more resistor talk...

[R.G. Keen]

>  What are the criteria for choosing the wattage rating for resistors 
>in a project?  1/8, 1/4, 1/2 etc....   
The criteria that the older engineers (and company practice) taught me
was that you calculate the actual peak/average dissipation in the 
circuit, then pick a resistor from that. For carbon comp, you picked
the rating to be at least twice the average and relied on the high peak
power capability of carbon comp to make it over the peaks. For metal
films, you picked the rating just bigger than the peak, but never less
than twice the average. The rating of the resistor was supposedly based
on a 100C surface temperature at rated power, and the derating held
temperatures - and drift! - in check. For higher long term stability
you derated even further, to 1/4 average. If average is really low, just
pick what's most economical in terms of power rating. 

Just soldering a part in can cause drift. This is worst with carbon
comp, and gets better in order with carbon film, metal film, and
wirewound. Carbons can drift 2% with soldering, in the worst case.

>Also, I would think that for 
>application like oscillators and other precision thingies of ours, a 
>low tolerance would be best.  Are the benefits of using 1% over 5% 
>that much?  
Depends entirely on what you need the resistor for. In some
applications, the absolute accuracy doesn't make any material
difference, or is adjusted out by other things. Metal film, which is
FAR better for tolerance, also happens to be FAR better for drift,
which does matter. Metal film is the technology of choice.