Datasheet Errors...........

[jfiresto]

Yep i had a similar thing happen to be with a SPDT relay, where they show a bottom view in the PCB layout drawing. WHY....

That is common with relays (because of point to point wiring?) and never surprised me as I worked with an engineer who used many of them in the early 1980s. Here is the pinout of a mechanical relay that came out a few years ago:



The coil of this relay has 1.3mm pins and the contacts, 1.0mm. What surprised me about the part was its datasheet swaps the diameters. The maker discontinued the part before I could report the error.

[RFDx]

The RRC filter in the DS will never work for a Type 2 PLL, but was copied from the original DS to the Motorola DS and uncritically propagated to the next generation 74HC4046 DS.

All the datasheets for the (74HC)4046 PLL/VCO show the same or at least very similar type I, 2nd order PLL examples. There is no mention of type-II PLLs and with voltage output PDs/PFDs and passive filters to my knowledge something that isn't really feasible anyway. The first order RC/RRC filters used are admittedly not particularly effective but technically there is imo nothing wrong. What/where is the error in the datasheet exactly?

[tooki]

https://www.kycon.com/Pub_Eng_Draw/STX-3000.pdf

... Also WTF square PCB holes with 90deg corners?

Well, gazillions of those cheap jacks are used in cheap devices whose (often extra-thin, often phenolic) PCBs are manufactured in such large volumes that the holes are not drilled and milled, but punched. And for those, rectangular holes with square corners are no problem.

[wobbly]

My favourite datasheet failure is one that bit me a couple of years ago.  I wanted to use a Microchip MCP2221 part, which is a simple USB to RS-232 interface.  Couldn't be simpler right?

So the chip did work a couple of times, but then it quickly became obvious that it was complete garbage.  A chip that requires no manual configuration completely failing at it's ONE JOB!   

So I dug deeper and found there was an errata document for this chip.  This document basically said that "the chip doesn't initialise it's internal registers properly.  Don't use it!  Buy this one instead!"

Errata for the MCP2221: https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/Errata/80000742A.pdf

As of right now, some big distributors are still unaware of this problem! 

The MCP2221A DOES work properly and totally solves the problem.



Funnier still, it turns out that this part is actually just a PIC microcontroller (their own bread-and-butter product!) and it had been programmed and then been locked down and had it's markings changed to disguise it's nature.  Unbelievable!

[exe]

Funnier still, it turns out that this part is actually just a PIC microcontroller (their own bread-and-butter product!) and it had been programmed and then been locked down and had it's markings changed to disguise it's nature.  Unbelievable!

I actually was expecting this. Why bother with making a custom design if one can just use a cheap OTP mcu for that? Esp. with such a profit margin. $2 in qty1000, c'mon!

[Smokey]

My favourite datasheet failure is one that bit me a couple of years ago.  I wanted to use a Microchip MCP2221 part, which is a simple USB to RS-232 interface.  Couldn't be simpler right?

So the chip did work a couple of times, but then it quickly became obvious that it was complete garbage.  A chip that requires no manual configuration completely failing at it's ONE JOB!   

So I dug deeper and found there was an errata document for this chip.  This document basically said that "the chip doesn't initialise it's internal registers properly.  Don't use it!  Buy this one instead!"

Errata for the MCP2221: https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/Errata/80000742A.pdf

As of right now, some big distributors are still unaware of this problem! 

The MCP2221A DOES work properly and totally solves the problem.

Funnier still, it turns out that this part is actually just a PIC microcontroller (their own bread-and-butter product!) and it had been programmed and then been locked down and had it's markings changed to disguise it's nature.  Unbelievable!

I use the A part all the time for automated test boards.  It's great pushing all the software PC side and not having any firmware on the boards.  Plenty of I2C chips to implement all sorts of test stuff.
My only gripes are that the python driver libraries have extremely poor USB HID error handling.  I had to spent a fair bit of time adding all the exceptions so HID errors didn't crash everything.  But that's a python problem not a chip problem.  The chip works great

[wobbly]

I actually was expecting this. Why bother with making a custom design if one can just use a cheap OTP mcu for that? Esp. with such a profit margin. $2 in qty1000, c'mon!

Sure, but they didn't have the engineering chops to get it right the first time of asking?  Even when they INVENTED the PIC?  That's what grinds my gears.

The number of times I've had no trouble implementing serial ports myself in micro's.  AND they had the nerve to charge a premium for these busted, janky parts and STILL DO!

[Smokey]

...
The number of times I've had no trouble implementing serial ports myself in micro's...
...

Just to be clear, we are talking about a USB - UART bridge (+I2c, +ADC, +DAC, +GPIO).  I've "implemented serial ports" too, but I would be insane to think I could write a whole uC USB stack and actually get CDC working reliably let alone all the other stuff they are pushing over HID. 

[wobbly]

@Smokey,
All of the features of this chip are "solved problems" as it were.  Nobody would bother implementing the entire thing from scratch, especially when low cost is an important feature.
You'd just use your company's existing IP and add a bit of glue logic.

Microchip knew this product was faulty and didn't recall it - sloppy customer care on to of sloppy engineering.

[exe]

The number of times I've had no trouble implementing serial ports myself in micro's.  AND they had the nerve to charge a premium for these busted, janky parts and STILL DO!

Holly molly, they still sell them?! It seems the bug was discovered in 2017... Not cool, microchip, not cool.

My entry for DS errors: for LT1468, on the last page the description for LT1468 and LT1468-2 are swapped: https://www.analog.com/media/en/technical-documentation/data-sheets/14682fb.pdf . LT1468 supposed to be 90MHz opamp, not 200MHz. I reported that to AD a couple of years ago, they didn't bother acknowledging or fixing it. The mistake cost me ordering wrong parts and several hours of troubleshooting.

Oh, and that naming. I'd assume that lt1468-2 is a dual version of lt1468, but it is not. The dual version is lt1469. That "-2" suffix may mean minimal gain.

[mikeselectricstuff]

My favourite datasheet failure is one that bit me a couple of years ago.  I wanted to use a Microchip MCP2221 part, which is a simple USB to RS-232 interface.  Couldn't be simpler right?

So the chip did work a couple of times, but then it quickly became obvious that it was complete garbage.  A chip that requires no manual configuration completely failing at it's ONE JOB!   

So I dug deeper and found there was an errata document for this chip.  This document basically said that "the chip doesn't initialise it's internal registers properly.  Don't use it!  Buy this one instead!"

Errata for the MCP2221: https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/Errata/80000742A.pdf

As of right now, some big distributors are still unaware of this problem! 

The MCP2221A DOES work properly and totally solves the problem.

I can see an argument for using a new part number for bug-fixed versions, e.g. where workarounds have been implemented which would be incompatible with the fixed version, or to make it easy to differentiate the fixed version, but in cases like this it seems insane to not obsolete the old version - sounds like something stuck in a corporate policy of never obsoleting things.
They could at least put a big warning banner on the datasheet.

[Smokey]

My favourite datasheet failure is one that bit me a couple of years ago.  I wanted to use a Microchip MCP2221 part, which is a simple USB to RS-232 interface.  Couldn't be simpler right?

So the chip did work a couple of times, but then it quickly became obvious that it was complete garbage.  A chip that requires no manual configuration completely failing at it's ONE JOB!   

So I dug deeper and found there was an errata document for this chip.  This document basically said that "the chip doesn't initialise it's internal registers properly.  Don't use it!  Buy this one instead!"

Errata for the MCP2221: https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/Errata/80000742A.pdf

As of right now, some big distributors are still unaware of this problem! 

The MCP2221A DOES work properly and totally solves the problem.

I can see an argument for using a new part number for bug-fixed versions, e.g. where workarounds have been implemented which would be incompatible with the fixed version, or to make it easy to differentiate the fixed version, but in cases like this it seems insane to not obsolete the old version - sounds like something stuck in a corporate policy of never obsoleting things.
They could at least put a big warning banner on the datasheet.

Not non-A versions are all more expensive (at digikey at least).  That's one way to limit sales.

[T3sl4co1l]

TI does it too.  I had to press an FAE for answers, they wouldn't post as such in the e2e thread, but turns out the UCC256303 is internally NRND.  Why they don't tell marketing to say the same, I have no clue.  It sure goes further to paint them as an incompetent company.

I haven't heard any good reasons why they shouldn't.  Mind, I have heard possible reasons. Just not good ones.

But, perhaps those same (bad) reasons, are why MCP doesn't for this one.

Tim

[Smokey]

TI does it too.  I had to press an FAE for answers, they wouldn't post as such in the e2e thread, but turns out the UCC256303 is internally NRND.  Why they don't tell marketing to say the same, I have no clue.  It sure goes further to paint them as an incompetent company.
I haven't heard any good reasons why they shouldn't.  Mind, I have heard possible reasons. Just not good ones.
But, perhaps those same (bad) reasons, are why MCP doesn't for this one.
Tim

https://www.ti.com/product/UCC256303
Pretty clearly NRND.

https://www.digikey.com/en/products/filter/power-management-pmic/dc-dc-switching-controllers/715?s=N4IgTCBcDaIKoGEFgKwDYDMAGDIC6ANCClKAA5QCMRZFklWWAvk0A
One part marked Obsolete, the other NRND

[coppice]

TI does it too.  I had to press an FAE for answers, they wouldn't post as such in the e2e thread, but turns out the UCC256303 is internally NRND.  Why they don't tell marketing to say the same, I have no clue.  It sure goes further to paint them as an incompetent company.

I haven't heard any good reasons why they shouldn't.  Mind, I have heard possible reasons. Just not good ones.

But, perhaps those same (bad) reasons, are why MCP doesn't for this one.

Tim

This can be a difficult situation for a silicon vendor. You have part B that is both better and has a higher profit margin than part A. Should you mark part A as NRND, keep fullfilling orders for existing designs, push B hard, and move on? Well if someone is currently designing part A into something high volume, you might want to hold off on the NRND notice for a while. Rocking the boat makes no sense, unless you are getting to the point of being unable to produce the thing.

Industry wide NRND means all sorts of things. It can be "We have something that's better value, which we think you should be using. On the other hand, its a perfectly serviceable product from a mainstream fab", and it can mean "the only fab qualified to make this is about to become unserviceable".

[T3sl4co1l]

TI does it too.  I had to press an FAE for answers, they wouldn't post as such in the e2e thread, but turns out the UCC256303 is internally NRND.  Why they don't tell marketing to say the same, I have no clue.  It sure goes further to paint them as an incompetent company.
I haven't heard any good reasons why they shouldn't.  Mind, I have heard possible reasons. Just not good ones.
But, perhaps those same (bad) reasons, are why MCP doesn't for this one.
Tim

https://www.ti.com/product/UCC256303
Pretty clearly NRND.

Hah, shame on me for not checking!

To be clear, this was a year or two ago -- at which time it was "active" in clear and bold print.

At least they finally moved it

Tim