Sidebar 3: Measurements
I used DRA Labs' MLSSA system with a calibrated DPA 4006 microphone to measure the behavior of one of the JBL 4329Ps in the farfield and an Earthworks QTC-40 mike for the nearfield responses. Before I started the testing, I connected the 4329P to my network router with an Ethernet cable and opened its local webpage. The review sample was running firmware version 0100.1685.0x1010f. I performed a factory reset and checked to see if a firmware update was available. It wasn't.
The Roon app's audio settings window indicated that the network-connected speaker had not yet been certified by Roon. However, it did allow the 4329P to be used as an Airplay device, which limits playback to 44.1kHz and a 16-bit word length. Apple's AudioMIDI utility indicated that the loudspeaker's USB input accepted 16- and 24-bit integer data sampled at all rates from 44.1kHz to 192kHz. The USB Prober app identified the USB-connected speaker as "JBL 4329P" from "JBL" and revealed that the USB input operated in the optimal isochronous asynchronous mode.
Sending white noise recorded at –20dBFS to the JBL speaker with Roon, with the volume control set to its maximum, gave an spl of 90dB(B) at 50". With the 4329P's balanced analog input fed MLSSA's single-ended pseudorandom-noise signal at 246mV peak–peak, the speaker's volume control again set to its maximum and its analog input sensitivity set to "–10dB," the speaker produced an spl of 92.7dB(B) at 50". When I repeated the measurement with the analog input sensitivity set to "+4dB," the spl dropped by 14dB, as anticipated. The JBL 4329 offers excellent dynamic range capability.
The JBL speaker's farfield output, averaged across a 30° horizontal window centered on the tweeter axis and taken with the balanced analog input (fig.2, black trace above 300Hz), is impressively even, with a narrow suckout between 900Hz and 1kHz and some inconsequential, narrow peaks at the top of the tweeter's passband. This graph was taken without the skeletal grille, which only covers the woofer and the ports. Repeating the measurement with the grille resulted in almost no difference.
Other than the suckout and delayed energy at the top of the midrange, the JBL 4329P offers excellent measured behavior: The frequency response, horizontal dispersion, and low-frequency alignment are textbook.—John Atkinson
Fig.1 JBL 4329P, cumulative spectral-decay plot calculated from output of accelerometer fastened to center of sidewall (measurement bandwidth, 2kHz).
I investigated the vibrational behavior of the enclosure's side and top panels with a plastic-tape accelerometer and the loudspeaker producing an spl of 100dB(C), slow ballistics, at 1m, measured with the Studio Six app on my iPhone. I found resonant modes at 359Hz and 1004Hz on both surfaces (fig.1). However, the lower-frequency mode is relatively low in level and all the modes have a reasonably high Q (Quality Factor), which make it less likely that these modes will have audible consequences.
Fig.2 JBL 4329P, anechoic response on tweeter axis at 50", averaged across 30° horizontal window and corrected for microphone response, with the nearfield woofer (blue) and port (red) responses and their complex sum (black), respectively plotted below 300Hz, 1kHz, and 300Hz.
With the 4329's low-frequency boundary compensation set to "Flat," the woofer's nearfield response (fig.2, blue trace) has its reflex notch, which is when the back pressure from the port resonance holds the cone stationary, at 30Hz. The output of the twin ports (fig.2, red trace) peaks slightly higher in frequency. Their upper-frequency rolloff is initially clean, though some low-level resonant peaks are present between 500Hz and 900Hz. The high-pass rolloff of both the woofers and ports below the port tuning frequency is much steeper than the usual 12dB/octave; presumably there is a series high-pass filter in the woofer feed to minimize subsonic excursion.
The complex sum of the woofer and port responses with "Flat" boundary compensation is shown as the black trace below 300Hz in fig.2. The boost in the upper bass is an artifact of the nearfield measurement technique. The 4329P's reflex alignment appears to be maximally flat, though the output drops like a rock below 35Hz. Setting the boundary compensation to "–3dB" reduced the levels below 200Hz by 3dB, which will be optimal if the speaker is used close to the wall behind it.
Fig.3 JBL 4329P, lateral response family at 50", normalized to response on tweeter axis, from back to front: differences in response 90–5° off axis, reference response, differences in response 5–90° off axis.
Fig.4 JBL 4329P, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 45–5° above axis, reference response, differences in response 5–45° below axis.
Fig.3, which shows the responses to the speaker's sides normalized to the on-axis response, indicates that the 4329P's horizontal dispersion is superbly well-controlled and even. In the vertical plane (fig.4), a suckout centered at the specified crossover frequency of 1675Hz appears more than 20° above and 5° below the tweeter axis. The JBL should be used on a stand that places the listener's ears on or just above the tweeter axis.
Fig.5 JBL 4329P, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.6 JBL 4329P, cumulative spectral-decay plot on tweeter axis at 50" (0.15ms risetime).
Fig.5 shows the 4329P's step response at 50" on the tweeter axis. (Due to the latency of the A/D, DSP, and D/A circuitry, the signal takes 32.5ms to arrive at the microphone, almost a whole video frame later than the expected 3.75ms.) Fig.5 shows that both drivers are connected in positive acoustic polarity, with the tweeter's output arriving first at the microphone. The decay of the tweeter's step doesn't blend quite smoothly with the start of the woofer's step, and ripples can be seen in the decay of the woofer's step, correlating with delayed energy at the frequency of the suckout in the JBL's cumulative spectral-decay/waterfall plot (fig.6—ignore the apparent low-level ridge of delayed energy just below 16kHz, which is due to interference from the MLSSA host PC's video circuitry). The graph is very clean in the region covered by the tweeter.
