By necessity, I am a meter pontiff in my professional life. In my audio life, I prefer to use my ears as the primary measurement tool for the evaluation of good sound. Many of the desirable attributes of good sound can not be measured by currently available testing methodology and equipment. Quantities such as power, impedance, harmonic distortion, etc. can easily be measured. Other quantities, such as those responsible for imaging properties, can not be easily measured. Indeed, to my knowledge, those quantities have not yet been identified!
We do know that, generally, assuming good design and construction techniques have been followed, the more noise we remove from the signal path, the better the resulting sound staging, imaging and musical detail.
Although noise removal from the signal path must be given rigorous attention in order to achieve good sound, noise in the power delivery path must be given the same rigorous attention. Why? Because the power delivery path provides the electrical energy, the "raw material", that an audio component uses to recreate a musical event.
I wanted to obtain some data on the quality of power delivered to my two channel audio system. Over a one week period (Friday to Friday) at various times of each day, I measured the wall voltage at the dedicated audio outlets in my living room. I also measured the wall voltage at various other outlets throughout my home. During nights and early mornings (12am - 6am), the new dedicated outlets measured about 1 volt higher than the old dedicated audio and other household outlets. During the day and early evening, the dedicated audio and other household outlets measured very close to one another, with a typical difference of +/- 0.3 volts between them. A Radio Shack digital multimeter was used to measure wall voltage. During the measurement week, voltage levels ranged from a low of 118 volts (two afternoons at 5pm) to a high of 122.7 volts (two mornings at 4am). There were only two days where a voltage less than 120 volts was seen.
On the last day of the measurement week, between 4:30pm and 5:30pm, a Tektronix model TDS 2012 oscilloscope was used to generate sine wave plots and spectral plots of the line noise on the dedicated audio outlets.
Figure 1. An oscilloscope is a handy thing to have around the house...if you want
to see the noise gremlins hiding in your wall.
Sine Wave Plot
Sine wave plots are good for letting you see how distorted (or not) the power is coming out of your wall. Spectral plots (Fast Fourier Transform plots) are good for letting you see what is causing the distortions.
Figure 2. Power coming out of one of the original audio system dedicated outlets.
Each vertical division of the oscilloscope screen represents 50 volts. Each horizontal division represents 2.5 milliseconds. The sine wave coming out of one of the original dedicated outlets for the two channel system does not look too bad. Evidence of mild waveform distortion is seen in the "bumps" along the waveform and in the flattening near the upper and lower peaks. The oscilloscope measured a peak to peak (from the lowest point to the highest point of one sine wave) voltage of 340 volts. This corresponds to a "wall" voltage of 120.2 volts (340 volts divided by 2, then divided by the square root of 2 equals 120.2 volts). As we will see later, gremlins are quite adept at hiding behind walls and rather benign looking sine waves.
Figure 3. Sine wave plot of the power from one of the original dedicated outlets
(20A circuit) with wireless networking equipment plugged in.
Figure 4. Sine wave plot of the power from one of the original dedicated outlets with
wireless networking equipment unplugged. Measurement was taken from a Signal
Cable MagicStrip (10 AWG).
Figure 5. Power output from new dedicated 20A circuit terminated with a PS Audio
Power Port receptacle.
Figure 6. Power output from PS Audio Statement SC power cable connected to new
dedicated 20A circuit terminated with a PS Audio Power Port receptacle.
The sine waves for (1) one of the the original dedicated outlets (figure 2), (2) the Signal Cable MagicStrip (10 AWG) with wireless networking equipment plugged in (figure 3), (3) the Signal Cable MagicStrip with wireless networking equipment unplugged (figure 4), and (4) one of the new dedicated outlets terminated with a PS Audio Power Port receptacle (figure 5) are very similar, showing a fairly smooth sine waveform with a few distortion "bumps" along the waveform and some flattening near the upper and lower peaks. The primary differences were a slight variations in output voltage: 120.2 volts (340 volts peak to peak) from one of the original dedicated outlets, 120.9 volts (342 volts peak to peak) from the output of the Signal Cable MagicStrip connected to one of the original dedicated outlets, and 121.6 volts (344 volts peak to peak) from a Power Port outlet on one of the new dedicated circuits.
The plot in figure 6 was taken from the output of a PS Audio xStream Statement SC power cord (8 AWG). The distortion bumps are gone and the flattening near the peaks has diminished. The peaks show a more symmetrical shape on either side of the peak midpoints. This verifies PS Audio's claim that the Statement SC has a "cleaning" effect on the power signal. The voltage was raised a bit to 122.3 volts (346 volts peak to peak) due to the much lower resistance of the larger gauge cable.
Time domain plots (signal amplitude vs. time) are good for showing the net effect of noise components (those gremlins). However, residential noise gremlins are usually are not of sufficient size to cause gross distortions in the power signal coming from the wall. They usually are small in size and prefer to attack in large numbers over a large area. This makes it harder to pinpoint and eradicate them...or so they think. If we wish to find out exactly where and how big the gremlins are, we need to look at a frequency domain plot (signal power vs. frequency). Frequency domain (spectral) plots show the power contained in each frequency component of a signal.
Figure 7. Spectral plot of power coming from one of the original dedicated audio outlets.
Figure 7 shows the spectral plot for the power coming from one of the original dedicated audio outlets. Each vertical division of the spectral plot represents 10 dB. Each horizontal division represents 50 Hertz. The sampling frequency used for computing the Fast Fourier Transform was 1000 samples per second. The first spike on the left edge is the DC content in the power line. The next, and largest, spike, is the 60 Hertz, 120.2 volt AC power signal. The next three large spikes are the significant odd order harmonics at 180, 300, and 420 Hertz. Note all the trash (noise frequencies...gremlins ;)) grouped around the 60 Hertz spike. The spike at 60 Hertz is the only part of the signal useful in audio and video reproduction. All the other spikes, large and small, are NOISE and constitute the "noise floor". When noise is removed or attenuated, the height of the noise floor is reduced and more of the signal becomes apparent. Than is why lowering noise results in an apparently louder speaker volume, although the actual sound level remains the same.
When I initially noticed the last large spike at the right, I thought it was the 8th harmonic. But then, I thought that the 8th harmonic should have been obscured in the thick layer of low magnitude line noise at the bottom of the plot. Putting the oscilloscope cursor on that spike showed it to be at 470 Hz rather than the expected 8th harmonic frequency of 480 Hz. I knew that the significant 3rd, 5th, and 7th harmonics result due to the way power is generated by the utility company. I did not know where this 470 Hz noise was coming from. After some effort, I found out that the 470 Hz frequency is one of the control tones generated by the power company in order to communicate with residential power meters. Other tones are used to control or communicate with other devices on the power grid (street lights, etc.). Surprise....some noise gremlins actually work for the power company and have a useful purpose in life.
The voltage level of each of the fundamental and harmonics is calculated this way: The vertical axis does not start at 0 dB. It starts at a reference magnitude of -27 db (0.044 volts) The fundamental 60 Hz spike has a magnitude of 68.6 dB, therefore -27 dB + 68.6 dB = 41.6 dB, which is the absolute (real) magnitude of the fundamental 60 Hz frequency. A reference voltage of 1 volt rms (Vo) is assumed. The rms voltage (voltage coming out of the wall) is calculated by Vrms=Vo x 10^(dB/20) and
for the 60 Hz frequency,
Vrms = 1 x 10^(41.6/20) = 120.2 volts.
The voltages of the DC component and the 3rd, 5th, and 7th harmonics were 0.65 volt, 2.14 volt, 0.98 volt, and 0.91 volts respectively. The 470 Hz control tone was at 0.45 volts. Although the typical voltages and energy levels of individual AC line noise components are very small compared to the main 60 Hz frequency, they collectively cause significant signal distortion which obscures detail in audio and video signals.