PS Audio Stellar Strata MK2 Integrated Amplifier Review

We (Dr. Yongki Go) reviewed the first version back in December 2020, and here, in the summer of 2024, it’s the MK2, reviewed by JJ with lots and lots of revealing bench tests.

PS Audio is an audio icon in the industry. Paul McGowan has helmed it for decades, although his son is now becoming more in charge.

One of the first PS Audio products I obtained was a power regenerator that took the incoming 120 Volts AC, converted it to DC, and then regenerated the power by using a 60 Hz sine wave. The output was 2,000 Watts of 120 Volts, 60 Hz power with no noise.

Over the years, I corresponded with him and did a video interview with him during his early years at CES in Las Vegas.

Those power regenerators are now all Class D, as is the current product under review, the PS Audio Stellar Strata Integrated Amplifier. This is the MK2. We reviewed the first version several years ago. The MK2 has an MM/MC phono preamp section that the first version did not have, and the first version had a streamer capability, while the MK2 does not. Also, the DAC was completely redesigned for the MK2.

The PS Audio Stellar Strata is a medium power (100 Watts per channel into 8 Ohms) integrated design that includes a DAC, capable of 24/384 sampling, DSD 256, and an MM/MC phono preamplifier section.

The power amplifier is Class D. This means high efficiency, but also, it is compact enough that it really looks nice (hint: spouse approved). It only gets warm during use, never hot, so ventilation is easy.

The rear panel has a plethora of inputs and outputs. Click on the image below to see a full-sized version. There is one set of RCA and XLR analog inputs, MM and MC analog phono inputs, analog RCA preamplifier outputs, and a full set of digital inputs including RCA, HDMI, USB, and Optical. The two sets of speaker binding posts are heavy-duty. There is a separate USB-A input on the left side for software updates. The 120 Volt AC input socket is grounded, and there is a toggle on/off switch. The front panel has another on/off (standby) button on the left side. On the right end of the front panel is a 1/4″ stereo phone jack to plug in your set of headphones, using the built-in headphone amplifier (3.25 Watts into 16 Ohms).

The remote is small, with buttons for power, volume, phase, dim, mute, and input selection. HT is for the HDMI inputs. Volume control is by the unlabeled horizontal button in the middle, but you can also adjust the volume using the up and down arrows. I took this photo on my dinner table placemat. I had to find a space that didn’t have spaghetti sauce stains.

I used the PS Audio Stellar Strata Integrated Amplifier with an HP Laptop USB-A digital output to the digital USB-B input on an AURALiC ALTAIR G1 Digital Audio Streamer and a VPI HRX turntable with Sumiko Blackbird MC cartridge. Speakers included Aperion Verus III Concert V8B and Polk Audio R200AE (both of these models are stand-mount speakers). The cables were by Clarus.

Here are my standard test albums. I don’t use them all with every review, but I do use some of them.

It was a pleasant surprise to see that, at low output (2 Watts), into 8 Ohms, distortion was primarily 2nd-ordered (Figure 2 below). So, at background levels, you will get a very musical, warm sound. This test was with a 1 Volt input, Volume set to 48 (max is 100), and an output of 4 Volts. Unless otherwise specified, I used the XLR analog stereo inputs on the Strata.

With the same settings, but with a 4 Ohm load, the results were about the same (Figure 2).

With the same input, but the volume adjusted to 78, the output was now 20 Volts. Into 8 Ohms, THD+N was 0.037%. The 2nd-ordered harmonic was about the same height as the 3rd-ordered harmonic, which is good.

Same settings in Figure 4 as in Figure 3 except it was into 4 Ohms. The output voltage dropped a bit, and THD+N was almost twice as much as with an 8 Ohm load. And, now the 2nd-ordered harmonic is substantially lower than the 3rd-ordered harmonic.

Moving on to the use of 19 kHz and 20 kHz sine waves, this first spectrum (Figure 5) is 1 Volt input and 3.1 Volts output into 8 Ohms, Volume set to 48. There are four distinguishable side peaks on either side of the fundamentals.

And, with the same settings into 4 Ohms, Figure 6 below, the output was reduced to 2.5 Volts. There are still only five visible side peaks though.

Increasing the volume setting to 78 resulted in 15.6 Volts output into 8 Ohms. There are now nine visible side peaks. This is what happens with all amplifiers. The question is how many side peaks and how high are they? The results here are about medium for a power amplifier.

At 4 Ohms (Figure 8), the voltage dropped even more, to 12.85 Volts. However, the number of side peaks did not increase, although their height is a bit more. I am pleased and surprised at this result with such a reasonably priced amplifier.

A standard IMD test uses 60 Hz and 7 kHz sine waves. Here (Figure 9), we used 1 Volt which produced 3.34 Volts out into 8 Ohms. IMD was 0.045%.

Changing the load to 4 Ohms (Figure 10) reduced the output voltage a bit and increased the IMD. This is expected.

Figure 11, below, shows the IMD at higher voltage into 8 Ohms. IMD was 0.07% which is not bad.

With a 4 Ohm load, distortion was still quite manageable at 0.052%.

The Frequency Response into 8 Ohms is shown below in Figure 13. The response is within 1.5 dB, 20 Hz – 20 kHz with a rise in response above 20 kHz to a maximum at 60 kHz, then a decline due to the low-pass filter that is necessary with Class D amplification.

The response into 4 Ohms (Figure 14) shows a decline in response at 20 kHz rather than an increase that was shown in Figure 13 at 8 Ohms. This is typical of Class D. The sharp decline above 60 kHz is the low-pass filter. The peak at 70 kHz is the low-pass filter resonance frequency.

For THD+N vs. Frequency, first the spectrum for low voltage into an 8 Ohm load. You can see a rise in distortion above 4 kHz maxing out at 0.3% at 20 kHz.

Low voltage into a 4 Ohm load produced about the same results, Figure 16.

Higher voltage (20 Volts) into an 8 Ohm load (Figure 17) shows more distortion in the high frequencies.

And, with a 4 Ohm load, distortion is also high in upper frequencies. The spikes are caused by the Audio Precision occasionally not being quite able to handle Class D switching.

Class D amplifiers have to use a low-pass filter at the amplifier’s output before the speaker to reduce the amount of the switching frequency noise to be passed. The switching frequency of the PS Audio Stellar Strata Integrated Amplifier is 100 kHz. A generic low-pass filter schematic looks like this (Figure below). The output from the power amplifier is on the left, and the RL on the right is the speaker. The problem with putting the low-pass crossover frequency too close to 20 kHz is that the roll-off will include significant amounts of energy below 20 kHz since a crossover results in roll-off below the crossover frequency as well as above it. Putting the low-pass frequency out beyond 20 kHz allows harmonic distortion in the 20 kHz to 40 kHz region to pass through to the speakers.

If the filter is not strong enough, then a significant amount of the switching frequency will enter the speakers and be emitted as RF interference. On top of all this is the fact that the inductors in the low-pass filter will act as antennae and pick up the 60 Hz power supply frequency if they are not orientated properly. Even though Class D is an analog process and not digital, the switching signal does still plots sample voltages along the waveform, so the mathematical deconstruction of the waveform still produces harmonic distortion. For a switching frequency of 100 kHz, there are five voltages plotted for a complete waveform at 20 kHz. It’s all a tradeoff, but Class D is definitely in the future of audio in a big way.

Power supply decoupling and audio signal filtering for the Class-D audio power amplifier

THD+N vs. Power (Watts) output into 8 Ohms is shown below in Figure 19. The knee is at 75 Watts, and clipping (1% THD+N) is at 105 Watts (per channel).

With a 4 Ohm load, the knee is at 135 Watts, and clipping is at 180 Watts.

Testing the preamplifier by itself and using the RCA digital input jack, the results of activating the three available digital filters on the frequency response with 16/44.1 sampling are shown below in Figure 21. The digital input was 0 dBFS which means full DAC analog output into the preamplifier section, The Volume was set to 48. Each filter delivers a different roll-off response at 20 kHz. You just need to try them all to see which one you prefer. Filter 2 gives the most high-frequency roll-off, and this will make the sound a bit smoother. Since this Class D amplifier has quite a bit of high frequency distortion, start with Filter 2, and then switch to the others, using music that has a lot of high frequencies, such as violins.

Figure 22 below shows a 19 kHz, and 20 kHz test with the RCA digital input, 16/44.1 sampling, and the DAC set to minus 5 dBFS. Volume was set to 48. Output was 176 mV. There are only 2 side peaks (2 on either side of the 19 and 20 kHz fundamentals), indicating the DAC and analog preamplifier section are quite good.

Figure 23 below shows a 1/10th decade response of the preamplifier, using the XLR analog inputs, preamplifier outputs, 1 Volt input, and volume set to 48. Distortion is relatively low. and it is spread evenly throughout the audible spectrum.

And, finally, the phono preamplifier. There are two sets of phono preamp RCA input jacks on the rear panel, one set for MM (Moving Magnet) cartridge inputs, and a second set for MC (Moving Coil) cartridge inputs. Using whichever set you need, connect your turntable cartridge outputs to the appropriate set, then, pressing the menu button to the left side of the front panel display, you hold it down until the menu items appear, then click the button several times until you reach “Audio”. At that point, you have about 1 second to click the menu button twice, which brings up a set of sub-menu items. Click the menu button until you see MC or MM, whichever one happens to be currently selected. Then turn the volume control until you get the one you want, either MM or MC. Then press the Input 3 button on the remote control, which will take you back to the main display, showing “Phono” as the input. Turn the volume control down put on an LP, and adjust the volume up as needed.

Shown below in Figure 24 are the frequency responses for MM and MC inputs. MM has less gain because MM cartridges generally have higher output than MC cartridges. The volume control setting will also be different. You can see that the MC input has less gain in the lower frequencies than the MM input.

The Number of FFTs Affects the Shape of the Sine Wave Spectral Peaks

A bit about the shape of the sine wave spectral peaks. When a test signal is analyzed, an FFT (Fast Fourier Transform) is used. This is a digital mathematical procedure that deconstructs the analog waveform into its fundamental frequency(ies) and harmonics, whether the harmonics are from musical instruments or harmonic distortion peaks. In the figure below, connecting the output of a laboratory-grade soundcard to its input, a 1 kHz sine wave spectrum is shown, resulting from a 128 FFT analysis. Note that the base of the peak is wider than the top of the peak. This is an FFT artifact due to FFT analysis limitations.

We can increase the number of FFTs to 256 as shown below. Note that the base of the 1 kHz peak is narrower than with the 128 FFT analysis.

Let’s go to 512 FFTs. The base is even more narrow.

And, now, 1 Million FFTs. The base of the 1 kHz peak is really narrow. One can go on and on, even out to 4 million FFTs.

It all depends on what you want to show. The problem with a high number of FFTs is that it takes a lot longer for the analysis to finish. Also, the FFT size alters the resolution of the resulting spectra. The number of spectral lines is half of the FFT size, and the frequency resolution of each spectral line is equal to the sampling rate divided by the FFT size.

I use either 256 or 512 FFTs in my analyses. But, what I wanted to show here is that the wide base of a spectrum peak is not a Device Under Test (DUT) problem. It is just an artifact in the mathematical process of analyzing an audio signal. It can be broadened by noise that modulates (IM) the frequency peak, but it is still an artifact because the modulation peaks are not individually resolved.

For $3,499, the PS Audio Stellar Strata Integrated Amplifier is a very nice package. The Class D power amplifier has its advantages and disadvantages, but it will get the job done. The whole thing is very compact, and as such, looks great on a hi-fi stand. Its power amplifier doesn’t have the watts to blow the walls out, but that is not its purpose.

“Women like good hi-fi equipment too.”