John Curl's Blowtorch preamplifier part II

[jlsem]

First and foremost on the mind of a designer of high-end audio equipment in regard to component choice should be reliability. More often than not, resistors and capacitors with the highest reliability factor are well in line with boutique parts in terms of cost. I've known people who've owned $20,000 amplifiers that have experienced component failures that are inexcusable. I met a guy one time who designed the electronics for a turntable and when he found out what it was going to sell for, he said he was going to need at least a year to prove the reliability of the motor controller - and that was the end of the project.

John

[john curl]

It has been my experience with both expensive and cost effective designs, that 1/2W resistors have been replaced with 1/4 or 1/8W resistors, wherever possible. In fact Roderstein for the last 10 years or so, STOPPED making 1/2W resistors, even though they made them by the millions for several decades.
What happened to the Carver based amplifier earlier in this thread shows that somebody, somewhere, used a smaller than 1/2W resistor for a feedback resistor (the very WORST place) because it burned up!
Now, with the rare exception of absolutely needing a 1/2W resistor, I suppose that we would ALL chose metal film, except for Scott, who has some high value AB baked carbon resistors that he likes to use. Well, thank goodness it is AB and not another, cheaper brand of carbon resistor, because the EXCESS noise can be terrible with a SMALL, CARBON resistor, especially high values.
The chart on resistor excess noise on page 177 of 'Low Noise Electronic Design' by M and F, shows that carbon resistors have a Noise Index, or NI of -20 to +10. AB baked carbon has an NI range of -20 to -10, a FULL 20 dB difference worst case. Scott lucked out. Metal film, on the same chart, shows -35 to -15 for the whole range, some better than others. It is also stated that carbon resistors are worse, the smaller their wattage rating, and the higher the resistance value. Resistor terminations are also considered a noise source on page 178. They also point out that cheaper manufacturing methods make noisier resistors overall, so let the buyer be aware of this.

[scott wurcer]

Originally Posted by john curl

resistor excess noise on page 177 of 'Low Noise Electronic Design' by M and F, shows that carbon resistors have a Noise Index, or NI of -20 to +10. AB baked carbon has an NI range of -20 to -10, a FULL 20 dB difference worst case. Scott lucked out. Metal film, on the same chart, shows -35 to -15 for the whole range, some better than others. It is also stated that carbon resistors are worse, the smaller their wattage rating, and the higher the resistance value. Resistor terminations are also considered a noise source on page 178. They also point out that cheaper manufacturing methods make noisier resistors overall, so let the buyer be aware of this.

Except at 0 DC current there is no excess noise no luck needed only physics. Please find me some 56Meg Dale RN55's.

[john curl]

0 DC current, mosfet input?

[hitsware]

depletion mode ?

[scott wurcer]

Originally Posted by john curl

0 DC current, mosfet input?

Any FET, do the math. The excess noise goes as V squared at millivolts we are below the white noise by a large margin.

[john curl]

1 to 10 nA, I conclude. Any argument?

[hitsware]

especially with power mosfets
where the total xsistor is many
' cells ' in parallel
statistic cancelling
as the # of cells increases
the noise goes down

[john curl]

Mosfets are very, very noisy. No hope, yet for them.

[PMA]

Originally Posted by scott wurcer

Except at 0 DC current there is no excess noise

Excess noise of resistors: