On the issue of documentation

(Translator’s PP note - I was hoping that the problem was only with us, and if I don’t see any parameters in the datasheet, then either I just don’t know how to read them, or this is the wrong date and there are real ones somewhere. But since it writes Jack, who, I hope, has access to the most correct date, means, really All Very Bad).

Does the supplier have to guarantee the delivery of components that match the description? And vice versa, how to make the description match the real component?

Recently, I received several emails from engineers who discovered that various components simply do not work as they should. This phenomenon can not be called new: I remember how in the 1970s, NEC made the first floppy disk controller. The description of several registers was incorrect. At that time, there was a joke (an example of laughter through tears) that the most frequently used bits should be described with inversion or moving. This problem was faced by everyone who worked in our industry for more or less long.

In the old days, the components were simple, and the descriptions were brief. It is not difficult to completely describe a simple component. Today, even the 7404 step inverter has a 27-page technical description!
')
MCU and SoCs are somewhat more complicated. A small 8-bit device can have a specification of hundreds of pages; for more complex devices, it can take thousands. People are not perfect, so how can we expect perfection in documentation?

However, it is absolutely necessary that the technical description be guaranteed. If you used a component as specified in the documentation, the supplier must ensure its operation.

Data tables are usually incorrect. The components themselves often have flaws; One correspondent complained that in a normal 8-bit MCU, one (out of several dozen) instructions did not work correctly. The vendor replied that "no one else uses assembler, and the C compilers do not use this instruction."

If the technical description is no longer a guarantee of operation, then why is it even necessary?

Customers who detect errors can report this to the supplier, which (we hope) will either fix the component / documentation or issue a list of errors. But another common complaint is that at least some providers never respond to requests for help. Today, we are often redirected to the support forum, where most of the “support” is provided by other clients (PPs - in many forums there are company representatives, at least they designate themselves that way, but their messages and answers are still not official). Customer communication may not even be taken into account by the company. Forums are very effective for the supplier, but often do not meet the needs of users.

If I use 7404, where the date states that max tPLH is 22 ns, then I can certainly expect that the component, if used correctly, will never be slower than the specified value. Shouldn't we also be confident in the behavior of, say, MCU timers in any mode?

How do you feel about what is happening? And I would like to hear the opinion of the manufacturers (it is possible on condition of anonymity).

(PP is my typical conversation with the developers of one of the domestic firms "You have a mistake in the description on page XXX. - Thank you, we know." They are fucking aware of the error for 3 years now and have not done anything to fix it, but found time to redesign the site)

(PP but it would be cool against the background of universal pofigism, suddenly someone, and preferably a domestic manufacturer, suddenly rolls onto the market a complete, comprehensive and absolutely correct description on its component and everything, including abroad, happily starts to apply it, other manufacturers start to envy themselves wildly and are forced to raise the quality of their documentation).