Fullscreen HTML5Med Res (??MB)HTML4

Build a low-cost, calibrated Measurement Microphone If you have ever wanted to characterise or build loudspeakers but couldn’t justify the cost of a fancy microphone, or you want several microphones you can tailor for performance or recording, this project is for you. It’s a phantom-powered, balanced, calibrated microphone you can build for much less money than a commercial equivalent. Project by Phil Prosser T his project aims to build a lowcost measurement microphone using an inexpensive electret condenser microphone (ECM) and a few other bits and pieces. The WM61A and alternative ECM capsules listed below are only a few dollars each. If you recycle parts for the housing, you can make a good microphone for under $40, which is ideal for getting started. With the calibration files we provide, it will let you measure frequency response to within about ±2dB from 20Hz to 20kHz. This Microphone uses phantom power, where the power for the microphone is provided over the signal lines from your microphone preamplifier or mixer. Our Speaker Test Jig (published in the June 2023 issue; see siliconchip. au/Article/15821) can provide this, as can several other Silicon Chip projects and most commercial microphone preamps. This avoids the need for batteries and is widely supported. If you want to build this as a measurement microphone, plenty of ECM capsules with calibration files are available from the Silicon Chip Online Shop at a modest cost. The capsules are numbered and you just need to match up your number with the downloaded file to get accurate calibration data for that capsule. We also have instructions to tailor the frequency response of a microphone for vocal or instrumental use. Aiming for a flat response How well does it work? Fig.1 compares the raw performance of two $2 WM61A capsules to our reference Dayton EMM-6 microphone. This is before the application of the calibration file. The curves’ 10-12dB offset is simply due to these capsules being more efficient than the EMM-6; note how the responses barely go outside the 9-11dB/11-13dB ranges that represent ±1dB from the average. To achieve this comparison, we placed the microphones within a couple of millimetres of the same point as the reference microphone. We feel the performance shown is pretty good for such a simple and low-cost design. Here’s a collection of the types of Measurement Microphones you can build. 68 Silicon Chip Australia's electronics magazine As mentioned earlier, the capsules we’re offering come with calibration data that allows the 1-2dB error to be corrected. The calibration accuracy is limited by our Dayton reference microphone, although we are confident that above 50Hz, it is flat within a couple of decibels. A Behringer ECM8000 runs about $80, while the Dayton EMM-6 starts at around $140. As mentioned earlier, you can probably build the Microphone described here for around $40, possibly a bit less. Note that the ECM8000 doesn’t come with a calibration file, while this one does, making it even better value. So you can achieve pretty good performance at a very competitive cost with this project. To get the best from your Microphone, the design incorporates a phantom-powered preamp and a balanced output buffer based on an industry-standard design, the ‘Schoeps transformerless design’. This harks back to the 1960s and is used in a vast range of professional and The Panasonic WM-61A microphone. siliconchip.com.au Fig.1: a comparison of the performance of two of the $2 WM61A microphones capsules to our reference Dayton EMM-6. This is uncalibrated performance; we can supply ECM capsules with calibration files that will reduce these errors. The offset of about 10dB/12dB for the two samples means those capsules are significantly more sensitive than the Dayton EMM6, which is rated at -40.3dBV/Pa. measurement microphones. We have added an input and filtering section to suit the ECM capsules we present here. The design is quite conventional, so you can make a general purpose phantom powered electret condenser microphone using this project. As you will see later, we have included the ability to tune the Microphone’s response. In our application, this is to get a flat response, but nothing is stopping you from using that capability to adjust the microphone response to suit vocals or instruments. So, how can you really get a good electret microphone for two bucks? ECMs are very simple devices and are made in huge volumes. As shown in Fig.2, they work by sound moving a very thin diaphragm relative to a backplate that is connected (typically) to the gate of a FET. A charge is created between these, and the capacitance between the diaphragm and backplate changes as the sound moves the diaphragm. The formula is C = ε0 × A ÷ d, where d is the separation between the diaphragm and the backplate, A is the area of the plate and ε0 is a mathematical constant. The charge between the plates Q is constant, and since C = Q ÷ V, as the capacitance changes due to the sound, so does the voltage between them (V). This drives the FET. As the capsules are tiny, and the siliconchip.com.au diaphragm extremely light, these devices can have excellent frequency response to very high frequencies with little resonance. The Panasonic WM-60A and WM-61A microphones are legendary examples and have an exceptionally usable frequency response from 20Hz to 20kHz. In the past, they were the mainstay of DIY measurement microphones. They were a workhorse component used in a wide range of devices, including telephones, which meant they were made by the million and thus cheap. Panasonic stopped making these in the early 2000s, which some ascribe to the demise of the old-fashioned ‘phone. Panasonic capsules can still be found, but many sellers list generic 6mm capsules as WM-61As. We bought a large quantity of real ‘new old stock’ (NOS) parts, all from a single batch, measured their response, and are offering them for sale – see Table 1. Before we found a batch of old stick WM61As, we bought and tested a huge number of microphone capsules. Our experience has been that ECM capsules that are ‘flat’ to 20kHz tend to be 6mm diameter units; they are pretty small. The larger 10mm ones generally exhibit a significant peak in the response between 5kHz and 10kHz, making them less than ideal for measurement applications. Therefore, all our recommended Australia's electronics magazine Fig.2: the structure of an electret condenser microphone (ECM). The internal FET amplifies the small AC voltage generated by the diaphragm moving in relation to the charged backplate. ECM capsules are 6mm. We also learned that the majority of capsules available cannot be used in this project as they exhibit peaks or dips, many over 10dB, that we are not comfortable addressing by calibration. Virtually all the satisfactory mics we found will be available from the Silicon Chip Online Shop, including the required SMD calibration components, all for similar prices. Another thing we learned is that there is no ideal ECM capsule that will give acceptable performance without calibration or at least some equalisation of the native response of the capsule. The old Panasonic WM61A capsules tend to be more consistent than most modern alternatives, but there can still be significant differences in frequency response from batch to batch. Manufacturers present typical frequency response plots for their ECMs, but there is significant variability in their response above 10kHz between batches. The Primo EM258 capsules are excellent, but at £6.10 (around $11.25) plus shipping, they are starting to defeat our goal of a low-cost design. We eventually concluded that calibration of each ECM capsule is essential. So we have done a couple of things: ● We designed a circuit that allows you to add a peak or dip and either a ramp up or down to the frequency August 2023  69 response. We have determined the required combination for each type of ECM we tested to get a reasonably flat response. ● Each ECM capsule we supply has a serial number matching a set of calibration corrections to make it perform even better than just with the frequency response adjustment. The calibration file can be loaded into the Room Equalisation Wizard (REW) or Speaker Workshop software to get your measurements as close as possible to ideal. For those who want to build a vocal or instrument microphone, we will show you how to tune the circuit’s response to get the ‘colour’ you want in the microphone you build. If you are making a vocal microphone, you don’t need one of the calibrated ECM capsules from our store; you can save money by buying a similar one from an internet vendor. Which capsule do we prefer? The NOS Panasonic devices still stand out. The best still-officially-available type is the Primo device. The CMC2742PBJ-A is pretty good with compensation (and still available). With compensation, all the types we’re selling are within a decibel or so of our reference mic to at least 10kHz, and with calibration, will be within ±2dB (or better) of our reference mic. Performance We are proud of the performance achieved, especially in a low-cost project. Fig.3 shows the compensated (but not calibrated) frequency response of 10 of the ECM capsules we tested. Some things we noticed are: ● The CMC6027-24T family of devices are very sensitive. That could be beneficial under certain circumstances, but using these for very close measurements or in very loud settings will result in potential compression and distortion. ● All microphones are within ±3dB of their average before the application of calibration over the range of 50Hz-20kHz ● All are pretty flat through the region where you would put a bassmid and midrange-tweeter crossover (although the JLI61A has a bit of a bump). So you could use these mics for such purposes even without calibration. ● The WM61A lot 4A14 microphones are brilliant. The great news 70 Silicon Chip is that we have lots of these available for constructors! Circuit description The electronics to drive the microphone capsule is not complex, as shown in Fig.4. The circuit has three main parts: buffers for driving the balanced output lines, a gain stage which includes some cunning frequency compensation and a power supply for the gain stage. The first thing to keep in mind when looking at this circuit is that pins 2 & 3 of CON1, the XLR socket, act as both 48V DC power inputs and AC signal outputs. The 48V DC is ‘phantom power’ from the upstream equipment like a mixer or microphone preamplifier. It is dropped across the 6.8kW resistors in the phantom power source, allowing the Microphone to vary the voltages on these pins to feed the signals back. PNP transistors Q1 and Q2 are emitter-­ follower buffers with 6.8V zener diode clamps between their collectors and emitters. The DC bias point for Q1 and Q2 is established by the 150kW resistors between their bases and collectors. The current flowing from their emitters to their collectors provide the supply current to the rest of the circuit via R12. Once power is applied, as the collector voltages of Q1 and Q2 increase, the base current through the 150kW resistors falls until DC equilibrium is established. For AC signals, Q1 and Q2 act as emitter-followers with the AC signals being coupled to their bases through 1μF electrolytic capacitors. Is this really balanced audio? By driving the hot pin with the microphone output and the cold pin from ground, we provide a differential output from the Microphone. The balanced line receiver for the Microphone will subtract any signal on the cold line from the hot, providing the immunity from noise pickup in the cable we seek. The 48V DC phantom supply is dropped across the 6.8kW series resistors in the microphone preamplifier and 5.6kW resistor R12 to the 6.8V limit set by zener diode ZD2. The collectors of Q1 & Q2 will sit at around 32V, as exlained below. This voltage (and the current that establishes it) supplies power to the amplifying NPN transistor, Q3, and the ECM itself, in both cases via 5.6kW resistor R12. The circuit includes 1nF and 2.2nF capacitors from pins 2 & 3 of CON1 to ground, with 47W resistors between them, to increase the immunity of the circuit to radio-frequency interference (RFI). These parts do little to affect the low-frequency audio signals or phantom power but will heavily attenuate ultrasonic signals. Additionally, 470pF ‘Miller’ capacitors across the base resistors of Q1 & Q2 roll off the frequency response of these buffer transistors above audible frequencies. Fig.3: the frequency responses of a selection of ECM capsules, including their recommended frequency correction parts, but without calibration corrections. These curves themselves form the calibration correction files. The vertical offsets represent differences in sensitivity, but we are mainly interested in the flatness of each curve (flatter is better from a measurement perspective). Australia's electronics magazine siliconchip.com.au Power supply The power supply for the ECM is very simple but includes plenty of filtering to get a stable DC supply from the hot and cold lines carrying our audio signal. We mentioned the 6.8V derived from the phantom power across ZD2. This is low-pass filtered to remove noise by the 100µF capacitor across ZD2, in combination with the source resistances (6.8kW & 5.6kW). It is further filtered by another lowpass filter (330W/10µF) before being applied to Q3 and the ECM. This is because the signal from the ECM is so low in amplitude that any noise getting through could seriously degrade our signal-to-noise ratio (SNR). Table 1 – Tested Microphone Capsules Model Source Notes Panasonic WM-61A – AliExpress 1005004118951415 – Silicon Chip SC6760 Gives the flattest response overall. Panasonic WM-61A – eBay 164187904055 – Silicon Chip SC6761 “Lot 4A14” – large quantity available; also gives a very flat response. JLI-61A – www.micbooster.com – www.jlielectronics.com – Silicon Chip SC6762 “Lot 3” – needs compensation for good performance. JLI-61AY-102 – www.micbooster.com – www.jlielectronics.com – Silicon Chip SC6763 Better than the JLI-61A but still needs compensation. CUI CMC-6027-24 – Mouser – Silicon Chip SC6764 Can have suffixes “T” or “L100” (same performance). They are the most sensitive of the tested types and among the flattest response with compensation applied. Frequency compensation Finally, we have the ECM interface and frequency compensation. This part of the circuit can be as simple as a bias resistor (R8 or R14) and an amplifying transistor (Q3). During our tests, we found several microphones that required either boosting their output at high frequencies, attenuating at high frequencies, or a little of both to give a flat response. Therefore, all our compensation is targeted at higher frequencies. Boost is achieved by R10/C12. These parts are in parallel with the emitter resistor of Q3 and thus increase the gain of Q3 at higher frequencies. We can set the corner frequency and the ultimate boost level by choosing the values of these parts. CUI – Mouser CMC-2742PBJ-A – Silicon Chip SC6765 Requires compensation and calibration, giving a reasonably flat response but with roll-off below 50Hz & above 15kHz. Kingstate KECG2740PBJ – element14 Requires compensation for good performance. Kingstate – element14 KECG2742TBL-A Requires compensation for good performance. Primo EM258 Excellent performer; expensive, no compensation required. – www.micbooster.com High-frequency attenuation is achieved by R13/C14, which are effectively in parallel with Q3’s 2.2kW collector resistor. Again, these parts can set the corner frequency and ultimate attenuation. This modification of the simple transistor amplifier (Q3) provides a powerful tool to tailor the response of a capsule. By implementing these corrections inside the Microphone, we achieve a respectably flat frequency response and leave only ‘fine-tuning’ to a calibration file. Fig.4: pins 2 & 3 of CON1 supply DC power (nominally 48V with source resistances of ~6.8kW) and are also the balanced audio signal outputs. PNP transistors Q1 & Q2 drive the audio signals onto those pins; their collector-emitter currents (and any current shunted by parallel zener diodes ZD1 & ZD3) also provide a power supply for amplifier transistor Q3 and the electric mic. The transistors shown are for the SMD version. Note that R8 is only fitted with 3-wire ECMs. siliconchip.com.au Australia's electronics magazine August 2023  71 Fig.5 shows how the compensation works. The green trace is the frequency response of the circuit using a JLI61A ECM with no compensation; note the ~7dB peak at about 7.5kHz. The red curve shows the compensation achieved with R10 = 220W, C12 = 12nF, R13 = 2.2kW & C14 = 15nF, and the blue curve is the much-flatter ultimate frequency response achieved. There is still a small peak of about +3dB, but we can’t knock it down further without overly attenuating signals at about 2-6kHz and 10-20kHz. It isn’t much bigger than some other peaks after compensation, anyway. Most ECM capsules within a batch behave similarly. During our calibration process, we set aside any parts that were outliers. Thus, you are guaranteed to get a pretty good response without the compensation file, and a very flat response with it. If you source your own ECM capsules, you will need to optimise the response and generate a calibration file. This project provides everything you need to do that, except a calibrated microphone against which to make the required measurements. Two PCB options If possible, we recommend you build the SMD version where all parts are on the top side. However, we have also laid out a through-hole version and managed to squeeze it into a Why bother with analog frequency compensation? If we are supplying a calibration file, why not just leave all the corrections to that file, and omit R10/C12 and R13/C14 from the circuit? If the microphone would only be used in a measurement system with a calibration file installed, there would be no reason to care that the Microphone itself had significant errors in its inherent frequency response. However, we wanted to make a microphone that, in itself, was quite respectable, leaving calibration via the associated file for fine-tuning. That means you could use it with other software without calibration support and still get reasonable performance. We also wanted to make a microphone that could be used for recording, with the possibility of tailoring it for vocal and instrumental use. By including these parts, we can do both. Because our calibration files are generated with the specified frequency compensation parts installed, if you use one of our ECM capsules and calibration file, you must load the recommended parts to get optimal performance. 13mm wide PCB, but it is 99mm long rather than the 64mm of the SMD version. The two versions are shown in Figs.6 & 7. Both these boards have been made thin enough to fit in a ‘skinny’ microphone case. Neither is hard to assemble, but we reckon the SMD one is less fiddly than the through-hole version due to all the parts mounting on one side. The smallest parts on the SMD board are the SOT-23 transistors and zener diodes, which are not that hard to solder. We hand-built about 20 prototypes and, without a doubt, soldering the ECM capsule pins is fiddlier than anything on the SMD PCB. So, unless you have plenty of room to house the Fig.5: the frequency compensation for a JLI61A microphone. Here we have set the compensation (red curve) to push down the peak in its response (green curve) while limiting attenuation at high frequencies. This is not perfect, as we need to match a batch of microphone elements with these parts, but we reckon ±2dB across most of the band is a good result for a microphone. 72 Silicon Chip Australia's electronics magazine through-hole PCB, we recommend you make the effort to build the surface mount version. SMD board assembly The SMD version of the board is coded 01108231 and measures 64 × 13mm. Start by fitting the resistors and ceramic ‘chip’ capacitors. There are variations depending on whether you have a 2-pin or 3-pin ECM and what compensation components are required. If you have a 2-pin ECM, fit R14 (2.2kW, near CON2) and leave off R8 (10kW). If you have a 3-pin ECM, fit R8 (10kW) and leave off R14 (2.2kW, near CON2). The compensation components are R10, R13, C12 and C14; they are all between Q3 and ZD2. Refer to Table 2 to determine which of these you need to fit for your ECM (if you purchased it from our shop, it will come with these components). Next, mount the three transistors (one NPN, two PNP) and three zener diodes. Watch out as these are all in SOT-23 cases. If you get them mixed up, you will find a code engraved on the top of the devices that identifies each. Unfortunately, this can vary depending on the manufacturer, so you might need to check the data sheet. Still, they will probably be one of these (a question mark ‘?’ represents any letter or number): BC849C: 2C?, 49C or 8DC BC860: 9EA/B/C, 4F? or 4G? BZX84C6V8: Z5, ?61, D4P, WC or KB Failing this, you can use a DMM on diode test mode or our SMD Test Tweezers (siliconchip.au/Series/396) to find siliconchip.com.au Fig.6 (left): this is the SMD version of the PCB. Note that the values (and presence) of R10, R13, C12 and C14 are varied to match your ECM capsule. Either R8 (10kW) or R14 (2.2kW) is fitted depending on whether you have modified your capsule; for an unmodified (2-pin) capsule, leave off R8 but fit R14. Fig.7 (right): to avoid making it too much bigger than the SMD version, the through-hole (TH) PCB has parts mounted on both sides. In most cases, the solder joints are still accessible should you need to make changes or repairs. It is the same width as the SMD version but about 50% longer, meaning it won’t fit in the inexpensive plastic case described in the article. the base/emitter pins of the devices. With the single pin at the top, the base will be at lower left and the emitter at lower right. If you get a ~0.65V reading with the red probe on the left, it’s an NPN transistor (BC849), or on the right, it’s a PNP transistor (BC860). If you get neither, it’s likely a zener diode. They will give a similar reading with the red probe on the lower left pin and the black probe on the top pin (that forward-­biases the zener diode). The three remaining SMDs are the three non-polarised 1µF electrolytic capacitors. These come in metal cans mounted on plastic bases. Like polarised electros, the bases have chamfered edges on two corners that normally indicate the positive end. Because they are not polarised here, it doesn’t matter which way around you mount them. Since two of these capacitors could be polarised types, we’ve left polarity markings on the PCB, but we’ve specified all three as NP caps to make things a bit easier. In terms of components on the board, that just leaves the two throughhole capacitors, which are both 100μF parts but with different voltage ratings. Solder them laid over on their sides, as shown in our photos, so that the assembly will fit in a small-diameter tube. The striped negative end must go towards the bottom of the PCB, with the longer positive leads to the pads marked with + symbols. Through-hole assembly The through-hole version of the board is coded 01108232 and measures 99 × 13mm. This can be assembled as usual, but it’s easier to fit all the components on one side (ideally the top side) before starting on the other. Fit the axial parts first (resistors and zener diodes, watching the zener diode’s cathode stripe orientation), then the MKT and ceramic capacitors with some laid over, as shown in Fig.7. Leave the electrolytic capacitor off initially to provide better access to the remaining solder joints. Table 2 – microphone capsule calibration component values Manufacturer Part R10 C12 R13 C14 Panasonic WM61A (AE) N/A N/A 100W 5.6nF Panasonic WM61A lot 4A14 N/A N/A 100W 6.8nF JLI JL61A 220W 12nF 2.2kW 15nF JLI JL60A-V02 220W 12nF 10kW 6.8nF CUI Devices CMC-6027-24T 220W 18nF 3.9kW 18nF CUI Devices CMC-6027-24L100 220W 18nF 3.9kW 18nF CUI Devices CMC2742PBJ 820W 4.7nF 2.2kW 8.2nF Kingstate KECG2740PBJ 10W 12nF 3.9kW 6.8nF Kingstate KECG2742TBL-A 100W 8.2nF 3.9kW 6.8nF Primo EM258 N/A N/A N/A N/A siliconchip.com.au Australia's electronics magazine Refer to the section above regarding which of the optional resistors and capacitors to install (R10, R14, C12 & C14). Next, fit the transistors as shown, pushing them fully down before soldering and trimming their leads, then flip the board over and solder the axial components (resistors & zener diode) on that side. Again, see the section above for what to do about R8 and R13. Follow with the single 1µF MKT on this side of the board, laid over, then the two electros, laid over and orientated as shown. Note that the 100µF 50V electrolytic capacitor is specified in the parts list as having a maximum diameter of 8mm. A 47µF 50V electrolytic capacitor is also fine to use, as long as its 8mm in diameter. Finally, flip the board back over and fit the last electrolytic capacitor (100µF) on that side. Capacitor selection Like the other low-value capacitors What if your phantom power is <48V? Phantom power for microphones is an old standard. Like many standards, it is not particularly well followed. Most phantom power systems operate at 48V. For 48V, your preamplifier/mixer will have 6.8kW series resistors from the 48V supply. However, if it has a 24V supply instead, they will be 1.2kW, or 680W for a 12V supply. R12 should be 5.6kW to suit 48V systems or 1.5kW for systems delivering 24V DC bias or less. Our calculations show that the Mic will work with 12V & 24V DC supply systems with R12 set to 1.5kW. August 2023  73 The SMD (left) and through-hole (below) versions of the Calibrated Measurement Microphone shown enlarged. Both have their XLR sockets fitted. (<1μF), the compensation capacitors, which range from 4.7nF to 18nF (if present) must be plastic film (eg, MKT) types for the through-hole board or NP0/C0G ceramics for the SMD board. Don’t be tempted to use cheaper X5R, X7R or Y5V ceramic capacitors. They have a high voltage coefficient and thus are highly non-linear; definitely not what we want as part of a filter network! The microphone housing Regardless of which PCB you’ve assembled, the remainder of construction proceeds in much the same manner. The connection to the XLR socket will depend a lot on the approach you have to construction. In many cases, you can push the PCB between the XLR pins and simply solder the PCB to the pins directly. How this fits depends on your chosen connector and how you house the PCB. If you are using a metal housing, add a wire link from the PCB ground pin on the XLR to the housing. We want the ECM insert in ‘free space’ and with minimal reflections to get flat performance. All the ECM inserts we recommend are 6mm in diameter. We will present two ways to achieve the required mounting, one based on metal pipe hardware and the other using plastic pen cases. Photo 1 (below) shows the collection of metal parts we used to build our Microphone, while Photo 3 (overleaf) shows the parts to make the plastic version. How you go about this comes down to what you can find in your shed and parts drawer. The three key goals are: ● We want the ECM insert mounted at the end of a 100-150mm tube that it just fits inside. ● We want a section that can house the PCB. Both PCBs are just under 13mm wide, but the electros are quite thick, so a tube with an inner diameter of 18-20mm is ideal. ● We want an XLR connector at the other end. If you have a vocal or musical instrument application, you might take an alternative approach to the housing. Copper housing We used a K&S #9825 brass tube for the ECM, which is 7mm outer diameter with 0.45mm wall thickness. An alternative is K&S #8132 brass tube, which is 9/32 inches (7.14mm) in diameter with 0.014-inch (0.36mm) wall thickness. These are available from hobby shops in 305mm lengths for about $7, enough to make two or three microphones. The challenge is to expand from the 7mm tube to the 20mm or 3/4-inch (19mm) tube that houses the PCB and XLR connector. You will likely find your own approach by looking through your parts bin. We adapted between the two different diameter pipes by first using the backshell from an Altronics P0192 RCA socket, which the brass tube just squeezes into, then fitting this to the small end of a 15mm to 20mm copper capillary adaptor. This might sound complicated, but it is not hard; Fig.9 and the photos show how it came together. The SMD version of the PCB fits into the 20mm tube easily; the throughhole version is no wider, but it is quite a bit longer. In more detail, the 7mm tube was a tight push-fit into the RCA backshell. We then wrapped the backshell in 1mm bare copper wire, making it a tight fit into the 15mm to 20mm reducer. Because these parts are all copper and brass, we simply soldered them together. There are many ways to do this, but after some thought, we assembled the parts using liberal amounts of solder paste (see Photo 2) and baked it in our reflow oven at 230°C for a few minutes. You could use any oven you don’t cook food in. We also successfully made microphones using a butane torch to heat the parts and literally soldered them using regular solder wire. We won’t present exact instructions here, as your parts will likely vary. Some ingenuity and finding surplus or recycled parts from your shed will save you a lot of money and hopefully be a fun challenge. The key parameter is that you adapt the XLR section to the 7mm tube 100-150mm long. Photo 1: we made our ‘high-end’ microphone housing from a 150mm length of 7mm brass tubing with a collection of copper pipe fittings, 3/4-inch (19mm) copper pipe and an XLR male-to-male adaptor. 74 Silicon Chip Australia's electronics magazine siliconchip.com.au Tuning your microphone response Photo 2: we pushed the 7mm tube through the RCA backshell, which was a tight fit. We then wrapped 1mm copper wire around this, which makes this a close fit to the 20mm to 15mm capillary reducer. The grey substance is solder paste. The assembly process was to pull the microphone wiring through the 7mm tube, with the ground and output wires soldered to the ECM capsule (see Fig.8). We felt confident nothing would short, so we simply tacked the tips of the hookup wire to the pads/ pins on the ECM capsule. At the plug end, we snipped the microphone wires off about 30mm past the opening and connected them to the PCB. The green (ground) wire goes to the ground pin, and the black wire (microphone output) to the middle pin. We then wrangled the wires into the microphone housing, and once everything was lined up, we fixed the plug to the housing. If you are using the Altronics XLR male-male adaptor, it is a simple matter of pushing the board in until the Fig.8: how to wire up a regular two-wire ECM (left) and modified 'Linkwitz' three-wire ECM (right); note the differences in R8 & R14. The arrangement is the same for the through-hole board. Our goal with a measurement microphone is a reasonably flat response before calibration and a flat response after calibration. If you purchase a calibrated ECM capsule from the Silicon Chip Online Shop, we will provide the necessary parts to load for response tuning. You will also get a calibration file, giving as close to a flat response as we can achieve with our equipment. Alternatively, you may want to tailor the response of your Microphone. In that case, you can download an LTspice model from the Silicon Chip website (associated with this article). This can be used to model your response while varying the tuning components. The following is a general guide to tweaking the response: ● C12 and R10 provide control over high frequency gain, with C12 setting the corner frequency. C12 increases the gain with frequency by reducing the emitter resistance, which is initially 1kW. R10 allows the ultimate gain of this combination to be set. Conceptually, if R10 is set to 1kW, then, at very high frequencies, this results in two 1kW resistors in parallel for a final gain of two times or 6dB. ● R13 and C14 set the gain roll-off at high frequencies. While these go to ground, they are effectively in parallel with the 2.2kW collector resistance. This is reduced by R13, which directly reduces the gain of this stage. R13 sets the ultimate attenuation of this stage, and C14 the corner frequency. You will also find the gain model in our “Analysis.odt” spreadsheet. While this is simpler to work with than LTspice, this spreadsheet is very much an engineering tool, so use it with caution. While the concept of how R10, R13, C12 and C14 interact is simple, getting the response you want can be tricky. The values shown in Table 2 are what we found to be effective with batches of capsules we purchased. These will be a good starting point for you to experiment if you have the ability to check your calibration. Reflowing the solder on the enclosure can be done with any regular oven by baking at 230°C. However, you shouldn’t use an oven that you cook food with. The final result is shown in the photo on the right. Fig.9: we used an Altronics XLR adaptor for the plug, which is a decent fit into a 20mm diameter copper pipe. We then used a capillary reducer and RCA socket shell to adapt that to the 7mm brass tube for the ECM. They came together very well with a few shims and some solder. siliconchip.com.au Australia's electronics magazine August 2023  75 A close-up of the interior wiring required for the microphone. Photo 3: the very inexpensive microphone housing is made from a whiteboard marker and Biro pen case. An epoxy glue (Araldite) was used for the XLR housing joint. screw hole in the plug lines up and inserting the screw. You are then ready to go. Plastic pen based housing As mentioned at the start, a major driver of this design was to keep the cost low. Copper pipe is great if you have off-cuts in the shed, but buying it is pretty expensive. So we looked for a cheap and accessible way of mounting the 6mm capsule at the end of a thin tube, and something suitable for housing the electronics. During one of the author’s less lucid moments, likely due to ingesting an How we generated calibration data for hundreds of ECMs Our calibration process generates a calibration file for Speaker Workshop that allows us to measure the error of an ECM capsule from a flat response. To do this we: ● Measure the SPL of a speaker at an exact location relative to that speaker using our calibrated Dayton EMM-6 microphone (without its calibration coefficients). ● Subtract the calibration coefficients for our Dayton microphone from the measured values and export the result as a “CAL file”. Using this as this synthetic calibration file, we will generate the calibration correction file for the connected microphone if we measure at the same location. We verified that this worked by running a measurement on the same Dayton EMM-6 microphone and confirmed that it produced the expected calibration values. We can then substitute our ECM capsules, and providing we get them in the exact same spot, generate suitable calibration files for those capsules. By labelling each ECM with a number that matches the file saved, anyone who purchases that module can find and use the calibration data we generated. We made a special spring-loaded jig that allows ECM capsules to be popped in and measured easily, speeding up this process. We also created a simple jig to ensure we always made the measurements at the exact same location relative to the speaker. 76 Silicon Chip Australia's electronics magazine unhealthy amount of coffee, the seemingly silly idea of using a mix of plastic pens popped into his head. He found some cheap Biros at Officeworks and some whiteboard markers that, with a bit of drilling and gluing, made an inexpensive microphone housing. If you use whiteboards (eg, at work), you will likely have a ready supply of dried-up markers. The SMD version of the board fits in these perfectly, although the through-hole version is too long. Even better, if you take an Altronics P0823 XLR plug and throw away all but the plug section, it fits perfectly into our whiteboard marker case, as shown in Photo 4. The assembly process is similar to that for the copper tubes but quite a bit easier. First, strip the whiteboard marker apart and clean it out. Cut the tab off the XLR connector with side cutters to allow you to solder to the An example setup of the Measurement Microphone with our previous projects, the Super Codec and Loudspeaker Test Jig. siliconchip.com.au Parts List – Calibrated Measurement Microphone SMD version – electronic module The XLR socket wiring on the SMD version of the Microphone. PCB. Cut the top of the whiteboard marker off and drill the end so you have a tight fit for the Biro tube, then fix the Biro in place with super glue. See the first and last pages of this article for the final result. Testing and using it Using the Calibrated Microphone should be as simple as plugging into a microphone preamplifier that supplies phantom power. We suggest that you check it out before gluing the case shut. If you don’t get a signal on power-up, here are some things to check: 1. Check your solder joints and that you have the PNP and NPN transistors and zener diodes in the right places and with the correct orientations. 2. Apply power by plugging it into the preamp or providing 24-48V DC from a power supply with equal resistors in series with the Hot and Cold (+ 1 double-sided PCB coded 01108231, 64 × 13mm Semiconductors 2 BC860 45V 100mA low-noise PNP transistors, SOT-23 (Q1, Q2) 1 BC849C 30V 100mA low-noise NPN transistor, SOT-23 (Q3) 3 6.8V ¼W zener diodes, SOT-23 (ZD1-ZD3) [BZX84C6V8] Capacitors (all SMD M2012/0805 50V X7R unless otherwise noted) 1 100μF 50V radial electrolytic (maximum 8mm diameter) 1 100μF 10V low-ESR radial electrolytic 1 10μF 16V X5R 3 1μF 50V non-polarised SMD electrolytics, 4mm diameter [Altronics R9600; Würth Elektronik 865250640005] 2 2.2nF 5% NP0/C0G 2 1nF 5% NP0/C0G 2 470pF 5% NP0/C0G Resistors (all SMD M2012/0805 size 1%) 2 150kW 1 100kW 1 39kW 1 10kW 1 5.6kW 2 2.2kW 1 1kW 1 330W 2 47W Through-hole version – electronic module 1 double-sided PCB coded 01108232, 99 × 13mm Semiconductors 2 BC560 45V 100mA low-noise PNP transistors, TO-92 (Q1, Q2) 1 BC549C 30V 100mA low-noise NPN transistor, TO-92 (Q3) 3 6.8V 400mW or 1W axial zener diodes (ZD1-ZD3) [eg, 1N754] Capacitors 1 100μF 50V radial electrolytic (maximum 8mm diameter) 1 100μF 10V low-ESR radial electrolytic 1 10μF 35V radial electrolytic 3 1μF 63V/100V MKT 2 2.2nF 63V/100V MKT 2 1nF 63V/100V MKT 2 470pF 50V C0G/NP0 ceramic Resistors (all axial 1/4W 1%) 2 150kW 1 100kW 1 39kW 1 10kW 2 2.2kW 1 1kW 1 330W 2 47W 1 5.6kW Copper-housed version 1 assembled electronic module (SMD or through-hole) 1 ECM capsule with calibration components [Silicon Chip SC6760-5] 1 60mm length of 20mm or 3/4-inch diameter copper pipe 1 150mm length of >6mm inner diameter brass tube (eg, K&S #8132 brass tube) [hobby store] 1 20mm straight capillary coupler [Bunnings 0252161] 1 20-15mm reducing capillary coupler [Bunnings 0252162] 1 RCA backshell [Altronics P0192] 1 XLR male-male adaptor [Altronics P0972] 1 200mm length of 1mm diameter bare copper wire (stripped from some spare solid-core mains wire) 1 300mm length of two-way ribbon cable or light-duty figure-8 Plastic pen-housed version 1 assembled electronic module (SMD version) 1 ECM capsule with calibration components [Silicon Chip SC6760-5] 1 whiteboard marker [Officeworks] 1 ball-point pen with unscrewable ends [Officeworks] 1 XLR plug [Altronics P0823] 1 300mm length of two-way ribbon cable or light-duty figure-8 siliconchip.com.au Australia's electronics magazine August 2023  77 What is this “Linkwitz Mod”? Most Electret Condenser Microphones use a FET in a common-source configuration. In this arrangement, the source is connected to the capsule case, and the 2.2kW resistor in series with the drain is the load across which the output voltage is generated. Linkwitz realised that if you can cut between the FET source pin and ground (a track that is accessible on the outside of the capsule), it is possible to rearrange the circuit as a source follower. This gives less gain but a lot more headroom. We tested it using our mics and found that all the frequency correction parts remain valid. This modification is very fiddly indeed, and it is easy to kill a mic doing this. We feel this is for ‘power users’ and something you might try once you are confident in making measurements. There are various references on the internet regarding this. A good place to start is at Siegfried Linkwitz’s own web page: www.linkwitzlab.com/images/ graphics/microph1.gif Assembled Calibrated Measurement Microphones in both the copper and plastic-type housings. Kits & Capsules SC6755 SMD Kit ($22.50) Includes the PCB and all onboard parts besides the XLR socket. and −) lines. Use 6.8kW for a 48V supply or 1.5kW for 24V. With this applied: a. Check the voltage on the microphone side of the resistors. This should be well over 10V, and the voltages should be about equal. If not, check for shorts and correct part locations on the board. b. Check the voltage across the power supply zener diode, ZD2. It should be close to 6.8V. Check the voltage at the collectors of Q1 and Q2, which should be well above 10V. If not, check the base voltages of these transistors. Also verify that each has a 0.6V base-emitter voltage drop. c. Check that you have installed R14 fitted (or R8 in if you’re using a “Linkwitz Mod” on the ECM) but not both. d. Check the voltage at pin 2 of CON2, the ECM output for two-wire mode. This should be somewhat less than 6.8V, and if you look with a ‘scope, you should be able to see the microphone signal. If not, check that you have the ECM connected the right way around. Also check for shorts on the capsule. e. If you still have no signal, but the DC voltages at the input and capsule are OK, check the voltage at the base of NPN transistor Q3. This should be about 1.9V, and the voltage on its emitter about 1.3V. The voltage at its collector should be around 3.9V. If these don’t make sense, check that you have the right transistor in the circuit. Using the calibration files Calibration files for all the ECMs we sell are available for download from the links in the ECM shop items. Your ECM will come in a bag with a number on it. Download the file for that specific type of ECM, then look for the files tagged with that number. The calibration files match specific capsules. You cannot use them for similar microphones and expect a great outcome. The file with the FRD extension, starting with your ECM serial number, is in the Speaker Workshop format. You can import it into Speaker Workshop and select it as the microphone calibration. This file contains 4096 rows with Frequency, Gain and Phase figures (the Phases are set to zero). Load this, and you are all set! 0dB in the calibration files equals -40.3dBV/Pa. Given that 1Pa is 94dB SPL, that means that 0dB is 53.7dB SC SPL. Happy measuring. SC6756 Through-Hole Kit ($25) Consists of the PCB and all onboard parts besides the XLR socket. SC6760/1/2/3/4/5 ECMs ($12.50) See Table 1 for the various options. Each comes with the required SMD compensation components, as shown in Table 2. If building the through-hole version, you can source the compensation components (resistors & MKT or greencap capacitors) from Jaycar or Altronics. 78 Silicon Chip Photo 4: the SMD board fits a treat into the whiteboard marker case after it has been stripped apart and cleaned out. The XLR connector will need the tab cut off with side cutters to allow you to solder to the PCB. Photo 5: the assembled Biro-cased Microphone, ready to have the ECM pulled in and glued to the tip. Australia's electronics magazine siliconchip.com.au