Resistors drifting in value with age?

[bikerhifinut]

As a postscript, I have noticed some component suppliers making a big thing of selling things like "Genuine Mills Carbon composition resistors" , and at a fair old premium. (plus other long gone manufacturers).
After reading this topic I shan't be worried about avoiding them and buying cheaper, quieter, more stable and close tolerance resistors for a tenth (or less) of the price!

A.

[Lucien Nunes]

A few of us have investigated this phenomenon, there's some relevant info here:

Why do resistors go high?

Moisture and high voltage, perhaps especially when combined, seem to aggravate drift. Low values occasionally go low, especially if overheated.

the 1-Watt type where the carbon 'stick' is encapsulated in a white ceramic tube with a tan-coloured cement plugging the ends

These are Erie products, I agree that they are particularly prone to drift and unfortunately I have over 50,000 of them - mostly high values - in various organs, which was one of the factors inspiring my interest in the subject.
Info and scans of datasheets are here: Erie resistor data

When replacing resistors, my favourites for appearance include the Arcol RCC050 carbon comps which look just like 1960s Allen Bradleys and much like Morganites and 2W or 3W carbon films that are about the right size and colour to stand-in for 1/4W & 1/2W Eries. For performance, of course metal film is the best bet.

Always watch out for voltage rating of newer parts, as well as power rating. It is often lower for the same combination of resistance and power and might be the limiting factor on higher HT circuits.

PS please do not junk NOS components especially if boxed! I am looking for stock to equip my reconstruction of a vintage service shop at Electrokinetica. The parts will never be used so the performance is immaterial, but I can't stock the shelves and drawers with pulls.

[stuarth]

I have an original Simpson 260, a 20k/v meter made from 1937 on. This is the very early version with the flat metal front panel rather than the later molded/engraved panel. It is accurate on current ranges, but hopelessly inaccurate on voltage ranges. Each voltage range resistor is made up of a pair of high value carbon composition resistors, presumably paired up on test for the correct value.

This meter is completely unmolested, all the innards are varnished, so since I have dozens of other analogue meters, including more recent Simpson 260s, I shall leave it unmolested.

In defence of Simpson, I should point out this instrument pre-dates the AVO 8, I believe Simpson switched to film resistors by the time the AVO 8 was introduced at the end of the 1940s. My later Simpsons do not suffer degraded accuracy.

Stuart

[turretslug]

Recently working on a Chapman S6BS tuner, it struck me that most of the resistors were either 4k7 (anode feeds) or 270k (AGC-g1 decouplers and others)- evidently these values either chimed nicely in the designer's mind or came cheaply by the sackful. Measuring them once the associated 'orrible 'unts had been evicted showed the 4k7s without exception to be either around 4k or 5k5, the 270ks were approximately 210k or 330k. My assumption was that they were actually not far away from their original values but fell into the 20% band, the closer ones having been sifted into 10 and 5% categories. I left them all in, being decouplers and very uncritical, I reckon that the 4k7s could be anything from about 1k to 10k, the 270ks anywhere from about 100k to 1M. As always, it's worth a judgement over exactly what each component is doing.