How is Chipageddon affecting you?

[Kasper]

About 3000 CP rail workers just went on strike / lockout.  CP rail is not the biggest railway in Canada but owns about 20Mm of track.  Not sure if this strike will impact chipageddon but it's an example of more possible decreases in productivity.  I wouldn't be surprised if we see more strikes as workers are essentially getting pay cuts while inflation outpaces wages.

[floobydust]

CP Rail is doing great, stock ever-climbing and shipping record amounts of crude oil. Endless profit but flat wages for employees, well what could go wrong.

[TimFox]

When I moved to my house in the Northwest Side of Chicago, I was close to the junction of the Chicago, Milwaukee, St Paul, and Pacific Railroad, and the Chicago and North Western Railway.  Now I live near the junction of the Union Pacific and Canadian Pacific Railroads.

[IDEngineer]

Suddenly I feel like I'm playing a session of Rail Baron. One of my all-time favorite board games!     

[TimFox]

By the way, after I posted above, two heavy freight trains (one in each direction) passed my house on the CP.  I think the laws concerning railroad work stoppages differ between the US and Canada.

[Kasper]

I'm guessing the union that is striking just covers Canadian workers: Teamsters Canada Rail Conference, which represents some 3,000 engineers, conductors, yard workers

[floobydust]

Maybe we need a new video game, Semiconductor Tycoon? Oh wait you need semi's to run a computer to play it.
Strategies wouild be stomp on engineers trying to design in your stuff lol, hoard neon, sell to scalpers.

Intel is spending big money $19B on a fab in Magdeburg, Germany as well other places in Europe. It won't be running until 2027 at the earliest.
We're in for at least 5 more years of shortages, unless a recession hits.

[SiliconWizard]

Intel is spending big money $19B on a fab in Magdeburg, Germany as well other places in Europe. It won't be running until 2027 at the earliest.
We're in for at least 5 more years of shortages, unless a recession hits.

That's good but not everything will be handed to an Intel fab.
And, that would help assuming Europe is still stable enough in 5 years. I honestly don't know.

[peter-h]

This is not a shortage of fabs.

This is simply due to everybody with more than 2 cents to rub together spending it on loads and loads of stock. No different to people going crazy on stocking up with toilet rolls during coronavirus lockdowns.

It was all triggered by the virus, same as the Russian invasion of Ukraine (via Putin's self imposed isolation and going mad as a result), same as so much else.

Eventually it will come to an end, loads of stock will end up with surplus dealers, Ebay, etc, and a few designers will have learnt a few hard lessons.

Hopefully JIT will be treated with the contempt it deserves, although I doubt it because the practice of f******g the smaller companies in the supply pipeline is an essential part of business, along with an MBA in "supply chain management" from the Univ of Upper Chippingham.

[Kasper]

About 3000 CP rail workers just went on strike / lockout.  CP rail is not the biggest railway in Canada but owns about 20Mm of track.  Not sure if this strike will impact chipageddon but it's an example of more possible decreases in productivity.  I wouldn't be surprised if we see more strikes as workers are essentially getting pay cuts while inflation outpaces wages.

Doesn't look like this will last long.  Today's headline is:

"Workers back on the job at noon after CP Rail and union agree to final arbitration"

[coppice]

This is not a shortage of fabs.

Really?

This is simply due to everybody with more than 2 cents to rub together spending it on loads and loads of stock. No different to people going crazy on stocking up with toilet rolls during coronavirus lockdowns.

It was all triggered by the virus, same as the Russian invasion of Ukraine (via Putin's self imposed isolation and going mad as a result), same as so much else.

Shortages are mostly triggered by real problems. People pilling on, and grabbing stock is a consequence of a real problem, which makes things worse. Initially factories being shut down by COVID reduced the demand for semiconductors. Then, the surge in demand for anything related to people's new activities, like working from home, created real surges in demand. Then the pile on made it worse. As a vendor you need to look really carefully at these orders flooding in, because in most of these panic buying periods many of these orders will be cancelled before they are fulfilled. You can end up with massive stockpiles you will never shift if you are not careful.

Eventually it will come to an end, loads of stock will end up with surplus dealers, Ebay, etc, and a few designers will have learnt a few hard lessons.

Surprise, surprise. Surges in demand don't persist, and are followed by slumps. Whoda thought? This is why many semiconductor vendors are not that enthusiastic about making massive new investments during demand surges. Its more reliable to put efforts into squeezing out a little extra production from existing capacity during the duration of the boom. Interestingly the general global economic cycle is about 11-12 years, but cycles in the semiconductor industry run at twice that rate.

Hopefully JIT will be treated with the contempt it deserves, although I doubt it because the practice of f******g the smaller companies in the supply pipeline is an essential part of business, along with an MBA in "supply chain management" from the Univ of Upper Chippingham.

JIT has HUGE upside. Over the long term bit players are not going to care about the short term problems it may be causing right now.

[peter-h]

No, JIT is just another word for "screwing a supplier who is less powerful than you are, and forcing him to keep a load of stock at no cost to you".

This is because JIT doesn't exist all the way back to the raw materials which god put in the earth The pipeline needs buffering at various stages, and the weaker members get f-ed to keep the buffer stock.

Shortages are mostly triggered by real problems. People pilling on, and grabbing stock is a consequence of a real problem

That's what I said, so why disagree?

[coppice]

Shortages are mostly triggered by real problems. People pilling on, and grabbing stock is a consequence of a real problem

That's what I said, so why disagree?

You said there is no shortage of fabs. There is a shortage, but panic buying as soon as the shortage becomes obvious makes the problem appear a lot worse than it really is.

[IDEngineer]

It's not all just "shortage of fabs". For example, unless Bosch had a fab fire I didn't hear about, their problems appear to be 100% self inflicted. Nobody can get any data out of them- not large customers, not their own authorized distributors, not their online support people. Yet they're quick to chastize anyone who dares breathe a word of "problems" in their forums. Reminds me of that Iraqi authority on TV insisting there was no invasion happening while you could literally see US tanks rolling down the street live in the background.

I don't mean to obsess over Bosch, but honestly I simply cannot believe the degree of incompetence coming from them. I've seen crazy stuff from semi manufacturers but this is just mind-numbing. We've used LOTS of Bosch products in the past but....

[IDEngineer]

JIT has HUGE upside.

Yes, on paper, JIT is very efficient. But in the real world it only "works" if everything goes according to plan. "Properly" implemented, JIT leaves you vulnerable to a single glitch from a single vendor. I've been in this industry long enough to admit that I cannot predict the unpredictable.

I've never - not once - seen a "JIT" shop that hasn't been burned. Most telling, a large customer of ours did a careful financial analysis of their JIT/stockroom/vendor stuff for CY2020 and to the shock of upper management (but absolutely nobody below them, least of all Production and Purchasing) they were able to numerically prove that JIT was costing them money. By the time they included expedited shipping charges, the costs of change orders to substitute alternative parts, interruptions in flow on the floor, etc. they were net negative. At that point, upper management allocated additional funds to Purchasing and instructed them to start building minimum threshold inventories on critical (especially single source) parts. Since that time, well over a year now, they haven't had a single interruption.

Manufacturing companies only make money when they're actually shipping and billing for products. JIT sounds great, but Job Number One must always be "keep production online". If JIT risks that, it must be backed off until it doesn't.

[T3sl4co1l]

Just more examples of fundamentally misunderstanding the concept of JIT.  As managers are wont to do.  To reiterate: it's the process of applying statistical methods to control inventory levels.  You keep just enough inventory on hand to cover expected surges in production use, or droughts in supply.  Tie in supplier deliveries / contract timing too, why not.  Anything that correlates, bring it in.

Put another way: there's

seen a "JIT" shop

and there's

JIT sounds great,

.  Real JIT includes:

but Job Number One must always be "keep production online". If JIT risks that, it must be backed off until it doesn't.

so it's not that it's "backed off", it's that -- as is usually the case -- the greedy C-levels see shiny buzzwords and reduction of inventory and dollarsigns in their eyes, and punch the button before they see there's actually work involved in doing the thing.  And then they blame underlings, investors, regulators, whoever when it all falls apart, also as is usually the case. 

Tim

[SiliconWizard]

Well, of course. Because JIT is one of those terms that can mean anything depending on some underlying property that is hidden if you don't mention it.
The "in time" must be defined: in time for what? For keeping a safe stock level, as you just explained, or for getting the closest to catastrophe as possible? The latter is unfortunately not uncommon. I don't know if it comes from a misunderstanding, or from a deliberate choice of playing with fire while mnimizing actual stock, which is often considered "dead money" by those accountants.

[Kasper]

Just more examples of fundamentally misunderstanding the concept of JIT.  As managers are wont to do.  To reiterate: it's the process of applying statistical methods to control inventory levels.  You keep just enough inventory on hand to cover expected surges in production use, or droughts in supply.  Tie in supplier deliveries / contract timing too, why not.  Anything that correlates, bring it in.

Put another way: there's

seen a "JIT" shop

and there's

JIT sounds great,

.  Real JIT includes:

but Job Number One must always be "keep production online". If JIT risks that, it must be backed off until it doesn't.

so it's not that it's "backed off", it's that -- as is usually the case -- the greedy C-levels see shiny buzzwords and reduction of inventory and dollarsigns in their eyes, and punch the button before they see there's actually work involved in doing the thing.  And then they blame underlings, investors, regulators, whoever when it all falls apart, also as is usually the case. 

Tim

So you are saying JIT means 'just in time when everything goes wrong'?

I often push back against unnesisarily tight tolerances.  They require extra planning and increase risk of small problems turning into big problems. 

A good system makes it hard for users to screw up.  As you say JIT invites the C-levels to make mistakes and blame their underlings.  Us underlings ought push back on the very notion of JIT.  Sure it could work well in a perfect world but if it invites improper implementation then it's not a great system.

What costs more: accurate JIT planning and updates and ocassional delays or larger inventory?  I'm guessing JIT planners aren't cheap.

[IDEngineer]

Certainly. Any process operated open loop is going to hit at least one extreme. "Buy too much inventory" and you have a cash flow crisis where you're essentially "property poor", having tied up too much of your asset base. "Buy too little inventory" and you become hypersensitive to glitches in the supply chain. The problem is that your "greedy C-levels" only see their spreadsheets and the debt service and inevitably want to push the ratios to the limits to "maximally optimize" (what they believe to be) the processes. As you point out, they lack understanding and/or appreciation of how the complete process works. So they run it open loop and - surprise! - it saturates.

This is an insideous temptation. I once attended what was supposed to be a financial education class given by the VP Finance to the Engineering senior staff. The theory was that "if we only understood the numbers" we could help the company work better and be more profitable. The guy started his first class by describing a hypothetical company with a hypothetical product; the company had done the R&D and knew how to build it, done the marketing and knew how to sell it, and had existing margins and ratios which they understood. His first questions: "How far can we scale this? How many times can we increase production and keep these multiples the same?"

His answer: "Infinity. Forever." No one else in the class dared say anything. So I raised my hand and pointed out that there were all sorts of variables he wasn't considering. Some would improve the numbers (example: further economy of scale) and some would limit them (example: market saturation). Until he knew these numbers he could not authoritatively state that they could scale infinitely. As a final nail in the coffin, I pointed out that there HAD to be an upper limit because the Earth's population is finite - there were only so many buyers even possible.

This VP actually dismissed my points, and claimed with a straight face that a company could, in fact, scale infinitely. He wasn't angry, he just literally could not see past his MBA training. That was the first and last class of that "series" that I bothered to attend.

Sometimes the folks who "punch the button" don't know enough to do so accurately.

[T3sl4co1l]

Well, of course. Because JIT is one of those terms that can mean anything depending on some underlying property that is hidden if you don't mention it.
The "in time" must be defined: in time for what? For keeping a safe stock level, as you just explained, or for getting the closest to catastrophe as possible? The latter is unfortunately not uncommon. I don't know if it comes from a misunderstanding, or from a deliberate choice of playing with fire while mnimizing actual stock, which is often considered "dead money" by those accountants.

If you just take it at face value, I guess.  But like so many phrases, there is a particular meaning coupled to it.  JIT is a Japanese creation, Toyota specifically I think.  It is the process they developed, to minimize inventory given expected fluctuations in supply and demand.  You don't eliminate inventory.  You size it based on the variance.  It's your bypass capacitor.  Obviously if power goes out, your cap discharges and you're SOL at some point.  If outages are on average only so long, you plan to that.  That's it.  I don't know how much more understandable I can make it.

When western companies picked it up, they greedily saw the inventory reduction part, while ignoring the statistical motivation underlying it.  They got into trouble, they strong-armed their suppliers, which has more or less always worked, and so they got by most of the time.  They eventually got burned, from not following and understanding the entirety of the process.

So you are saying JIT means 'just in time when everything goes wrong'?

If you set out to intentionally misunderstand it.  Sure.

A good system makes it hard for users to screw up.  As you say JIT invites the C-levels to make mistakes and blame their underlings.

Bad C-levels invite mistakes they blame on their underlings.

If they're using buzzwords to advance their mistakes, that's not the fault of the buzzwords and what they mean to others.

Postwar Japan didn't have much choice but to innovate the entire system, bottom to top, from manufacturing to management.  And their gains were incredible.  They outpaced the western world through the 70s and 80s.  Notoriously good cars.  Well, notoriously rusty some of them, but many excellent engines and transmissions.

Tim

[jmelson]

Now Digi-Key says the AD2S1200WSTZ will not be available until NEXT JANUARY!  But, I ordered directly from Rochester Electronics, and got the parts in a couple of days! Digi-Key shows stock at Rochester, but when I placed the order, they did NOT get the parts from Rochester!  I have no IDEA why?  What's the POINT of this marketplace system if they won't get the parts from their marketplace partners?  I DID place orders for several other parts that were at marketplace partners, and those DID come in in a few days.  EVEN from Rochester in one case.
I'm just totally confused!
Jon

[Kasper]

It is the process they developed, to minimize inventory given expected fluctuations in supply and demand.  You don't eliminate inventory.  You size it based on the variance.

So you are saying JIT means 'just in time when everything goes wrong'?

If you set out to intentionally misunderstand it.  Sure.

Is this better? JIT: Just in time when a reasonable amount of things go wrong?

[IDEngineer]

Is this better? JIT: Just in time when a reasonable amount of things go wrong?

Strictly speaking, yes. Think of it as fault tolerant design.

Look, I'm no fan of JIT. I and my customers have been repeatedly burned by it. But it's not (necessarily) an all-or-nothing game. Properly implemented, as others have noted here, the decisions are on a continuum where riskier components are more deeply stocked. Contrary to an MBA's default mindset, optimization is not always minimization.

The problem is that we let MBA's assert too much authority. That's mostly our fault. They're like that fabled scorpion... we knew what they were when we hired them. Their inherent nature means they require careful management, a very short leash, and persistent mistrust. MBA's are one of those rare examples where minimization (of their authority) really IS optimization.

[coppice]

So you are saying JIT means 'just in time when everything goes wrong'?

Who but an idiot would say that? You can't allow for every eventuality. When Asahi-Kasei's factory burned, the only thing people could do was reengineer their audio products around different ADCs and DACs, most of which make performance suffer compared to the excellent AKM products. There have been a number of disasters taking key suppliers out of a business for years like that. Some people have multiple very similar product designs ready to roll when they launch a product, so if one design hits a production roadblock they can keep shipping the others. That takes a lot of extra up front investment, though. In many cases a single supplier's offering is so compelling nothing else will really do.

[IDEngineer]

In many cases a single supplier's offering is so compelling nothing else will really do.

As a side note, this is one of my primary complaints about connectors. Other than a very few exceptions like D-Subs (which almost no one uses in new designs anymore), I swear it seems every connector is proprietary. There's almost no second sourcing. It's not that "nothing else will really do", it's that "nothing else is physically compatible". If you're designing wire harnesses, I guess you can substitute some other free-hanging connector. But when you're designing PCB-level products a connector change requires an artwork redesign. And likely an enclosure redesign. And that gives you incompatible stock in the field, where version B cannot act as a replacement part for version A.

Connectors: Proprietary, expensive, labor-intensive (read: even more expense), and don't generally enjoy decreasing cost with age like most other electronic components. What's not to love?!?