This review page is supported in part by the sponsor whose ad is displayed above
Reviewers: Marja & Henk
Sources: CEC TL5100, Audio Note tube DAC, Philips DVP 5500S SACD/DVD player
Preamp/integrated: TacT RCS 2.0 room control system, Lamm L2 Reference [on loan], modified Audio Note Meishu with AVVT, JJ or KR Audio 300B output tubes; Moscode 401HR [in for review]; Trends Audio TA-10
Speakers: Avantgarde Acoustic Duo Omega; Avantgarde Acoustic Solo in HT 2.0 setting; Audio Note AN/Jsp silver-wired
Cables: Audio Note AN/Vx interconnects; Siltech Paris interconnects; Gizmo silver interconnect; Qunex 75 reference interconnect; Crystal Cable CrystalConnect Reference interconnect, CrystalDigit S/PDIF RCA/RCA and RCA/BNC, Y-cable, Crystal Cable Piccolo iPod to XLR, CrystalPower Reference AC-Eur/IEC’ CrystalSpeak Reference; Audio Note AN-L; Gizmo silver LS cable; Virtual Dynamics Revelation power cords [in for review], Bocchino Morning Glory [in for review], Harmonic Technology Magic Woofer, Magic Tweeter & Pro AC11 [in for review]
Power line conditioning: Omtec PowerControllers
Equipment racks: Two double sets of Solid Tech Radius; Acoustic System amplifier shelf
Sundry accessories: IAR carbon CD damper; Denson demagnetizer CD; Nanotech Nespa #1; TacT RCS calibrated microphone and software; Exact Audio Copy software; Compaq server w/Windows Server 2003 and XP; wood, brass and aluminum cones and pyramids; Xitel surround processor; Manley Skipjack; Boston Audio Design TuneBlocks and CD mat
Room treatment: Acoustic System Resonators; Gizmo's Harley Davidson cap
Review component retail: $6800
Contrary to popular belief, a so-called Class D amplifier is not a digital amplifier. Class D is simply the next available letter after C. It does not stand for digital. When starting with Class A amplifiers, their efficiency to convert incoming DC power to outgoing RF power is very low. A lot -- often 90% -- of the incoming power is merely dissipated as heat since the transistors or tubes used are always on. On the plus side, a Class A amplifier has very low output signal distortion to be more linear, making this class very suitable for high-end audio.
In Class B amplifiers, the ratio of incoming DC power versus outgoing RF power is much better. In a Class B amplifier, the signal is cut in half. Now the positive half of the signal can be fed to a separate transistor or tube, the negative half to another transistor or tube. One 'pushes' while the other 'pulls'. This allows the transistors or tubes to be on only 50% of the time. Though more energy efficient, a Class B amplifier is inherently more cross-over distortion. Each time the signal crosses zero, one transistor or tube switches off while the complementary transistor or tube needs to switch on. Here in this transitional state, a slight timing error can occur that is perceived as signal distortion.
To compromise between both classes, Class A/B operates in Class A at low levels before converting to Class B mode at higher levels. The end result is an amplifier that runs in inefficient but low distortion mode for its first few watts and only during dynamic peaks of the incoming signal goes into Class B overdrive. The exact point where the amplifier shifts into Class B is determined by its bias current. The higher the DC bias current, the longer a circuit runs in Class A - and thus dissipates more heat. Think of an average 20% efficiency for a Class A/B amplifier.
Class C provides an even higher efficiency than Class B at roughly 75%. Where Class A handles the full 360 degrees of the signal with one transistor or tube, Class B handles merely 180 degrees of that same signal per output device. In Class C a transistor or tube handles less than 180 degrees of the incoming signal. Though the output is a very loud signal, it is nothing more than a pulse and far from linear. For audio let alone high-end audio, Class C isn't suitable.
Class D is the next step. Where a Class A circuit can be compared to a faucet that regulates the water -- here the outgoing signal -- in a continuously variable way by gently opening and closing its valve, Class B uses two faucets -- say for warm and cold water -- and is switching between them to get an even temperature at all output levels. Class C in this faucet comparison is more like a finger on a garden hose giving a not so accurate squirt every time you take the finger off. Class D is that same finger on the hose but now highly regulated. The water or output of the transistor is always full on or completely off.
|The trick becomes to make the incoming audio signal steer the switching of the transistor. The continuously variable analog audio signal is treated by a technique called pulse width modulation or PWM. The amplitude of the incoming signal is encoded as pulses representing the ratio of time between the on or off state. The positive phase of a signal is thus represented by a sequence of 'on' states and the negative phase by a stream of 'off' states. The more 'on' states follow each other consecutively, the longer the 'on' state persists and the higher the amplitude of the amplified signal. Ditto for the off sequence.
In a Class D amplifier, the incoming signal is first buffered and from that buffer, the signal is converted into a PWM signal. This signal now switches the output Mosfet on or off. Only Mosfet transistors are capable of switching at the required speeds of more than 100kHz. This switching itself generates a lot of noise that must be filtered out before the signal is passed to the speaker.
It is mostly this output filter which determines the overall sound of a Class D amplifier. A filter too steep may introduce phase shifts. A gentler filter will introduce less phase shift but will pass ultrasonic frequencies with the output signal and in the worse case, turn the amplifier into a radio transmitter. Another problem with the output filter is of course that it is connected to the loudspeakers. The inductance and capacitance of the loudspeaker becomes part of the output filter and since each loudspeaker make and model has its own inductance and capacitance characteristics variable with frequency, it is hard to predict how a Class D/speaker combination will interact and perform. Furthermore, only a few designers are capable of designing and building a Class D amplifier from scratch. Hence many commercial Class D amplifiers rely on OEM building blocks like ICEpower. At most, they may add their own power supply - and that's basically it for invention. In conclusion, good-sounding amplifiers have been and continue to be built in all of these classes and as with most things audio, the secret to success has more to do with implementation than dogma.
Kharma does it a bit differently. When Charles van Oosterum contemplated the various assets and liabilities of Class D topologies, he went to the well and hired Bruno Putzeys. Belgium-born Putzeys worked at Philips' Applied Technology laboratory on digital and analog-controlled Class D amplifiers. Here he designed the UcD circuit that he now exploits in his own company Hypex. The collaboration between Charles and Bruno led to the birth of Kharma's Matrix MP150 Class D monoblocks.
One of the design goals was to develop an amplifier that would offer the same characteristics as a SET tube amplifier but with the added distinction of being a Class D amp. Bruno Putzeys claims to be able to bend the sound of any Class of amplifier, be it A, A/B or D, to his will. Thus the MP150 was made to sound like a tube amp in the mid and high frequencies. Where many early Class D designs were found to sound very different from preexisting amplifiers, this was basically due to the new ways in which they distorted. Later designs fixed the problem and left early admirers and adopters in doubt.
Kharma's website has a couple of graphs for their monos, showing a very flat frequency response, essentially a straight line from 20Hz to just before 20KHz where is gently slopes down. With a 100-watt power output at 8 ohms and 150 at 4, these amplifiers have plenty of power to drive almost any decent loudspeaker. The documentation further states that the design only takes feedback from the speaker terminals for its pulse control and not from the switching outputs. This creates low output impedance to give the MP150 a damping factor of more than 400 up to 2000Hz, maintaining better than 100 at 20kHz. Combined with a rated frequency response of DC to 50kHz (0,-3dB) and a THD and IMD of less than 0.02% make for some rather impressive figures. Jitter problems in Class D topologies are here controlled by eliminating a separate oscillator altogether and using self oscillation based on the phase shift of the feedback network.
Charles hand-delivered our pair of MP150 review loaners. He wanted to watch our reaction as we opened the custom-made briefcase with matching anodized aluminum covers to get at the amplifiers inside. In our case -- go figure [Marja & Henk wear black exclusively - Ed.] -- it was all black. Charles' customization also showed in the amps' temporary serial numbers - M&H 1 and M&H 2. Measuring only 15 x 12.5 x 5cm (5.9 x 4.9 x 2 inches) and milled from a solid aluminum block, the MP150s are very heavy. This is due to a substantial transformer inside each of them as we shall see later. At the front there's only a tiny blue light next to the Matrix logo. More action shows up at the back where a custom Cardas-inspired single cap screw tightens down over two spades (bananas may not apply) once fitted. When you order a pair of MP150s, you may specify either an XLR or RCA input. Below the power switch is the AC receptacle and that's it. The amps come with their own four coned feet.
After Charles had left, we started setting up these two elegantly demure amps. First they would play from the pre-outs of our Audio Note Meishu integrated and drive the Avantgarde Duo Omega hornspeakers. Their 17-ohm, 107dB sensitivity would be a nice looking glass on these Dutch marvels. For cables, we used a complete set of Crystal Cable. The first music played came compliments of one of the best live recordings we know of - Paul Simon's Concert in the Park. On an average system, this album can sound mediocre and shallow. On a system that handles great dynamics and superior details, the recording acts like Jekyll and Hyde and sheds its somatic hide for the cerebral joys of great music. The percussion on the CD is phenomenal and even though -- or precisely because -- the recording is from 1991, compression is completely absent. The MP150s project an outstanding image of this memorable event in the room. Soundwise, the MP150s behave like well-nourished SET on a healthy diet, not on steroids or other chemicals. From the lower midrange to the very top of the highs, the music flows very naturally and the MP150s need very little play time -- a couple of hours are sufficient -- until the top end opens up completely. No signs of harshness are to be found. The amazing speed these amplifiers offer proved a great match for the equally agile horns. Where "real" SET amplifiers then betray their weak spot in the bass, the Class D bricks from Breda excel. Never before did we enjoy such rock-solid bass without any bloominess or blur shy of some very big, heavy and hot-running Class A transistor amplifiers. Yet here this bass combines with a very tubey character higher up in the frequency range.
Alas, our happiness would be rudely wrecked the next day when we want to listen some more. One of the monos switched itself on and off constantly. When we swapped M&H 1 for M&H 2, the problem stuck with the same channel. So we suspected the preamplifier. Swapping out the Meishu for a Lamm L2 Reference which stayed a few days with us, the problem now spread to both power amplifiers. Time to call Charles.
Charles had the answer the very moment we mentioned the issue. The tubed preamps were leaking DC through the interconnect. This can act up all of a sudden or be always present. The standard wiring of the MP150 bypasses a buffer capacitor in order to keep the circuit as simple as possible. However, now this cap was needed to block DC. All we had to do was unscrew the bottom plates and take out two jumpers per amplifier. That remedied the auto reset protection. Ten minutes later, the amps were playing upside down without bottoms for easy access just in case. Gone was the problem but there was something else. The sound has changed into a just slightly rounder version of the sound before the miniature surgery. We closed up the aluminum boxes again and proceeded to listen. Again something had changed. The roundness was gone again. A little gentle bite was reintroduced, acoustic guitars seemed to be played with a harder nail or pic, violins bowed more on the bow's edge. Was it the amplifier's chassis? We removed the bottom plate again. Back was that roundness. In the end, we left a few screws out from the bottom and got what to our ears was the best sound.
Is it the enclosure that has so much influence? It must be. Other Class D amplifiers we heard had some zing in their sound, a metallic edge. Few sounded different in a positive way, most notably the Red Dragon Leviathan. Is it the latter's wooden chassis that exerts such a positive influence? Now that we thought about it, the big brother of the MP150, the MP350, does have a lot of wood covering its electronic innards. An interesting test would be to wrap a Trends Audio Class D miniature amp in wood.
While reversing the cap bypass, we had a good look at the M150's innards. Next to the PCB, there's a large custom-made Tortran toroidal transformer, a fair amount of BlackGate and Kharma capacitors and wherever possible, Kharma's own silver cable with gold inclusions. Playing the MP150s in combination with our highly sensitive hornspeakers proved beyond a doubt how deadly -- quiet this amplification circuit is - quiet as a (dead) mouse. The Lamm L2 Reference's gain structure made for a very limited usable range in this combination but without playing music, we could turn its two attenuators up quite high while remaining noise free. With music, that amount of output was far too high for prolonged listening but the noise floor remained exceptionally low regardless.
In conclusion, the cooperation between Bruno Putzeys' technical excellence and Charles van Oosterum's pushing the audio boundaries birthed a very competitive amplifier. Without any effort, the MP150 will challenge and defeat many far more expensive SET and solid-state power amplifiers. On top of that, the MP150 makes the choice between tube and a transistor amplifier easy. It's neither, really, since Class D in this special Kharma rendition combines the very best features of both into a couple of very musical bricks indeed.