IEC 60730-1/60335-1 selftest libraries for the MSP430 family

Some test libraries / code samples would be good have developed.

I suspect it would be a handy sample and test application for your various evaluation boards, as well as being useful for people that need to have such code for production or QC hardware test purposes.

I also suspect that the work involved in creating this would not be very large when amortized over all devices in that some well crafted "C" language routines for generic purposes like memory test could be ported across many platforms like MSP, C2000, DSP devices, ARM, OMAP, et. al. 

Assembler based tests for things like registers, peripheral units, et. al. would still be portable across many of the devices within a given platform and series of parts.

That's too bad, however, as a resource, may I point out that the self-diagnostic software modules developed by NEC Electronics for their 8-bit and 16-bit microcontroller families are available, free of charge, with documentation. All of us who are being paid "the big bucks" to develope systems, can now continue to develop safe appliances, like we already have been.[;)] The only thing is that this time we have to do the initial work ourselves.[8-|]

G**buster,

I am sure TI would do this for a fee - maybe a 3rd party consultant can do this as well - at the system level.

Regards,
johnw

TI is working on a IEC60730 biliotheque fpr MSP430, but it is still in process. All I've seen is a early draft and this is now over one year old.

Not only NEC has released a free bliotheque for it's uC, but also Renesas, Toshiba, Luminary Micro, Freescale, Atmel.

However, 'self test' does not necessarily mean 'testing all internal processor components, whether used or not' but rather 'test all relevant functions'. And this is a mostly project-specific test. Due to the complexity and the many possible uses of the MSPs internal hardware, only the programmer can determine what kind of tests needs to be performed an, more important, how to interpret the feedback of the hardwar emodules, as this varies with the external circuitry and the software-based configuration based on this circuitry and the intended use.
Any pure internal self-test will only result in "well, i can write to this register and I can read the same value back and I don't know at all what it may cause outside".

However, many (not all) European regulations lack even a minimum of common sense. But those politicians need some excuse for their salaries.

But nethertheless you are not allowed to sell any commercial product without compliance. 

IEC60730 defines the minimum requirements for "self-test" based on software class.

Therefore it makes sense to group this required test in a bibliotheque, because it is needed by everyone who develops for household devices.

All I want to know is when TI releases this bibliotheque, because the development of it started long time ago and ohter suplliers are still on the market with versions for their devices. This could be a decision reason for a TI microcontroller or not.

TN.Richard said:

But nethertheless you are not allowed to sell any commercial product without compliance.

IEC60730 applies only to a certain group of devices and only for certain functionalities in these devices. Any device that does not control possibly dangerous mechanics is not affected at all. Only if a malfunction may lead to overheating/explostion or similar hazards, then this kind of self tests and plausibility checks is required (and I'm pretty sure that this kind of checks was always already implemented even before IEC60730)

However, a generic library cannot do much more than some basic self tests (like the typical power-on-self-test evey PC makes when the BIOS starts): check the CPU registers and the ram.
Anything beyond this highly depends on the usage and the external hardware, so a generic library is useless. You cannot even check the timers without knowing which kind of crystals is attached (if at all) and other such specific information. The bandwidth of generic tests you can perform increases with the limitation of the target CPU. So the CPU with the most tests available will most likely be the least one you want to use, becuase it is the most limited one.

However, advertising an MCU by the existence of such a library is plain marketing without any useful content. And falling for it sheds a bad light on your qualification (which sadly is true too often for those in the management who make the decisions :( )