has the 'ART accelerator' - on chip cache, that reduces flash reads from n wait states (say 3-5 waits) to 0 waits if it is in cache. That acceleration is quite substantial.
And it has a good host of pheriperials 3x pretty fast ADC, DAC then other IO stuff like spi, i2c, uart, usb etc are there. pretty much packed with peripherals.
https://www.st.com/en/microcontrollers-microprocessors/stm32f405rg.html
this shortage will hit quite a lot of players, small and large ones alike.
stm32 mcus are also popular in 3d printers e.g. the more recent 'Reprap clones' are all using stm32 F103 / F4xx and NXP chips

Begging doesn't work, sorry. You can't build anything when raw materials are in short supply. Supply constraints outside of actual internal fabrication are starting to damp output predictions.

 

how soon would chip producers get so short that they tell you they can't ship mosfets, power regulators, ac-dc flyback converters (all your mobile phone chargers), buck converters and just about any part of that 'chip spectrum'?
would the chip shortage squeeze mobile phone supplies too? I think the fad of changing phones is less these days, but nevertheless are phone makers going to be stopped from making phones as there is *no chip*? this would just about extend to any sort of consumer electronics, washing machines, lcd tvs, monitors, computers, laptops, keyboards, mice, motor drivers etc.

for computers you may still be able to buy your $500-$1000 8-32 core processor, but just imagine that there is no chips for the motherboard !

 

It's not just MCU's. It's every integrated circuit. I own a small business and we make things on the order of 100 pcs for the lifecycle of the product. We are ordering parts before we even design them into a product, all at risk of them being gone when we need them.

My understanding is that TSMC can't keep up with demand. Nearly all manufactures use TSMC for their manufacturing. Few exceptions, TI has their own fab. TSMC can't keep up with demand. I've heard through colleagues in the chip industry that some companies are selling their time at TSMC because their time-blocks are more profitable than their product. Si wafers are also in short supply, so I'm told. This doesn't make sense to me since Si wafers are grown and they were keeping up with things before. Perhaps it's in short supply because the manufactures can't staff due to COVID, but I don't think that's true either.

In the meantime, companies are panic buying, just like we are. And companies have the cash and space to buy parts they may not use, so it takes longer for supply to rebound, unlike in the case of toilet paper at the onset of COVID.

Really, it just doesn't make sense to me that this is continuing, and I suspect geopolitical effects to be an additional contributing factor, although I don't have any solid proof nor do I want to discuss politics.

 

for silicon wafer shortage, there are a few fronts, I'd guess the highest quality types could be short as those are used for the top notch 5nm-14nm semiconductors, all that processors and gpus. The 2nd front is 'most other chips' e.g. 25nm-110nm a broad spectrum of it, my guess is most microcontrollers falls in this category, I'm not too sure about the power chips such as mosfets , regulators, buck converters, op amps etc. I'm thinking this may be the 3rd front, i.e. minor defects won't damage the functionality, this could be the mosfets, regulators, buck converters and arbitary transistors, even diodes etc, they could well go as coarse as the 1um levels and smaller wafers are used for these, then the last category is actually solar cells, these are the ones that use tonnes of silicon and energy and the purity would likely be the most slack. I'm not sure how this plays out as in could it that there is extreme demands for miles square or kilometers square / hectares of solar cell requirements that has compromised on the higher quality silicon? it would be a bit strange if it is true as 'sand' is pretty much available.

 

Flat out, there are no more mature machines (new or surplus) or mature manufacturing capacity available at any reasonable price for 2nd and 3rd tier semi products. The technical generation of workers that enabled that technology is retiring or soon to retire and there is just nobody to replace them. The run-up to this has been happening for a decade but the last two years have accelerated the IMO unavoidable chip shortage in those tiers.

 

Makers should also get access to semiconductors. Maybe copyright the term because semiconductor manufacturers area all about the marketing and PR "look what we do for innovation and makers" - yet they are only selling to their biggest customers for 2 years now. They've lost the right to claim they work with innovators, makers, schools etc.

ST and GlobalFoundries just announced they plan to build a $5.7B 300mm fab in France. Maybe STM32's will be available in 2026 lol but the fab is apparently for FD-SOI tech for FinFETs despite the IoT claims.
I can't believe ST executives are so stupid to take 2 years to engage in this no brainer. It must be the government funding they waited for.

The US CHIPs act I'm afraid will never get the $52B funding approved. Congress is jammed in a bad way, politicians bickering to our demise over this. Good that Intel aborted ground breaking on their new fab, they won't get the grants, tax concessions etc. so 1,000 acres of land just sits there. Unbelieveable.