How is Chipageddon affecting you?

[IDEngineer]

I think one of the reasons it's so hard for manufacturers and distributors to give estimates is they have no idea how much hoarding and over bidding is going to happen.

I'll give the distributors a pass for that reason, but not manufacturers. They 100% control what they manufacture. If you check their websites they often reveal some of their manufacturing schedules. When they say something isn't going to be available for 52 weeks, unless they can't get leadframes or something, that's entirely their decision about how to allocate time on their fab lines. I don't begrudge them the ability to make those decisions - it's their investment in equipment - but they are also responsible for the shortages those decisions create.

[tom66]

You have to realise a lot of the shortages there are down to manufacturers supplying their preferred customers: big auto, big computer manufacturers, and so on, first, because they have huge contracts with them.

The likes of Digi-Key sell mostly to the small fry, so we are on the lower end of priorities unfortunately.

You can see there is not really that much of a shortage when e.g. GM can still make millions of cars. They make less than they want to, but still millions of cars, so they can get chips just fine now.

[SilverSolder]

You have to realise a lot of the shortages there are down to manufacturers supplying their preferred customers: big auto, big computer manufacturers, and so on, first, because they have huge contracts with them.

The likes of Digi-Key sell mostly to the small fry, so we are on the lower end of priorities unfortunately.

You can see there is not really that much of a shortage when e.g. GM can still make millions of cars. They make less than they want to, but still millions of cars, so they can get chips just fine now.

But, e.g. GM is selling LESS cars than before...   new car production in North America is down about 3.4 million vehicles in the first three months of this year, and the car makers blame chip shortages...   seen in that light, does it really make sense for the chip manufacturers to blame the car makers?   They can't both be right! 

[VK3DRB]

Which model of Bosch IMU's?

BNO055.

We caught it with 182 left, but couldn't buy in time.

They're $4 parts normally, but WinSource etc. have them at $60 each...

I also got caught out with the BNO055. I was able to source some in China for an engineering build. The BNO055 is essentially a good chip, made by a bad company.

Botch changed the hardware functionality of the BNO055 (a pinout change) without changing the part number of the chip! Hello, McFly.

According to Rev 1.2 datasheet, pin 10 has to be grounded, but in more recent datasheets, it is a Do Not Connect for something called a boot load indicator output. So what happens if an earlier design has this grounded as instructed, but Botch now says Do Not Connect? So you buy in new chips for an old PCB revision. What do you do? How do you know if you have the latest chips or not? Was pin 10 grounded because it was an input to the onboard MCU, but it is now an output changed in firmware? Is a re-spin of the board required to remove the ground? That's for Botch to know and you to find out! There are a few other pin-out changes too. The functions column in the pin-out table looks like a high school kid created it. In fact the entire datasheet is arse-about. TI, AD and LT's datasheets are far superior. Obviously SensorTec has serious internal problems with component supply, it's management and it's technical writers.

There is a certain banned episode of Fawlty Towers that reminds me of Botch. I won't be using Botch brand of chips in any design again.

Rev 1.8 https://www.bosch-sensortec.com/media/boschsensortec/downloads/datasheets/bst-bno055-ds000.pdf
Rev 1.2 https://cdn-shop.adafruit.com/datasheets/BST_BNO055_DS000_12.pdf

[EEEnthusiast]

I can't even find the popular FT2232HL-REEL which was available almost everywhere. Not to mention some exotic CPLDs and FPGAs which have disappeared forever. Either I change my design with new components and validate them or buy grey market stuff from China, which I am not sure of the quality.
The chip shortage is really killing small companies.

[tom66]

You have to realise a lot of the shortages there are down to manufacturers supplying their preferred customers: big auto, big computer manufacturers, and so on, first, because they have huge contracts with them.

The likes of Digi-Key sell mostly to the small fry, so we are on the lower end of priorities unfortunately.

You can see there is not really that much of a shortage when e.g. GM can still make millions of cars. They make less than they want to, but still millions of cars, so they can get chips just fine now.

But, e.g. GM is selling LESS cars than before...   new car production in North America is down about 3.4 million vehicles in the first three months of this year, and the car makers blame chip shortages...   seen in that light, does it really make sense for the chip manufacturers to blame the car makers?   They can't both be right! 

Right, there's over-allocation to preferred industries, and of course the car manufacturers got hit badly by cancelling all the orders.  There is still a shortfall.

But I expect it those guys who are taking most of the supply.  Digi-Key gets whatever's left over, which isn't much.  And the preferred suppliers are guaranteed cash, for some parts Digi-Key may work just on consignment for some of these parts which means they don't get paid until 30 days + past the sale time. 

Thing I've learned unfortunately is the big semiconductor manufacturers (Bosch, Xilinx, Microchip etc.) really don't care about the little guy.  They push the support requirements onto their FAEs and sales/distributor agents, but these guys aren't interested in providing any support if they can't sell any parts.  We have been trying for months to get support from Xilinx despite buying £60k of parts from them in FY20, no interest.

[tom66]

According to Rev 1.2 datasheet, pin 10 has to be grounded, but in more recent datasheets, it is a Do Not Connect for something called a boot load indicator output. So what happens if an earlier design has this grounded as instructed, but Botch now says Do Not Connect? So you buy in new chips for an old PCB revision. What do you do? How do you know if you have the latest chips or not? Was pin 10 grounded because it was an input to the onboard MCU, but it is now an output changed in firmware? Is a re-spin of the board required to remove the ground? That's for Botch to know and you to find out! There are a few other pin-out changes too. The functions column in the pin-out table looks like a high school kid created it. In fact the entire datasheet is arse-about. TI, AD and LT's datasheets are far superior. Obviously SensorTec has serious internal problems with component supply, it's management and it's technical writers.

I know a lot about BNO055.  That BL_IND pin is just open drain, so safe to ground.   It's a good chip in many regards, with some very interesting limitations.  It's also going NRND at some point, I think the gyrometer in it is NRND already. 

Bosch sell the IMU that's used in the iPhone 12, so I imagine that's where most of the supply is going right now.  It's fabbed in Korea as a MCM, not sure whether Bosch make all the composite sensors or it's a co-marketing agreement.  The ATSAMD20 in the chip is also a pretty powerful little processor.

What else I do know can't really be shared unfortunately.

[IDEngineer]

Botch changed the hardware functionality of the BNO055 (a pinout change) without changing the part number of the chip! Hello, McFly.

Not sure if "Botch" was an intentional misspelling, but it sure fits.

I've written here before about how they completely screwed up another IMU chip of theirs. We consume a fair number of BMI160's, which we selected after extensive evaluation. We wanted a very recent part (so it would be in production for a nice long time), we electrically and firmware tested it in a variety of conditions (because we've seen horrific operational reliability problems in similar parts), etc. The BMI160 checked the boxes, so we designed it in. And all was well - for a while.

When this availability nonsense came along, suddenly the BMI160 1) was out of stock everywhere, and 2) had a recommended replacement on Bosch's website! After maybe a whopping two years of production. But Bosch would not give a final production date... in their "support" forums they screamed at me for even suggesting that the BMI160 was going out of production, calling me out for spreading lies when all I asked was what did they think would be the BMI160's retirement date. I kid you not. Keep in mind that their own BMI160 webpage was recommending a newer part for current and future designs. Interpret that how you will.

As for that replacement part... yeah, some replacement. Utterly incompatible in every conceivable way, including - again, I must stress I'm not making this up - its requirement that you download a full 8KB firmware image every single time it powers up. That's right, the part is nonfunctional at powerup and requires its host to program it with an 8KB firmware image which they courteously make available at GitHub. News flash: That doesn't work well for embedded applications, where 8KB can represent a significant percentage of the MCU's entire program memory space.

They had lots of alternatives here. They could have used flash memory in the device and programmed it at the factory with a basic set of firmware so it could be used straight off the reel. If yours is an embedded application, you could thus treat it as a normal device. Meanwhile those applications wanting to update the firmware, and having the memory space and/or Internet access to do so, would still have that flexibility. A win-win for everyone. But no, this "recommended replacement" is fundamentally incompatible with a huge percentage of embedded environments. And this comes after just a couple of production years of the original device.

We spent a huge amount of time and money qualifying and approving this Bosch (Botch?) device. And then they pulled this. I warn everyone I can with this story so they don't get burned like we did. Companies need lots of negative feedback to discourage such behavior.

Yeah, I'm still angry about it. Can you tell? 

[Kasper]

Wow, thank you for the warning about botch!

Im working on a new product. First build had bmi160. 2nd build I couldnt find bmi160 or bmi270 so I went with STM.  Now Im planning 3rd build all 3 are hard to find.

The aweful replacement you mentioned, is it bmi270? 

I guess I should stick with STM or more realistically, whatever I can find.

[IDEngineer]

The scary part is that we qualified Bosch to replace an STM part that had incredibly bad reliability. As in, it would simply cease working and only a power cycle would recover it. That board still has a PFET controlled by the MCU that switches power to the IMU chip as a backup in case we ever run into that problem again.

Now Bosch is failing the industry with bad product design choices backed by unavailability.

The MEMS market seems particularly cursed.

[tom66]

Can confirm the BNO055 has some, er, interesting 'bugs' that we've found numerous workarounds for.  It's also very easy to get into firmware update mode if you write to undefined registers, which resets the chip (thankfully it doesn't write the firmware pages unless you write exactly 256 bytes...)

I think we're going to qualify a 3 chip solution ourselves next time and just roll our own sensor fusion.  That way we can requal for a different mag, gyro, etc. if a part goes NRND or hard to find.

[Kasper]

I think our STM IMU does that too. Stops working until power cycle.

[VK3DRB]

Botch changed the hardware functionality of the BNO055 (a pinout change) without changing the part number of the chip! Hello, McFly.

Not sure if "Botch" was an intentional misspelling, but it sure fits.

I've written here before about how they completely screwed up another IMU chip of theirs....

...Yeah, I'm still angry about it. Can you tell? 

Yes it was intentional. It appears Colonel Klink is running Botch Sensortec and Sergeant Schultz is their technical writer.

Oddly enough Bosch do make relatively good quality dishwashers and clothes washing machines. Would I buy their white goods in future? Maybe. Would I buy their Sensortec chips again? Nope.

[Howardlong]

I’m adding LCSC to my growing list of specualtive scalping outfits to avoid.

[JPortici]

I’m adding LCSC to my growing list of specualtive scalping outfits to avoid.

the biggest increase i've found yet is on dsPIC33EP256MU806 and company. (microchip itself increased the price by 20%+ but that was ridiculous) 
Guess what i won't be using in the new board restilying

[fourfathom]

If you're actually looking for the Si5351, there is a clone available, and it tests out quite well (I don't know if this is an authorized copy or not):
https://lcsc.com/product-detail/Clock-Generators-Frequency-Synthesizers-PLL_Hangzhou-Ruimeng-Tech-MS5351M_C1509083.html
  A few months I purchased these through Chipmall, but the LCSC price is decent: $0.71 qty 100.

[TimFox]

Botch changed the hardware functionality of the BNO055 (a pinout change) without changing the part number of the chip! Hello, McFly.

Not sure if "Botch" was an intentional misspelling, but it sure fits.

I've written here before about how they completely screwed up another IMU chip of theirs....

...Yeah, I'm still angry about it. Can you tell? 

Yes it was intentional. It appears Colonel Klink is running Botch Sensortec and Sergeant Schultz is their technical writer.

Oddly enough Bosch do make relatively good quality dishwashers and clothes washing machines. Would I buy their white goods in future? Maybe. Would I buy their Sensortec chips again? Nope.

In “Wallace and Gromit and the Curse of the Were-rabbit”,  Gromit is shown assembling crates to hold rabbits using a cordless tool labeled “Botch”.

[VK3DRB]

I’m adding LCSC to my growing list of specualtive scalping outfits to avoid.

Definitely. I have seen it in a few of their parts. One micro I used was about $1 a few months ago. Now it is around $4. Quantities available "coincidentally" had not changed, meaning they were likely on their shelves all along.  Even so, LCSC does have components that Dodgykey has long had zero stock of.

[VK3DRB]

In Australia, we produce one type of chip ourselves for which there are no supply chain problems . But some countries are starting to find acute shortage of these chips...

https://www.bbc.com/news/business-59750613

[harerod]

Quote from: harerod on 2021-12-13, 12:57:36Quote from: VK3DRB on 2021-12-13, 12:19:13>... Rochester Electronics seems to be prominent... any good?
For delivery to Germany, free shipping doesn't apply to Rochester Electronics. In addition to another ~US$40 for shipment expect "handling fees" from the carrier plus duty. I can't tell you more at this point, since I am still waiting for the additional charges. (in short - different INCOTERMS)
Update to the Rochester-topic:
In December things were a bit hectic, because we have to replace most of our silicon by what is still available. So I ordered LM53602 as sample for a replacement of LM53601AQ from Rochester via Digikey.
"Auftragsbestätigung" is "Order Confirmation". I just checked the price (€539.02), freight (€38.21) and total. So the emails and the pdf looked fine at first and second glance. I even noted the last line, which told me about a 1-3 days delay, because of reasons.
Later, when I did inventory/book keeping, I found out that this particular shipment had not arrived. I took me a while to figure the actual problem out. By that time the components were gone. Today I removed the LM563*-series from three designs and ordered a 200 samples of a replacement component.
Regarding Digikey/Rochester and overseas - make sure to thoroughly read the multilingual communications and keep "US DOLLARS/GERMANY" ready.

[Martin Miranda]

lcsc keeps running out of micros too.


https://lcsc.com/products/Microcontroller-Units-MCUs-MPUs-SOCs_11329.html?brand=457

[Bud]

Ironically, Canon had to tell customers how to bypass their own warning  because of shortage of chips for cartriges

https://www.techdirt.com/articles/20220110/06555048260/chip-shortage-forces-canon-to-issue-workarounds-own-obnoxious-drm.shtml

[IDEngineer]

Ah, sometimes there is karma.

[Sal Ammoniac]

With my usual suppliers (Digikey, Mouser) completely out of nearly everything, I've resorted to buying eval boards and removing the chip(s) I need. Now even eval boards are getting scarce.

[Nortek-Chris]

Scalpers and hoarders appear to be causing an even bigger problem with stock from Manufacturers -  T.I to name just one of a few I've experienced. Several times i've tried to buy stock of Class D amps during the 'more units available' period shown on their website - logging in about 20 times a day (and night) for a week just to try and grab some stock to the point it becomes an OCD, only for it all to be sold out whilst I'm entering in the Credit Card details!. Even the Automated email based notifications from the site always seem to arrive 6 hours after the stock has gone!.

I seem to get similar issues on Microchip Direct too. I think this is going to be a growing problem in the cases where stock inevitably appears first on 'the "Direct" outlet websites of some manufacturers, months or a year before it gets shipped to Distributors - so everybody is all watching the same source. In relation to T.I you can't even back order, to get your PO in the queue!.

Magically there always seems to be tens of thousands of stock on Winsource and similar sites though. Provided of course, you are able to budget for six times the normal price!.

If I was being selfish, I would be hoping for a short period where stock was only made available to bonafide OEM's and Manufacturers  . Not like it will ever happen, but it would at least even up the scales a bit and give us the opportunity to actually make something. Or at least spread out the stock a bit as it becomes available from the Factory across several different distributors rather than seemingly selling every bit of stock directly from their own site.

Mouser, at least put an upper limit on some of their lines, in order to stop one PO buying up all of the stock.