Need Clarification on Embedded Chips

greenspun.com : LUSENET : Electric Utilities and Y2K : One Thread

Several distingusihed individuals have testified before congress that embedded chip systems must be tested individually. They cannot be "type tested". Is individual testing being conducted in the electric power industry?

-- Anonymous, February 08, 1999

Answers

If "type testing" means testing a sample of a make/model for Y2K Readiness, the answer is yes and no. The debate over the reliability of type testing results is ongoing.

Here's the example. If a utility has 100 units of a make/model, how many individual units should they test given the following: The vendor has issued a "Y2K Readiness Statement" suggesting the device is Y2K Ready, the utility tested 20 units and they all passed, and every test conducted by other utilities has been successful. Should that utility test the other 80, or move on to type testing other systems?

-- Anonymous, February 08, 1999


It is my understanding, that the reason that "type testing" doesn't work, is that evidence has shown that a system's model/serial (type) does not have a certifiable connection to the programs that are "burned" onto the individual date sensitive chips. In order to be absolutely sure that you don't have a date problem in a system, those date-sensitive chips must be individually tested/verified because this question exists. You only need to do this if you want to be absolutely sure that there is no problem with a given system. If it is only important to "gestimate" the problem, then "type testing" should work. I would feel better if the utility industry did a complete test of mission-critical (what we need for power) systems, and "guestimate" the non-critical. They could then say : "We have tested every chip in the critical path of system XXX and found it to be compliant." It is my understanding that this is what would be required in order to say that a system is compliant for a specific function.

-- Anonymous, February 09, 1999

This is a terrific site. They've come up with a testing approach that may be a big help in many fields.

Their FAQ has some excellent info on the characteristics of embedded systems:

Embedded Science corp. Delta-T probe FAQ

If I am summarizing accurately: Embedded chips with the potential for date related failure are not "everywhere".many systems can be judged "safe" with a high degree of confidence simply by examination. But where they exist, they are extremely critical and a often a bear to fix. Worth a look if you are interested in the technical details about embedded system remediation.

-- Anonymous, February 09, 1999


tesloid: Thank you for posting this link. It is a big help.

-- Anonymous, February 10, 1999

The Delta-T Probe site also mentions the fact that rollover tests cannot be performed on some embedded systems. It also explains that some problems may reside in memory of systems that don't have an RTC (real time clock).

-- Anonymous, February 10, 1999


One of the shortcomings of type testing is that it does not address application-specific uses of an embedded device. If you have non- critical applications, then using type testing for the embedded systems involved would be logical and acceptable.

But certainly not in safety of operation or single-point-of-failure uses. In such critical uses, application-specific testing is the most appropriate method for determining whether the device is suitable or not.

WW

-- Anonymous, February 13, 1999


Moderation questions? read the FAQ