I've just read that Vishay brochure. It's full of misconceptions - goodness knows what sort of person wrote it, but it wasn't an engineer. As for the "major objection to wirewounds is the inductance that chops the peaks and fails to replicate the higher frequencies of the second and third harmonics..." 
Someone talked to the writer about excess noise due to a DC voltage applied across the resistor. Naturally, you could consider the signal itself to be the DC but (by Vishay's own admission) the effects are small even for poor quality resistors. Bear in mind that they're referencing everything to noise expressed in uV per V of applied DC. In other words, their reference is already at -120dB.
Perhaps I'd better just put my volume control in a bucket of liquid nitrogen...
Someone talked to the writer about excess noise due to a DC voltage applied across the resistor. Naturally, you could consider the signal itself to be the DC but (by Vishay's own admission) the effects are small even for poor quality resistors. Bear in mind that they're referencing everything to noise expressed in uV per V of applied DC. In other words, their reference is already at -120dB.
Perhaps I'd better just put my volume control in a bucket of liquid nitrogen...
Hi,
I used 294K and it confirms that the 100k pot will give noise at about -111db below the 2Vac input signal when it is attenuated to -6db. The noise is effectively 91db below the average level coming off a CDPlayer when the attenuation is set to -6db.
Reducing the volume to an attenuation level of -40db and the noise becomes -71db below the average level off the CDPlayer, yes the volume drops by 34db but the noise only drops by 20db.
But there is something else helping out here.
Let's take a hypothetical amplifier/speaker combination that can reproduce 130db at the listening position, assume the amplifier has an excellent -120db noise to maximum output ratio. Set the pot to -6db and the pot noise is at about 13db, the amplifier noise is at about 10db and the 16bit CDP noise at about 28db. We can hear neither the amp noise nor the pot noise for the racket coming off the CDP.
Reduce the volume to -40db, the respective noise figures become:
pot noise -1db (below threshold of hearing), amplifier noise still 10db, CDP noise about -6db. Now it's the amplifier noise that is dominant.
It appears that 16bit could be improved noise wise by going to 18bit or 20bit (I am not considering resolution of low level signals here). Much more than 20bit is not going to help with noise because we will not be able to hear the improvement.
It appears that the pot value is also inconsequential in respect of noise for our current level of technology.
That still leaves the cable/input capacitance as the major determinant for choosing pot value.
I have some thoughts on resolution but I need to gather them into something comprehensible.
I used 294K and it confirms that the 100k pot will give noise at about -111db below the 2Vac input signal when it is attenuated to -6db. The noise is effectively 91db below the average level coming off a CDPlayer when the attenuation is set to -6db.
Reducing the volume to an attenuation level of -40db and the noise becomes -71db below the average level off the CDPlayer, yes the volume drops by 34db but the noise only drops by 20db.
But there is something else helping out here.
Let's take a hypothetical amplifier/speaker combination that can reproduce 130db at the listening position, assume the amplifier has an excellent -120db noise to maximum output ratio. Set the pot to -6db and the pot noise is at about 13db, the amplifier noise is at about 10db and the 16bit CDP noise at about 28db. We can hear neither the amp noise nor the pot noise for the racket coming off the CDP.
Reduce the volume to -40db, the respective noise figures become:
pot noise -1db (below threshold of hearing), amplifier noise still 10db, CDP noise about -6db. Now it's the amplifier noise that is dominant.
It appears that 16bit could be improved noise wise by going to 18bit or 20bit (I am not considering resolution of low level signals here). Much more than 20bit is not going to help with noise because we will not be able to hear the improvement.
It appears that the pot value is also inconsequential in respect of noise for our current level of technology.
That still leaves the cable/input capacitance as the major determinant for choosing pot value.
I have some thoughts on resolution but I need to gather them into something comprehensible.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.