SMT passive worldwide shortage

[schmitt trigger]

Not a week passes by without the purchasing people advising me about some chip resistor or  MLCC shortages. They either go on allocation or their lead times have been extended to many, many months.

Are you suffering a similar situation?

[PTR_1275]

Yep. Got a email from future electronics the other week with typica leadtimes and trends. Some parts are so far backlogged that it was saying allocation full, meaning there is no more allocation for production available until Q3 2019.

Some parts we ordered in October 2017 have been pushed back several times now and the latest update was for delivery in Feb 2019.

[schmitt trigger]

What would you think are the main contributors for this situation?

Most people blame it that the passive market has literally grown exponentially, with even the lowliest gadget having some embedded electronics.

But there must be something else?

[CatalinaWOW]

This has happened at least four times in the last forty years.  Look at the worldwide economy.  Doing quite well thank you.  Everybody is building stuff and selling it successfully.  That is on top of all of the new applications.  Component manufacturers aren't keeping up.  Their financial watchdogs are reminding them of the last time they built capacity and got it on line just in time for the downturn.  So they are cautious about building up capacity.  And even with this caution they end up overbuilding when the next slowdown occurs.  And reinforce the caution for the next time around.

[NorthGuy]

I certainly noticed it with capacitors. You find something in stock, put in into your cart, press checkout, and it is already backordered. Happened to me few times.

I didn't notice it with resistors or anything else.

[DerekG]

Component manufacturers aren't keeping up.  Their financial watchdogs are reminding them of the last time they built capacity and got it on line just in time for the downturn.  So they are cautious about building up capacity.

Quite right.

During the GFC, the world economy shrank & component manufacturers shrank their inventory as they closed up excess plant capacity.

Now, they simply want the consumer to order on allocation. They will make the parts after they have received enough forward orders.

One part that has been hit really hard is the STM8S Value Line micros. These are probably the best 8 bit micro value in the world & delivery times have now blown out to 35 weeks.

https://www.digikey.com.au/product-detail/en/stmicroelectronics/STM8S003F3P6TR/497-18051-2-ND/4357536

https://www.digikey.com.au/product-detail/en/stmicroelectronics/STM8S003K3T6CTR/STM8S003K3T6CTR-ND/3455602

For medium to high volume production, Microchip & Atmel (now also Microchip) just can't compete.

[Bassman59]

Component manufacturers aren't keeping up.  Their financial watchdogs are reminding them of the last time they built capacity and got it on line just in time for the downturn.  So they are cautious about building up capacity.

Quite right.

During the GFC, the world economy shrank & component manufacturers shrank their inventory as they closed up excess plant capacity.

Now, they simply want the consumer to order on allocation. They will make the parts after they have received enough forward orders.

One part that has been hit really hard is the STM8S Value Line micros. These are probably the best 8 bit micro value in the world & delivery times have now blown out to 35 weeks.
For medium to high volume production, Microchip & Atmel (now also Microchip) just can't compete.

Seem to me the OEM has a choice: wait 35 weeks and manufacture and sell nothing in that time, or redesign for a part you can get in a reasonable time frame and bite the bullet on cost.

[DerekG]

Seem to me the OEM has a choice: wait 35 weeks and manufacture and sell nothing in that time, or redesign for a part you can get in a reasonable time frame and bite the bullet on cost.

Quite correct. No other choices. Remember to search www.octapart.com for any suppliers holding stock.

[rx8pilot]

Stick with modern parts. Forget about 0603/0805/larger unless you need double digit uF capacitance or more than 50V.
I don't see supply issues with 0402/0201.

There is truth in that.....
I have some power designs that use 1210 10uf/50v and 22uf/50v X7R and they are rather hard to find and very expensive now. Looking at switching those to polymer if I can take the size hit.

[DerekG]

Looking at switching those to polymer if I can take the size hit.

Yes, I can see the SMD robot stopping much more often .... not to mention increased rework.

[DerekG]

And in the meanwhile, solid tant are sometimes more expensive than polymer tant... Go figure!

It is due to a combination of volumes manufactured & the cost of the tantalum material itself.

[schmitt trigger]

Stick with modern parts. Forget about 0603/0805/larger unless you need double digit uF capacitance or more than 50V.
I don't see supply issues with 0402/0201.

That is also what I have heard.

Additionally, does anyone remember capacitor arrays? Those that would save you board space and assembly time? Not that long ago.....
Well...... the duals are being phased out. And the only manufacturer standing, AVX, has increased its lead time tremendously.

[calexanian]

This all started with Yaego taking on no extra orders for 2018 and the ripples went out from there. Also the auto industry and smart home automation is sucking up tons of supply on basically speculation. It is also a case of them as a community of manufacturers not wanting to add more production and therefore increase prices due to demand so its double in their favor. Sucks for us small guys. I have been seeing some real junk come in as substitutes from our vendors. I send it back and just find something suitable from Digikey. I just saw a case of through hole 25 volt at 1000 uf caps today from Taitron. I have not seen them since the great cap shortages of the 90's. They usually slum around and brand whatever they can get cheap and off spec. I tell our purchasing guy only buy specific brand and series caps from nichicon, panasonic, rubycon, and a few others, but they don't want to go searching for current stock and have the distributor send whatever they have.I have tried to explain ripple current, ESR, and life span on caps, but all they see is being lazy and cheap. A dangerous combo.   

[DerekG]

I tell our purchasing guy only buy specific brand and series caps from nichicon, panasonic, rubycon, and a few others, but they don't want to go searching for current stock and have the distributor send whatever they have.

Don't forget that in times like these, https://octopart.com can be a really good friend.

[rx8pilot]

I have stopped production on a number of my boards and re-designing some of the newer ones trying to avoid some of the worst offenders. I was using a lot of fairly large-ish MLCC's which are getting out of hand - even for small qty's.

Nuts.

[KEVBULL]

I got quotes today from authorised  distributors for Vishay MLCC and resistors with L/T - 120 weeks and 90 weeks resp.

[thermistor-guy]

I tell our purchasing guy only buy specific brand and series caps from nichicon, panasonic, rubycon, and a few others, but they don't want to go searching for current stock and have the distributor send whatever they have.

Don't forget that in times like these, www.octapart.com can be a really good friend.

I think you meant Octopart: https://octopart.com/

A few years ago, my boss at the time ordered me to throw out full reels of SMD passives and semiconductors that were duplicates - only keep one reel of each component. I was so uneasy about it, he confirmed more than once that this was what he wanted. Ended up with a pallet loaded high with reels. Out they went - diodes, BJTs, MOSFETs, JFETs, ceramic and tantalum caps, resistors, 0603, 0805, 1206, 1810. All reputable brands.

[DerekG]

I think you meant Octopart: https://octopart.com/

Quite right. I have corrected the typo.

I have stopped production on a number of my boards and re-designing some of the newer ones trying to avoid some of the worst offenders. I was using a lot of fairly large-ish MLCC's which are getting out of hand - even for small qty's.

Quite right. You can find out more about the shortages here:

https://www.ttiinc.com/content/ttiinc/en/resources/marketeye/categories/passives/me-zogbi-20180302.html

[D3f1ant]

Also redesigned a few products to get rid of the large value MLCCs. They are either unobtainable or very expensive compared to the physically larger Electrolytic polymer caps that would do the job adequately. Have also been stock piling 0.01, 0.1, 1.0 and 10uf in various sizes and voltages when I see a reel available to mitigate production risk going forward.
Some values of resistors are also getting hard to find.

[tszaboo]

Also redesigned a few products to get rid of the large value MLCCs. They are either unobtainable or very expensive compared to the physically larger Electrolytic polymer caps that would do the job adequately. Have also been stock piling 0.01, 0.1, 1.0 and 10uf in various sizes and voltages when I see a reel available to mitigate production risk going forward.
Some values of resistors are also getting hard to find.

Not everyone has the luxury to do this. If I have to respin a board, it starts with 10K+ in certification and 2Months+. So production should just stop for 2 months, right? And you cannot trust the suppliers anymore, because they confirm your backorder, and then halfway they tell you that they dont know it anymore.
So you order the same part from multiple vendors, just to make sure. Ant this is done by others also, not just you. So my need for 1 million 100nF 0402 just created 3 million in orders at 3 different places. I mean imagine the company shutting down production, because you didnt get MLCCs. Stupid right? And since everyone does this, they see an enormous increase of orders, snowballing.

[schmitt trigger]

We are like that, too.
Several of our customers are automotive and they want a full validation for even the slightest change.

A case in point. We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper. Our purchasing people told us that they had found a large cache of 25V caps from the same series from the same reputable vendor.

Well, the supplier quality engineer at our automotive customer wanted to perform all the validation tests, including EMI. Talking in excess of $25K and 8 weeks minimum.
So we had to procure from distributors, gray market components, and paying twice as much.

[thm_w]

We are like that, too.
Several of our customers are automotive and they want a full validation for even the slightest change.

A case in point. We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper. Our purchasing people told us that they had found a large cache of 25V caps from the same series from the same reputable vendor.

Well, the supplier quality engineer at our automotive customer wanted to perform all the validation tests, including EMI. Talking in excess of $25K and 8 weeks minimum.
So we had to procure from distributors, gray market components, and paying twice as much.

That is a bit shortsighted. Imagine the chance of getting counterfeit components would go way up. At least with the 25V part it should be reputable.
It was very common, in the airline industry at least, to write up a report on substituting a part and argue that it is equivalent enough to avoid testing. Maybe automotive is more strict? Or someone is being anal.

[DerekG]

We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper.

Unless these are in timing circuits, you can often replace them with 0.12uF, 0.15uF, 0.18uF or 0.22uF caps.

[rx8pilot]

We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper.

Unless these are in timing circuits, you can often replace them with 0.12uF, 0.15uF, 0.18uF or 0.22uF caps.

For bypass applications - that is almost always true but only if you are authorized to make the substitution. For many - that kind of stuff triggers a whole slew of test and validation / documentation.

[thermistor-guy]

We are like that, too.
Several of our customers are automotive and they want a full validation for even the slightest change.

A case in point. We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper. Our purchasing people told us that they had found a large cache of 25V caps from the same series from the same reputable vendor.

Well, the supplier quality engineer at our automotive customer wanted to perform all the validation tests, including EMI. Talking in excess of $25K and 8 weeks minimum.
So we had to procure from distributors, gray market components, and paying twice as much.

That is a bit shortsighted. Imagine the chance of getting counterfeit components would go way up. At least with the 25V part it should be reputable.
It was very common, in the airline industry at least, to write up a report on substituting a part and argue that it is equivalent enough to avoid testing. Maybe automotive is more strict? Or someone is being anal.

Not always shortsighted. Yes, some component substitutions can easily be justified. But if you rely on certain caps to meet EMI standards in a harsh environment, and then you change the caps, then I too would want the EMI testing redone.

[mbless]

We had shortages of 0.1uF, 16V, X7R, 0603 MLCC. You know, the ones that you sprinkle on a board like salt and pepper.

Unless these are in timing circuits, you can often replace them with 0.12uF, 0.15uF, 0.18uF or 0.22uF caps.

Which would still require all of the validation they want to avoid...

[AlfBaz]

Seems to be happening with resistors as well
0603, 1%, 1/4W, 1k, not a problem right?
Mouser list 5 manufacturers, they're out of stock for 4 of them with varying volumes on order. They have just over 10 thousand in stock for Vishay (which will disappear if someone buys 2 reels) and then it's a 76 week wait from vishay

If ordering parts wasn't tedious enough

[schmitt trigger]

I agree that certain timing/filtering/EMI applications do require a precise capacitance value.

But the 0.1uF cap, when used as a generic Vdd decoupling cap, is not a magic value.....

In addition, going from 16 to 25 volt means that the capacitance reduction due to DC bias will be minimized.

However my experience has been that many -but not all, supply quality engineers do not really have a profound grasp of electronic circuits and components. And it is also true that many times the trivial changes are the ones which cause the largest headaches.

So...they play it safe and request the full validation route. The onus is in the supplier to prove that the requested change won't have problems.

[tszaboo]

I agree that certain timing/filtering/EMI applications do require a precise capacitance value.

But the 0.1uF cap, when used as a generic Vdd decoupling cap, is not a magic value.....

In addition, going from 16 to 25 volt means that the capacitance reduction due to DC bias will be minimized.

However my experience has been that many -but not all, supply quality engineers do not really have a profound grasp of electronic circuits and components. And it is also true that many times the trivial changes are the ones which cause the largest headaches.

So...they play it safe and request the full validation route. The onus is in the supplier to prove that the requested change won't have problems.

Please all remember this once and for all, it doesnt work this way.
The voltage rating of the capacitor has nothing to do with how it loses capacitance.
Take a look at this:
https://www.we-online.de/redexpert/#/module/13/productdata/=885012209014/V/asc/C/gte:10uF+lte:10uF/type/X7R

The 1210 10V part loses 20% at 10V, the 16V part loses 24% at 10V, the 1206 10V part loses 37% at the rated voltage.

[rx8pilot]

I agree that certain timing/filtering/EMI applications do require a precise capacitance value.

But the 0.1uF cap, when used as a generic Vdd decoupling cap, is not a magic value.....

In addition, going from 16 to 25 volt means that the capacitance reduction due to DC bias will be minimized.

However my experience has been that many -but not all, supply quality engineers do not really have a profound grasp of electronic circuits and components. And it is also true that many times the trivial changes are the ones which cause the largest headaches.

So...they play it safe and request the full validation route. The onus is in the supplier to prove that the requested change won't have problems.

Please all remember this once and for all, it doesnt work this way.
The voltage rating of the capacitor has nothing to do with how it loses capacitance.
Take a look at this:
https://www.we-online.de/redexpert/#/module/13/productdata/=885012209014/V/asc/C/gte:10uF+lte:10uF/type/X7R

The 1210 10V part loses 20% at 10V, the 16V part loses 24% at 10V, the 1206 10V part loses 37% at the rated voltage.

What?

6v part vs 25v part
The 6v part rolls off C value much faster with a rising voltage.....
If you are operating at 1V or so - who cares? I think the point is that in the face of a shortage, a higher voltage rated part is usually not a problem. If the original designer was allowing more than 50% rated voltage - the circuit would see an increase in capacitance. If it was designed under 50% of the rated voltage the C change based on V bias would be negligible.

[thm_w]

Not always shortsighted. Yes, some component substitutions can easily be justified. But if you rely on certain caps to meet EMI standards in a harsh environment, and then you change the caps, then I too would want the EMI testing redone.

You are not changing the capacitance or temperature rating, if the circuit was designed appropriately, only the maximum voltage. If you look closely at the parts they may literally be the same part remarked as various voltages.
Of course if this is some ultra harsh environment it could deserve retesting, but that would be the case even from switching vendors. That's more of a reliability test than EMI though.

[MT]

Ooh! The good ol' "spot market" days from the 90'ies is back i supose!

[ubbut]

It was my understanding, that the voltage rating does not influence the capacitance decrease (at least not very much). The main factor is the physical size:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527

[Kjelt]

So how do you guys handle this in your new designs? Checking current stock is no guarantee for future stock.
At the time you are ready for production ( half year later ?) some parts might not be in stock anymore, what are you gonna say ?
Asking the logistics dept to order the parts in advance is a political nightmare in it self (against company policy?).

[tszaboo]

Asking the logistics dept to order the parts in advance is a political nightmare in it self (against company policy?).

It shouldn't be. You should plan deliveries of products with like a year in advance. How else do you plan production? Spot buying everything, not talking to vendors, and manufacturers?
It is a political nightmare though that I need to explain this to upper management. Like they cannot just start a new product all williy nilly next week.

[Kjelt]

In my previous company there were preferred part lists and those were bought in huge quantities in advance and dynamically stocked up.
If you design in a new part you needed a very convincing story, but than indeed only after the engineering validation runs so 9 to 12 months later the components would be stocked in production quantities. If the components is from a preferred supplier the contracts were already done, even with cancellation clauses, but with a non preferred supllier it did happen we had to postpone production over 6 months due to component shortage.

[GreggD]

For 100nF (0.1uF) I only use the most common 50V parts. They are cheaper than 16V as there sold in larger qu.
The voltage rating of a X7R/X5R has little to do with capacitance reduction due to DC bias. Case size 1206 vs: 0402 has a very large effect on capacitance retention. Use larger caps.

[schmitt trigger]

In my previous company there were preferred part lists and those were bought in huge quantities in advance and dynamically stocked up.

I have asked that same question many times. However...........
Depending on how the accounting and tax structure is set up, inventories have very significant financial implications.

Don't ask me to explain them, I don't fully understand them myself.

[Kjelt]

Yes that is true, inventory can financially be quite expensive. Writing off components that are beyond the very conservative shelf life due to moisture mostly can also be a big factor.
Not being able to produce anything for a half year costs nothing ofcourse   

[MT]

Asking the logistics dept to order the parts in advance is a political nightmare in it self (against company policy?).

It shouldn't be. You should plan deliveries of products with like a year in advance. How else do you plan production? Spot buying everything, not talking to vendors, and manufacturers?
It is a political nightmare though that I need to explain this to upper management. Like they cannot just start a new product all williy nilly next week.

Back in the 90ies at that time  purchase department just bought whatever they could get hold of and called us instead and asked if such and such can be a resonable substitute, later we just got  their list and went through and X'ed whatever was OK. But it was one of those super big multinational companies they had resources. Despite their several year spaning contracts with the big manufacturers they had to go into the spot market.

[tszaboo]

In my previous company there were preferred part lists and those were bought in huge quantities in advance and dynamically stocked up.

I have asked that same question many times. However...........
Depending on how the accounting and tax structure is set up, inventories have very significant financial implications.

Don't ask me to explain them, I don't fully understand them myself.

If you order something, you dont pay it ASAP. For example, you order a million resistor for your doodad, you only pay for the resistors 60 days later. In that 60 days, you assemble the doodad, and send it to a customer, and get the money for it. From this money, you pay the resistor supplier. It is called company credit.
If you buy a large stock of components, which you dont use it right away, you need to pay it before you got payed for it. It means that the company actually has to have that money, or get it by other means, investors, bank credit, etc. It doesnt sound like a big deal, but it is. It means that a small company can do big business. If you spot buy on Farnell, not only do you pay potentially 10-100 times the price for the component, it also means, that only big companys can do small business.
Of course there is also the possibility to have a lawyer sitting on meetings with suppliers, and write contracts, that are very expensive to break. I've seen a PCBA company assemble a board with loss for years, because breaking the contract would cost more. Big players are doing this, basically when you order, you also state a penalty for late delivery is x% every week. Sometimes it is even just setup in a system like SAP, if you didnt get the parts, they get a penalty invoice with no human interaction. Of course production will see it, but that is another story.

[schmitt trigger]

Sometimes it is even just setup in a system like SAP, if you didnt get the parts, they get a penalty invoice with no human interaction. Of course production will see it, but that is another story.

That has happened to us.
The material arrived at our customer's dock the very last allowable day, but nevertheless on time.

However, the customer's receiving department did not scan the boxes that day......for all purposes, the material had not arrived to the stockroom and we got penalized.
The charge was disputed and later revoked, but you had to have a person spending a lot of time clearing it up.

[Nortek-Chris]

I don't see supply issues with 0402/0201.

It may be happening soon with 0402. Murata issued a bulletin a few months ago, giving notification that various value of some older series' are being discontinued and 0402 was included in one of the range listed in their EOL notice. Last time buy from them is March 2019. No Values are actually listed on the Bulletin, but I hope hope that the more popular values like 100nf continue to be produced, but i'm not holding my breath for the longer term.

https://www.ttieurope.com/content/dam/ttiinc/products/PCN/Murata/Murata-Obsolescence-PCN-032018GR-ZRA.pdf

This is also part of the problem, some key MLCC manufacturers are ramping down production of 0402, 0603, 0805 etc, in favour of 0201, which is the package size favoured by their larger customers, especially Automotive, and when you think that there can be upwards of 10k pcs of 0201 MLCC parts in an average Electric vehicle then you can see why its in their interests to discontinue the older parts and use the extra manufacturing capacity to produce the 0201.

Unfortunately, as manufacturing capacity and output of 0402, 0603 etc versions drop, it puts the pressure on the remaining manufacturers to pick up the demand, and lead time continues to increase - rinse and repeat. So I can't see any improvement happening soon.

[Kjelt]

What is the logic behind this?
0201 MLCC has a huge derating. The result is hardly any gain in surface space or am I missing something?
For resistors yes but for caps I don't get it.

[ubbut]

Soooo..... Any suggestions on what we could do to get hold of at least the most necessary parts?

Spent about 2 days comparing 5 distributors to find 10u 0805, 25V at reasonable prices. Mouser had some, but the next day they were gone of course.

[tszaboo]

Soooo..... Any suggestions on what we could do to get hold of at least the most necessary parts?

Spent about 2 days comparing 5 distributors to find 10u 0805, 25V at reasonable prices. Mouser had some, but the next day they were gone of course.

Order in advance?
BTW you shouldn't select components based on the spot market price.
The issue I'm seeing now is that they dont have it even in prototyping quantity anymore. There are zero 0402 1uF X7R capacitors in the mayor distributors.

[Nortek-Chris]

What is the logic behind this?

I've worked in a purchasing capacity for an Electronics Manufacturer for over 20 years and still haven't figured it out!, in the early days I drove myself old and grey trying to work out the logic behind the component manufacturers. Back then it was Amp / Tyco (TE) connectors. I never quite worked out the fact that you could get a free plug assembly Ex-Stock but had to wait 10 weeks for the mating PCB connector. That is a bit like going into a shoe shop and buying the left shoe immediately, but having to wait 10 weeks for the right shoe to turn up  . These days I don't bother getting stressed over it, it is what it is.

Soooo..... Any suggestions on what we could do to get hold of at least the most necessary parts?

All you can do is to act on early warnings and buy stock or schedule call off orders in advance before the price increases (The potential for the current problem was highlighted in some distributor blogs and newsletters back in 2017). If you buy in Volume then distributors like TTi or Rutronik are your best bet, and if the design allows also look at 'unknown' brands like Hitano which can often be overlooked by other volume 'brand' users. There is also a Uk based distributor called VTM  who seem to be able to offer shorter lead times on their "Tancap" brand.

I suppose it depends on which distributor you use and your relationship with them, however I find that for production volumes and despite the aforementioned EOL notice, buying Murata currently from TTi seems to yield faster delivery times although given the EOL notice, that may change soon!. Samsung seem to be the worst for lead times.

[schmitt trigger]

And, let's not forget:
Trump's trade war with China.

Adding more fuel to the fire.

[rx8pilot]

And, let's not forget:
Trump's trade war with China.

Adding more fuel to the fire.

Sadly. True.