Description. My Red Dragon S500 loaners came in silver but black anodized is an option. The Red Dragon logo is engraved on the front and red light illuminates it when turned on. Inside this smart aluminium chassis resides a Pascal Audio module. This Danish company have been in business since 2006, just shy of 10 years. They have enhanced and patented a new Class D technology called UMAC. Launched in 2013, their Pascal S-PRO2 as used also by Gato Audio is said to be a major leap forward due to its compactness, being the smallest 1’000 watt professional amplifier module to market. Jeff Rowland have been one of Pascal’s earliest customers among high-end manufacturers to release a few ‘entry-level’ products based on that module like the bridgeable 525 stereo power amp and the integrated Continuum S2. Red Dragon’s S500 beats with the same heart in a less high-end chassis for roughly half the price. Jesper Hansen, co-founder of Pascal, explained that "the design of our power supply was inspired by looking at the approach of other industries such as LED lighting. This led to a process of optimization to produce a new audio-specific power supply that was far more efficient and therefore smaller." As is the case with Abletec/Anaview amps, the power conversion/power supply and switching output stage integrate on one and not separate boards like recent Hypex Ncore NC1200/SMPS1200 modules. That makes implementation easier and more streamlined. Pascal-based amps generally require an input buffer because the input impedance of the modules alone is very low.  Hence the input buffer of Red Dragon raises it to over 50KΩ for an easy load. This is an especially comfortable feature for tube preamps with relatively high output impedance. Ryan Tew’s input buffer allows for balanced or single-ended connections thanks to a rear switch. The mono/bridge function is controlled from the input stage along with a push button. The latter intentionally recesses into the chassis to prevent accidental engagement. Single-ended signal is first converted to balanced before being processed by the power module.

Obviously each input buffer will exert some (a lot of?) sonic influence on the final result and I would not presume that different Pascal implementations sound the same or that Jeff Rowland’s small boxes are 1:1 comparable to Red Dragon. Pascal’s architecture calls for little modifications from various OEMs and I assume that most of the price difference between Jeff Rowland’s 525 and Red Dragon’s S500 is driven by the fact that the latter sells direct whilst Jeff Rowland sells through a complete dealer network. Let's focus now on the main protagonist, the Pascal S-Pro2 integrated module. It's useful to highlight the difference of this new high-tech circuit against the usual pulse-width modulated amplifiers. The first PWM amplifiers like B&O’s IcePower were based on triangular generators. Their triangle waves were typically compared to an input signal by a comparator creating a PWM signal on the comparator’s output. This triangle-wave solution has several drawbacks. The modulator didn't use feedback correction, leading to nonlinearities in the derived PWM signal. If a control loop was applied to the amplifier, it was restricted in bandwidth and loop gain to comply with the Nyquist stability criteria. This resulted in significant levels of distortion along with high load dependence of the amp’s closed-loop transfer function. One way to avoid these drawbacks became self modulators with increased bandwidth and loop gain. The modulator and control system in a switching amplifier impact parameters such as distortion, closed-loop bandwidth, step response, load dependence, load step response and output impedance. Self-oscillating local loop modulators still lack the control of the demodulation filter in the power converter. This results in high output impedance and distortion from the output filter’s nonlinearities. Self-oscillating global loop modulators exhibit lower output impedance, better step response and higher linearity.  Their low sensitivity in some cases is considered a kind of trade-off with the modulator's linearity. One way to increase sensitivity is to cascade the global loop as nested loops but this creates further transient stability issues. Pascal tried to overcome the major drawbacks of the modulator and control loop as used in standard PWM architectures.

Jesper Hansen first worked on the design of a modulator with increased linearity to reduce the total amount of self-generated distortion in the power stage. Then his efforts focused on a control and modulator system which would achieve very low error sensitivity in his global loop. This had to be extremely simple and avoid complicated cascade structures. Particular attention was paid to minimize variance in the switching frequency so that the power converter would see a higher switching frequency at high modulation indexes for more efficient use of the power supply rail voltage and lower interdependence between switching frequency and load. Finally, Pascal wanted to reduce the power seen by the resistor in their Zobel network to reduce its size. Their new control system is called hybrid control and based on the measurement of two different feedback signals. A voltage or current feedback is measured at the power stage input, one voltage feedback signal at the amplifier's output. These signals are summed to obtain linear modulation along with very high loop gain and low output impedance.


Modulation behaviour is defined by loop instability in the forward path including a regulator, comparator, hysteresis loop, power stage, demodulation filter and one each nested and global feedback loop. Total loop gain in the pass band is mainly determined by the global loop whilst modulator linearity is achieved through high-frequency summation of the local and global loops. This control system is said to achieve very low distortion, low output impedance, linear frequency response along with a low noise floor and a high power supply rejection ratio. A hysteresis loop controls the switching frequency along with the pole and zero placement in the regulator and in the local and global loop transfer functions. The hysteresis loop is applied to reduce the switching frequency for acceptable efficiency and to obtain linear modulation of the power stage output signal. The modulator and control system provide for a variable switching frequency with higher efficiency along with spread-spectrum HF noise. The switching frequency variation is limited to create an amplifier capable of running at very large modulation indexes utilizing the power supply rail very efficiently.

Due to high stability of the modulator and control system, the Zobel network simplifies drastically for both higher efficiency and smaller parts. Due to high global loop gain around the filter’s cut-off, small signal bandwidth can be extended much further and reduce residual ripple on the output. In this implementation, the amp is claimed to approach an ideal voltage source. This leads to some impressive measurements for the Pascal S-PRO2. They include 26dB of voltage gain, >120sB S/N (full bandwidth, A-weighted, 8Ω); <0.05% THD (full bandwidth, 3dB below full rated power, 8Ω); +0/-0.25dB response from 20Hz – 20kHz (8Ω, 3dB below full rated power); <1kHz damping factor of >1000 into an 8Ω load; and low 0.35W stand-by draw to comply with the Green Energy Star and ErP 1275/2008/EC specs. Out of the box, the S-PRO2 module is optimized for 4Ω loads in standard mode and 8Ω in bridged mode. But the latter can still be set to ‘4Ω BTL Mode’ for better compatibility with low-impedance speakers.

Sound. Bridged-Tied Load or BTL mode was undoubtedly a big step forward when used with demanding loads like my Vivid G1. Using two bridged mono units instead of one stereo amp created more efficient control and a wider deeper soundstage. It didn't at all change the sonic fingerprint of the Red Dragon S500 and timbres remained very accurate. This quality of timbre in fact was a very robust feature and reminded me of the surprising triode-like sound of my small Tripath-based Trends Audio integrated amp with perhaps more honesty and transparency. Like most class D devices, these were very silent in operation and didn’t seem to have any compatibility issue with tube preamplifiers with their friendly input impedance.

They also relied on—and this becomes really important if your speakers are not easy loads—the quality of their associated power cords. With the stock cords, they produced an accurate detailed sound lacking in pace, rhythm and timing. The sound felt highly accurate but too thin. This obvious flatness made them a bit boring in the long run. To unveil excellent PRAT, one sadly must spend extra on more ambitious power cords. That was the complete opposite of my Luxman amps which have never yet sounded better than with their stock cords. With big snakes like Esprit’s Lumina, the two dragons became very dynamic and involving. Of course it seems completely crazy to add two power cords which cost more than a pair of these amps. I assume that it was primarily a matter of avoiding small conductor diameters. I had satisfying results also with more affordable cords but none as brilliant as the big Esprit Lumina.