How is Chipageddon affecting you?

[Messtechniker]

I'd rather take delivery times of important
components like 0.1 uF Cs, for example, as the indicator.

You have found an indicator that there are no shortages. 

Just looking at DigiKey, they have over 800 million stock of 0.1 uF ceramic chip capacitors in 0201, 0402 or 0603 case, with nearly all of their normal SKUs in stock (~90%).

Likely these are available in large numbers because there are no chips available to use them with 

Seems, I've suffered a bit of a brain fart. Sorry.

[Ed.Kloonk]

I'd rather take delivery times of important
components like 0.1 uF Cs, for example, as the indicator.

You have found an indicator that there are no shortages. 

Just looking at DigiKey, they have over 800 million stock of 0.1 uF ceramic chip capacitors in 0201, 0402 or 0603 case, with nearly all of their normal SKUs in stock (~90%).

Likely these are available in large numbers because there are no chips available to use them with 

Seems, I've suffered a bit of a brain fart. Sorry.

I think it's good that we can, as engineers and gentlemen, postulate theories and solutions. We all benefit.

[Smokey]

Thanks Zero-Covid lockdowns!
This isn't chip specific, but my boards from PCBWay just turned from the DHL-2-day shipping I paid for into DHL-6-day-maybe**

**(or whenever it gets there, whichever comes first).

[chickenHeadKnob]

GPUs come off of the high end TSMC fabs and graphics cards also have memory and power bus components.
What sayeth the EEVBLOG forum braintrust?

The thing to keep in mind is that semiconductors are, to a significant extent, non-fungible. There are probably only one or two fab lines in the world that ONSemi has qualified to produce your particular MOSFET, and none of them are in common with anything nVidia or AMD sells. All that means is the odds a relaxation in demand for one part used on a graphics card will bring forward the production schedule the specific parts you need are low.

This is a valid and cogent counter-argument and one that I considered. Yes individual chips and fab processes are non-fungible, but I would say they are fungible-adjacent in the sense that the whole industry is a kind of moveable feast which transitions to smaller process nodes over time. For industry sectors that have short product cycle times  like smart phones, PC's, memory, ect. the move to smaller nodes is baked in and relentless so they will be redesigning in new parts anyway.

For stodgy slow market segments like industrial controls there will be some customers that wont budge and painfully wait for micro-controller X to become available again while others who need X will redesign for micro-controller Y - the new die-shrunk hotness in the X series just to get back in business again. That relieves some demand pressure on the older parts.

I was recently looking at the AVR DA's and their availability and  wondering if something like this isn't already in progress.
You can say "well all that takes time to happen", sure but we have been at this for  about 2 years.

[AndyC_772]

There are plenty of components for which the relentless drive to newer, finer geometry parts really doesn't exist, though.

For example, a linear voltage regulator is what it is; chances are that the device in your existing design gets warm at a rate determined entirely by the voltage drop and current, and dissipates that heat through a package that's governed by geometry and the thermal conductivity of copper. It's essentially 'problem solved', there's no benefit to be had by redesigning a product to use a newer one, and a smaller die certainly isn't inherently 'better'.

The same might be said of many other components; once a design is proven, swapping to, say, a newer op-amp doesn't represent an opportunity to improve the product, it represents the need to re-test and qualify something already known to work. A change (even to a 'better' part) is almost entirely a burden for the manufacturer, not a benefit.

These parts might not be exciting or high value, but almost every design needs them. And so what if your CPU has been replaced by a new one that's half the die area, and as a result, fractionally cheaper? You've still got to redesign the board to use it, and fingers crossed you can get the other parts that it needs in order to be useful.

[coppice]

There are plenty of components for which the relentless drive to newer, finer geometry parts really doesn't exist, though.

For example, a linear voltage regulator is what it is; chances are that the device in your existing design gets warm at a rate determined entirely by the voltage drop and current, and dissipates that heat through a package that's governed by geometry and the thermal conductivity of copper. It's essentially 'problem solved', there's no benefit to be had by redesigning a product to use a newer one, and a smaller die certainly isn't inherently 'better'.

The same might be said of many other components; once a design is proven, swapping to, say, a newer op-amp doesn't represent an opportunity to improve the product, it represents the need to re-test and qualify something already known to work. A change (even to a 'better' part) is almost entirely a burden for the manufacturer, not a benefit.

These parts might not be exciting or high value, but almost every design needs them. And so what if your CPU has been replaced by a new one that's half the die area, and as a result, fractionally cheaper? You've still got to redesign the board to use it, and fingers crossed you can get the other parts that it needs in order to be useful.

As you say, a large amount of analogue stuff just doesn't scale, but the issue it broader than that. You have an ultra low power digital device you want to shrink? How are you going to do that when the ultra fine processes leak like a sieve? Most really ultra low power applications spend a lot of their time idling, waiting for the next burst of activity to be triggered. The run current is less significant than the leakage during the idle periods. You have a relatively simple digital device with a lot of I/O pins it controls? How are you going to shrink that when the I/O pad ring limits the die area? People are using multiple rows of pads in the I/O ring now, but pad limiting is a major issue for many things as they shrink.

[madires]

Some chip-starved manufacturers are scavenging silicon from washing machines, as global shortage shows no sign of easing (https://www.scmp.com/business/article/3175018/some-chip-starved-manufacturers-are-scavenging-silicon-washing-machines)

[tom66]

We've been buying PIC24 dev kits and desoldering the chips off them to make €x0,000 imaging systems, because that's cheaper than buying the chips from Win Source (can't get them anywhere else).

[SiliconWizard]

I've seen this being increasingly done indeed. Pretty sad days.

[b_force]

Besides the fact that I (we) had to postpone 90% of the projects planned for this year (and last year), I decided to do some smaller projects on the side.

apparently the vast majority of precision resistors (like 0.01%) are just also being affected by all of this.
Barely any value available anymore (if anyone has an handful 500R and/or 250R, 0805 or 1206 in stock, let me know)

[floobydust]

I believe we're in a second wave of the semiconductor shortages, it has gotten much worse.

95% of Microchip SAMD21, 51 etc, Mega328's even, STM32 MCU's out of stock with 12-18 month lead-times for fall 2023. Then, when you go to design in any other available processor, the ICD is unobtainium i.e. Segger J-Link, PIC ICD4 etc. is "not available". It's more than a drought.
No sign of 28nm Raspberry Pi's either for many months, unless you don't mind the scalper's 400% markup. CEO update is pretty much useless and full of the usual "hope mode" narrative.

For the smaller businesses, I predict a mass extinction event - they can't get enough parts to make products and stay afloat. It's going to cause a die off and only a matter of time before even the PCB assembly houses and anything downstream are twiddling their thumbs.
Automotive electronics manufacturers I know are dealing direct with NXP, Infineon etc. and not through any distributors, parts have been ordered 2 years in advance. They can remain operational albeit at reduced volumes.
But the small fish, has no voice to the semi manufacturers. Who wants to be an EE right now?

[AndyC_772]

Who wants to be an EE right now?

Actually I have a lot of work on right now, business is booming. The ability to redesign products to use parts that do happen to be available is a skill that's in high demand, so it's really not a bad time to be an EE at all.

That's true in the near to medium term. Long term, of course, if parts continue to be unavailable, my customers will inevitably suffer.

[b_force]

Who wants to be an EE right now?

Well it does force you to open up the book of tricks and creativity, as well as dusting off those old 70s,80s and 90s books about discrete circuits.

Simple transistors are still available.

[rx8pilot]

Who wants to be an EE right now?

In general.....not me.

With that said, opportunities to repair high-end commercial electronics has become a 'name your price' situation. Since people cannot get new stuff, they will pay $$$ to keep existing gear running. I am not, however, configured to be a repair house.

I have been paid to re-design a few products in an attempt to keep them on the market. Then, the newly chosen parts become unavailable and I do it again. Not sustainable, but at least a short burst of paid work.

My primary push at the moment is 100% mechanical devices that I can machine myself. So far, that is working WAY better than any EE projects. Of course that only works because I spent a very long time doing that before going into electronics and I have the tools and software to do it.

[IDEngineer]

You have a relatively simple digital device with a lot of I/O pins it controls? How are you going to shrink that when the I/O pad ring limits the die area? People are using multiple rows of pads in the I/O ring now, but pad limiting is a major issue for many things as they shrink.

We went 3D to shrink transistors... maybe we need to think outside the box for high pin count devices too. I wonder if we could put pads on both sides of the die, and build a "via" on each layer on the way up. Yeah, call me crazy, but fab technology has successfully done some crazy sounding stuff in the past.

[floobydust]

[...] Simple transistors are still available.

I'm not talking about hobby projects, and actually looking at discretes you can see no stock from many usual manufacturers. They are disrupted as well but not to the point of unobtainium.

The shortage is very complicated and perusing industry news, it's all over the place. For every article sounding optimistic hurrah 10 new fabs this year, there's others saying trouble ahead. Many new fabs are going to be EUV which doesn't help us - that's not the bread and butter EE's need. "European industry, in general, does not require many of the cutting edge, sub-10nm chips, says Julia Hess at SNV, who adds, "The demand in Europe is basically focused on industrial and automotive demands and these kind of chips do not rely on cutting edge fabrication."

I'm doing more firmware work instead, instead of the panic and grind for alternates. But I worry about the small fish companies, right now they can't make much product.

[cdev]

Here in the US  there seems to be a lot of complaining in the media about allegedly "zombie" companies. Including electronics companies but not specifically. As they put it these "zombie" companies are hanging around too long and lowering profits by staying alive despite being cut off from various income sources, when they were supposed to just die. .

TPTB seem to be agitated about this. Maybe what's really happening is an pre-engineered Greed Crisis.

Graphic:  This figure plots the share of publicly listed firms (left panel) and the share of private firms (right panel) that are in zombie status according to alternative measures. The financial press measure considers firms that are mature and have ICRs below one to be zombies. The metric used by the academic literature adds to a low ICR requirement the condition that firms have high leverage and rely on cheap bank credit (that is, firms pay interest rates that are below those applied to the most creditworthy (AAA-, AA-, or A- rated) companies).

Source: Authors' calculations based on S&P Global, Compustat, Wharton Research Data Services, http://wrds.wharton.upenn.edu/; Federal Reserve Board, Y-14. See also:  This figure plots the share of publicly listed firms (left panel) and the share of private firms (right panel) that are in zombie status according to alternative measures. The financial press measure considers firms that are mature and have ICRs below one to be zombies. The metric used by the academic literature adds to a low ICR requirement the condition that firms have high leverage and rely on cheap bank credit (that is, firms pay interest rates that are below those applied to the most creditworthy (AAA-, AA-, or A- rated) companies).

Source: Authors' calculations based on S&P Global, Compustat, Wharton Research Data Services, http://wrds.wharton.upenn.edu/; Federal Reserve Board, Y-14. See also: https://www.federalreserve.gov/econres/notes/feds-notes/us-zombie-firms-how-many-and-how-consequential-accessible-20210730.htm#fig3

[b_force]

[...] Simple transistors are still available.

I'm not talking about hobby projects,

I wasn't either.

[tom66]

Actually I have a lot of work on right now, business is booming. The ability to redesign products to use parts that do happen to be available is a skill that's in high demand, so it's really not a bad time to be an EE at all.

That's true in the near to medium term. Long term, of course, if parts continue to be unavailable, my customers will inevitably suffer.

Similarly I'm not in contracting but I'm seeing offers for contractors & salaries in my field well above the 2019 median.  I've seen jobs in Leeds for EE's advertised at £60 ph - that used to be £37-40 ph for top candidates.  Salaries are good right now too.

So despite the shortages(?) the market is booming -- which suggests these businesses are not struggling to get the parts they need.

[cdev]

God forbid they could not get those Broncos.. ha ha.. Pickup trucks tend to be a very fuel inefficient kind of vehicle..

It's ugly with chipmakers playing favorites.
Thousands Of ‘Unfinished’ Ford Bronco Pile Up, Can’t Be Delivered Due To Chip Shortage
Ford's approach is to keep building and advise customers "just 3 more months"... which is nothing in semiconductor fab times.

[Ed.Kloonk]

God forbid they could not get those Broncos.. ha ha.. Pickup trucks tend to be a very fuel inefficient kind of vehicle..

It's ugly with chipmakers playing favorites.
Thousands Of ‘Unfinished’ Ford Bronco Pile Up, Can’t Be Delivered Due To Chip Shortage
Ford's approach is to keep building and advise customers "just 3 more months"... which is nothing in semiconductor fab times.

What gets me is they never explain exactly what the missing chips do.

[floobydust]

Canada has more zombie companies on the stock exchanges at around 16% verses the global average 10%. I worked for one, it was surreal they lost money but used investors for life support. Executives got the million dollar wage plus bonuses, they were quite happy. Put the cash into marketing hype.
Lordstown Motors, Nikola same game. I wonder when they'll have product to sell. Would you be an EE there? Constant flogging, stock options... and of course a fantasy to maintain.

Bronco now has a "navigation removal" change due to the semiconductor shortage. Who needs a nav/infotainment anyway lol. They sure look cheaply made and what a tacky dashboard.

I say let the automakers suffer, they have no loyalty or interest, no nationalism- as they source the cheapest parts from anywhere in order to maximize profit.
The Bosch Dresden plant makes total sense for the German car makers to have semi's.
I say Intel is out to lunch building EUV fabs that won't be pounding out the bread and butter semiconductors.

[Ed.Kloonk]

But why would a car manufacturer let cars pile up with pieces missing? Knowing Ford, they would sell cars without doors if they had to. There must be a regulatory reason they can't let these cars onto the road.

Maybe they could do a George Lucas and sell the empty box and promise to send the toy when they're available.

[peter-h]

Automotive electronics manufacturers I know are dealing direct with NXP, Infineon etc. and not through any distributors, parts have been ordered 2 years in advance. They can remain operational albeit at reduced volumes.

In that case, who is using up all the chips?

They can't be going into thin air.

None of this doom and gloom stuff is adding up.

It has to be simple hoarding by somebody. But who? I reckon the phone makers and medium size industrials are hoarding it. The car makers are powerful but got caught out because they ran a too-tight supply pipeline and always working on the principle that f*****g their suppliers will always be possible.

The industry media are publishing crap; they always were.

[tom66]

It's *everyone*.  Parts are getting bought as soon as they are in distributors hands, because the demand is so high.  It's just like the toilet roll shortage.  Shelves get stocked,  product gets bought.   There's no significant "shortage" really - just really high demand. 

For semiconductors it's a little bit of column A and a bit of column B,  there is a mild parts shortage caused by COVID aftereffects, fabs running at lower capacity,  ST micro had a strike recently at one of their plants ... but, the real problem is high demand.

Companies are buying stock whenever it appears because they need this stock to make their products.  They're also buying more than they need for that month's production, for fear of the very shortages their actions are creating.  Rational self-interest leading to irrational outcomes.

And yes, scalpers are a factor too, no doubt... but I highly doubt that Win Source's buddies can buy the millions of PIC24's that happen to be available in any given year. They're just getting lucky with the ones that run out at distributors first.