In this blog post I compare distortion performance between four different microphones.
From left to right I have the following microphones.
I've used the UMM-6 for about a decade now and it has served me well. It is the only USB microphone in this test. The microphone retails for $137.00 USD
The EMM-6 is similar to the UMM-6 however it uses an XLR connection instead of USB. Both the EMM-6 and UMM-6 use an 6mm electret condensor which include a built-in preamp. I purchased this microphone for this test to see how it performs. It retails for $88.00 USD.
The SM58 is a vocal microphone that is well regarded for it's sound quality. So I've included it in this test. The SM58 is a dynamic microphone which does not require phantom (external) power. It retails for around $100 USD.
I purchased this microphone after a recommendation from a reader that has been using ACO Pacific professionally for over a decade. I was hesitant to purchase this microphone because of it's high cost and zero return policy. At $1,200 USD it is not cheap, however I decided to take a leap of faith on this one based on extensive reading. The 7052PH uses a 1/2" titanium diaphragm 7052E Electret capsule and 4048 ACOtron Preamp Body. The 4048 provides <14 dBA noise floor, the lowest in the industry.
Interpreting Test Results
Below is the test result for the Dayton UMM6. To interpret the results I compare the peak of the test signal against the black "grass" in the noise floor. The grass represents the intermodulation side band products. Normally this would be generated by the loudspeaker, however as we shall see, it is actually the microphone. The specific distortion performance for the UMM6 is -60dB (10kHz). Converting this from dB to percent is equal to 0.10%.
Below are the test results for the EMM-6. There are some unusual side band products across the spectrum from this mic. The 2kHz region shows performance in the -50dB region. The 10kHz region improves to -70dB but quickly deteriorates to only -50dB at 20kHz.
Below are the test result for the Shure SM58. As you can clearly see, things improve considerably over the previous two mics. Distortion performance across the frequency spectrum is consistently -80dB. This converts to only 0.01% which is the distortion performance for many amplifiers. An example of this can be found here and here.
Below are the test results for the 7052PH. Distortion is the best yet, at -85dB at 10kHz. This converts to 0.0056% distortion.
Conclusion on the above test results
It is clear that the each microphone has a unique performance profile. It is unclear at this point what bottleneck exists to improve beyond the -85dB threshold. However it is clear that it may not be the loudspeaker.
The following is a little more discourse on the topic and relevance of my test.
Over the course of the past year I have realized that my measurement microphone is a bottleneck in terms of distortion. In other words, the microphone that I've been using has significantly higher distortion than the loudspeakers I'm trying to measure. For those of you that don't know, I develop loudspeakers for the home audio industry and one of the metrics for sound quality that I use is the multitone intermodulation distortion test. This test closely correlates to subjective sound quality.
Phillip Newel put it best in his book "Loudspeakers for music reproduction and recording"...
"Over 50 years later, intermodulation distortion is still a significant problem, and we still have no simple way to measure it in an easily interpretable way which intuitively relates to all its audible implications".--
The 12 band per octave multitone test signal used in my testing is the most effective test I've been able to find, although it is a little difficult to interpret.
Intermodulation Distortion (IMD) is a type of distortion measurement where the test signal is comprised of multiple tones. The affect of multiple tones stresses the device under test (DUT) much more than a simple harmonic distortion sweep which is comprised of a single test tone that is swept across the frequency band. A multitone signal can be comprised of two or more test tone signals. I've taken the test even further by using 12 test tones per octave. This has the affect of really fleshing out weaknesses in a device. It should be noted that a multitone signal such as this closely mimics the signal found in music.
IMD is more audible than harmonic distortion since the side band products (grass) are not harmonically related to the test signal tones in the same was as a harmonic test.
More study is needed to assess the audibility of IMD. However there are a few things that should be pointed out.
The digital music signal that we listen to has musical information well below -85dB. If we cannot ascertain the audibility of IMD, we can at least be confident that more musical information is at least present in the digital domain.
During my testing, I need to ensure that all sources of environmental noise has been eliminated. This includes the low level noise from refrigerators, computer fans and the like. Even when the test signal is playing, I can still hear broad spectrum noise such as a computer fan, even though it barely registers in my test results. So the audibility of IMD is not a stretch for me to consider. This correlates with my own experiences where I need to have absolute silence to fully evaluate loudspeaker sound quality with music. If there are environmental noises, then I cannot know the true potential sound quality of a loudspeaker. I also find background noise extremely irritating when trying to enjoy music. I am fortunate to live in a rural setting where I can create a quiet environment. Most sources of noise for me are the children, the dogs, and the wind outside since we live on a hill that is not sheltered by trees.
The mic is connected to my Focusrite Scarlett Solo preamp. The source is the Hypex FA501 supplied by the Topping D10S which is connected to my laptop running ARTA measurement software. I inserted a -10dB fixed resistor L-Pad to eliminate amplifier noise at idle. The amplifier gain knob on the Hypex was set to -10dB with windows volume set to maximum. The loudspeaker used for the test was a single Dayton Audio AMT4-PRO Tweeter. The mic was placed about 5cm from the driver. The test signal SPL was set to 85dB at 1m using my handheld decibel meter. For each of the four mic tests, I simply replaced the microphone while keeping the test signal unchanged. The AMT tweeter had an analogue 2nd order high pass filter centered around 2kHz. There was no DSP filtering used on the Hypex.
Below are the software settings and signal generator setup for ARTA.
I am relieved that the ACO 7052PH performs as well as it does and I did not waste my money. The microphone has a flat response from <3 Hz to > 25kHz typ.(capsule) and so I will continue to use this mic for future testing. I will continue to investigate this topic including how to improve my test setup to fully flesh out loudspeaker performance in terms of IMD. I am convinced that loudspeaker distortion is on par with that of upstream components (Amplifiers, DACS, Streamers) since it is very common to hear differences in these devices through loudspeakers. The entire goal is to develop a measurement system that is one step closer to our own listening ability. I hope the positive implications of this are obvious to most readers.