- Written by John E. Johnson, Jr.
- Published on 24 December 2013
On the Bench
Distortion measurements were made within a bandwidth of 80 kHz, and into a load of 100 kOhms for the preamplifier and 8 ohms for the power amplifier, unless otherwise specified.
First, the Xs Preamplfier:
At 1 kHz and 2 volts, THD+N was 0.002%. Notice that there are many harmonics, but they are all below -100 dB, which means, no audible harmonics at all. Fantastic performance, as 2 volts is the range you would most likely be using to drive power amplifiers.
Even at 5 volts output, only one harmonic (the third) is above -100 dB, and it is just marginally so at -90 dB.
The side bands around these 19 kHz and 20 kHz test tones are essentially also below -100 dB with 2 volts output. The B-A peak at 1 kHz is also below -100 dB. Inaudible. Incredible!
At 5 volts output, most of the side bands are at -100dB or below, with just a few at -80 dB. The B-A peak at 1 kHz is at -98 dB.
IMD at 2 volts output was only 0.001%.
At 5 volts output, IMD rose slightly, to 0.003%. This is truly landmark performance!
Distortion vs. Frequency is shown below. It shows the typical rise at the high end of the spectrum, but does not go above 0.007% even at 5 volts output.
Distortion vs. Output Voltage resulted in a graph where all the lines overlapped, whether at 2 volts output, 5 volts output, 100 kOhms load or 600 ohms load. I have never seen performance like this. It means the Xs Preamplifier is built to drive any load, and there will be no sagging of the deep bass at high power, which is exactly what I experienced when I cranked the volume.
The measured Frequency Response was 20 Hz - 50 kHz, -0.5 dB.
And now, the Xs 300 Monoblock Power Amplifier:
At 5 volts output, THD+N was 0.02%, with the second harmonic being the major peak.
At 20 volts output, distortion rose to 0.1%, with the second harmonic again being the largest, but the third also contributing a major peak.
Using 19 kHz and 20 kHz sine waves, at 5 volts output, the noise floor was at -100 dB instead of -120 dB as it was in the previous graphs. I had an issue with noise contamination due to using a dedicated 240 volt outlet to power the Xs 300's. This outlet has its own dedicated ground as well. The other components in the test (including the Audio Precision) were connected to 120 volt outlets, which have a different ground. At first, I eliminated the PC and used a laptop with battery power. This helped a bit, and then I connected the grounds of the 120 volt outlet to the ground of the 240 volt outlet. This helped even more, but for the 19 kHz, 20 kHz test, below about 7 volts, this -100 dB noise floor was visible. I could see the noise fluctuating on the Audio Precision, but at 7 volts output from the amplifier, and above, the noise seems to have been suppressed.
Moving the output up to 7 volts resulted in the normal noise floor of ~ -120 dB. There are quite a few side bands, and the B-A peak is at -63 dB. This is part of the sculptured sound that the Pass Labs created for the Xs 300. They listened, made adjustments, listened, made more adjustments, etc., and when they were all happy, that was the design, with the distortion peaks falling wherever they might.
At 20 volts output, the B-A peak is at -42 dB. If I had just looked at the graph, but never heard the amplifier, I would have predicted some audible distortion. But it didn't occur. Here is the reason. Pass Labs used extremely high quality parts, a massive power supply, and more than 100 output devices. They created the sound by changing a capacitor or resistor here and there, but these were not band aids to fix problems. They were there to create a pleasing sound, one that was similar to a Pure Class A triode power amplifier (50 watts) that they designed beforehand, and decided that they would build the Xs 300 to mimic that sound, but not be deficient in the bass or high frequencies. A mass market amplifier (such as in an inexpensive receiver) that has distortion peaks like these, has those peaks because of trying to keep the costs down, using lower quality parts, and using a lot of negative feedback to fix issues in a weak power supply. So, the amplifier produces high distortion peaks from stress, not because the amplifier was designed to sound a specific way. The Xs 300 produces them because it was engineered to do so. It plays the music stress-free, just coasting along, with the peaks added for "flavor" by changing the value of a few parts and listening to the results. Secondly, the Xs 300 is operating in Class A, which makes it very fast. So, short transients, like the leading edge of a ride cymbal "ping" come through clearly.
At 5 volts output, IMD was 0.09%.
And at 20 volts output, IMD was 0.34%. Again, these are results that were "built into" the design.
THD+N vs. Frequency is shown below.
At 8 ohms, with 5 and 20 volts output, distortion maxed out in the audible spectrum at 0.4% and 20 kHz at 20 volts output.
At 4 ohms and 20 volts output, distortion was right at clipping (1% THD+N) at 20 kHz.
Distortion vs. Output at 8 ohms indicates that the Xs 300's hard knee is exactly at the rated output of 300 watts. I'm sure this is a reflection of the Pure Class A bias at full output. Clipping was at 340 watts. The jagged part of the graph at low wattage output reflects the noise contamination from using a 120 volt supply and 240 volt supply for the various components that I described earlier. It's not the amplifier's problem. It's my problem, and I am working on solving it. I have purchased a 240 v to 120 v step-down transformer that I hope will do the trick. Both the Xs Preamplifier and the OPPO BDP-105 will be powered with the transformer (and the Audio Precision when I run the tests). This will result in all of the components having the same ground.
At 4 ohms, the hard knee was at 500 watts, and clipping occurred at 550 watts.