[Radio Wrangler]

Back in the day of the Leak amps, basic quality resistors were +/-20% tolerance.
10% tolerance parts got a silver band. 5% parts "High stab" got a gold band and that was about it. You didn't dare to ask what their drift over time was like. These percentages were simply how accurate they were at time of manufacture. 2% was for very special parts in measuring instruments.

Modern materials and manufacturing methods have made 1% parts routine and affordable. What's more they have the stability over time and temperature to make that tolerance meaningful.

The two sites you linked to are hilarious. What makes the prices so high is exclusivity or have I got it the wrong way round? the high prices exclude most people from having them.

They seem to make a big thing about noise performance. I've been involved a bit with noise from resistors. In the RF world, the noisiness of components is of great interest and gets expressed in several ways. "Noise Figure" compares the noise created in some device, system or component to the noise power density created in a perfect resistor at a temperature of 290 Kelvins (about room temp, and an arbitrarily picked convenient number) and expresses the difference in deciBels. Another way is to state the temperature an ideal resistor would have to be at to create the same noise power density. This is easier to visualise. Resistors get noisier the hotter they happen to be. An ideal resistor has a characteristic which is linear, and if you project the line downwards, passes through no noise at all at absolute zero. Bad resistors suffer from excess noise and produce more than ths line at real-world temperatures. Good quality metal film resistors are very very close to the ideal line. They aren't expensive and they aren't made out of unicorn hide and unobtainium. Seeing fancy resistors made out of exotic materials just for fashion and marketing reasons is very funny. Looking at the prices, I suppose they couldn't get real unobrainium and so they faked it with unaffordium.

Straight forward metal film resistors without pretensions will do fine. There are many firms who make dependably reliable ones for pennies. Vishay make lots. Holsworthy are a specialist manufacturer of very close tolerance parts to custom designs when you need them. They appear on the hifi collective webpage as 'Holco' though the prices seem to have inflated in the retailing. About 50p for a 0.1% part if you need them, if you get them from industrial rather than boutique sources.

The industry standard equipment for measuring this is the Agilent N8973A family (up to N8975A) of Noise Figure Analysers, along with the N4000A series of 'smart' noise sources. This equipment can measure the noise created in a near ideal resistor down to about 10 degrees above absolute zero. I originated it, led the hardware development team and handled the metrology aspects and liaised with the National Physical Laboratory over transfer standards for calibration. It was a lot of work, but I think it earned me the right to have a laugh at some of the claims attached to fancy resistors. Pick the right types and you are so close to the state of the art for pennies.

By the way, treating noise as power density gets rid of having to state bandwidth or even resistance value in the formulae. Noise power density from an ideal resistor is kT Watts per Hertz. T is the temperature in Kelvins, k is the Maxwell-Boltzmann constant.

David