|Elliott Sound Products||Electret Microphones|
Of all the microphones ever devised, the electret has taken the #1 position by a significant margin, and in a remarkably short time. First appearing in the 1970s, they are used in the cheapest PC microphones, the vast majority of all new telephones, high quality recording applications and fully certified noise measurement systems. MEMS (micro-electro-mechanical systems) microphones are now starting to make serious inroads, but we can expect electret mics to remain dominant in many fields for some time to come.
No other mic has covered such a wide range of applications or had the same range of prices - from perhaps $1.00 or less right through to $1,000 or more. Once considered the 'poor man's' solution, even very cheap electret capsules can give higher performance than very expensive dynamic microphones. There are limitations of course, but this applies to every microphone type - none is perfect for all applications.
This article looks mainly at the myriad powering schemes that have been used. Quite a few are already described in other ESP pages, but the purpose of this collection is to examine the different schemes to give the user a better idea of the options available. We will also look at the advantages and disadvantages of some of the schemes.
Some people's ideas are very well engineered, while others are incredibly complex for no expected benefit. It is also extremely difficult to determine where some of the ideas first appeared, and who was responsible. This makes it hard to give credit because I wasn't able to determine the original designer in several cases.
Early electret mics used a 'pre-polarised' diaphragm, with a vacuum deposited metallic coating to make the diaphragm conductive. These mics were unreliable, and often lost their pre-polarisation charge. This rendered the mic useless. The current mic capsules are almost exclusively 'back electret' - the diaphragm backing plate is both the second part of the capacitor and holds the electret 'charge'.
These mics are available in a wide range of sizes, and although the most common are omni-directional (pick up sound more or less equally regardless of direction), directional versions are also available. The back electret principle keeps the electret material away from potential contaminants, and the latest capsules have a long life and stable operating conditions. They are so good that they are steadily replacing traditional high voltage DC polarised capacitor microphones in even the most demanding applications.
The primary drawback of electret mics is the internal preamp. The best measurement mics do not use an internal FET preamp, but expect the microphone preamp to have an input impedance of at least 1 Gigaohm, and often more. The electret capsule is connected directly to the preamp using a standardised thread and connection scheme. In most respects, the preamp is identical to that used by a true capacitor mic, except that there is no requirement for a polarising voltage (typically around 200V).
An external preamp can be configured to handle high signal voltages - typically up to 4V RMS. Most measurement mics are around 50mV/Pascal (i.e. 50mV output at 94dB SPL). The maximum output level is reached at a SPL (sound pressure level) of 132dB.
By contrast, the typical electret capsule we buy from the local electronics supplier has an inbuilt FET, and is intended to be operated from as little as 1.5V from a single dry cell. Since these capsules operate from a low voltage, their ability to handle high SPL is limited in the extreme. Even if the supply voltage is increased, the internal FET limits the ultimate level - usually dramatically.
It doesn't help the beginner that electret capsules have their sensitivity commonly quoted as (for example) -35dB (±4dB) referred to 0dBV at 1 Pascal. This demands that the user calculates the output level to get something sensible. The above specification reduces to ...
V = 1 / antilog ( db / 20 )
V = 1 / antilog ( 35 / 20 ) = antilog ( 1.75 )
V = 1 / 56 = 0.018 V = 18mV @ 1 Pascal
Therefore, a mic with a sensitivity of -35dB referred to 1V/Pascal has an output of 18mV at 1 Pascal or 94dB SPL. With cheap inserts, this varies quite widely though, and the maximum SPL is generally rather limited. Those I've tested are ok up to around 100dB SPL, but after that their distortion rises quickly. Distortion at 114dB SPL is usually too high, so these cheap mics must only be used with comparatively low levels (singers, close mics on a drum kit or right in front of a guitar amp will be badly distorted, for example). The same process is used for any other specification where the reference is 1V/Pascal.
The output level of microphones should be rated in millivolts per Pascal (mV /Pa), although there are many variations. Other conventions used include dBm or dBu (referred to 775mV) or dBV (referred to 1V) at 0.1 Pa (this will always be a negative number). The older standards persist in some countries and with some manufacturers. There doesn't seem to be any logical pattern, but it's very annoying to have to convert units all the time.
1 Pascal = 10 micro-Bar = 94dB SPL
0.1 Pascal = 1 micro-Bar = 74dB SPL
1 dyne/cm2 = 0.1 Pascal = 1 µbar
There are also noise ratings (which vary widely, both in output noise and the way it is specified), output impedance, recommended load impedance, polar response, frequency response, etc. Frequency response claims are meaningless without a graph showing the actual response, and for directional mics this should also indicate the distance of the mic from the sound source. Cheap microphones are particularly bad in this respect, and it is not uncommon to see the frequency response stated as (for example) 50 - 20,000Hz. Because no limits are quoted (such as ±3dB) this is pointless - any microphone will react to that frequency range, but may be -20dB at the frequency extremes, with wide variations in between.
Even cheap electret capsules usually have very good response, but only for omnidirectional types. Cheap directional capsules are a lottery at best and like all directional mics, have generally poor low frequency response unless used very close to the sound source. In this case, the bass is often heavily accentuated (due to proximity effect).
This section expands on the information provided in Microphones. I do not intend to cover capsules that require an external FET preamp, because these require the constructor to have access to resistors of at least 1 Gigaohm (1,000 Meg ohms), and often more. It also helps to have clean-room facilities, because even a tiny amount of contamination can cause reduced impedance, noise, or even failure to function at all. Mic capsules without inbuilt preamps are also generally at the very top end of the price structure. They are also rather delicate, and all too easy to damage.
Consequently, I will look at the more common types - bear in mind that some of this material is duplicated in the Microphones or Project 93. This primarily looks at powering the microphone as a complete system, but there are schemes that appear to present a complete mic system with only the capsule and a few other parts.
Figure 1 - Basic Microphone Capsule Powering
Figure 2 shows two of the most basic possible powering schemes, and these cannot be recommended for any serious use. There are many variants, with some using an inductor to increase the available output. At 1.5V (Version 'A'), the available supply is simply too low to be useful, and it really needs to be upgraded substantially to be useful for anything other than casual amateur recordings. PC sound card microphones (Version 'B') use a similar scheme, except the supply voltage is 5V from the PC supply, and some of these are almost useful for low quality low level speech recording.
As shown in Figure 2, the standard PC microphone connector is a stereo mini-jack (3.5mm diameter). Earth and shield is the sleeve as always, the signal is on the tip, and DC is applied via the ring. Presumably, the signal and DC were separated to prevent possible problems caused by DC on the mic input circuit, but IMO the whole idea was somewhat misguided from the outset.
Apart from a few simplified examples, this article will concentrate on phantom power (DIN 45595). In all cases, phantom power should be provided at the nominal 48V. There are many pieces of equipment available now that rely on the fact that many phantom powered mics will operate fine at (often much) less than 48 Volts. This is an extremely poor practice, because there are also phantom powered mics that will not operate at voltages that are much less than the nominal value. It is perfectly alright for the P48 voltage to be as low as 43V or as high as 53V, as this is within a tolerance of ~10%.
Traditionally, P48 is delivered to the two signal lines of a balanced connection via 6.8k resistors. You will often see these specified as 6.81k - the extra 10 ohms is immaterial, but implies that the resistors should be close tolerance. It has been claimed (although I don't recall where) that the resistors should be no more than 0.4% tolerance, but it's easy to select them to be much closer than this. I would suggest that 0.1% is more appropriate - this means they should be within 13 ohms of each other. Closer matching means better common mode rejection, but there is a practical limit imposed by everything else in the signal chain.
Some mics use an internal cell or battery, and do not require phantom power. Most of these are hobbyist mics, and are also unbalanced and are not suited to professional applications. For those mics that use a 1.5V cell as their power source, as you can imagine the maximum output is extremely limited, and they distort readily even with normal speech at close range.
Figure 2 - Microphone Powering Methods
Figure 2 shows the two main mic powering methods in use. Phantom power (aka P48) is by far the most common, and is recommended for all applications. The alternative T-Power should be avoided as it is incompatible with P48 (although adaptors exist, they may or may not work), and it is all too easy to plug in the wrong mic type and cause damage. As you can probably guess from the tick and cross, I have a pretty strong opinion of the two powering schemes.
Phantom power uses equal voltage on pins 2 and 3 with respect to earth, but T-Power systems use 12V DC between pins 2 and 3. In some systems, pin 2 is +12 volts with respect to pin 3, but there is always a chance that the polarity may be reversed. The DC voltage on these pins is usually earth (ground) referenced, but not always! There are also systems where the DC supply is floating - it's not referenced to earth at all.
In general, I would have to recommend that T-Power be avoided wherever possible. It is capable of providing up to 33mA through the voicecoil of a dynamic mic (P48 power does not put any current through a floating voicecoil or transformer). In addition, T-Power can provide as much as 66mA between the positive lead and earth limited by 180 ohm resistors on each signal line). In comparison, P48 is limited to a short circuit current of 14mA, which is only available if both signal leads are shorted to earth. Each lead is limited to a short-circuit current of 7mA ( 48V / 6.8k ).
The term 'T-Power' is from the German 'Tonaderspeisung'. This is also known as A-B Powering and is covered by the DIN 45595 specification, but in some circles you might hear it called by other names as well (not all are for polite company, especially if you just killed a mic by using the wrong powering scheme). Unlike phantom power, T-Power may damage dynamic and phantom powered mics (and possibly others as well) not designed for it, and is thankfully becoming less and less common. Predictably, phantom power will very likely damage a T-Powered microphone.
T-Powered mics are still commonly used with some film sound equipment, and for 'ENG' - Electronic News Gathering for radio or TV. Sennheiser still makes a range of RF 'condenser' microphones that are available in both T-Power and P48. T-Power systems are their own worst enemy in many respects. Not only is there no strict convention for polarity (a potentially disastrous situation in itself), but in some cases the power supply may be fully floating and doesn't use the shield (earth/ ground/ pin 1) connection at all, while in others the supply is earth referenced. The electronics don't actually care either way, but it's another level of abstraction that only gives people something to argue about, but has no benefit either way. The alternate connection is shown in grey in Figure 2 (the connection shown dotted is not used with a floating supply).
Some of the older 'condenser' (capacitor) microphones had their own special power supply, and used a multi-pin connector for the different voltages. This was especially true of valve (vacuum tube) microphones, which were unable to use phantom power because their current demands were well above what can be supplied. These power supplies are used in-line with the mic, and typically present a standard XLR output with no voltages present. Many use a transformer to provide full galvanic isolation thus preventing earth loops.
Finally, many test and measurement mics use a 4mA current loop supply. This is a completely different approach from the other methods, in that it is unbalanced. Despite claims to the contrary, an unbalanced system can be just as quiet and reject just as much noise as a balanced system, although in some extreme cases high frequency interference may cause problems. A complete 4mA microphone system using an electret capsule is described in Project 134. This system typically uses a 24V supply, and a microphone 'conditioner' provides a constant 4mA current to each connected mic.
This is where we actually start to look at the many different schemes that have been used. Remember, this article is devoted to electret mic capsules with an inbuilt FET preamp, so some of the more exotic schemes are not applicable. Most of these are already discussed in Microphones, which explains the different types and has a lot more generalised information.
Many of the published schemes for powering electret capsules via phantom power have tried very hard to ensure that the circuit is symmetrical. While many of these schemes may appear to be perfectly balanced, this may not be the case.
Figure 3 - A Selection Of Microphone Powering Circuits To Be Avoided
The schemes shown above are some of those you may come across on the Net. Unfortunately, after finding this particular set of drawings (which I have redrawn and changed slightly), I couldn't find it again to give credit. While 'C' and 'D' look nice and symmetrical and would probably work well enough, their impedance is too high to allow a reasonably long cable to be used. 'A' and 'B' are (IMO) unusable - while there is a convenient formula shown, there is nothing to indicate where the impedance figure of 492 ohms came from, and it is seriously doubtful that this is real. I was unable to verify the claimed value by calculation or simulation, and it will vary depending on the FET characteristics. Although these circuits appear to be impedance balanced, in reality they are no such thing and the two upper circuits should be avoided. The other two circuits should be avoided too, because of their excessively high output impedance.
In addition, no measures have been taken to protect the capsule against high transient voltages created when phantom power is switched on. This is a group of circuits that should never be used. To make matters worse, the mic capsule's case is not at earth potential, and cannot be connected directly to the housing. This increases the likelihood of hum pickup.
Rather predictably, if you need to use an electret capsule with phantom power, I suggest Project 93, not only because it's my design, but because it is a proven circuit, is well behaved and it works very well. The capsule is earthed to minimise hum, and although it uses impedance balancing only (the signal only appears on one lead), no-one who has built it has had the slightest problem with the design.
There is no benefit to using a fully signal balanced circuit, and once the necessary protection is included they can become quite complex. The important thing for noise rejection is not signal balance, but impedance balance. If the impedance is exactly equal on the two signal wires, then noise rejection will be as good as the receiver can manage.
There is a vast amount of info around about the benefits of balanced systems, but in many cases this has been misconstrued - often to the point where original reason has been lost completely. For anyone who has not done so, I strongly recommend that you read the article Design of High-Performance Balanced Audio Interfaces, because a proper understanding is important.
As noted in that article, there is no requirement whatsoever that a balanced circuit be symmetrical or even that signal be present on each conductor. What is important is that the impedance of the two conductors is equal over the full frequency range. I have had countless email questions that demonstrate that this point is not understood, with people insisting that "surely the circuitry should be symmetrical". Totally unnecessary in all respects - especially so because of one simple fact - symmetrical circuits aren't symmetrical at all. Just because every NPN transistor has a matching PNP transistor does not constitute symmetry, because the two devices are sufficiently different due to manufacturing processes that a perfectly symmetrical circuit is impossible. It's not necessary either - it may please the eye, but it makes no difference to the sound.
As already described briefly, one of the most critical applications of all often uses unbalanced connections. This is in the area of noise measurement, which is critical not because it really matters, but because there is legislation behind it. I'm not about to launch into a diatribe about the noise measurement industry, but it is important to understand that measurements may be taken to extreme accuracy and the results used in court, yet unbalanced cables are considered perfectly alright. This is easily proved of course, and if balanced connections were found to be superior, they would be used.
Unbalanced connections are regarded as inferior by most professionals, but they are every bit as good as balanced if done correctly. The signal travels along the inner conductor, and this is protected from external noise by the shield. High quality coaxial cable is readily available, and it may have a far better shield than many balanced microphone cables.
Provided the impedance is low and high quality cable is used, almost no microphone needs to have a balanced connection. The balanced line is really based on convention, but it also adds a secondary means of reducing external noise. Because microphones are a floating source (having no secondary connection to other equipment), the balanced connection is overkill. Of course, it does absolutely no harm either, and the vast majority of all professional equipment uses balanced interfaces as a matter of course. Balanced connections are needed for phantom powering because the DC voltage is common mode (present equally on each signal line), and for this alone there's a good case for using all mics in balanced mode.
Balanced lines became common because of the telephone system (which uses unshielded twisted-pair (UTP) cable). While fixed line telephones are considered to be rather 'old hat' these days, the phone network provided a vast amount of technique, nomenclature and convention, much of which has endured in audio even though the need or reason may no longer be apparent. Even the standard 48V phantom voltage is taken directly from the phone system, which has used 48V since phones were first implemented on a large scale.
Completely beside the point, but interesting, is the reason that the phone system uses -48V. The negative phone line (with respect to earth/ ground) is used to prevent corrosion of the phone lines. If the lines were positive with respect to earth, electrolytic action would create oxygen on the phone lines, leading to conversion of the copper wires to copper oxide, which is a (poor) semiconductor, and the wire would eventually be eaten away completely. This was found during research into corrosion by Sir Humphry Davy for the British Navy in 1834. It's called 'cathodic protection' when applied to pipelines, ships, etc.
Figure 4 shows the P93 mic capsule amplifier. This circuit is used by many people worldwide, and has extremely good performance for such a simple amplifier. The transistors are arranged as a Class-A opamp, with the microphone connected to the non-inverting input. Open loop gain is over 60dB, and open loop frequency response is within 1dB from 2Hz to just under 30kHz. It will outperform most electret mic capsules easily.
Figure 4 - ESP P93 Electret Capsule Powering Circuit
Normal operating gain as shown is about 10dB (3 times) but it's easy to have unity gain. Just reduce R8 to 1k (note that R1 may need to be increased to get symmetrical clipping - try ~82k). Frequency response extends from below 8Hz to over 100kHz within less than 0.5dB, and the output voltage can be as high as 2V RMS, with distortion typically below 0.02%. When gain is greater than unity, there is a little more output level available before clipping. The output is pseudo-balanced, which in this case means that it is balanced for impedance, but not signal.
There are other circuits circulating on the Net that are also high performance, but you do need to be careful to make sure the circuit you choose will work as claimed. Many professional mics use comparatively simple circuits, and there are a few 'ready-made' electret mics that are phantom powered. Some will accept 'phantom' powering with voltages of 15V or so rather than the usual 48V. Several circuits require that the mic capsule is modified to make it 3-wire. While this certainly works with a (genuine) WM61A capsule, it's less certain with substitutes.
Figure 5 - Fully Balanced Electret Capsule Powering Circuit
The circuit above is published in a few places with various changes - this is my version, which is quite different from most of the others. It's based on a circuit that's claimed to be the schematic for a Behringer ECM8000 microphone. I can't comment on that one way or another, because very similar schemes are used by several manufacturers, with some having a JFET front end (rather than the bipolar transistor shown). These are often used with conventional capacitor capsules, and bias the JFET and mic capsule via 1G resistors.
As shown, the circuit has a gain of two, because Q1 is operated as a unity gain 'phase splitter', similar to those used in valve amplifiers. It's quite a good circuit overall (at least as simulated). Note that the positions of Pin-2 and Pin-3 are reversed compared to Figure 4, because of the connection of Q1. I've not built one, and can't comment on its noise performance. Q1 should be a low noise transistor, but how it compares with the P93 circuit shown above is unknown. Many similar circuits show the negative end of C3 connected to earth/ ground, which reduces output and increases noise. Should anyone build the circuit, you are essentially on your own. Feel free to let me know how well (or otherwise) it works in practice.
While electret mics are often thought to be at the low end, they are now very common for the highest quality measurement mics, and are also common for nature recordings and elsewhere where high sensitivity, relatively low noise and wide response are required. 'True' capacitor (aka 'condenser') mics will usually out-perform most electrets, and for the very lowest noise levels it's almost impossible to beat a large diaphragm capacitor microphone.
However, for the price, nothing else comes close to an electret capsule. Where it was once common to struggle by with a moving coil mic (in cheap sound level meters for example), now an electret is used which has more output, wider response, and will usually have lower noise. The simple fact that electrets are now common in very expensive sound monitoring and measuring equipment is testament to the fact that they are no longer the 'cheap and cheerful' devices they once were.
It is somewhat regrettable (to put it mildly) that the Panasonic WM61A electret capsule is no longer made, as this was one of the great bargains of all time for its performance. While there are countless on-line vendors claiming that they have WM61A capsules for sale, unfortunately most are substituting whatever they can get in the same form factor (6mm diameter) and claiming it's the real thing. I have a small number of the real thing and quite a few 'fakes', and there is no comparison - especially at very low frequencies. For speech the substitutes are ok, but not for measurements where good LF response is required.
It's unknown if MEMS mics will ever be able to equal a good electret for noise measurement or recording applications. They are certainly getting better all the time, but it may be a challenge to get frequency response from 0.1Hz to 20kHz - something that is easily accomplished for under $100 with an electret capsule. Most that you'll see are limited to a lower frequency of around 100Hz, but some claim 20Hz (but typically at as much as 20dB down which isn't exactly inspiring). Many also have a resonant peak at 4-6kHz, and while this is usually fine for voice applications it's of no use for accurate recordings or noise monitoring.
Electronics is changing all the time, so at some stage in the (probably) not-too-distant future we may see MEMS mics taking a greater share of the market in more demanding roles. In the meantime, electret mics still give by far the best value for money of anything that's currently available.
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2015. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|