Just need a bit of help understanding a past paper.

What is the difference between a defect and a fault?

Can someone explain the single stuck fault model?

And what are the limitations of this model?

Thanks
Jez

 

A defect is something that prevents a component or system from ever working correctly.

A fault is something that happens to a working system that prevents it from working correctly. A fault may be temporary and recoverable, or it may be permanent, which I guess turns it into a defect.

The single stuck fault is simply the logic value on a particular net that is forced to a zero or a one regardless of any other consideration.

The limitations in a system with a large number of nets is that it takes a while to evaluate the results of all the possible "stuck at faults" one at a time.

 

A defect is something that prevents a component of system from ever working correctly.

The defect need not prevent operation, it may be merely cosmetic. It may also be hidden or 'latent' and not come to light for a long period of time.

The boundaries between the terms defects and faults are blurred but a rule of thumb might be:

Defects arrive with the item from new. That is the new item is imperfect (defective), but may be functional.
Faults appear over time with use as a result of wear or accident. The new item may have been perfect or defective.

Both defects and faults may or may not prevent or inhibit proper action, but neither need do so.

 

I respectfully disagree with the previous answer. In the sense in which the term is normally used within the industry, that is, in the context of "fault coverage" for example, a "fault" is a conceptual or computer model, whereas a "defect" is a physical reality. http://billgascoyne.wordpress.com/2009/02/02/fault-coverage-vs-defect-coverage/ (OK, shameless plug for my own blog; sue me!)

 

Well here's a short extract from your link, Bill

many people fail to realize that a fault is only an abstract model of a defect

Rarely have I heard so much claptrap from engineers. That sort on nonsense usually comes from politicians.


You obtain a new audio amp and connect two different inputs that you can switch between.

Consider the following situations.

1) You hear sound from input A but when you switch to input B you hear nothing.

1.1) Upon studying the drawings you see that the amplifier was correctly designed, but when manufactured a wire connecting input B to the mainboard was omitted. Was this a fault or a defect?

1.2) Upon studying the drawings you see that there is no connecting wire from input B shown to the mainboard and look inside to find the amplifier was correctly manufactured according to the drawing so input B was never connected. Was this a fault or a defect?

2) When you connect up the amp everything functions perfectly, however:-

2.1) The amp functions correctly for a short while but after 2 weeks you hear no sound from input B. Upon looking inside you see that inadequate heatsinking has caused components to melt or burnout. Was this a fault or a defect? and do you take the amp back for a refund.

2.2) The amp functions correctly for some years but one day you hear no sound from input B and look inside to see that the connection wire has fallen from the switch. Was this a fault or a defect? would you take the amp back for a refund?

3.1) The amp functions correctly for some years but one day you hear good sound from input A but when you switch to input B you hear distorted sound. You look inside and find the switch contacts have become dirty. Upon cleaning good sound is restored. Was this a fault or a defect?

 

Sir,

Do you understand the word, "respectfully"? I must take umbrage at being compared to a politician, nor am I in any way talking nonsense.

I did not know we were talking about audio amplifiers soldered together with wire and figuring out who to blame and whether or not it's the manufacturer's responsibility (yes, I avoided the word "fault" right there). Given that the original poster also inquired about the Single Stuck-at Fault model, I presume that the topic is with regard to digital integrated circuit design and manufacture, in which the Single Stuck-at Fault model is widely used. In this context, "fault" has nothing to do with blame or whether we're talking about a design flaw or a manufacturing flaw. I've been in the semiconductor industry for over twenty years; trust me, this is how the terminology is used.

 

I've been in the semiconductor industry for over twenty years; trust me, this is how the terminology is used

While I am sure you are quite correct in a limited sense, the larger meaning is different, as is a definition of those terms in software design.

Posts #2 & 3 are entirely correct in their contexts. Bieing a technician in background (40 years), I would even throw in a composite definition of a "design fault", meaning a poor or compromised design that lead to a failure over time, but not immadiately. It is entirely possible to discover other shadings of "faults" and "defects".

The exact context of the original question is unknown, so holding out for any one definition may not be a complete answer.

 

Hi Bill, thank you for sharing your expert knowledge in a particular area with us.
I have certainly learned something about a field that is not my own.

Equally I was not getting at you personally, or comparing you to a politician as you are not personally responsible for the terminology. That honour was addressed at the author of the article you linked too.

Having said all that I stand by my comments. The audio amp was merely an example with perhaps wider recognition than your area? Suppose for instance we went completely digital, perhaps even with the components you are talking about, and replaced the audio amp with a switched two input frequency counter.

Exactly the same statements could be made about this digital device, except that we would now see the output on the counter screen (or not) instead of hearing it through a speaker. The counter may even be subject to the stuck condition you mention.

But that is only one possible fault condition, lots of others still apply and need to be classified as faults or defects if we are to make a distinction.

Communication difficulties often arise when one small group redefines a common word.

 

Does the context of posts 2 & 3 include the concept of single stuck at faults?

 

Studiot,

I am the author of the article I linked to.

 

So how do you recommend we deal with a piece of digital equipment which could be subject to the (fault?) condition you describe or to the many fault conditions I described?