Larsen HiFi 8 loudspeaker Measurements

Sidebar 3: Measurements

I was intrigued by the Larsen 8 loudspeaker. More than 40 years ago, I had auditioned a pair of the late Stig Carlsson's unique Sonab speakers and come away with mixed feelings: While Carlsson's original designs, with their arrays of tweeters pointing in all directions, seemed to me innovative attempts to design an omnidirectional mono speaker, I felt they just didn't work when used in pairs for stereo.

But that was a long time ago, so I unpacked from their wooden crates the Larsen 8s that Art Dudley had shipped to me and set to work measuring sample 1505157. I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the frequency response in the farfield, and an Earthworks QTC-40 for the nearfield responses.

The Larsen 8's voltage sensitivity is specified as 88dB/2.83V/m. However, my estimate was significantly lower, at 84.5dB(B)/2.83V/m. Over most of the audioband the Larsen is an easy load for the partnering amplifier to drive, with an impedance that remains above 7.33 ohms from 10Hz to 6kHz (fig.1, solid trace). However, the impedance drops sharply above that range to reach a minimum value of 3.4 ohms at 13kHz, and there is a current-hungry combination of 5 ohms and a –42° electrical phase angle at 8kHz. These suggest that a 4 ohm–rated amplifier will work best with this speaker.


Fig.1 Larsen 8, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).

The impedance traces are free from the small discontinuities that would suggest the presence of panel resonances, and when I investigated the enclosure's vibrational behavior with a plastic-tape accelerometer, I found very little untoward. Fig.2 was taken with the accelerometer fastened to the center of the front baffle. Some low-level modes can be seen between 250 and 500Hz, and these were present at a higher level on the sidewalls, but I don't think they will color the Larsen's sound.


Fig.2 Larsen 8, cumulative spectral-decay plot calculated from output of accelerometer fastened to center of front baffle (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz).

With its array of angled drive-units—the baffle on which are mounted the midrange unit and the front-firing tweeter is angled at 40° from the horizontal, 50° from the vertical—I had a hard time deciding on which axis to measure the Larsen 8's farfield behavior, or where to place the microphone. The manual was not much help, saying that "the drivers should be facing inward towards the listening position. The direct sound from the tweeters should 'hit' slightly in-front and above your seated listening position for optimal soundstage." This is a difficult problem in solid geometry to resolve. In the end, in order to achieve repeatable measurements, I placed the microphone on the tweeter axis, which is 34" from the floor, with the front of the speaker's enclosure facing straight ahead.

The black trace in fig.3 shows the response on this axis of the tweeter section (front-firing tweeter and the two upward-firing tweeters). It is relatively flat, other than being disturbed above 8kHz by interference among the three drive-units, and the tweeter crosses over with a steep filter slope to the midrange unit (green trace) at 2.3kHz. The midrange driver rolls off rapidly above this frequency, but its lower-frequency output is marred by a narrow suckout centered just above 1kHz and a large peak between 400 and 900Hz. The side-firing woofer (blue trace) duplicates the midrange unit's output in the upper bass, and each has a minimum-motion notch in its output at 23Hz, which suggests both that the port does load the midrange unit as well as the woofer and that the port has a very low tuning frequency. The port's output does peak between 20 and 40Hz, but is too low in level to fully extend the Larsen's output to its tuning frequency without the benefit of low-frequency "room gain."


Fig.3 Larsen 8, acoustic crossover on tweeter axis at 50", corrected for microphone response, with nearfield responses of midrange unit (green), woofer (blue), and port (red), plotted in the ratios of the square roots of their radiating areas below 350, 800, and 500Hz, respectively.

Fig.4 shows the Larsen 8's farfield response, averaged across a 30° horizontal window on the axis of the front-firing tweeter. Above the region marred by the large peak—which I'm sure is the culprit behind AD's finding voices to be affected by a cupped-hands coloration—and the deep suckout, which would have affected my sensitivity measurement, the treble is relatively flat. However, it does roll off rapidly in the top octave, which is probably why AD felt that the high frequencies sounded "soft." The low frequencies extend at full level in this graph to 50Hz or so. However, remember that the measurement of the speaker's output in this region is taken in the nearfield, which, with a speaker having a flat low-frequency response, will produce an apparent boost in the mid-upper bass. The Larsen's bass lacked such a boost, which suggests that its low frequencies will be shelved down to an extent dependent on the proximity of the room's boundaries. I note that AD felt that 40Hz was the practical limit of low bass in his room.


Fig.4 Larsen 8, anechoic response on tweeter axis at 50", averaged across 30° horizontal window and corrected for microphone response, with complex sum of nearfield midrange, woofer, and port responses plotted below 300Hz.

Because the behavior in the horizontal plane of a speaker with laterally disposed drive-units will be complex, I have not followed my usual practice of showing a speaker's dispersion, in which I subtract the on-axis response from each of the off-axis responses in order to emphasize the differences between them; instead, I have shown the actual responses. You can see in fig.5 that the peak and suckout in the upper midrange are maintained at all off-axis angles, and that while the Larsen 8's dispersion in the treble is complex, the overall pattern is relatively even. You can also see that there is very little top-octave energy present off axis, again supporting AD's finding that the Larsen's treble sounded "soft." However, as you move higher, there is a little more high-frequency energy present (fig.6).


Fig.5 Larsen 8, lateral response family at 50", from back to front: responses 90–5° off axis on tweeter side, reference response, responses 5–90° off axis on midrange side.


Fig.6 Larsen 8, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 15–5° above axis, reference response, differences in response 5–10° below axis.

I wondered whether the problems with the loudspeaker's quasi-anechoic behavior would be ameliorated in-room. I therefore measured the Larsens' spatially averaged response in a rectangular vertical grid centered on the position of my ears in my listening seat, which is 36" from the floor. I placed the speakers as recommended in the manual, close to the nearest sidewall, and pointing straight ahead so that the front-facing tweeters were firing over my head. You can see from the resultant graph (fig.7) that the Larsen's bass, its lower midrange, and the entire treble region are in very good balance with one another. Compared with the quasi-anechoic response in fig.4, the upward-firing tweeters add more top-octave energy. However, that upper-midrange peak is still very much in evidence and I found it quite audible.


Fig.7 Larsen 8, spatially averaged, 1/6-octave response in JA's listening room.

In the time domain, the Larsen 8's step response on the tweeter axis (fig.8) shows the output of its two upward-firing tweeters preceding those of its front-firing tweeter (the sharp up/down spike just before the 4ms mark) and midrange unit. All of the drive-units appear to be connected in positive acoustic polarity, but the midrange's output is marred by what looks like a strong reflection half a millisecond later. This may well be the root of the speaker's problematic frequency response. To create the cumulative spectral-decay plot for the Larsen 8 (fig.9), I subtracted the output of the top-firing tweeters, as they obscured what was happening. Other than that problem in the upper midrange and something untoward in the high treble, the Larsen's decay is relatively clean.


Fig.8 Larsen 8, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).


Fig.9 Larsen 8, cumulative spectral-decay plot on tweeter axis at 50" with upward-firing tweeters gated out (0.15ms risetime).

AD wrote: "once I'd sat down and acclimated myself to the Larsens' style of sound, the music itself was extremely satisfying. I never felt at a loss"; and "When I dropped the Larsen 8 loudspeakers into my system and set about enjoying them as a normal person might, I approached that ideal: I was immensely happy." In my considered opinion, however, the Larsen 8's measured performance reveals its audio engineering to be flawed. Like the legendary curate's egg, this is a speaker that is "good in parts," I feel.—John Atkinson

Larsen HiFi
US distributor: Audio Skies
4602 Greenwood Place
Los Angeles, CA 90027
(310) 975-7099

NeilS's picture

Do Stereophile reviewers have access to John Atkinson's measurements when they write their reviews?

John Atkinson's picture
NeilS wrote:
Do Stereophile reviewers have access to John Atkinson's measurements when they write their reviews?

No. It is very important for them not to see the measurements, because there would be the danger that they might then hear what they see.

John Atkinson
Editor, Stereophile

NeilS's picture

Much appreciated.

Jason Victor Serinus's picture

There is a very interesting Manufacturer's Comment on this review's measurements in the current issue, along with a reality check by JA.

corrective_unconscious's picture

Speakers which are theoretically bizarre are almost always interesting, and this one appears to have you listening to every single driver off every single driver's axis, (although it's hard to see where the one that is _partially obscured by a metal plate in its head_ is aiming.)

I was once pleasantly surprised by a little, costly FJ "The Ohm" in a short audition, but at least that one had the tweeter aiming forward. Not sure how many of the drivers in the multi way Shahinians aim at the listener.


Everybody remember the Yamaha speaker shaped like an ear?

Hi-Reality's picture

I listened to the Larsen 4s (the little brother of Larsen 8) at T.H.E. Show Newport and was very impressed. The acoustic guitarist on the playback track was in-the-room and playing.

Independent of the gear price, I didn't quite succeed to experience this level of realism at the show in any other room!

Thank you for this review.

Regards, Babak
Hi-Reality Project