Paradigm Millenia HT Speaker System

Viewing page 6 of 8 pages. Previous 1 2 3 4 5 6 7 8 Next


SPCR is a couple of major steps ahead of almost all other PC hardware review sites in acoustics metrology. We have...

  • our own home-built, hemi-anechoic chamber, an environment that is extremely quiet and almostly completely without reverberation above ~150 Hz.
  • a lab-reference, calibrated, ultra-low noise microphone with ruler-straight frequency response that cost over $2,000
  • a sophisticated PC-based sound measurement system

SPCR's audio measurement system can be employed to provide basic loudspeaker measurements such as on/off axis frequency response, maximum SPL, and total harmonic distortion. The audio measurement / spectrum analyzer system consists of...

For testing loudspeakers, a signal generator is needed to drive the speakers. As the speakers have built-in amplifiers, this was provided with software via the integrated sound card of a second PC, a silent PC with no moving parts, inside the anechoic chamber.


1. SPL: The sound pressure level at which measurements are done is extremely important. A common procedure is to provide the sensitivity with 1W input, and also test the frequency response at the same power input. For a typical passive speaker (one that does not have a dedicated amplifier built into it), this might be something like 90 dB/W, which means when driven with 1W input at, say, 1 kHz or with white noise, the speaker output measures 90 dB SPL one meter away. In fact, 90 [email protected] is a fairly common level for frequency response measurements.

Two considerations:

What is the right baseline SPL? I have already tested some other speakers in the chamber using 85 dB SPL at 1m as a reference, so it makes sense to continued using this level. It's 5 dB lower than the usual 90 dB used for hi-fi speakers. 90 dB is much louder than you might think: Typical SPL scales suggest that 90 dB is about what you hear from a diesel truck 10m away, inside a moving subway train, or from a food processor directly in front of you.

What is a realistic volume for actual use? A check of SPL levels was done at the listening position ~10' from the speakers in the large audio-only room playing a variety of music at various volumes, and movies and other video material in the small home theatre room, with the seated position 6' from the speakers. The results are summarized below.

SPL @ 1m, Typical Use w/ Paradigm MilleniaOne/Sub speakers
Music only system Pop music, background 75~80 dB
Pop music, "loud" 85~95 dB
Other music, background 70~80 dB
Other music, "loud" 80~95 dB
HT/TV Drama 70~85 dB
Action 75~95 dB
Documentary 75~85 dB
*Other Music: Jazz, Classical, Folk, etc.

The measured SPL in the large room was naturally higher because the speakers have to play at a higher amplitudes to reach the same subjective volume in the bigger space, seated farther away from the speakers. Still, it was rare for the SPL to exceed 95 [email protected] in any application. Even 90 [email protected] was not that common in the big room, reached mostly during brief peaks. The SPL @1m in the TV room averaged about 5 dB lower, reflecting both the smaller space and the closer distance between speakers and listener. This could be simply a reflection of my listening habits, but spot checks with other family members, friends and visitors confirmed that this is fairly typical: It is not likely that many people actually listen to music in their living room or watch videos on their TV with higher SPL than indicated above. There are always exceptions; remember, I am just trying to establish a reasonable baseline.

2. Frequency Response: This is the single most widely cited specification in audio, especially with mechanical devices like loudspeakers, which traditionally have the greatest deviations from flat frequency response. It is best shown in a frequency vs sound pressure level graph. In the simplest terms, frequency response tells the ability of an audio device to reproduce sounds of different frequencies at the correct relative levels (loudness). (Here is a good primer on the topic.) A perfect device has a frequency response that looks like a ruler straight line; hence the term "flat" (not flat as in B-minor flat.) Alas, there are many complex issues around this much-cited parameter.

It is highly dependent on the acoustics of the room, the position of the speaker(s) in the room, and the position of the microphone. If this test is performed in a live room, then sophisticated calculations must be used to remove the effect of room reflections (echoes). Otherwise, it must be performed in an anechoic chamber.

Frequency response of a loudspeaker generally does not stay constant with loudness level. Typically, there is a range of SPL in which a speaker is most frequency-linear; go outside this range, especially above it, and the speaker will exhibit frequency non-linearities that lower fidelity.

Frequency response also changes with the angle of perception, both vertically and horizontally. How smoothly the frequency response changes as one moves off axis is a key to better sound loudspeakers.

Given the complexities, dozens of frequency response graphs could be plotted and posted... but their usefulness would be questionable for most readers. So... this is the procedure established for frequency response testing:

  • Place the speaker at the front edge of the 28.5" (72cm) tall table in anechoic chamber.
  • Place the microphone 1m directly in front, at the same height.
  • Set the output level to 85 [email protected] SPL using white noise.
  • Capture the frequency response graph at 1m distance, on axis, and at 30 degrees laterally off axis
  • Treat one satellite + subwoofer as a single speaker, with bass unit directly under the satellite.

NOTE: Only a single speaker (or single speaker + center woofer) is tested, as there are many complications that arise when trying to measure a stereo pair together at the same time.

3. Harmonic Distortion: This is a relatively easy parameter to measure. A pure sine wave tone is fed into the speaker, and the spectrum analyzer sums up all the aspects of the signal that are not this pure tone, expressed in a percentage of the total signal. Harmonic distortion is not particularly important, however, as it occurs naturally in music and is thus difficult to perceive. Even 10% THD at 50 Hz is probably not audible to most people if it occurs in the context of music — and I am not talking here only about a heavy fuzz bass guitar. Amplifiers and other electronics are capable of miniscule levels of harmonic distortion (say 0.01%), but it is much higher in mechanical devices like speakers, especially when large cone excursions are involved (necessary for high volume at low frequencies). Again, the SPL at which HD is measured has a serious impact on the result, as does the frequency. Simply put, the louder and lower the test tone, the harder a speaker has to work to reproduce it. Longer cone excursions almost always result in greater signal anomalies.

After much experimentation, this is the procedure established to test for harmonic distortion:

  • Place the speaker at the front edge of the 28.5" (72cm) tall table in anechoic chamber.
  • Place the microphone 1m directly in front, at the same height.
  • Set the output level to 85 [email protected] SPL using white noise.
  • Leave the gain unchanged while running test tones at the following frequencies: 10kHz, 5kHz, 2.5kHz, 1kHz, 500Hz, 250Hz, 100Hz — and lower, to the lowest frequencies where distortion does not exceed 20%.
  • Tests were kept as short as possible: It is easy to damage speaker drivers with steady state pure tones, even at low power levels!


There is no shortage of tests that can be run on speaker systems. These include impulse testing, intermodulation distortion, resonance/decay testing of the enclosures, phase response, etc. With time and effort, all of these tests can probably be brought to bear on SPCR speaker reviews. The real question is, for what benefit? Once the basics of frequency response, dynamic volume capability and distortion are covered, in-use details and good subjective descriptions are more important for a buying decision. Many other technical parameters are simply too complex for even geek-tech readers to correlate with the listening experience. Finally, keep in mind that speaker measurements are not infallible, nor do they tell the whole story. Like careful listening with the right ancillary equipment, acoustic measurements are just tools by which the performance of a speaker system can be more fully understood.

Previous 1 2 3 4 5 6 7 8 Next

Audio|Video|Misc - Article Index
Help support this site, buy from one of our affiliate retailers!