In this blog post I would like to compare two woofers and provide both objective and subjective test data. I would also like to extend my testing to include a proposed test method to support the intermodulation distortion test.
Woofers Under Test
For this test I am using the Eton Orchestra 12-612 12" woofer against the Dayton Audio RSS315HFA-8 12" woofer. A side by side comparison of the specifications is shown below.
Why compare these woofers?
If you are designing a speaker that requires the low frequency driver to cover 30Hz-250Hz you are left with a myriad of choices. How do you decide? Should you go for a light, fast, paper cone woofer for clean mid-bass? Or should you go for a subwoofer that digs deep and hope that it sounds good in the mid-bass as well? For this test I selected two drivers that represent opposite ends of this spectrum in the hopes that we can determine the winner through objective test data, and hopefully that data correlates with our listening impressions.
Also worth noting is that the Dayton has the following features:
- 2-layer coil for reduced back EMF
- Triple shorting-ring motor for ultra-low distortion
The Eton does not have these features which are intended to lower distortion in the motor. Will this show up in my results?
My testing is also in the context of home hifi (audiophile) listening. Why does this matter? The ultimate goal in this application is for sound quality at normal listening levels. My testing is configured to reflect these goals and nothing else. So for this test I decided to look at the driver performance at what would typically be the normal SPL at the listening chair. I decided on 100dB SPL at 1 meter since this relates to 90db at the listening chair with about 3.5dB of headroom for dynamic peaks in the music. This also assumes a 3 meter listening distance from the two speakers.
100dB SPL at 1 meter requires about 9 watts of power for the Eton (Sinewave Signal at 100Hz). This means that that at 40Hz the woofer diaphragm is moving about 2mm one way. This is about 25% of the overall xmax available. So we are well within the linear operating range of the driver in terms of power handling.
Objective Test Data
In this test I will measure the following:
- Frequency response
- Step response
- Burst decay
- Harmonic distortion
- Intermodulation distortion
Frequency Response
The frequency response is shown below as a side by side comparison between the two drivers. Each driver is mounted in a 50 liter sealed enclosure with non-parallel walls, and 100% polyfill. My measurement mic was placed at near field with the driver dust cap. I mention the non-parallel walls since the Eton is a paper cone woofer, which can be more susceptible to allowing internal resonances inside the enclosure to creep through the thin paper cone of the driver, which would effect my results. The Dayton has an aluminum cone and dust cap with a relatively thick rubber foam surround, so it would be less susceptible to acoustical reflections within the cabinet from escaping through the diaphragm.
The amplifier is the Hypex FA122 class D unit. No EQ was applied to the drivers, so this is the raw response.
The Eton shows a F3 of 56Hz and a F10 of 37Hz. The Dayton shows a F3 of 39Hz and a F10 of 25Hz.
Step Response
Below is a side by side comparison of the step response between the two drivers. The step response decay is half that of the Dayton.
Burst Decay
Below is a side by side comparison of the burst decay between the two drivers. Both drivers show similar burst decay.
Harmonic Distortion
For this test I raised the SPL to 95dB at 1 meter and conducted a harmonic distortion sweep. At first glance both drivers appear to have very low distortion figures, with the Eton creeping above 0.50% below 40Hz. Is there nothing to conclude here?
Intermodulation Distortion
For this test I changed the operating mode in ARTA from 'impulse response' to 'spectrum analyzer'. For my test signal I raised the level to 100dB SPL and ran a series of dual sine wave continuous signals.
Choosing specific test frequencies for F1 & F2
I was not able to find a standardized test method that spelled out what dual frequencies should be used for this type of test. So I devised my own!
I deviced my own formula for determining F1 & F2. I figured the goal should be to ensure that F1 and F2 were not harmonically related. Even though the sub-harmonics would not be related, I felt that if F1 & F2 were not harmonically related than it would provide an even more aggressive and revealing test.
To ensure the two frequencies used in the test were not harmonically related I decided to simply multiply the first frequency (F1) by the golden ratio (1.618). The equation is F1 x 1.618 = F2. This method is my own creation and is not a standard method, but allows for a robust check across the frequency spectrum.
The dual frequencies of F1 and F2 then became the nine tests shown below:
For each of the nine tests I plugged F1 and F2 into the Signal Generator Configuration within the ARTA software. The resulting real-time spectrum analyzer display is shown below. The two peaks are the dual frequencies of 137Hz and 222Hz. The IMD % is shown circled in red at 0.74%.
I conducted the nine IMD tests for both the Eton and Dayton and displayed the results in chart form below.
The data shows that the Dayton clearly performs better if looking at Intermodulation Distortion especially in the lower bass and even in the mid-bass.
The real time spectrum analyzer shows these non-harmonically related side bands as higher in level on the Eton (red lines below)...
Subjective Listening Impressions
I struggled with this part of the test simply because there was no way to compare the two drivers in a fair manner. If I attempted to match the frequency response between the two drivers I would end up with new variables such as increased group delay on the driver receiving additional EQ. Additionally, applying EQ boost also dramatically increased distortion on the Eton at SPL levels above 90dB@1m. So for my evaluation I decided to play a variety of tracks with cello or electric bass guitar. I wanted to hear the difference in the way the driver follows the amplitude modulation of the strings, and the ability to follow the pitch of each note. I felt like this would be a good way to evaluate both the bass and mid-bass portion of the frequency spectrum in terms of intermodulation distortion. The rationale being that spurious side tones generated by IM distortion could likely raise the noise floor, and thereby harming a perceived sense of dynamics. The spurious side tones could also harm ones ability to follow the pitch of each note.
I found that the Dayton did indeed sound better on bass guitar specifically when it came to amplitude modulation. The silence between amplitude peaks was lower resulting in a somewhat more dynamic contrast. I was also able to follow the pitch of each note better with the Dayton, although I could tell the Dayton simply sounded stronger, which could have been due to the different frequency responses between the two drivers. So take my subjective test worth a grain of salt, however in the end all things considered I preferred the overall sound of the Dayton. There was considerable heft which really created a foundation for the music. I do not believe the Eton could have matched the performance of Dayton simply by applying EQ. The resulting increase in group delay and distortion would have further harmed the sound quality rather than improve it.
Conclusion
There seems to be a correlation that suggests low intermodulation distortion along with a flat frequency response results in good subjective bass and mid-bass sound quality. The higher moving mass (mms), low mechanical compliance (QMS), and low sensitivity of the Dayton did not seem to result in a 'slow' bass or mid-bass as some would speculate or expect.
This blog post was not merely a comparison between two drivers, but also a study on how to acquire test data that's relevant to sound quality. My struggles were confirmed afterwards when reading Philip Newell's book on 'Recording Studio Design 4th Edition'...
"Intermodulation distortion in a non-linear system is therefore frequency-dependent, level-dependent, waveform-dependent … in fact, it is very difficult to devise any simple test signal that could yield a realistic description of how the intermodulation performance of two systems could be compared."
To burden my point also from Voishvillo, Alexander, ‘Assessment of Loudspeaker Large Signal Performance – Comparison of Different Testing Methods and Signals’
"Since the dynamic reaction of a complex non-linear system such as a loudspeaker cannot be extrapolated from its reaction to simple testing signals, such as a sweeping tone, the thresholds expressed in terms of loudspeaker reaction to those signals (total harmonic distortion [THD], [individual] harmonics, and two-tone intermodulation distortion) may not be valid."
I do not want to measure IM distortion simply because I can. I will continue to conduct this test in an attempt to find a subjective correlation. If I'm convinced over time then I will use this metric to drive future designs.