Tweakers' Asylum

Posted by Al Sekela (A) on January 15, 2008 at 15:00:17

In Reply to: RE: Thanks for sharing your results. Follow up posted by Awe-d-o-file on January 15, 2008 at 09:11:47

The primary purpose of the resistors is to damp resonances on the house wiring. The ideal way to stop a cable from resonating is to terminate it with a resistor that matches its characteristic impedance. The capacitors are there to keep the 60 Hz power waveform out of the resistors, as they would have to dissipate on the order of 120 watts if connected directly to the line. The larger the net capacitance, the lower the corner frequency, below which the filter has diminishing effect and above which it couples the resistor to the line.

In your case, the capacitors are closely spaced in value, so they will have overlapping effective frequency bands. The effective R seen by the line will be one-third of 120 ohms for frequencies above the smallest capacitor corner frequency and below the largest capacitor resonance frequency. This may be a good thing, as there will be ultrasonic noise on the line that is below the house wiring resonant frequency range. The lower the resistance this noise encounters, the better.

My house, and two of my dedicated audio circuits, are wired with Romex-type plastic jacket cable. I measured the inductance and capacitance of a sample of similar cable and came up with the 120 ohm estimate for the characteristic impedance. Other types of wire, such as armored cable, would have different values of characteristic impedance.

Making R too large (compared to the type of wire used for the house circuit) results in underdamping and some ringing from reflection of the noise. Making R too small results in overdamping and reflection of noise.

I'm doing experiments with different values of R. At this time, 110 ohms sounds best. To me, overdamping sounds closed-in, while underdamping results in excessive treble artifacts. That is just my system, though, and other systems may have different results. The TI-Shield bartc mentioned may also affect the performance so that the optimum value of R may change.

These are small performance changes. I would not worry about experimenting with different R values unless you have a lot of time and interest. The TI-Shield has to have at least a week of burn-in before you can hear something close to the final effect.