This is a review and detailed measurements of the PS Audio Stellar S300 Stereo Power Amplifier. It is on kind loan from a member. The S300 costs US $1,499 from the company.
Fitting the design language of the rest of the PS Audio and using similar enclosure, the S300 is quite hefty considering that it uses class-D amplification:
There is just a power switch and light that indicates whether the unit is on, or has shut down due to protection circuit kicking in.
The back panel shows very high quality and beefy connectors:
As you see, there are redundant pairs of speaker sockets which I assume is an invitation to use bi-wiring. Or feed the subwoofer using them as Paul McGowan advocates although a warning tells you to be careful as these are floating signals, not ground referenced.
XLR inputs are nicely provided which I used for all of my testing.
The internal amplification is provided by ICEpower 300AS1 modules. These are integrated power supply and single channel amplifier modules.
PS Audio uses its own input buffer on these which they call "Analog Cell." This got the owner worried that it would measure poorly as did the PS Audio Stellar Gain Cell Preamplifier and DAC. I must say it got me worried reading the description of it too:
Seeing how the class-d module uses ample feedback to linearize its response, it is puzzling that he thinks the input stage should be without one. Either feedback is good ro bad. Can't have it both ways.
Anyway, back to the operation of the unit, it ran cool and never got upset in my testing. Overall, my subjective impression of the unit is positive.
Amplifier Audio Measurements
As always, we start with 5 watt output into 4 ohm and see where we stand:
Ah, that is respectable! FFT spectrum just shows two distortion products with the second one at or below -100 dB. There is however noise though which pushes the combination of those distortions and noise to -88 dB resulting in SINAD of same amount, only positive. This puts the S300 amplifier above our median for all power amplifiers tested so far:
Our frequency response is a bit variable at the high end:
I looked up the 300AS1 response as shown in the inset and is consistent with that. While most of us won't hear that rise in response, it does show load dependency which means each speaker will interact with it differently.
All my tests use an AES-17 filter which has a flat response to 40 kHz and beyond. But for the purposes of finding the actual high-bandwidth spectrum of the unit, I take it out for this test of 1 kHz tone at 5 watt:
Switching frequency is around 450 kHz which is fine. Typical of many of these class-d modules, the level is high at just -15 dB below signal.
@DonH56 recently asked me if the level of switching noise tracks the signal itself so I decided to test for that by upping the power by a good bit to 180 watts and see what happens:
As you see in blue, there is good news and bad news. Good news is that the level doesn't scale with output power much. Bad news is that its spectrum becomes quite wide so total energy may have increased that much. Certainly scary to the eye but likely not to the ear.
Anyway, back to our regular programming, here is our signal to noise ratio:
Not that good at 5 watt as we could have predicted from the SINAD in the dashboard. At full power the situation improves but still not very competitive with top of the class switching amplifiers which can clock as high as 125 dB. So be careful with sensitive speakers.
Most important graph for amplifiers is power versus distortion+noise. Here it is into 4 ohm:
We have tons of good power to the tune of 256 watts. Distortion starts to rise early though at just 10 watts.
Here is the response into 8 ohm:
Perhaps more revealing is what happens if we perform the same sweep but at different frequencies starting at 20 kHz and working our way down to 20 Hz. An ideal amplifier would have all the graphs landing on top of each other. But this is what we get here:
At frequencies up to 1 kHz, the graphs indeed nicely land on top of each other (bottom group). But at 5 kHz and higher, we get very odd responses indicating multiple non-linearities taking turn to dominate. I thought this was an issue with PS Audio implementation but the same response is shown by ICEPower in their data sheet.
Peak power shows disappointing level of headroom available for short durations as would be typical in music:
The switching power supplies in class-d amplifiers are tightly regulated which means the power you get in steady-state, is what you get in peak. Standard class AB amplifiers tend to have variable supply rails which if you don't tax much, will maintain higher voltage and produce more power in short bursts. For this reason, you may want to get a more powerful class D amplifier to replace a class AB amp.
Conclusions
It is a sigh of relief that PS Audio had not managed to reduce the performance of the ICEPower modules. Response and sonic abilities are completely determined by those modules unlike what PS Audio marketing says. The ICEPower module used here has a distinctly lower performance than Hypex modules we have tested in the past. So you are not going to get state-of-the-art performance. For that, you need to go to amplifiers higher in the SINAD graph.
Still, this is a competent, well-built and well-supported amplifier based on proven amplification modules within. Build quality is much higher than DIY or garage shop amplifiers using Hypex modules. $1,500 is not a ton of money given these attributes. For these reasons and good looks of the unit, I am going to put the PS Audio Stellar S300 amplifier on my recommended list.
------------
As always, questions, comments, recommendations, etc. are welcome.
US tax season is upon us so I had to tabulate my expenses and donations for last year. Needless to say, the former number was far, far higher than the latter due to expensive equipment purchases and becoming on first name basis with my local post office. Depression has set in as a result. Please consider making me happy by donating using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
Fitting the design language of the rest of the PS Audio and using similar enclosure, the S300 is quite hefty considering that it uses class-D amplification:
There is just a power switch and light that indicates whether the unit is on, or has shut down due to protection circuit kicking in.
The back panel shows very high quality and beefy connectors:
As you see, there are redundant pairs of speaker sockets which I assume is an invitation to use bi-wiring. Or feed the subwoofer using them as Paul McGowan advocates although a warning tells you to be careful as these are floating signals, not ground referenced.
XLR inputs are nicely provided which I used for all of my testing.
The internal amplification is provided by ICEpower 300AS1 modules. These are integrated power supply and single channel amplifier modules.
PS Audio uses its own input buffer on these which they call "Analog Cell." This got the owner worried that it would measure poorly as did the PS Audio Stellar Gain Cell Preamplifier and DAC. I must say it got me worried reading the description of it too:
Seeing how the class-d module uses ample feedback to linearize its response, it is puzzling that he thinks the input stage should be without one. Either feedback is good ro bad. Can't have it both ways.
Anyway, back to the operation of the unit, it ran cool and never got upset in my testing. Overall, my subjective impression of the unit is positive.
Amplifier Audio Measurements
As always, we start with 5 watt output into 4 ohm and see where we stand:
Ah, that is respectable! FFT spectrum just shows two distortion products with the second one at or below -100 dB. There is however noise though which pushes the combination of those distortions and noise to -88 dB resulting in SINAD of same amount, only positive. This puts the S300 amplifier above our median for all power amplifiers tested so far:
Our frequency response is a bit variable at the high end:
I looked up the 300AS1 response as shown in the inset and is consistent with that. While most of us won't hear that rise in response, it does show load dependency which means each speaker will interact with it differently.
All my tests use an AES-17 filter which has a flat response to 40 kHz and beyond. But for the purposes of finding the actual high-bandwidth spectrum of the unit, I take it out for this test of 1 kHz tone at 5 watt:
Switching frequency is around 450 kHz which is fine. Typical of many of these class-d modules, the level is high at just -15 dB below signal.
@DonH56 recently asked me if the level of switching noise tracks the signal itself so I decided to test for that by upping the power by a good bit to 180 watts and see what happens:
As you see in blue, there is good news and bad news. Good news is that the level doesn't scale with output power much. Bad news is that its spectrum becomes quite wide so total energy may have increased that much. Certainly scary to the eye but likely not to the ear.
Anyway, back to our regular programming, here is our signal to noise ratio:
Not that good at 5 watt as we could have predicted from the SINAD in the dashboard. At full power the situation improves but still not very competitive with top of the class switching amplifiers which can clock as high as 125 dB. So be careful with sensitive speakers.
Most important graph for amplifiers is power versus distortion+noise. Here it is into 4 ohm:
We have tons of good power to the tune of 256 watts. Distortion starts to rise early though at just 10 watts.
Here is the response into 8 ohm:
Perhaps more revealing is what happens if we perform the same sweep but at different frequencies starting at 20 kHz and working our way down to 20 Hz. An ideal amplifier would have all the graphs landing on top of each other. But this is what we get here:
At frequencies up to 1 kHz, the graphs indeed nicely land on top of each other (bottom group). But at 5 kHz and higher, we get very odd responses indicating multiple non-linearities taking turn to dominate. I thought this was an issue with PS Audio implementation but the same response is shown by ICEPower in their data sheet.
Peak power shows disappointing level of headroom available for short durations as would be typical in music:
The switching power supplies in class-d amplifiers are tightly regulated which means the power you get in steady-state, is what you get in peak. Standard class AB amplifiers tend to have variable supply rails which if you don't tax much, will maintain higher voltage and produce more power in short bursts. For this reason, you may want to get a more powerful class D amplifier to replace a class AB amp.
Conclusions
It is a sigh of relief that PS Audio had not managed to reduce the performance of the ICEPower modules. Response and sonic abilities are completely determined by those modules unlike what PS Audio marketing says. The ICEPower module used here has a distinctly lower performance than Hypex modules we have tested in the past. So you are not going to get state-of-the-art performance. For that, you need to go to amplifiers higher in the SINAD graph.
Still, this is a competent, well-built and well-supported amplifier based on proven amplification modules within. Build quality is much higher than DIY or garage shop amplifiers using Hypex modules. $1,500 is not a ton of money given these attributes. For these reasons and good looks of the unit, I am going to put the PS Audio Stellar S300 amplifier on my recommended list.
------------
As always, questions, comments, recommendations, etc. are welcome.
US tax season is upon us so I had to tabulate my expenses and donations for last year. Needless to say, the former number was far, far higher than the latter due to expensive equipment purchases and becoming on first name basis with my local post office. Depression has set in as a result. Please consider making me happy by donating using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
