Microphone Distortion Higher than Speakers? Yes? No? Maybe?

[MakeMineVinyl]

Show us the distortion plot just for grins.

Here you go. This is of the Bruel & Kjaer microphone. Both this mic and the UMK-1 had essentially identical distortion with differences too small to bother to post. The distortion was overwhelmingly 2nd order.

 

[Sancus]

Really? That’s not been my experience. There’s virtually no difference between indoor and outdoor assuming both environments are quiet and the mic is relatively close to the driver.

Erin posted about this on his YouTube channel recently... though I wish it wasn't a YouTube post, the differences are not always large, but they can be, for example:

So, look at Speaker C, for example. At 80Hz the in-room THD is approximately -20dB from the fundamental. Or ~10% distortion. But with the anechoic data - mitigating the room - you see the THD is closer to -30dB; 3% THD. So, you have a difference of about 7% higher distortion with the in-room.
Now, let's look at Speaker B at 40Hz. In-room THD is -25dB (~6% THD). Anechoic THD is -38dB (~1% THD).

There are some examples of significant differences from 500hz-3khz as well. Above that, it does seem like there's little to no difference. But distortion above 1khz is rarely relevant anyhow.

If you're measuring everything in the same conditions though then the measurements are more likely to be comparable, for sure.

 

[Joseph Crowe]

Erin posted about this on his YouTube channel recently...

Can you kindly share the video link? I’m having trouble finding it. Thank you.

 

[amirm]

If you perform the speaker distortion measurements near-field and have a large room (as I do), the effect is only pronounced at very low frequencies. But it is so small that I don't bother eliminating the reflections anymore for THD results. Room gain would screw with that response anyway.

 

[Joseph Crowe]

Are there any microphones that anybody would suggest? What about the Shure SM58? From my live sound mixing I know this mic to sound excellent, and the larger diaphragm area may have lower distortion.

 

[DonH56]

Are there any microphones that anybody would suggest? What about the Shure SM58? From my live sound mixing I know this mic to sound excellent, and the larger diaphragm area may have lower distortion.

The Shure M58 is a vocal mic with limited frequency response (50 Hz - 15 kHz), a large upper-midrange peak, and relatively high self-noise. It is rated to 94 dB SPL so not a great choice for low-distortion speaker measurements.

The UMIK-1 is decent as others have said. It actually has lower noise and distortion than most of its competitors in the ~$100 USD or so range. Moving the mic a little further from the speaker will help significantly though you have to watch getting the room response into the measurement (treatment and gating can help). At one point in the primordial past I set up OC-703 panels (sometimes doubled or tripled up, 4"~6" thick) in a "V wing" so I could place the mic about 1 m away to take measurements without (or reduced) room interaction.

I use an Earthworks M30, a much more expensive mic (~$700). Its specs are 3 Hz to 30 kHz, 140 dB max SPL, and 20 dBA self noise. Compared to my CSL-calibrated UMIK-1, frequency response was essentially identical, M30 noise was lower and dynamic range greater (higher overload), and I did not characterize distortion (was not a focus at the time). For a commercial mic with relatively flat response and reasonably low distortion try a Shure SM81 (~$400 USD, 20-20 kHz, 136 dB max SPL, 16 dBA self noise).

 

[AnalogSteph]

I noticed the same kind of issue on my Monacor measurement mic (pronounced H2, and a ton of noise, too), and ultimately resorted to a large-diaphragm condenser, the Thomann SC 400, which happens to have a fairly flat on-axis response - definitely not intended as a measurement mic but workable if you keep a certain distance (to keep proximity effect at bay, it's a cardioid after all) and aim the mic precisely. At least since it's a more directional mic, it can also be placed further away than an omni would. And with plenty of space inside the mic body, there is space for a slightly more than bare minimum amplifier circuit with actually decent level handling, something not necessarily a given for pencil condensers (the all single-ended Oktava MK012 circuit comes to mind, but really it's likely to be quite common given the cramped quarters inside of the things; a "jangling keys" test may be revealing, the IMD was quite audible with my Thomann SC 140 pencil condensers).

Using an SM58 would be sort of like that (its 1" element is right in that territory size wise) except response on dynamics tends to get pretty ragged up top and fall off starting somewhere around 100 Hz on the bottom end (if that). You may be able to calibrate some of that out but getting enough resolution for a precise match may be tough.

The one I would try in terms of "not actually sold as a measurement mic" is a Line Audio OM1. It's not even all that expensive. They've got very limited distribution though.

I don't think any inexpensive mic with a 1/4" electret capsule is going to be a distortion king. The typical 2-pin capsules are generally going to use a JFET in common source configuration, which is what prompted the Linkwitz capsule mod in the first place. (The more electronically inclined may get a kick out of the cascoding approach to JFET linearization as suggested by John Conover. No idea how much use this has seen.) Something using a Primo EM273 10mm capsule would be a lot more interesting, as that exposes both JFET drain and source among its 3 connections. (Perhaps said OM1 is using that as a basis plus some electronics to iron out the response.) I've also seen 3-pin capsules from Sennheiser IIRC, but again I don't expect those to be bargain basement priced.

MEMS microphones have come a fairly long way and may also be worth considering. I've seen some larger ones with both better level handling and lower noise than traditional small electret capsules, even rivalling lesser SDCs. Ever noticed that the microphones on smartypants phones these days are actually pretty good?

 

[Joseph Crowe]

I noticed the same kind of issue on my Monacor measurement mic (pronounced H2, and a ton of noise, too), and ultimately resorted to a large-diaphragm condenser, the Thomann SC 400, which happens to have a fairly flat on-axis response - definitely not intended as a measurement mic but workable if you keep a certain distance (to keep proximity effect at bay, it's a cardioid after all) and aim the mic precisely. At least since it's a more directional mic, it can also be placed further away than an omni would. And with plenty of space inside the mic body, there is space for a slightly more than bare minimum amplifier circuit with actually decent level handling, something not necessarily a given for pencil condensers (the all single-ended Oktava MK012 circuit comes to mind, but really it's likely to be quite common given the cramped quarters inside of the things; a "jangling keys" test may be revealing, the IMD was quite audible with my Thomann SC 140 pencil condensers).

Using an SM58 would be sort of like that (its 1" element is right in that territory size wise) except response on dynamics tends to get pretty ragged up top and fall off starting somewhere around 100 Hz on the bottom end (if that). You may be able to calibrate some of that out but getting enough resolution for a precise match may be tough.

I’ll be trying the SM58 which I recently purchased. I have a Tascam US-1640 USB interface for the mic pre, but it may not be low noise enough and may purchase the Scarlett Solo since the specification seems much better.

 

[Joseph Crowe]

Today I tested the Shure SM58 against the UMM6. As expected like dozens of other times, the UMM6 never gets better than -65dB. The SM58 blew it away with -80dB at 1.5kHz.

 

[DanielT]

I compared my Bruel & Kjaer instrumentation 1/2" microphone (very costly) to the MiniDSP UMK-1. Here are the frequency responses of my left channel speaker at the listening position (11' away) overlaid. Note that I measured using the B&K mic (no cal file) at both 0 degrees and 90 degrees incidence and the UMK-1 was at 90 degrees, using the appropriate cal file. As expected, the 0 degrees measurement with the B&K microphone had a buildup in the high frequencies.

I turned off my subwoofers for these measurements.

View attachment 205028

From another forum. Please note, I am not an expert so what I quote below I do not know if it is true but I thought it seemed interesting.

"The distortion in the microphone output signal is largely dependent on this load capacitance and other scattering (if the translation via google translate is correct) capacitances. This provided that the diaphragm moves like an "ideal drum skin" below the resonant frequency (in the mechanically controlled, "stiffness controlled" range). In addition, the amplitude is very small in relation to the distance to the back plate.

These premises are met very well for "normal HiFi measuring" and e.g. a Bruel & Kjaer 4191 microphone has 0.01% 3rd ton dist at 148dB (SPL) and 0.01% 2nd ton dist at 115dB (SPL).
- 2: ton distortion increases in proportion to the sound pressure
- 3: tons distortion increases with the square of the sound pressure

The "absolute value" of distortion is in practice relative to the movement of the membrane, which results in small membranes distorting less (or rather membranes with lower sensitivity). Therefore, a 1/4-inch microphone has a lower dist, at the same sound pressure, than, for example, a ½-inch mic (type mentioned B & K4191 above which is a 1/2-inch mic).
The 1/4-inch mic B & K4938 has 0.01% dist at 153dB (SPL) for 3rd tone and 118dB (SPL) for 2nd tone.

If, as above, it is clear that eg mic B & K4938 can withstand a maximum of 182dB due to the fact that there is then a risk of the membrane hitting the back plate (peak 20um), it can be stated that the membrane movement is very small at eg 153dB and this is very favorable for low dist values."

faktiskt.io • Visa tråd - Mätutrustning - erfarenhetsutbyte

 

[AnalogSteph]

These premises are met very well for "normal HiFi measuring" and e.g. a Bruel & Kjaer 4191 microphone has 0.01% 3rd ton dist at 148dB (SPL) and 0.01% 2nd ton dist at 115dB (SPL).

These would be very good values that extrapolate to 1% THD at ~155 dB SPL or 0.5% THD at ~149 dB SPL, with H2 being dominant each time (155 dB SPL: H3 = 0.05%, H2 + H3 rms = 1.0012...%). They are basically consistent with the performance of good LDC microphones when accounting for the difference in sensitivity brought about by capsule size - the distortion is virtually entirely on the electronics.

 

[DanielT]

These would be very good values that extrapolate to 1% THD at ~155 dB SPL or 0.5% THD at ~149 dB SPL, with H2 being dominant each time (155 dB SPL: H3 = 0.05%, H2 + H3 rms = 1.0012...%). They are basically consistent with the performance of good LDC microphones when accounting for the difference in sensitivity brought about by capsule size - the distortion is virtually entirely on the electronics.

But .. ~ 155 dB SPL .... 149 dB SPL ....

What? I do not get it. These are very high dB numbers, what is the practical significance of this (for normal home HiFi measurements)?
It's a beginner's question,I admit that, when it comes to measuring microphones, but still. There may be many who wonder about that?

 

[Blumlein 88]

But .. ~ 155 dB SPL .... 149 dB SPL ....

What? I do not get it. These are very high dB numbers, what is the practical significance of this (for normal home HiFi measurements)?
It's a beginner's question,I admit that, when it comes to measuring microphones, but still. There may be many who wonder about that?

Well in use they may be very close to drums or cymbals or someone singing. If a singer hits 90 db way across the room from you think of how loud it is 2 inches from their lips. Much less a few inches from a drum or from a guitar amp's speaker.

 

[DanielT]

Well in use they may be very close to drums or cymbals or someone singing. If a singer hits 90 db way across the room from you think of how loud it is 2 inches from their lips. Much less a few inches from a drum or from a guitar amp's speaker.

Okay, I failed to add. Measurements on voices, musical instruments, measurements that come directly, live I understand that, but if we stick to measurements on speakers?

SPL speaker:

Speaker, highest SPL with a maximum of 1% distortion

Do you have any suggestions for such speakers? You can also guess at some that you think have potential.:) I'm just generally curious about it. I added the criterion max 1% distortion because it can then be considered the loudest speaker that reproduces the signal in a good way (not just...

audiosciencereview.com

Edit:

I take another quote from Jansch that I quoted earlier (because I know that, before retiring, he worked professionally with microphones):

Question (not from me):
"Can you really measure such low levels with computer / REW? When it comes to microphone amplification, will your microphones completely dominate both noise and distortion when measuring?"

Jansch:
"The noise from a "1/4 inch microphone" (ie all standard electret microphones) is misleadingly low in A-weighted measurement. However, if measured "linearly", a noise floor of over 50dB (SPL) is usually more realistic down to 20Hz.
I therefore never use 1/4 inch microphones for frequency sweeps when measuring on speakers as it requires too high a sound volume for it to be healthy for the speaker in question.
For example, a bookshelf speaker and LTAS measurement at a reasonable sound level then show more microphone noise than tone curve at the lowest frequencies.

A 1/2 inch mic (which unfortunately is 20-30 times more expensive) can handle levels down to 20-25dB (LIN) and thus shows the correct tone curve at 20Hz.

As for distortion, I do not know as it is not specified for electret microphones.For "real" capacitor microphones for measuring purposes and with a tensioned diaphragm, the distortion is immeasurable below 100-110 dB (SPL), ie below about 0.01% dist."

faktiskt.io • Visa tråd - Mikrofonförstärkar EMA2201

 

[egellings]

I'm not a mic user and have no experience with them, but my guess is that mic distortion would be much lower (if the mic's not exposed to excessive sound pressure) than that of speakers because mic powers are microscopic, and diaphragm excursions are tiny. It's pretty easy to stay in the linear region. I'm assuming that the mic preamp poses no limits to the mic's performance.

 

[fineMen]

In room distortion measurements are not accurate anyhow.

But in-room distortion counts, right? What is accurate?

My Panasonic WM62 (?) is good for about 0,1% of IM at interesting levels close to the speaker, maybe at 100dB or so.

How do I test it?

Low bass 30Hz from my trusty JBL2206, voice tone at 10kHz from my as well trusty BMS4550 in horn. Must be the mike, right?

Where does it come from?

The mike has a membrane that is literally stretched by excursion, and additionally it has a pre-amplifying FET transistor in it. My mike is as good, because the FET is driven with home-made negative feed-back.

 

[KSTR]

I therefore never use 1/4 inch microphones for frequency sweeps when measuring on speakers as it requires too high a sound volume for it to be healthy for the speaker in question.
For example, a bookshelf speaker and LTAS measurement at a reasonable sound level then show more microphone noise than tone curve at the lowest frequencies.

Well, with long sweeps I never had noise issues when measuring down to 20Hz with my Gefell MK301 1/4" and it's rather noisy preamp. And several measurements can be averaged (in the time-domain with the raw sweep recording) if that isn't enough.
If the convenience of fast measurements is important then of course a larger capsule is the way to go.

 

[DVDdoug]

The SM57/58 isn't a measurement mic and it doesn't come with a calibration curve but if distortion is related to sound level, distortion should be low.

The Shure M58 is a vocal mic with limited frequency response (50 Hz - 15 kHz), a large upper-midrange peak, and relatively high self-noise. It is rated to 94 dB SPL so not a great choice for low-distortion speaker measurements.

94dB (1 Pascal) is the just standard level for sensitivity measurements. Most mics are measured at 94dB. The SM57/58 puts-out 1.6mV at 94dB SPL. A typical studio condenser mic (with its built-in head-amp) puts out about 20mV at 94dB.

And since it's passive it doesn't generate any (electrical) noise, or technically, no more noise than the thermal noise of a resistor. However it's low output (compared to a condenser) means more gain and more preamp noise after the preamp-noise is amplified along with the signal.

The SM57/58 is almost impossible to overload and the maximum it puts-out about line-level so you don't need a preamp! From the Shure Website:

#4 You can turn it up to 11. Maybe even 12.
How much SPL can the SM58 handle? At what point will the sound distort? The answer is much higher than would be safe for your ears: somewhere in the 150 to 180 dB SPL level, close to the noise level of a space shuttle launch. A well-designed dynamic like the SM58 is unlikely to reach its distortion point under normal circumstances.

P.S.
Most (analog) condenser mics are limited by head amp clipping so they usually can't handle the same SPL levels as a dynamic mic. But many have a "pad" switch that cuts-down the signal into the head amp to allow-for higher SPL levels.

 

[fineMen]

Jansch:
"The noise from a "1/4 inch microphone" (ie all standard electret microphones) is misleadingly low in A-weighted measurement. However, if measured "linearly", a noise floor of over 50dB (SPL) is usually more realistic down to 20Hz.

In my environment, as a home speaker DIYer, the ambient noise is the limit. Even at night it never goes below 30..35dBs in the very low frequencies. Once amplified the tone like characteristic is clearly perceivable. All those motorcars on the highway a mile away, and the city. I average multiple measurements to cope with that, -3dB for every doubling of measurement count.