Theta Digital Prometheus monoblock power amplifier

Several seconds after I began listening to it, I knew that Theta Digital's Prometheus monoblock amplifier ($12,000/pair) was different from other amplifiers. The violins and brass were more dynamic, and had more pace. The orchestra sounded more three-dimensional, depicted in relief by a degree of hall ambience I hadn't heard when I played the same recording through my reference solid-state stereo amplifier, a Mark Levinson No.334.

I've experienced other such "revelations" on first hearing a new audio component. Sometimes this first impression is accurate, as with the unique transparency of Day Sequerra's Reference FM tuner. Sometimes it's misleading, such as the strong room lock in the bass response with pedal notes of pipe organs I heard through Escalante Design's Fremont loudspeakers—and later discovered was a serious coloration that troubled higher-frequency sources like voices and piano.

My first impression of the Theta was all the more interesting to me because the Prometheus has a class-D output section. In his review of MBL's Corona C15 class-D monoblock, John Atkinson noted that the "unique selling propositions of a class-D amplifier are that it is efficient, lightweight, and inexpensive." In his review of Anthem Statement's M1 class-D monoblock, Kalman Rubinson wrote that "everyone acknowledges the advantages of the [class-D] technology in terms of efficiency and ecological benevolence, but so far, neither has captured the souls of aficionados who demand cutting-edge sound regardless of the cost, whether in dollars or degrees." Although the MBL Corona C15 ($24,000/pair) won JA's qualified praise, other class-D monoblocks have not fared so well, including the relatively inexpensive Anthem Statement M1 ($6998/pair) and the cost-no-object Mark Levinson No.53 ($50,000/pair).

Theta Digital & the Hypex Ncore NC1200 amplifier
Because class-D amplifiers generate high levels of ultrasonic and RF energy, such designs typically employ a passive low-pass filter between their output devices and output terminals. This filter needs to be tuned to a single load impedance. Speakers that have a different impedance may cause the amplifier to prematurely roll off the top octave—or, much worse, the filter's maximum effect peaks above the audioband.

To address this problem, in 2001, while working at Philips, Dutch engineer Bruno Putzeys developed a class-D output stage as a cheap drop-in replacement for linear amplifiers. "What Bruno did," Jeff Hipps, marketing director of Theta Digital told me, was to find "a way to include [the] load in the filter circuit." As a result, this amplifier design, which Putzeys dubbed Universal class-D (UcD), was "completely load agnostic [because its] frequency response is load invariant." The UcD, with its six-transistor comparator circuit, served as the basis of the more developed Linear Analog Switching Amplifier (LASA) output-circuit modules found in MBL's Corona C15 amplifier.

Later, while working at Hypex, Putzeys developed a second-generation design, the Ncore, in which the six transistors of the UcD's comparator circuit grew to 10. Putzeys named his new amplifier circuit the NC1200 because its maximum power rating is 1200W into 2 ohms. It is said to achieve 93% efficiency at full power, has a 38A output current capability, comfortably handles 98V rail voltages, and employs extensive, microprocessor-controlled error protection. It is faster and handles smaller signals than the UcD controller, and uses what Putzeys calls "higher order control loops" (footnote 1). Interestingly, the loop gain throughout the audioband never drops below 53dB. The NC1200's improved comparator circuit matches the output to the input, and a control loop circuit allows more open-loop gain thus a greater amount of feedback. It was designed, Putzeys wrote, to "have essentially frequency-independent distortion with 32dB loop gain from DC to 20kHz."

Theta Digital combined Putzeys's Ncore NC1200 class-D module with a linear power supply created by Theta's David Reich. Much of the weight of each Prometheus monoblock's 54.5 lbs comes from its 1.44kVA toroidal power transformer. Theta rates the amplifier's harmonic distortion at less than 0.01% at 500W RMS full-band into a 4 ohm load. Hipps was quick to point out that, at <0.001%, the Prometheus's total harmonic distortion plus noise (THD+N) at 1W is lower than it is at full power by a factor of 10. The Prometheus's output impedance is also extremely low, meaning that its damping factor is higher than is usually the case with a conventional class-D design.

The Theta Prometheus is the shape and size of a small desktop tower computer and measures 13.5" high by 9" wide by 19.8" deep. The curved front panel is of aluminum, with a small built-in handle; the side panels have screened cutouts to dissipate heat. Unlike most monoblocks, the front panel is narrow and the case is deep.

The amplifier is manufactured in mirrored pairs: the handle is on the right side of the faceplate of the left-channel unit, and vice versa. However, there's no reason not to reverse the amps' positions in your room.

You press the single button on the front panel to exit Standby mode (in which the Prometheus draws less than 1W from the wall and its Standby LED glows red) and enter operational mode (the LED glows green, then blue). A second LED will turn red if the amplifier exceeds its maximum operating temperature.

On the rear panel is a column of connectors. At the top are single-ended RCA and balanced XLR input jacks, separated by a toggle switch for selecting between them. Below these are two Cardas speaker binding posts, for use in biwiring. There is also a jack for the standby remote trigger. The Prometheus can be turned on remotely with a 5–12VDC signal. Below this, toward the bottom, is the Main Power switch, which connects AC to all circuits. Flipping this on illuminates yellow Standby LED on the front panel. Below that is a fuse bay, and at the bottom is an IEC inlet for the detachable AC power cord.

The Prometheus is specified as outputting 250W into 8 ohms, 500W into 4 ohms, or 850W into 2 ohms.

The two Theta Digital Prometheuses were easily unpacked. I attached balanced interconnects to their inputs, and the spade lugs of my PSC speaker cables to their binding posts, then flipped the Main Power switch on. As always, I used Stereophile's Test CD 2 to ensure that the channel assignments and phase were correct.

Footnote 1: This and subsequent quote: Bruno Putzeys, "Ncore Technology White Paper" (Hypex Electronics: February 2, 2012).
Theta Digital/ATI
1749 Chapin Road
Montebello, CA 90640
(323) 278-0001

remlab's picture

..and I think he is a genius, but when all these companies are using the same basic modules(I think that Jeff Rowland uses it also), aren't they all building pretty much the same basic amp?

dce22's picture

sys2722 has AES17 filters use those to test Class D amps stop using AUX-0025 on AP2.
Theta Digital Prometheus Distortion 0.01 at 15 watts?
Is that a joke? John Fix Your Measurements Chain!
I have tested alot of ICE power amps and Stereophile data always has more than 10x distortion than the real world data.
AP1 and AP2 inputs will not distort from switching residual only the Autorange circut will trigger wrong scale and you use AES17 filter to fix that, and for frequency response you do not use filters of any kind only Distortion Measurments needs AES17.

PS. IM graph is not correct and Freq Response is from the AUX-0025 not the amp.

John Atkinson's picture
Thank you for your comments, especially concerning the auto ranging. I have tried measuring both with and without the AES17 brick wall filters. The frequency response is genuine, BTW, and is not due to the AUX0025.

John Atkinson
Editor, Stereophile

dce22's picture

John i did a prototype with NC1200 modules and got identical graph as Hypex datasheet i dont use AUX0025 at all only AES17 the switching residual will not induce slew rate distortion (unless it has so badly design filter that will not pass FCC regulation), maybe Theta Digital has that bad input stage that adds high frequency distortion, noise and roll off ill try to borrow one to test it.
The distortion nubmers on 4 ohm are terrible for this kind of amp maybe the power supply transformer is too small and the noise performance is crap.
I can upload images but they are the same as the official datasheet. Cheers J.

PS. I need to mention for non technical audiophiles when using AES17 filter distortion data are only valid up to 6 khz because AES17 filter is a 20khz filter design to measure Sigma Delta DAC's, from 6 - 50khz is covered by 19+20kHz Intermod distortion measurments (you can use 19+20khz IMD data as equivalent to 20khz sine wave distorion measurments only instead of looking at the spectrum above 20k the distorion commponents will mirror back inside the audio band and you can use 20khz bandwidth for less noise there is no need to use some new "super hyper oscilloscope" thats the beauty of it) and switching residual is not somekind of wierd noise, it looks like a parabolic shapes up and down string together something like sinewave but with parabolic halfs and is in milivolts the tweeters are up to hundreds of ohms at 400khz cant even fell anything let alone distort so no worries.
When you Measure the amp with AP2 and you start at microvolts the Autorange circut will see mili volts and yell to his buddy DSP hey there is voltage comming get going change the scale! and the low level data is not usefull, there is no distortion no wierd thing going on people need to relax AP recommends AUX passive filter because there is a nutjobs that run Class D without filters and try to measure it that way that amp my frends is broken, that can be a good test for John if it need AUX-0025 That amp is Broken :) and AP1 needs AUX-0025 because it does not have proper AES17 module slot. I forgot to trasmit 400khz you need 750 meters of speaker cable so dont worry about RF offcourse there are horrible designs that wipeout all the radio staions in the area but it is not because off the speakers cable emmits at 400khz its because amp pushes Mhz's into it, amps modules like B&O Ice power and Hypex have non of the problems.

dmusoke's picture

John.. thanks for the detailed review. Its nice to know that Class D technology has matured to the point of being considered in high end audiophile amplifiers. I have one concern in your measurements, but by placing an external 20kHz filter that doesn't exist in real life, you make the measurements of the amplifier look better than they should be. The effect of these high energy ultra-sonics on the distortion numbers is masked. What effect would they have on actual loudspeaker systems? Why not make the measurements with the default 80kHz bandwidth of the AP test system?

John Atkinson's picture
Why not make the measurements with the default 80kHz bandwidth of the AP test system?

First, because with hundreds of millivolts of ultrasonic noise present, any THD+Noise measurement will be dominated by that noise, even with the 80kHz measurement bandwidth. Second, that ultrasonic noise will drive the AP's input stage into slew-rate limiting, generating additional distortion.

John Atkinson
Editor, Stereophile

dmusoke's picture

So should we have no concerns for these ultra sonics on our loudspeaker systems or radiated emissions to other electronics in our AV systems? BTW, what is the frequency of this ultrasonic noise, which I presume would be the switching frequency of the amplifier?

John Atkinson's picture
dmusoke wrote:
So should we have no concerns for these ultra sonics on our loudspeaker systems or radiated emissions to other electronics in our AV systems?

Not necessarily. Remember that hooked up to the amplifier's class-D output stage is an antenna - the speaker cable. This, of course, is loaded by the speaker, which presents a low impedance at audio frequencies. But above 100kHz, the speaker may well present a very high impedance, due to the voice-coil inductance of the tweeter. So depending on the length of the cable and the ultrasonic impedance of the speaker, the residual switching noise may well be re-radiated into the environment, as well as finding its way into the system ground, which will depend on the amplifier design and circuit layout. All these factors are unpredictable.

dmusoke wrote:
BTW, what is the frequency of this ultrasonic noise, which I presume would be the switching frequency of the amplifier?

It tends to be between 400kHz and 500kHz.

John Atkinson
Editor, Stereophile

dmusoke's picture

I appreciate the answers John. Thank you :)

IgAK's picture

An interesting article, thank you. A comparison to the similarly priced Merrill amp using the same rather popular N-Core module would have been an interesting implementation comparison.

MerrillAudio's picture

I would be happy to provide a pair of the Merrill Audio VERITAS Monoblocks for the comparison. I have also offered Stereophile the Merrill Audio VERITAS Monoblocks for review.

Perhaps there could be an independent, neutral place where the A/B comparison is done side by side. I understand some have done the side by side comparison and hope they will share their findings.

dcvibe's picture

Would love to get a non biased opinion about these two amps?