Theta DS Pro Generation III digital processor Measurements

Sidebar 3: Measurements

Theta processors have always put out a very high signal voltage, and the Generation III is no exception. From the unbalanced outputs, the unit produced a whopping 6.98V when decoding a 1kHz, full-scale sinewave. This is 10.8dB higher than the standard 2V output. I measured a huge 14.03V from the balanced outputs. These are extraordinarily high levels. According to Mike Moffat, this high output voltage results in better sound.

Potential purchasers are cautioned, however, about using the Gen.III with preamps with unusually low input-overload voltages. Additionally, preamps with lots of line-stage gain—more than about 20dB—are to be avoided; even a tiny opening of the volume control will result in high sound-pressure levels from the loudspeakers. Another factor to consider: preamplifier volume controls typically have their worst channel balance at the low end of their range. With such a high output level from the Gen.III, the volume control will most likely be used in the 8:00 to 10:00 position—where the worst channel-tracking area lies. Bottom line: Be cautious about what preamp is used with the Gen.III. Finally, it is essential that levels be matched between components when performing side-by-side auditioning. The Gen.III's huge output level will give it a decided advantage if the levels aren't carefully matched.

One benefit of a high output level, however, is that it makes the unit ideal for driving a power amplifier through a passive control unit. Additionally, the Gen.III has a very low output impedance—2.7 ohms across the band from the unbalanced outputs (balanced output impedance was 5.3 ohms at 20Hz, rising to 6.8 ohms at 20kHz). The combination of high output voltage and extremely low output impedance make the Gen.III a good choice for use with a passive control. I had excellent results driving the Mark Levinson No.23.5 power amplifier through the Electronic Visionary Systems Balanced Ultimate Attenuators, even through 20' of interconnect (AudioQuest Lapis).

Unless otherwise noted, the following measurements were taken at the balanced outputs.

Frequency response, shown in fig.1, was extremely flat. In fact, the treble rolloff—seen in all digital processors—is very slight (less than 0.1dB) and unusually shaped. It starts to roll off sooner, but looks as though it maintains its passband amplitude beyond 20kHz. This is no doubt due to the Gen.III's custom digital filter. As would be expected from a processor that performs de-emphasis in the digital domain, the Gen.III's de-emphasis error (shown in fig.1 along with frequency response) was virtually nonexistent at audible frequencies. It did, however, have a 0.4dB negative error at 20kHz.

1092Theta3fig01

Fig.1 Theta Generation III, CD frequency response at –12dBFS into 100k ohms, with de-emphasis (bottom) and without (top). (Right channel dashed, 0.5dB/vertical div.)

Interchannel separation (fig.2) was excellent, measuring nearly 130dB at 1kHz, decreasing only slightly to 118dB at 20kHz. In the bass, the crosstalk is below the lowest horizontal division (–130dB) on our standard plotting. This is outstanding channel separation.

1092Theta3fig02

Fig.2 Theta Generation III, channel separation (10dB/vertical div.).

Fig.3 is a 1/3-octave spectral analysis of the Gen.III's output when decoding a dithered –90dB, 1kHz sinewave. The plot is exceptional: low noise level, absence of power-supply–related junk, and good linearity. The tiny spike at the power-line frequency (60Hz) is very low. A spectral analysis of the Gen.III's output when decoding digital silence (all data words are zero) is shown in fig.4. The curve's shape is unusual: besides a very low noise level, the curve has a sway through the upper midrange and treble. Most processors exhibit a rising noise level with frequency, sometimes with converter artifacts visible as spikes at specific frequencies. The peaks at 60Hz and 180Hz are probably related to the power-line frequency.

1092Theta3fig03

Fig.3 Theta Generation III, 1/3-octave spectrum of dithered 1kHz tone at –90dBFS, with noise and spuriae, 16-bit data (right channel dashed.)

1092Theta3fig04

Fig.4 Theta Generation III, 1/3-octave spectrum of silent track, with noise and spuriae, 16-bit data (right channel dashed.)

The Gen.III's excellent low-level linearity is confirmed by fig.5, the unit's departure from linearity. The DACs are virtually perfect to below –100dB, superb performance by any measure. This is among the best linearity I've measured, regardless of conversion technology. Note that because the Burr-Brown PCM63 DAC requires no MSB trimmer, the Gen.III will maintain this excellent performance over time and temperature variations. Fig.6 shows the Gen.III's reproduction of a 1kHz, –90dB, undithered sinewave. The three code transitions at this level (+1, 0, –1) are clearly evident. There is a remarkable absence of audioband noise, and the quantization step sizes are uniform.

1092Theta3fig05

Fig.5 Theta Generation III, departure from linearity (right channel dashed, 2dB/vertical div.)

1092Theta3fig06

Fig.6 Theta Generation III, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit data.

The noise modulation plot (fig.7) revealed a very low level of noise-floor modulation with signal level. The plots (five traces, from –60dBFS to –100dBFS) are tightly grouped, and there is an overall low noise level. This is the second-best performance I've measured on this test. The best was the Mark Levinson No.30, but by a slim margin.

1092Theta3fig07

Fig.7 Theta Generation III, noise modulation, –60 to –100dBFS (5dB/vertical div.)

The Gen.III's squarewave response (fig.8) was unusual in that it looked like that of most other frequency-domain optimized digital filters, with leading-edge overshoot and "Gibbs Phenomenon" ringing. The overshoot is slightly less than I've seen from the NPC filter chip, however.

1092Theta3fig08

Fig.8 Theta Generation III, 1kHz squarewave at 0dBFS.

Driving the Gen.III with the code representing a combination of 19kHz and 20kHz at full scale and performing an FFT on the output produced the plot of fig.9. The sidebands around the test tones are fairly high (just below –80dB), and the 1kHz difference product reaches the –80dB level. This is not atypical, but higher than I've seen from some other processors.

1092Theta3fig09

Fig.9 Theta Generation III, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).

The Gen.III had no problem locking to any of the three sampling frequencies, but the output waveform was distorted when the unit decoded a 32kHz or 48kHz digital signal. Driving the Gen.III with the code representing a 1kHz sinewave with a 32kHz sampling frequency produced a 1kHz wave at the output, with some 8kHz junk superimposed on it. Using the Audio Precision, I captured the waveform shown in fig.10. The printout, showing much less detail than the 'scope screen, doesn't reveal the extent of the spuriae. When driven by a 48kHz signal from a DAT machine, however, I heard no distortion.

1092Theta3fig10

Fig.10 Theta Generation III, 1kHz sinewave at 0dBFS, 32kHz sample rate.

I measured no DC at either the unbalanced or balanced output jacks. The unit doesn't invert absolute polarity from either the single-ended or balanced outputs.

Overall, the Generation III performed very well on the bench. Its low-level linearity, crosstalk, and noise modulation were exemplary. I was disturbed, however, by the fact that it produced a distorted output when driven by sampling frequencies other than 44.1kHz, even though there didn't appear to be audible distortion.—Robert Harley

COMMENTS
Mars2k's picture

Interesting product of course I agree about differing DAC qualities..huge. (Just for fun compare to Lumin please) Would like to add the following point about DAC investment.

The digital reproduction space is where the movement has been in audio and video. Think of it...

Hi Res catalogs are growing not just in selection but also quality and resolution as well as format.

Delivery options are growing Tidal, Qubuz, and more download sources. Compression/decompression protocols MQA and don't forget Music server...Roon Plex, Serviio take you pick.

Think of DAC evolution. Processing power,sample rates and word length are now exponentially greater just in the last 5 years. DAC selection is exploding and price is moving in a direction that puts very hi quality reproduction in more hands than ever before ....case in point Theta gen3 with all the bells and whistle $5500 . Compare and price a specific vinyl rig turntable, MC cartridge, phono stage that competes and beats.

Consider all that and and the fact that I would have to get up from my sofa walk to my system and turn the record over or change the disk from a finite selection of those albums that are physically present to keep listening vs swiping and tapping on my iPad to access what is essentially an infinite library at my finger tips.

X