 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.161.54.196
In Reply to: Reducing Noise by Choosing Impedances posted by Charles Peterson on December 29, 2005 at 17:15:59:
Charles,the noise from a shunt resistor does decrease as the resistor value is increased as in your example of having an almost infinite shunt resistor. If the resistor was placed in series with the output then the noise voltage due to that resistor would add directly to the signal and would increase as the resistor value was increased (noise voltage power is proportional to R). In the shunt case, however, it is easier to think of the equivalent noise current of the resistor that will flow in the signal path. Noise current power is inversely proportional to the resistor value so increasing R gives less current noise.
When thinking about the impact of noise at different places in any signal processing chain it is important to consider the signal to noise ratio at those points. Usually the signal is amplified along the chain and input noise is amplified also. Each block will add some noise of its own but if that is much lower than the noise at the input then it is of little consequence. This is well known to radio engineers as Friis' formula - the noise performance of the first block in the chain is most important so that it adds least to the inherent noise of the signal source. At the output of that first block the signal and noise levels are much higher so the noise performance of subsequent processing blocks in less important (though the overload requirement goes up as the signal is getting bigger).
If you have the power amp pot always on full and adjust the 5k pot then I would expect you to get little benefit from the change - as you're still basically using a 5k pot. Removing the 5k pot from the x-over shouldn't hurt it and you can add a shunt resistor if you like - a higher value will not give more noise.
Regards
13DoW
Follow Ups:
Thanks to all so far! This noise current/voltage thing is definitely expanding my mind.With regards to the Friis' law consideration, one has to thing that a modern hifi system is an odd beast. While one might assume that the preamp-> amp connection is of least significance as regards to noise (and hum) of all interconnections, in most cases now it is the MOST important, because it actually carries the LOWEST level signal.
The output from most "digital" sources is 1-2 volts peak with an average not far from 6dB below that. The output from most FM tuners, tape decks, etc., is also in that range. The only exception would be the direct output from a phono cartridge, and often people now use separate preamplification devices close to the cartridge to handle that.
On the other hand, I may be driving my speakers with an average power of 5 watts, which is about 30dB below the maximum level of the amp, and the maximum level of the amp is achieved with about 1-2V.
Thus, while the source device CD player is sending about 1V average to the preamp, the preamp (better described as a buffered attenuator, and I like mine, thank you very much) is typically sending only about 50mV to the poweramp! When dealing with a 50mV average signal, issues of noise, EMF rejection, etc., become very important! (In fact, I'm beginning to think EMF rejection is truly the important issue here. But more about that in my reply to John Curl later tonight or tomorrow.) The 33000 - 100,000 ohm input impedance of the amplifier isn't helping this! Plus, it's currently quite typical (as in my system) to have the longest interconnect (mine is 4m) between "preamp" (or electronic crossover actually) to poweramp.
If we had to design a system from scratch, blank sheet of paper, we would never do it like this. We would have much higher driving level for the power amp, and much lower input impedance (possibly even 1K or 600 ohms, sound familiar?), and balanced connections. That might sound familiar as it's basically what professional equipment does, with a current defacto peak level of +22da (around 8 volts).
Now, back to Friis' law, it's true that there are much lower levels inside components. Inside my CD player or DAC there is a very tiny level of current from or inside the DAC chip, which then has to be very carefully converted to voltage and amplified. But that is all done very well with a careful engineering analysis, and in fact contributes hardly anything to my system noise (and hum!). On the other hand, the ad hoc interconnections between components are where most of the hum comes from, and I suspect much of the noise comes from related issues, chief among them the poor choices of driving voltage levels and impedances, I still believe.
Charles
13DoW wrote:
When thinking about the impact of noise at different places in any signal processing chain it is important to consider the signal to noise ratio at those points. Usually the signal is amplified along the chain and input noise is amplified also. Each block will add some noise of its own but if that is much lower than the noise at the input then it is of little consequence. This is well known to radio engineers as Friis' formula - the noise performance of the first block in the chain is most important so that it adds least to the inherent noise of the signal source. At the output of that first block the signal and noise levels are much higher so the noise performance of subsequent processing blocks in less important (though the overload requirement goes up as the signal is getting bigger).
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
