You are absolutely correct. Frankly it is a topic that would take a lot more words than you could cram into a single column or blog. One of the main points to be made is that (seemingly ... and I could be wrong) many design engineers are not aware of all the factors that affect the individual parts that we use. It is also true that many of these effects are small relative to the 'big' hitters but, until you get a more comprehensive picture, you don't know what you can ignore! In my 50+ years of electronics design, I have been constantly surprised by undiscovered factoids. Ah, life can be interesting!
Robert thank you for the very detailed response. You are correct there are a number of other paramters that one must consider for a compete design.
However, the article was mainly looking at worse case analysis, which typically covers such things as tolerance and drift and the effects upon the peformance of the circuit.
There is another blog which Max will hopefully be pulling across from the old programmeable logic design line which addresses the reliability and part stress anaylsis.
Typically the reliability analysis would look at the component failure rate, however this could be either a parts count or stressed reliability in which case the voltage, current and power rating of the resistor would be considered from the part stress analysis.
I should do a blog showing how all these are interconnected and analysed to demonstrate a reliable system.
With low value leaded resistors, the measurement point on the leads is sometimes specified since the temperature coefficient of the leads can contribute significantly to the measurement. With the old venerable Allen Bradley carbon composition resistors, humidty could affect the resistance value as the carbon comp material would absorb the humidity and change the value. Baking the resistors for several hours would drive out the moisture and reveal the resistor's true value.
Your discussion is essentially correct but does not include everything that should be included if you are going to do a really complete job. Just off the top of my head here is my list:
Resistor tolerance (which you covered nicely although the statiscal distribution may be of importance - I usually just assume a square distribution - i.e. it is equally likely that the resistor value will be anywhere within it's specified range).
Power Rating (including the derating factors for high temperature - some resistor types start to derate at 25 degC and are rated at zero dissipation at 100 degC). You need to do a check on the in circuit power dissipation of ALL your resistors.
Voltage rating (some resistor types are only good for 50 V or so - might not be a problem but you need to be aware of it).
Reliability (and the impact of all the stress factors on the reliability).
Current ratings - short impulse as well as long term (e.g. heating) (a 10A, 100 ns wide pulse may not exceed the power rating of a small resistor but it might very well blow it to bits).
Changes in value to stuff like the thermal shock of being soldered into your circuit.
Non-linearity (e.g. variations in resistance with applied voltage).
Effects of shock (on reliability, etc).
Probably more but it is late at night and I have some wood working to do.
You are correct this does assume we are working at DC, I was going to follow it up with one looking at AC but this would look at other discretes as well as once frequency is introduced you need to consider the parasitics
Replay available now: A handful of emerging network technologies are competing to be the preferred wide-area connection for the Internet of Things. All claim lower costs and power use than cellular but none have wide deployment yet. Listen in as proponents of leading contenders make their case to be the metro or national IoT network of the future. Rick Merritt, EE Times Silicon Valley Bureau Chief, moderators this discussion. Join in and ask his guests questions.