Measurements of an AR 5 and AR 302 woofer

[Guest richd]

This post shows 4 figures that may be of interest to readers of the AR newsgroup.I have performed close microphone measurements of the woofer of an AR 5 and then an AR 302 I got my pair of the AR 5's Robert Kuhn who restored them and then sold them to me. The were purchased for a comparison with the AR 302 in a review I wrote for Sensible Sound. The woofers of the AR 5 were re-foamed and level controls replaced. The woofers of both AR 5s measure the same. The AR 302s are new stock. The curves show only relative data. No SPL can be inferred from the curves.

Figure 1 shows the AR 5 woofer. Note that I measured the speaker with the microphone as close as could get it inside the speaker and 6 inches away from the speaker. The 6 inch measurement shows that microphone distance can effect the accuracy of the measurement significantly but only in the upper range of the speaker. The peak in the curve and low end rolloff is the same when the microphone position is changed. The 6 inch curve does show the effect of room noise as well as a small effect of the room itself. Moving the microphone back further increases these effect without changing the overall shape of the curves significantly. I would welcome any input from group members with regard to the measurement changes with microphone distance.

Figure 2 is the same as figure 1 for the AR 302. Note that the AR 302 does not have a biamp speaker connection thus the whole speaker is active. The midrange leaks into the measurements from about 400Hz up.

Figure 3 compares the performance of the AR 5 and AR 302 with the microphone 6 inches away from the speakers. In an earlier post I pointed out the more complex crossover topology of the AR 302/303 should not have an effect on the frequency response of the speaker inband. In comparison to the data supplied by AR the Q of my sample of the AR 5 looks high. Is this the change in materials properties that would occur in a 40 year old speaker or is it the result of the reforming? I would welcome input from other group members. Please note that a single Q number to describe the shape of a frequency response curve is applicable only to second order systems. An acoustic suspension speakers when modeled with the simplest speaker model is a second order system. Ported speakers are higher order system and the single Q parameter should not be used to describe the shape of the curve.

Figure 4 is the same as figure 3 except the microphone is as close as possible to the speaker. It is proved since the noise level of the measurement is lower.

Note that the Q of the AR 302 appears about the same as the Q of the AR 5 extrapolated from AR measurements. Also note the extension of the bass extension of the AR 302 relative to the AR 5. I believe the driver is different from the original AR 5. The box size and filling material as different. The speaker is specified as 1dB less efficient than the AR 5. The AR 302 is specified as 85dB at 1 meter with 2.83 volts at the input (1 watt into 8 ohms). The AR 5 was specified as 86dB at 1 meter with an input of 1 watt. In 1995 one could purchase twice the power at half the price so a trade between bass limit and efficiency made sense.

The Stereophile measurements of the AR 303 are often sighted to indicate that the speaker had a higher Q than the AR 3, In this post we have demonstrated that microphone placement can significantly effect the measurement results. Measurements using near field and anechoic chamber may yield significantly different results. It should come as no surprise that something as easy as placing a microphone close to the cone of a loudspeaker is not going to provide that kind of free lunch. It is unclear if near field measurements are more or less accurate in acoustic suspension speakers or ported designs where the summation of the port amplitude and drive amplitude provides a degree of freedom in producing the final frequency response curve.

Exactly what can be inferred from these measurement with respect to differences between the AR 3 and the AR 303 I do not know but it is clear that Ken Kantor made no attempt to increase the Q of the AR 302 relative to the AR 5 and he may have decreased it based on the measurements shown here. Other measurement of the AR 302 and AR 5 will appear shortly.

David Rich

[Pete B]

>The 6 inch curve does show the effect of room noise as well as

>a small effect of the room itself. Moving the microphone back

>further increases these effect without changing the overall

>shape of the curves significantly. I would welcome any input

>from group members with regard to the measurement changes with

>microphone distance.

The near field measurement method was published by Don Keele in the AES and he graciously gives credit to Raymond J. Newman of Electro-Voice "for first making the observation that nearfield measurements correlated well with anechoic measurements.":

"Low-Frequency Loudspeaker Assessment by Nearfield Sound-Pressure Measurement" D. B. Keele JR.

He covers the theory and gives experimental data. In fact his figure 10 a. anechoic, shows baffle diffraction loss as compared to the nearfield measurement b. in support of my theory for why there is peaking in the nearfield response of baffle step compensated systems. His figure 10 also shows good agreement with half space far field measurements as I also stated.

>Figure 3 compares the performance of the AR 5 and AR 302 with

>the microphone 6 inches away from the speakers. In an earlier

>post I pointed out the more complex crossover topology of the

>AR 302/303 should not have an effect on the frequency response

>of the speaker inband. In comparison to the data supplied by

>AR the Q of my sample of the AR 5 looks high. Is this the

>change in materials properties that would occur in a 40 year

>old speaker or is it the result of the reforming? I would

>welcome input from other group members.

>

>The Stereophile measurements of the AR 303 are often sighted

>to indicate that the speaker had a higher Q than the AR 3, In

>this post we have demonstrated that microphone placement can

>significantly effect the measurement results. Measurements

>using near field and anechoic chamber may yield significantly

>different results. It should come as no surprise that

>something as easy as placing a microphone close to the cone of

>a loudspeaker is not going to provide that kind of free lunch.

>It is unclear if near field measurements are more or less

>accurate in acoustic suspension speakers or ported designs

>where the summation of the port amplitude and drive amplitude

>provides a degree of freedom in producing the final frequency

>response curve.

>David Rich

>

Don Keele's work on near field measurment was so significant that he won an AES award for the work. It does have excellent agreement but one must understand what is being measured, volume velocity without any baffle effects which must be applied to the data if a full space measurement is needed. It is very useful to determine if a system is working in agreement with T&S simulation predictions which also do not take baffle effects into account. One must also keep in mind the limitations of this method, it is not useful or intended for full audio range response testing, it is a low frequency measurement only. Good to about 400 Hz for 12" woofers and 800 Hz for 6" drivers and the microphone must be very close, typically less than 1". Many on the internet who've probably never read the paper post measurements indicating that they "might" be useful above 1 kHz, this is uninformed wishful thinking and misleading.

Pete B.