Gear‎ > ‎Microphones‎ > ‎

Capsules

Large Condenser Mic Overview

This discussion is designed to bridge the chasm between the complex worlds of electrical engineering, physics, acoustics and our ears; how a mic "sounds". While much of this could apply to different types of mics, it is written with a focus on large condensers (LCMs).

Components
LCMs are composed of a capsule, head basket, a body containing a preamp and an XLR connector. The mics with tube preamps also require an "outboard" power supply just to power the tube's heater and high voltage plate voltages.

While there are many "classic" LCM product lines, and many more opinions about the sound characteristics of each and how they compare, one of the more frequent references is to the Nuemann U-47/U-67/U-87 line. These are often talked about as a "reference point" of sorts for a "classic, warm sound". This is partially because of the historical significance of the recording artists who used them, partially because so many feel their sound is consistently flattering to voices and partially because the human ear appears to enjoy their "sound". Yes, this is very subjective, but it shows that what is most pleasing to the human ear is not an idealized, flat-response microphone and electronics with no added "colors". This is why there are so many choices of high quality studio microphones and why the decisions in the studio about mic and placement is so complex and will always remain so subjective.

"Sound" Price
Reading through web commentary covering the last 10 years, there does appear to be a distinct correlation between "the classic Neumann sound" and the particular LCM preamp+capsule combination. The difference in sound is frequently exemplified by the comparison between a Neumann U47/U87 and medium-grad/medium-cost recording mic and a much lower cost LCM like the Studio Projects C1 as just one example. Both have essentially the same size capsules and similar broad-band performance with similar FET mic preamp designs. So why is the Neumann U87 a $3,200 mic (whose "warmth" artists crave) and the Studio Projects C1 a $250 mic (which is great for the price but many complain is too "bright). The U87 is 20Hz-20kHz and the C1 is 30Hz-20kHz; similar, so what's up? The details!


Capsules

From extracted data on the Web [read disclaimer here], a large part of this difference can be explained by the FET preamp design and the capsule selection. The Neumann U47 ("predecessor" of the U67 which was the "predecessor" of the U87) was based on their "K47" capsule and the U67 & U87 are based on a newer K67 capsule design. In the mean time, these capsule designs spawned a host of similar products manufactured by many companies including manufacturers in China and Australia. As expected, capsules are very complex acoustic devices and not everyone understood them equally which leads to significant variation in not only capsule construction and performance but capsule applications as well. At an initial cursory level, the K67 appears to have some peaking in the 6-7.5 kHz range compared with the k47. The resulting fallout has been interesting and down right contentious at times.

K47
Neumann's K47, first used in the tube-based U47, was a double-diaphragm capsule to provide multi-pattern response and provided a much smoother response than the previous K7. The U47-FET uses the K47 but changed the preamp from the Telefunken VF14 tube to a FET design. The U67 changed the K47 by splitting the backplate in two to make matched tuning for pattern symmetry easier and was re-named the K67. The K67 design also altered the capsule hole pattern to extend (pre-emphasis) the response. With only small iterations, the K67 design continued it's use in the U87/U87Ai.

This added "bump" in the upper end of the K67 response, provided pre-emphasis which was later removed by feedback de-emphasis in the preamps of the U67 and U87. This was to improve the dynamic range while keeping the preamp self-noise down. The correlation between K47/K67 and flat vs pre/de-emphasis is what seems to have been lost as the low-cost mic market developed.

Unfortunately, other LCMs which were designed to use the K67 capsule design, use linear preamps without the U67/87 negative feedback de-emphasis. This results in the commonly seen peaking in the upper range. Whether this was just a simple oversight or whether the designers thought a more "bright" sound would sell is not known. What is known is that there continues to be significant interest in "upgrading" these new "linear K67" designs to the older classic "warm" sounds of the K47 vintage.

There are numerous versions of the K47 available on the market, mostly made in China or Australia, which sell from $50 to $500 each. Some published curves suggest there is a correlation between price and response curve as related to a) extended low end, b) overall curve "smoothness" and c) a relatively small, broad peak with gentle roll-off at the high end. The MJE-K47 is even custom manufactured/selected for it's own "classic" sound.

It is difficult to find curves which provide a real comparison between a modern, flat response FET preamp and the range of K47 and K67 type capsules. So, again searched the web, the following curves have been extracted which provide a more subjective comparison by comparing a mix of mics and capsules but still using the Neuman line as a form of "reference point". (While all of this is in the public domain, I still credit to all those who produced or found and displayed these curves on web pages.) Following a Left-to-Right and Top-to-Bottom patterns it is easy to see the change from the original U47 to the SP C-1 to other mics using new generation K47 capsules. It is also clear that the U47 did have some peaking (but at lower frequencies than the peaking of the SP C-1) but that the "classic warmth" may be more a function of the Neumann low-frequency response.

 
 
Neumann U47 (Tube) [K47]
Neumann U47 (FET) [K47]
 
 
Neumann U67 [K67]
Neumann U87 [K67]
 
 
Neumann U87Ai [K67]
Studio Projects C-1 Mk-1 [K67]
 
 
MJE K47 Custom [K47]
Equinox Q47-D [K47]
 
 
FAR K47 [K47]
CM 47 [K47]


Mid-Range LCMs
The mics priced in the $500 and below range often use Chinese manufactured "knock-off" K67 style capsules and FET circuits which are low noise and quite flat but do not compensate for the K47-to-K67 change. In addition, some of the subtleties of the K47 original capsule design appear to have been lost in these versions. Most visible is the hole pattern; a somewhat random looking (but actually a specific precision pattern of hols forming an acoustic low-pass filter) radial pattern in the original K47/K67 and a regular 4x6 rectangular pattern in these later versions. Other differences such as membrane thickness and gold plating can also be seen. In addition, these knock-offs also exhibit a broad-range peaking in the 8-16kHz region -- exactly where we see the "bump" in the C1 frequency response. What this means is that there ARE reasons why the C1 sounds very good, but is considered too "bright" compared with the "warmer" Neumann U47/U87. So, what to do?

Capsule Replacement
There are several companies which have specialized in LCM modifications and, despite the online arguments, they have many satisfied customers. While the mods are not cheap, the price becomes competative if the other option is thought of as purchasing a Neumann U87 for example. However, there is another alternative for improving the sound of the much less expensive C1; replace the Chinese K67 knock-off capsule (with regular hole pattern) with a better-made K47 (with random-radial hole pattern) and leave the "flat" Schoeps-style FET preamp intact. This might not provide every possible optimization (so this is not a claim that the mic mod businesses have little to offer) it is just a very simple change (for someone with the skills, guts and plenty of coffee) to get a more "classic warm Neumann" sound from a mid-priced mic and reduce the "too bright" sound from a stock, K67 copy capsule sound.

Which Capsule?
Interestingly, there are also K47 copies available for cheap but this begs the question of why replace one cheap copy with another copy. What's the point behind that effort? The answer again lies in the details. In any manufacturing process there is a question of tolerances, variances and specs. If you are willing to accept every unit from production, then the average quality will suffer as some units won't meet spec. So, the same source of capsules can also provide capsule parts which are then final assembled under more strict controls and tighter tolerances and specs. All of this comes at a price, but if you really want to improve the performance of a lower cost K67 based, flat response preamp mic then higher tolerances and tighter specs on a more customized K47 will provide a superior result.

K67 to K47 Wiring
Electrically, the K67s (cardioid clones) are a unipolar device but the K47s are a symmetrical unipolar pair with a common backplate connection.
  • The "output" from these devices is the "center" connect which has a terminal on a wire screwed to the membrane (Hint: Never touch it or the membrane.) The wire may be Red or White or other color.
  • The K67 (cardioid) has only one (1) membrane "center" connect. The opposite side of the capsule is an insulated plate with an array of holes (acoustic filters).
  • The "bias connect" is the support "backplate/ring" which get screwed to the "saddle/mount". Often a wire with terminal must be added when screwing the capsule "ring" to the saddle to make electrical contact
The K67 (cardioid) has two (2) connections and the K47 has three (3). So, how are they connected?
  • The K67 (cardioid) is simple; the "center" (output) goes to the gate of the FET in the preamp and the backplate is biased with the +50-100 VDC bias
  • The K47 is designed for multi-pattern mics where both (2) "center" (output) signals are combined dependent on the selected pattern.
  • Using the K47 in a circuit designed for the K67 (cardioid) just means using one (1) "center" (output) wire and leaving the other "center" (output) wire floating (insulated from everything) to get the cardioid pattern.
  • It is suggested that the K47 (cardioid) can just be physically and electrically reversed (flip the capsule around 180 degrees and swap the "center" (output) wires to use the other membrane should the selected active membrane fail. So, initially, the second membrane is a bit of a waste in cardioid mode.

Headbasket

There is much less technical information published about the acoustics of the 1,2 or even 3-grill "headbasket" covers over LCM capsules. However, they are all metal and all are grounded to the preamp shield/case. Grounding is necessary because, in their simplest concept, capsules are a large "antenna" feeding a high gain, high impedance preamp which is susceptible to 60 Hz EMI ("hum" from your 110VAC power lines and equipment) as well as local radio stations. So, clearly the capsule needs RFI (Radio Frequency Interference) shielding. The remainder is acoustics.

Sound pressure must be able to reach the capsule diaphragm but the headbasket gets in the way. However, so-called "plosives" and "esses" from the voice up close can overload the diaphragm on peaks so the grills are helpful as an acoustic shield. In between all of that, is largely un-reported except for the plot below.


Trade-Offs
Sound pressure must be able to reach the capsule diaphragm but the headbasket gets in the way. However, so-called "plosives" and "esses" from the voice up close can overload the diaphragm on peaks so the grills are helpful as an acoustic shield. In between all of that, is largely un-reported except for this one plot from the web:

The Plot
This plot source image is a bit fuzzy but it reportedly shows the effect of the grill on a Neumann U-47 mic; Top Curve - the response with the grill in place as manufactured and Bottom Curve - with the grill removed. This suggest that the grill on a mic really does "color" the sound. Reading "between the lines" of the fuzzy curves, they extend from 2 kHz - 20 kHz with the grill causing about 2-3 dB differences to the normal response curve; Specifically: a "dip" at 8 kHz, a "peak" at 10 kHz and another "dip" at 15 kHz. It is logical that frequencies below 2 kHz would not be substantially affected because the wavelengths are much too long. The "dip" at 8 kHz has been conjectured as being responsible for smoothing many of the typical vocal transients.
 

Clearly the headbasket is necessary and reportedly it can improve the quality of the mic's response to voice. What is much less understood (maybe just published) is the value of multiple grills and/or the value of removing the "inner grill" on mics to give them "more air" in their sound. This is largely too subjective to draw any conclusions as it is all too easy to hear what we want or expect to hear.

It would be very informative to see some detailed measurements of the grill effects with a number of different condenser mics -- but, alas, that is an expensive test. However, one additional observation of this effect was attempted while measuring a Studio Project's C1 and the results are included in the Studio Measurements section below.


Studio Measurements

By no means the only choice (read disclaimer here) some of the better options should include the P-K47 capsule from Peluso Microphones (maker of high-end LCMs) and the MJE-K47 from Michael Joly Engineering -- and there are others other quality engineering sources as well. Again, from reading the web, some sources appear to start with overseas K47 manufacturers and then add a significant amount of selection, testing, grading and additional manufacturing steps to produce superior results; a) flatter and improved low-frequency response, b) a smoother high end and c) much better controlled upper-end peaking.

Experiment
While I don't have an anechoic chamber for testing mics, the studio acoustics have been significantly corrected as described elsewhere on this site. By placing the microphones in close, near-field (centered, 6-inches) of a studio monitor and by taking advantage of the directionality of the mics, the monitor's frequency response can be measured. The data will be valid ONLY for that specific near-field position, but it should be sufficiently accurate to either verify or reject the claims read on the web. Below are the plotted results.

Measurement Method
The basic concept behind these microphone measurements is to:
  • Generate a transfer standard which represents the monitor's sound field at the measurement point
  • Use comparative measurements as absolute values are of relatively little importance
  • Use both an omni-directional reference mic and a near-field cardioid mic which will help isolate the residual bass resonance of the room.
  • Rely on published typical specs if possible
  • Disclaimer - While neither lab-calibrated microphones, monitors nor anechoeic chambers are used, the curves do show there is sufficient validity of this method as the results for the Studio Project C1 are quite close to both published and independent measurements as found on the web. For this discussion and for this studio, these results are sufficiently meaningful to have been worth the effort.
To calibrate the studio monitor's output, at this specific measurement point, both the ECM8000 omni-directional reference microphone and a near-field SM137 mic were sequentially positioned at the reference position; centered between the cones of the Alesis Monitor One Mk2 studio monitor and at as distance of 6-inches.
  • Note that while this "sweet spot" is not the only possible position, it did appear to have a reasonable balance between bass and treble energy. What is more important, as will be shown by plots below, is that every mic placement be at the same position.
 
 
ECM8000 Measured / Default Cal / Corrected
Shure SM137 Mics #1, #2 & Average

Plot #1
  • This plot shows the measured response [Blue] of the studio monitor. The ECM8000 reference mic is quite flat so the published "typical" calibration curve [Green] was assumed. True, it will not precisely match this specific ECM8000, but it will be close enough as will become clear in later plots.
  • The monitor's corrected response [Red] is very close to the measured response because the ECM8000 has a very flat frequency response.
  • Notice that the nearly 10 dB range results from a) the mic's own response (a minor effect), b) the studio monitor's own response (a major effect) and c) the room's response (major effect - largely below 500 Hz).
  • Also remember that the ECM8000 is an omni-directional mic which does not perform well a rejecting residual room bass resonances. This will be addressed in a later plot.
Plot #2
  • Two Shure SM137 near-field small condenser mics were measured [Blue and Green]. These curves include the ECM8000's studio monitor correction curve from Plot #1 which is why they appear more "normal".
  • The [Red] curve shows the average of these two responses.
  • Note that all three curves are nearly plotted on top of each other -- this is good validation that:
    • There is good measurement repeatability
    • The room acoustics, whatever they are, are stable
Bass Resonances
Because the measurements are made just 6 inches from the studio monitor, frequencies above 1 kHz will have little effect on the response measurements. However, below 500 Hz, it is difficult to isolate the monitor and mic from the room reflections. This adds significant uncertainty to the lower end of the response curve.

However, by applying a second correction factor, from a near-field cardioid response mic, like the SM137 used here, the bass resonances can be at least "partially" removed. If the mic's directionality and reverse-side attenuation high enough, this will make a substantial improvement in the accuracy of the overall correction curve -- which will hold true if all measurements are performed at this same position.

 
 
Shure SM137 Average vs Default Spec
Final Frequency Response Correction Curve

Plot #3
  • The [Bue] curve is, again, the average of the SM137 measurements
  • The [Green] curve is the SM137's default response curve as published by Shure
  • The [Red] curve is the difference between them. This curve is the presumed error, most of which is at frequencies below 500 Hz where room resonance cause significant errors.
Plot #4
  • The [Blue] curve comes from Plot #1 and is the ECM8000 correction
  • The [Green] curve come from Plot#3 and is largely the room's bass resonance errors
  • The [Red] curve is the resulting studio monitor response curve which will be used to correct all other measurements with the C1 mics.
Measurement Position
A the normal listening distance from studio monitors, we hear a great balance between "tweeter" and "woofer" output because both height and angle because the high frequency cones are much more directional. However, these measurements are all near-field so we expect the position sensitivity to be accentuated. All mic placements for these tests have been made with care to assure consistency and an accurate comparison between curves. However, it would be good to know the position sensitivity.

As a simple test, the Studio Projects C1 was positioned at the "centered", 6-inches reference position and then Left and Right 2 inches, then In and Out 2 inches and then Up by 2 inches and Down by 1 inch. The results are below:

 
 
SP C1 Mk2 Corrected vs Left / Right / In / Out 2-in
SP C1 Mk2 Corrected Up 2-in / Down 1-in

Plot #5
  • The "center" position is in [Black]
  • Left and Right by 2 inches are shown in {Magenta] and [Red]
  • In (closer to monitor) and Out by 2 inches are shown in [Blue] and [Cyan]
  • Notice:
    • Left - Right position variation has low sensitivity
    • In - Out position is more sensitive -- but is easier to control in mechanical placement
Plot #6
  • The "center position is in [Black]
  • [Blue] is 2-inches Up and shows a large dip at 2 kHz and more peaking at 13-15 kHz
    • This is expected as more of the tweeter field is being intercepted
  • [Red] is just 1-inch down and shows a huge loss above 3 kHz
    • This clearly shows the C1 dropping out almost completely below the tweeter's acoustic field
  • Notice:
    • Clearly, the Up-Down positioning is the most critical
    • This sensitivity could be reduced at 12 inches but at the expense of greater room resonance error
SP C1 Plots
The "proof of the pudding" is the corrected C1 measurements and whether or not they match what has been published elsewhere.

 
 
SP C1 Mk2 Measured and Corrected
SP C1 Mk2 Flat / 75Hz LPF / 150Hz LPF

Plot #5
  • The [Blue] curve is the overall studio monitor correction curve from Plot #4 above.
  • The [Green] curve is the measured C1 curve. The general high frequency peaking can already be seen and has been frequently described on the web.
  • The [Red] curve is the corrected Studio Projects C1 response plot. Notice that is looks extremely similar to both reported and published plots.
  • Also notice that the degree of peaking more closely matches what has been widely reported as the measured peaking vs what has been published as the nominal response plot of the C1. This actually give more validation to the methods and measurements shown here.
Plot #6
  • The [Blue / Green / Magenta] plots correspond to the [Flat / 75 Hz LPF / 150 Hz LPF] filter settings which are selectable on the C1 mic.
  • These curves add further validation to these measurements and methods as the curve look very close to what one would expect.
  • Note the slight start to peaking at the bottom end of the curves. This is assumed to be some residual low frequency room resonances which are just too difficult to remove -- so acceptable here.
Grill Effect
While much harder to measure under these limited test conditions (not having an anechoic chamber), the effects of removing the grill can clearly be seen in plots #7 and #8.

 
 
SP C1 Mk2 (Flat) Effect of Grill
SP C1 Mk2 (150Hz LPF) Effect of Grill

Plot #7
  • This plot shows the Grill [Blue] vs No-Grill [Red] effect while using the Flat setting on the C1. These curves were measured first to verify there was a difference. There is!
  • The No-Grill "peaking" at about 400-500 Hz is thought to be caused by greater exposure to room bass resonances since the entire Grill cover was removed.
  • The No-Grill peaking at 2.5 kHz is thought to be a downward shift in the existing Grill peak just below 4 kHz.
  • The near 4 dB drop in the ~13 kHz Grill peak at the high end of the response likely has a direct relationship to the enclosed cavity the Grill creates around the capsule.
    • The Grill is 2-inches in diameter and the capsule membrane is offset by 1/4-inch.
    • The membrane-to-Grill spacing is thus about 3/4-inch
    • In air, the velocity of sound is 1128 ft/sec
    • For a half wavelength of 3/4-inch to cause a resonance, the frequency would be about 9 kHz
    • For the full grill width of 2-inches, a 1 wavelength resonance would be about 6.7 kHz -- notice that we find a "dip" in the response at about that frequency.
    • Considering the capsule Grill volume acoustics is considerably more complex than this brief analysis, a broad resonance at 12-14 KHz, as seen in the curve, could easily be reinforced by the Grill cover's own mechanical / acoustic conditions.
    • While plots of this effect are hard to find on the Web, the frequent discussions about "removing internal fine-mess grills" and the associated subjective commentary about the change in the mic sound provide additional support to the measurement plots shown here as being real.
    • However, the true acoustic complexity of this analysis will be left to others.
Plot #8
  • The results of Plot #7 were somewhat unexpected so the test was performed again but with the 150 Hz LPF. Clearly the difference shown in Plot #7 is real and, with the stock K67 capsule, the Grill plays an integral part in explaining the mic's sound.
Observations
  • The evidence suggests that the C1's Grill actually accentuates the "peaking" above 10 kHz. From various discussions, this might be caused by some mix of in-phase internal reflections within the capsule-Grill region.
  • The No-Grill peak at ~2.2 kHz appears to be a capsule peak; just shifted down with the Grill off
  • The Grill-On peaks at 4 kHz and 12-14 kHz appear to be directly related to the K67 capsule -- something that has been reported numerous times on the Web.

P-K47 Replacement
Given all of the above data, the obvious test is to replace the K67 type capsule of the SP C1 Mk2 with a K47 type and then possibly, test the effect of removing one layer of the Grill. However, the first step is only to replace the capsule as it is possible that the Grill will be less of an issue with the change in capsule.

 
 
SP C1 K67 type [Blue] vs SP C1 with P-K47 [Red]
SP C1 with P-K47 capsule: Grill [Blue] vs No Grill [Red]

Plot #11
  • The 6-15 kHz peaking of the original K67 type capsule is dramatically eliminated just by replacing it with the P-K47 capsule from the Peluso Microphone Lab.
  • Some peaking in the 3-4 kHz region remains which may not be caused by the capsule.
  • The P-K47 roll-off above 12-13 kHz is likely the capsule first and the preamp second.
  • The remaining small variations over frequency are assumed to be just measurement "noise"
Plot #12
  • These two plots attempt to identify the effect of the standard Studio Project's C1 Grill.
  • Previous plots suggested the grill has a significant effect on the upper frequency response but here, with the P-K47 capsule the difference is much less obvious.
  • There appears to be a little added peaking at about 9 kHz and a slightly reduced roll-off above 12 kHz -- by maybe as much as ~1.5 dB at 20 kHz.
  • The 4 kHz peaking is not made worse by removal of the Grill
Observations
  • Clearly the P-K47 capsule swap does reduce the 6-15 kHz 6 dB "brightness" of the stock C1
  • With the P-K47 installed, the Grill has far less impact on the upper frequency response
  • Removal of just the inner layer Grill would likely have questionable measurable benefit so the mechanical risks of its removal are not clearly outweighed by potential benefits.
  • The Web comments about removing the inner Grill as providing "more air" or "clarity" to the sound might be related to these subtle differences -- which could be more dramatic with other LCMs.
"Shoot-Out"
One comparative method often requested on the Web is a mic comparison using "stock" and "modified" mics positioned close together while simultaneously recording a voice or instrument. This is a very subjective test because all listeners easily hear what they expect to hear.

To be a little more objective, two C1 Mk2 mics, one "stock" and one with replacement P-K47 capsule were positioned "head to head" with near zero spacing. A Martin OM-JM, with an open turning, was strummed once (with a pick to ensure upper frequency content) as a recording example. To reduce the natural "boominess" of the sound hole, the mics were position about 6-inches from the 12-th fret; a common spot for recording acoustic guitar.

Spectral Plots
However, the listening differences in the tracks are so subtle that this one test is too subjective. So, the tracks were processes through a full FFT spectral analysis with long term hold and averaging set while cycling the track through a loop to accumulate the data. This allows a more reliable measure of total spectral content. Each mic's track was processed the same way with equal mixer levels as can be seen below:

 
 
SP C1 Mk2 K67 type - OMJM Open Tuning Strum
SP C1 Mk2 P-K47 capsule - OMJM Open Tuning Strum

Plots
  • The two spectra are very nearly identical -- same basic shape, structure and peak levels
  • The single most notable difference is the added 7-20 kHz range peaking of the K67 (Left) vs P-K47 (Right)
  • A slightly more subtle difference is the slightly "flatter" 1-3 kHz region of the P-K47 (Right) version
However, the above spectral plot detail is still too complex, so we need a bit of clarity to view the differences. The method is to plot the spectral band averages over the same frequency span. Remember, this is the spectrum of a pair specific audio tracks (Martin OMJM acoustic guitar) which includes all acoustic conditions. Comparison is possible because the tracks were recorded simultaneously with the mics as close to the same position as possible.

Two plots are shown:
  • The first (Left) is from simultaneous recording tracks of an acoustic guitar
  • The second (Right) is from simultaneous recording tracks of voice - including some proximity effects

 
 
SP C1 Mk2 K67 vs PK47 Spectrum
(Averaged from OMJM Open Tuning single strum)
SP C1 Mk2 K67 vs PK47 Spectrum
(Averaged from Voice 'She Sells Sea Shells')

Plot #13
This plot shows the spectral differences between a stock C1 (K67 type capsule) and a modified C1 (P-K47 capsule) from a single open-tuning strum, using a pick, with both C1s positioned 6-inches and centered from the 12th fret of a Martin OMJM.
  • The P-K47 shows a slightly better low-frequency response below 100 Hz
    • Note: The P-K47 mic was positioned on the neck-side of the 12th fret (the K67 mic on the sound hole side of the 12th fret) so the sound hole's influence cannot account for this additional bass response.
  • The P-K47 shows a slightly smoother response from 2 - 6 kHz
    • The P-K47 is assumed "smoother" because of the previous response plots.
  • The P-K47 plot also supports the previous plots showing less peaking above 8 kHz
Plot #14
This final plot is the C1 K67 vs PK47 comparison spectrum from a voice track ("She Sells Sea Shells..."). Even with the two mics placed side-by-side and aligned 180 degrees apart in orientation, the proximity effect is still hard to keep identical.
  • Notice that the 200 Hz to 2 kHz mid-range is virtually identical
  • Notice that again, the region above 8 kHz shows the same "peaking" with the stock K67 capsule
  • The biggest anomaly is the lower frequencies below 200 Hz
    • With a difference as much as 30dB 10 100 Hz and with the bass response giving the worse performance, there must be an explanation other than capsule frequency response
    • Since the room is not completely free of bass resonances, mic position could be a major issue
    • Proximity effect near a mic is very non-linear and balancing a voice accurately in position between two mics is more difficult than what is imagined simply by what is heard in the headphones

Conclusions
  • It is possible to make reasonably accurate microphone measurements in a small studio with good acoustics
  • Anechoic chambers are obviously better but reasonable data can still be acquired without them
  • Measurement plots can be validate by comparison with published plots
  • Mic changes, such as capsule and/or grill can be measured to quantify the effect of those changes
  • A better correlation can be achieved between measured and subjective results
  • Swapping the K67 type capsule for a P-K47 capsule in a C1 Mk2 provides measurable improvement
  • Removal of the inner Grill, with a P-K47 installed, is of questionable importance
  • Simple listening comparisons of recorded tracks can easily be much too subjective
  • Differences with voice, with proximity effects is completely different than recorded acoustic instruments
  • Simultaneous tracking and spectral analysis of "stock" vs "PK47 Mod" SP C1s confirms the results