In this blog post I would like to look at the effect amplifier choice has on a speaker's measured intermodulation distortion levels. My test involves measuring the acoustical performance of the loudspeaker using a measurement microphone. My test only involves changing the amplifier and matching the output level at the loudspeaker with the preamp gain. I used a sound level meter to match the output to within 0.1dB between tests.
It has long been understood that speakers have orders of magnitude higher distortion than modern amplifiers. Amplifier manufacturers quote harmonic distortion levels as low as 0.0001%. However consider that these figures are specified using a single test tone at 1kHZ (harmonic only). How do these amplifiers perform when it comes to multitone testing at 10kHz?
To highlight my point, below is the frequency/distortion characteristics for the Texas Instruments TPA3116D2 Class D Chip.
As we can see THD+N(%) rises to 0.30% at 6kHz. An intermodulation distortion test which uses a multitone test signal may highlight even more problems, however these are rarely published.
In my previous blog post which can be found here, I reviewed the SEAS FA122RCZ driver. This is one of the first driver tests where I've started to look at Intermodulation Distortion. From this test I found that the IMD was very low in the upper treble.
Below is the IMD sweep for the SEAS FA122RCZ at 90dB SPL@1m.
IM Distortion at 10.4kHz is very low at only 0.10%. If I physically moved the measurement mic to only 10cm of the driver cone these numbers dropped even further.
Below is a dual tone IMD test which shows just how low the IMD is on this driver. Distortion is only at 0.07% IMD. A side band at 6435Hz is -65dB down from the fundamental.
Switching to a class D amplifier that uses the aforementioned TPA3116 chip produces the the following IMD distortion level from the SEAS FA122RCZ driver. I was very careful in my testing to match the output SPL to 90dB at 1m. There was perhaps only about 0.10dB variation from test to test. As you can see IMD distortion rose from .072% to 0.12% simply by switching amplifiers.
I then decided to try my Emotiva UPA-700 Solid State Class A/B amplifier. IMD lowered down to 0.063%. A side band at 3978Hz is -58.4Hz from the fundamental.
I then decided to try my Meixing Mingda MC368-B902 KT88 Single Ended Triode Tube Amplifier. I should note that I've heavily modified this amplifier from stock form. It produced the lowest IMD of all the amplifiers at only 0.042% The most dominant side band is at 3978Hz at -68.2dB down from the fundamental.
Conclusion
My testing suggests that the amplifier is at least having a measurable effect on the speaker's performance when it comes to IMD. How much of an effect could possibly be confirmed by measuring the electrical output of the amplifier itself.
Are these distortion levels perceptible? I've had these amplifiers for years and the test results perfectly correlate to my own subjective preference for amplifier choice.
The treble quality produced with 0.04% IMD is world class and could legitimately be a viable performance standard for clarity. This could be one step closer to providing a guiding light in the hifi world dominated by subjective opinion.