Understanding microphones

[Chamber005]

So I've been trying to get more into the technicals regarding sound, (Hz, Khz, dbs, etc). I've never had any real problems recording sound (keeping things around 0 dbs and always below 6dbs so as to eliminate distortion), but I've never really understood much about microphones other than their recording pattern.

Here's the schematics from a 25 dollar cardioid mic (CAD CM217):

Sensitivity: -53dB +/- -4dB @ 1KHz (0dB = 1V/u bar)

Output Impedance: 200 ohms +/- 20%

Frequency Response: 30Hz - 18KHz

Signal to Noise Ratio: More than 60dB

Here's the schematics from a 700 dollar cardioid mic (NTG-3):

Sensitivity: -30dB +/- 2dB (31.6mV/Pa)

Output impedance: 25

Frequency response: 40Hz - 20KHz

Signal to noise ratio: 81dB

The first thing that jumps out at me is the "Output impedance". 25 vs 200 (not sure what the 20% is about) which as I understand it relates to the voltage? I'm just hazarding a guess that the CAD would use a lower voltage than the NTG (so then 200 might be better than 25), but it looks like a lot of crazy math to me, so any elaboration would be great.

The frequency responses are pretty close, and while the signal to noise ratio is definitley greater on the NTG (again, not sure if that's a bad or a good, but I assume that it's worse to have a "lower" signal to noise ratio as the CAD loses this one as well), their numbers aren't too terribly far off from one another.

There's a -23 difference between the front sensitivity numbers. Does that mean that the CAD is more sensitive?

These two mics are completely different (though both can be used to record voice), which is kind of why I wanted to compare their specs -- so that I can hopefully understand what makes one mic's specs increase the mic's value over another.

 

[Alex H.]

First, remember that specs on paper can be misleading, and you should trust your ears more than the sheet. And for fairness in comparison, these aren't two cardioids. One is a cardioid. One is a super cardioid but is a line+gradient shotgun, which modifies the polar response. The difference in polar patterns is going to impact how the mic performs in specific situations.

That said, there are some telling things here.

First is the difference in sensitivity. A higher number (we're in negatives here, so -30 is higher than -53) means that the mic has a hotter signal (or is more efficient at converting the acoustic SPL to voltage). The NTG-3 is a hotter mic, meaning it willnot need as much of a boost in the pre-amp stage.

Second is impedance. Both of these mics are "low impedance," which is 600 Ohms or less. This is how much resistance the signal meets as it travels down the line. High impedance mics meet a lot of resistance, are usually over unbalanced connections, and lose signal over longer runs. Low impedance doesn't meet much resistance, is typically over balanced lines, and can hold up to longer runs. The lower the impedance, the more robust the signal.

With frequency response, the range is only one part of the crucial information. Human hearing, generally speaking, goes from 20Hz to 20kHz. So the wider the frequency response, generally, the better. And the NTG-3 has a wider range. But you also need to look at a frequency response chart to see how colored that mic's response is. Are there major peaks and dips? Is it all over the place? Is it fairly flat? A mic that has a response of 40-18k but has a really wavering response across the spectrum may not sound as good as a mic that has 50-17k but is relatively flat.

As for signal-to-noise ratio, the higher the number the better (assume it is followed by ":1" which means XdB of signal to 1 dB of noise). So the NTG-3 is significantly quieter than the CAD.

 

[JordanBlock]

Here's a good book on mic design from Neumann that might interest you.

 

[Chamber005]

First, remember that specs on paper can be misleading, and you should trust your ears more than the sheet. And for fairness in comparison, these aren't two cardioids. One is a cardioid. One is a super cardioid but is a line+gradient shotgun, which modifies the polar response. The difference in polar patterns is going to impact how the mic performs in specific situations.

That said, there are some telling things here.

First is the difference in sensitivity. A higher number (we're in negatives here, so -30 is higher than -53) means that the mic has a hotter signal (or is more efficient at converting the acoustic SPL to voltage). The NTG-3 is a hotter mic, meaning it willnot need as much of a boost in the pre-amp stage.

Second is impedance. Both of these mics are "low impedance," which is 600 Ohms or less. This is how much resistance the signal meets as it travels down the line. High impedance mics meet a lot of resistance, are usually over unbalanced connections, and lose signal over longer runs. Low impedance doesn't meet much resistance, is typically over balanced lines, and can hold up to longer runs. The lower the impedance, the more robust the signal.

With frequency response, the range is only one part of the crucial information. Human hearing, generally speaking, goes from 20Hz to 20kHz. So the wider the frequency response, generally, the better. And the NTG-3 has a wider range. But you also need to look at a frequency response chart to see how colored that mic's response is. Are there major peaks and dips? Is it all over the place? Is it fairly flat? A mic that has a response of 40-18k but has a really wavering response across the spectrum may not sound as good as a mic that has 50-17k but is relatively flat.

As for signal-to-noise ratio, the higher the number the better (assume it is followed by ":1" which means XdB of signal to 1 dB of noise). So the NTG-3 is significantly quieter than the CAD.

Gotchya. And the CAD's diagram is almost completely flat in terms of response (which is probably why people suggested using it as a uber-cheap mic for recording voice).

So, basically, using a mic with lower sensitivity (negatives) means more "noise" in relation to its preamp stage, right? Is that regardless of gain? I wonder if this also comes into play regarding the phantom power, as I know that sometimes using a recorder's phantom power can add noise to the recording...?

Are lav mics also cardioid and super cardioid? Lavs are usually what I end up renting (Countryman B3s) instead of shotguns (primarily because no one in Chicago rents a decent shotgun mic and I do all of my shooting indoors).

Obviously I own a couple of these CAD mics. It would be nice if there was some way to get some use out of them. Any recording made with them would be noiser than a recording made with a mic of higher sensitivity, though, right? Even for ambient (or foley) stuff?

 

[Steve House]

...
Are lav mics also cardioid and super cardioid? Lavs are usually what I end up renting (Countryman B3s) instead of shotguns (primarily because no one in Chicago rents a decent shotgun mic and I do all of my shooting indoors).

...?

Yes, there are cardioid and a few hypercardioid lavs made but they are very rarely used for film/video. Normal movement of the talent's head will take them on and off mic when a directional lav is used, resulting in a very strange sound. Omni's are generally used. The directional lavs are pretty well limited to sound reinforcement applications where they're needed for feedback control.

 

[Gohanto]

Gotchya. And the CAD's diagram is almost completely flat in terms of response (which is probably why people suggested using it as a uber-cheap mic for recording voice).

So, basically, using a mic with lower sensitivity (negatives) means more "noise" in relation to its preamp stage, right? Is that regardless of gain? I wonder if this also comes into play regarding the phantom power, as I know that sometimes using a recorder's phantom power can add noise to the recording...?

Obviously I own a couple of these CAD mics. It would be nice if there was some way to get some use out of them. Any recording made with them would be noiser than a recording made with a mic of higher sensitivity, though, right? Even for ambient (or foley) stuff?

I'll say it, a lot of specs are either outright lies, or highly misleading. The sad part is that the companies listing their specs correctly look worse, and therefore don't get bought as often (with people who don't know to trust their ears yet).

As to the CAD mic being flat, yes it is but that's not always what you want ESPECIALLY with music. The U87, for example, is the most commonly used vocal mic in the world for professional recording, yet it has a notable peak between 5k-15k. Another mic may have that same peak and sound like junk. It all depends on what components are used, the design, ect... what the ideal response for the mic needs to be. I've used very noisy tube mics on vocals before, but because they worked on the singers voice no one cared about the noise.

All of this stuff is relative and the only way to know for certain what your mics are good for is to try stuff with them, and then play it back on a good system (ideally a studio).

 

[Noiz2]

So I've been trying to get more into the technicals regarding sound, (Hz, Khz, dbs, etc). I've never had any real problems recording sound (keeping things around 0 dbs and always below 6dbs so as to eliminate distortion), but I've never really understood much about microphones other than their recording pattern.

Here's the schematics from a 25 dollar cardioid mic (CAD CM217):

A few things. Not sure what you mean by "dbs". And how can you "keep them around 0" and yet "always below -6". I would guess your talking about dBfs, but -6dBfs is not close to 0dBfs. so? And those are "specifications" not "schematics". Schematics would be the circuit design.

The first thing that jumps out at me is the "Output impedance". 25 vs 200 (not sure what the 20% is about) which as I understand it relates to the voltage?

That would be plus or minus 20%, so 20% more or less the number stated. And no impedance is not voltage, it's resistance (sort of). Both of those numbers are low compared to most modern input stages so they really shouldn't make much of a difference.

I'm just hazarding a guess that the CAD would use a lower voltage than the NTG (so then 200 might be better than 25), but it looks like a lot of crazy math to me, so any elaboration would be great.

Not really. First a lower impedance would allow a greater current (more power usage), but impedance relates (now I know I'm going to get into trouble) to AC signals and should have nothing to do with phantom power.

The frequency responses are pretty close, and while the signal to noise ratio is definitley greater on the NTG (again, not sure if that's a bad or a good, but I assume that it's worse to have a "lower" signal to noise ratio as the CAD loses this one as well), their numbers aren't too terribly far off from one another.

Paper specs are of only slight use because of the way things are tested and reported (all in house and always to make the product look good). A bigger signal to noise ratio is better, BUT a lot depends on the type of "noise". Some noise is really noticeable and some isn't. So while the "bigger is better" is generally true it isn't always the case, you have to listen to the mic to know how significant the numbers are.

There's a -23 difference between the front sensitivity numbers. Does that mean that the CAD is more sensitive?

Maybe, but notice that they are not measuring the same way so you can't really compare them. This is the key problem with manufacturer specs., they are often not giving you apples and apples and you don't learn much from apples and oranges.

These two mics are completely different (though both can be used to record voice), which is kind of why I wanted to compare their specs -- so that I can hopefully understand what makes one mic's specs increase the mic's value over another.

I'm always interested in the specs. But it is rare that you get any really useful info. And the big catch is that the more "consumer" the mic the more likely that the specs are meaningless. For instance a "standard" for frequency response is the points where the level has dropped by 3dB. BUT it's not the law so you will see all kinds of consumer gear say 20 - 20K. The spec is I'm sure "true" but what they don't tell you is the quality at the extremes. At 20k it's probably buried in noise and may be 20 or 30dB down, or may be distorted and ugly sounding. High end gear tends to have more honest specs. since they know that people at that level are generally not buying on the specs. But since it's rare to see two things from different manufactures use the same testing it's hard to compare on paper.

The biggest price factors are quality, support, and marketing (not nec. in that order). People will pay more for the Schoeps name. They are great mics but compared to say an Oktava are they 10X better? It's kind of a judgment call but I would argue that they are probably twice to maybe four times as good, but your still paying 10X the price. The extra 5-8X is the Name and reputation. The same is true of a lot of audio equipt right now. Twenty years ago it cost a LOT to get into the top 5-10% (quality wise). Now with some good research you can get into the top 10% very reasonably. The is still a big jump to get into the top 1-2% but do you really need to? The benefits from moving from say a top 10% mic to a top 2% mic are going to be overshadowed by technique (or lack there of) for most people. In some cases your results might actually be worse with the "better" equipment because of technique. Mics are a good example. A very sensitive mic needs to be handled very carefully to get good results. Many cheap mics can be handled badly and still do OK. In this case OK beats BAD. Hand one kid an SM58 to do street interviews and another a schoeps, neither with shock-mount or wind screen and the 58 is the only one likely to come back usable. So "better" needs to be looked at in context.

 

[Chamber005]

Hmm. I was referring to the gain I guess when I was talking about being "lower than 6". Does that make sense?

Good to know about the specs overall. I'm pretty much set on what mics I'd like to rent (the NTG-3 or 416, and the MKH 50 or CMC641) the problem is not being able to find a place nearby that rents them. Zacuto blows (which is a huge surprise since they have literally every camera package you can imagine), AVChicago -- all of 'em. There's lots of places online, but with shipping, etc it might get pointless pretty quick. I can probably get through most everything I need to do using the Countryman B3s, but I was hoping to play around with some different mics (if my CAD mics are indeed unable to provide decent ambient/foley sound with little noise).

 

[j1clark@ucsd.edu]

Here's the schematics from a 25 dollar cardioid mic (CAD CM217):

Sensitivity: -53dB +/- -4dB @ 1KHz (0dB = 1V/u bar)

Signal to Noise Ratio: More than 60dB

Here's the schematics from a 700 dollar cardioid mic (NTG-3):

Sensitivity: -30dB +/- 2dB (31.6mV/Pa)

Signal to noise ratio: 81dB

The first thing that jumps out at me is the "Output impedance". 25 vs 200

The first things that 'jump out' at me are what I've chopped down your list to.

What 'jumps out' at me, is I don't know what "-53 +/- -4dB @ 1KHz (0 dB = 1V/u bar)" means interms of, "-30 dB +/- 2dB (31.5mV/Pa)"


Of course I understand the 31.5 mv/Pa is suggesting what the mic will produce in millivolts, given an 'sonic' input measured in terms of the pressure units of pascal...

But how does the '-53 dB ... 0 dB 1V/u bar relate to -30 dB ... 31.5 mv/Pa...

The next item is the SNR of 'more than 60 dB' vs. '81 dB'... boy that's 'vague' vs. 'exact'...

How I interpret the first spec, is the company didn't seem to want to give the sensitivity in terms of 'international standards', but rather used, as found on the internet, an older, 'american' standard. So, I'd look for what this spec means...

As for your other question about the impedence, the +/- 20% is indicating a tolerance on the 200 Ohms spec... that is the impedence could range between 160 Ohms and and 240 Ohms.

I also presume mic specs are not listed in terms of any noise that may 'creep' in via non-mic sources, or even ripple on the phantom power on the mic. The noise in a mic spec would be the 'noise' of the mic itself, because it is an electronic device.

The other issue in specsmanship is that the way in which the spec was determined, often in no way represents 'real world' use of the device... Unless you have your own private anechoic chamber, and that's where you are going to be shooting.

 

[gco]

A few things.......................
Not really. First a lower impedance would allow a greater current (more power usage), but impedance relates (now I know I'm going to get into trouble) to AC signals and should have nothing to do with phantom power.
......

That is correct.
Resistance (think resistor) is a fixed value no matter what kind of voltage is applied, AC or DC, it does not change. Impedance is similar in that it affects the passage of electrical current, however, it's value changes largely as a result of the Frequency (AC) of the applied voltage. Other factors are the value of inductance or capacitance and how those elements are connected in the circuit but again, it's largely a function of Frequency and that means alternating voltage or current. Bottom line: Resistance is fixed. Impedance is dynamic.

 

[Richard Crowley]

Written specs can do almost nothing to tell you how a microphone will sound or how it will perform in actual use. As j1clark observed, "specsmanship" has been a favorite corporate game for many many decades and continues with even more gusto today. A famous saying goes: the nice thing about "standards" is that there are so many to choose from.

Of all the specs you quoted, the output impedance is likely the least important (at least between those two microphone examples.) As long as you're not talking about using mic cables thousands of feet long, it is very unlikely that even 20 ohms vs. 200 ohms will make any discernible difference. Of course, lower impedance is usually preferred because it will drive longer cables with highest quality. But people rarely use mic cable lengths that will reveal any practical difference. And many (most?) mics have output impedances higher than 200 ohms, some very much higher.

The things that will make the most difference in a shotgun mic are probably the directional pattern and the actual frequency response, both on-axis and off-axis. Almost nobody, even the high-price, famous-name microphones shows actual single-microphone frequency response curves. At best the curves published for microphones use 1/3 or 1/2 octave "averaging", and/or the curves area statistical average over many microphones. But many are simply drawn with a ruler (or a big french-curve) by the marketing flacks to represent what they would like the microphone to perform like. They bear no resemblance to the actual curves. It is pretty scandalous that they get away with such blatant deception.

Every once in a while you will come across a microphone spec sheet that contains a pretty rough curve, but one that is an actual measured curve from a single production microphone. I saw one within the last few weeks, but I can't recall where I saw it.

Trying to compare the specs of even two similar microphones is a very iffy thing. Trying to compare the specs of two such dissimilar mics comes very close to meaningless. 10 minutes with a couple of mics will reveal orders of magnitude more than months of reading specs and comparing curves.

 

[Richard Crowley]

That is correct.
Resistance (think resistor) is a fixed value no matter what kind of voltage is applied, AC or DC, it does not change. Impedance is similar in that it affects the passage of electrical current, however, it's value changes largely as a result of the Frequency (AC) of the applied voltage. Other factors are the value of inductance or capacitance and how those elements are connected in the circuit but again, it's largely a function of Frequency and that means alternating voltage or current. Bottom line: Resistance is fixed. Impedance is dynamic.

Impedance is, by definition, the AC equivalent of DC resistance. AC impedance is almost always a complex number with a "real" and an "imaginary" part (vs a simple scalar number for DC resistance). And indeed, impedance generally varies with frequency.

http://en.wikipedia.org/wiki/Electrical_impedance

 

[DrDave]

Always look to see if they publish the self noise, A weighted. Most of my mics are in the 10-15 dB range self noise. On a cheap mic, the spec will be more of less meaningless, on a a mid to high end mic, they can be revealing.
When purchasing a mic, there are two rules. 1. Audition the mic at home with your own gear, and 2. don't buy anything the salesperson recommends.
One minute of ear time is worth ten hours of surfing.
Here are the specs from my MKH 80s

transducer type double push-pull pressure-gradient
transducer
directional characteristics omnidirectional, wide-angle cardioid,
cardioid, super cardioid, figure-8
frequency range 30 - 20000 Hz
sensitivity 40 mV/Pa ± ldB (-28 dB re lV/Pa)
output impedance 150 Ω
maximum SPL 136 dB SPL
(142 dB SPL with 6/12 dB preattenuation)
maximum output voltage 5 V
equivalent SPL due to inherent noise 20 dB SPL (CCIR-weighted)
10 dB SPL (A-weighted)
output voltage due to inherent noise 8 μV (CCIR-weighted)
2.5 μV (A-weighted)
S/N ratio at 94 dB SPL ( 1 Pa) 74 dB (CCIR-weighted)
84 dB (A-weighted)
dynamic range 116 dB (CCIR-weighted)
126 dB (A-weighted)

As you can see, the self noise figure is clearly indicated as 10 dB SPL (A-weighted) with no attempt to hide it or disguise it as something else. On the mics you show above, I don't even see that figure. I'm sure the manufacturer has it somewhere. Some of the CAD claim very low self noise figures, but 10-15 is a workable range, and you may want to go up to 18 for a ribbon. You have to figure in the sensitivity on a budget mic, on a pro level mic it will have hot enough output.
I'm sure Richard knows more about it than I do, but all of my Schoeps, DPA and Sennheisers came with a custom frequency graph with a serial number for each microphone and signed by the technician. Of course I bought them some time ago so the practice may have been discontinued, but I would be surprised if Schoeps did not do it. And the matched microphones are matched very well, although there are differences.

 

[Richard Crowley]

... all of my Schoeps, DPA and Sennheisers came with a custom frequency graph with a serial number for each microphone and signed by the technician. Of course I bought them some time ago so the practice may have been discontinued, but I would be surprised if Schoeps did not do it. And the matched microphones are matched very well, although there are differences.

Indeed, higher end microphones come with individual measured frequency response curves. But even most of those are "averaged" because the raw curves, even for the very best microphones are unusually "rough". That is the nature of the sound out in the real world.

Here is an example of a "real" frequency response curve for one of the classic old-school large-diameter tube condensor microphones (Sony C-37A, circa 1955)

The averaged curves give a somewhat better overview of the frequency response because the raw curves change dramatically even by moving the microphone a few millimeters. But most curves published in microphone specs are "idealized" or "optimized" at best and often they are pure fiction drawn by a marketing intern with a ruler.