Introduction: True Condenser OPA Mics
Let's build a couple of exceptional quality condenser microphones. These are true condensers as they are externally biased. They use an Operational Amplifier, (Op-Amp or "OPA") based impedance converter circuit and a hex inverter voltage multiplier for creating the bias voltage. One is your basic cardioid and the other uses a dual diaphragm that lets us choose the response pattern. But we are doing that differently than most microphones out there.
This Instructable is the culmination of a lot of microphone research, building, unbuilding, a few steps forward with a step or two back. Sound and recording have always fascinated me and as a kid, while my friends got guitars, I built a synthesizer. As their musicianship progressed, I was drawn to the tech and the mixing board. My first “real” microphone was a RadioShack PZM followed by a Sure SM57. Electronics always fascinated me and I built and designed a lot of analog studio gear through the 1990’s. All with the mindset of keep it simple, keep it quiet, and keep it clean. And with the thought of: Can I improve something a bit or do something different?
Then of course the digital world took over. I embraced it big time and knew it was the future for audio. I watched an album in 1993 being mastered to an Apple computer with outboard A/D converters. It was mixed down from a Studer 2 inch 24 track via an analog mixing board. I knew things were changing. We had a 2 track reel to reel running as well, but never needed to use it. I still feel that getting the audio into the digital realm as soon as possible is the best solution. Not that there aren’t very cool analog things still out there, and tube guitar amps will always exist. OK, I digress. Back to the: Keep it simple, clean, quiet, and oh yea, innovate.
That brings us to the two microphones we are building today. They are “true condenser” microphones that use transound capsules, the TSC-1 and TSC-2. Both are center terminated, meaning there is an electrical connection and wire right in the middle of the capsule. The other electrical connection is the backplate, which is the brass metal ring in the middle of the mic capsule. These are similar to the Neumann K87 capsules. The other major style is the CK12. Those are edge terminated. See the photos. The big difference between the TSC-1 and the TSC-2 is that the TSC-2 allows you to get a signal from both sides of the capsule. Normally that is used with the internal microphone electronics to allow you to change the pattern of the mic from Omni, to Cardioid, to Figure 8. We will be doing that, but not internal to the microphone. We are going to take both signals out of the mic body, allowing us to create whatever pattern we want in post. That lets us do some very interesting things. Both microphones use the OPA impedance converter I layed out in a previous instructable. For the TSC-2 we have a new board. This puts two of the OPA circuits on one board with a common ground. This lets us bring both signals out of the mic with a 5-Pin XLR.
UPDATE!!!! I have worked with JLI electronics to have fully populated surface mount versions of all three circuit boards available on their website. See here: https://www.jlielectronics.com/diy-accessories/
And now, JLI carries full kits that include everything you need to build these including a pre-populated PCB and a really nice body. If you can solder a few wires, you can build one!
Step 1: A Brief History of Microphone Capsules and Global Supply Chains.
There are plenty of great documents out there on the history of microphone capsules. See this site. Georg Neumann, yes the Georg Neumann, invented the multipattern mic with his team’s M7 capsule. This was 1948. So it has been a while. The basic design for both the center terminated and edge terminated capsules have been around a long time and both are great with subtle differences in sound. See Matt’s great description at Microphone Parts Without getting into the internal mechanical details of how they are machined, just know that there is a lot of precision required. And to repeatedly manufacture them well requires a solid quality system in place and sound manufacturing practices. I mention this as to why I chose these particular capsules. Globally there are several companies that either build directly or source condenser capsules. Here are some really good ones in no particular order:
Beesneez, Australia, Beesneez
Peluso Microphones, US, Peluso Mic Lab
Microphone Parts, US Microphone-Parts
Telefunken, Germany (repair parts) Telefunken
The prices vary and all the Telefunken ones list as unavailable. There are also a lot of very inexpensive ones on ebay and aliexpress, none of which I can vouch for. I can for the ones above. They are all good and you are free to use them as they will all work with the rest of this project.
I found the ones I am using today through the same company that makes the TSB2555B’s and other electret capsules used in my previous instructables, Transound. They are a Taiwanese company that is 20 years old and has a very robust quality system. As an OEM, they make a lot of microphone capsules annually. A year ago I discovered their external bias capsules and ordered a couple TSC-1’s. After a great build and some brainstorming, a couple TSC-2’s. Apart from their great sonic qualities, the fact that the company has multiple ISO certifications and has a history means that these should be available for a while. They are used in multiple commercial microphones as well. For publishing DIY projects, being able to get the parts is really important to me. I still get emails and comments from my original Modify A Cheap Microphone instructable because people are still building it. The standalone through hole FET transistors used are still available but are fading fast. There is a term used by engineers who have been around a while called “Unobtanium”, meaning that something, usually something critical, is impossible to get. That is the nightmare of service and sourcing teams everywhere. Here is my audio story of that: In my peak of analog design years a company called SSM (Solid State Music / Solid State Micro Technology for Music) existed that made great analog synthesizer chips. They were later acquired by Precision Monolithics, which was then bought by Analog Devices. Precision Monolithics released a dual dynamic range processor, the SSM2120 in 1990. It was a dual VCA with log level detectors built in. In short, it was the building block for a fantastic compressor or downward noise expander/gate. I was connected with the product manager for it, had engineering samples up front and helped design the reference circuits for the data sheet. I built many of them and put them into every channel of an analog console in Central Florida used by, among others, Deep Purple in the early 90’s. I also licensed my compressor design to PAiA electronics as a kit and wrote several DIY articles on how to build them. Then in the mid 90’s, they went obsolete. No notice. This was right about the time you could easily do compression digitally with a DSP chip. I mention this as I am very sensitive to supply chains and parts availability. Do not take them for granted.
The other thing we need to turn a capsule into working microphones is a bias voltage generator. All my other mic building Instructables have used electret capsules, meaning that an internal charge was supplied by the material and construction. In this case we need an external voltage to supply that charge. We are going to use a hex inverter voltage multiplier to take 12 VDC from the OPA boards and turn that into 80 VDC. Additionally the saddle to hold the capsules is 3D printed and available from Shapeways. The STL file is also included here for those that have access to a 3D printer.
Step 2: Theory and How It Works Part
These are condenser microphones. The heart of this is a capsule with a fixed backplate and thin flexible diaphragm that is “metalized” so it will be conductive. These form a capacitor typically in the 10-60 picofarad range. Smaller capsules, less capacitance. The flexible diaphragm will move due to sound waves in the air. Not a lot of motion, but enough that if there is a voltage applied to the capsule to charge it, we can pick up the changes in the voltage due to the sound waves moving the diaphragm. That is how a condenser microphone works. There are internal mechanical characteristics involved with them as well, but that is beyond the scope of where we are going today.
OK, how much voltage is needed? That is a great question that I set out to answer. Traditionally (as in today) capsules are typically 60 VDC. Historically some older tube microphones used 100 VDC. Brüel & Kjær use 200 VDC or more on some of their small ½” capsules. Changing the voltage does a few things. It changes the sensitivity of the microphone by the ratio of the voltage difference, larger voltage, more sensitivity. It can also change the frequency response by stretching the diaphragm. Brüel & Kjær goes into detail in their handbook here. RØDE microphones use 80 VDC in a few of their mics. For these mics we are using 80 VDC, giving us a bit more sensitivity. With the OPA circuit we are using we created a "virtual ground" meaning that it is low impedance to function as a signal return path, but it isn't actually at Zero volts. it is about 5.5-6V depending on the zener diodes actual voltage. Even though the zener is rated at 12V after filtering etc. we are down around 11.5 ish. Our virtual ground is 1/2 of that. End result the bias voltage on the condenser is about 75.
Traditionally this is provided by a small transistor oscillator using a couple of small coils and capacitor diode voltage doublers. I was trying to stay away from through hole transistors and definitely coils. I knew a couple mics out there, CAD in particular were using CMOS inverters to drive the voltage multiplier circuit, so that is what I decided to do.
We are doing this with a hex inverter using a 4584 or a 40106, either will work. The key specs are that they are Schmitt-Trigger and that they will work with 12VDC or higher. A “Schmitt Trigger” means that as the input changes on an input, the inverter will change state as you cross a certain threshold and it won't change back until the threshold is exceeded in the opposite direction. This provides hysteresis and provides a nice clean square wave for us. Both data sheets explain how this works. We are using one of the inverters as an oscillator with a 1nF capacitor and a 10K resistor. This gives us about 100Khz -ish for a clock frequency, well above the audio range. Then the next five inverter stages are used as voltage multipliers. Doing this with a CMOS inverter was invented in the 1980's. See this for background.
Here is how this works. One end of our diode string is attached to +12VDC. Assume that the output of the first stage is ground. The capacitor for that stage starts through the diode connected to the supplied voltage. When the output of the inverter stage changes from ground to 12 volts on the next clock cycle, that raises the “ground side” of the capacitor up by 12 volts. Assuming that the capacitor is fully charged, that brings the positive side of the capacitor to 24 volts, which will then conduct through the next diode to charge the next stage capacitor to up to 24 volts. Now as we move down each stage, on opposite clock cycles we are adding about 12 volts to each stage. The beauty of all this is we need negligible current. Just enough to apply a bias voltage to the capsule. After six stages of adding 12 volts we end up with 72 volts on top of our original 12 or about 84V max. Following the inverter stages we have an RC filter. This cleans up any ripple or noise left from the process. A 1M resistor supplies the voltage to the capsule. Taking into account voltage drops across the diodes, and other fun things, we end up with about 80 supplied to the capsule. The original 12 volts is supplied by the OPA board from the zener regulator circuit. Both the original OPA board and the new two channel one bring this out to a connection point. Interestingly, there are some mics out there that actually use the incoming phantom power to directly supply the bias voltage. In a perfect world, that would be 48 volts, which doesn't sound bad. The reality is that the voltage is always lower because the internal electronics draw power and lower the voltage supplied. Most are about 34 volts. Remember, the 48 volts is in series with two 6.81K resistors, one to pin 2, and one to pin3, of the input preamps XLR connector.
Now onto the impedance converter. We are using the same one from this instructable for the TSC-1 microphone and a two channel version for the dual diaphragm capsule. I can't stress enough how good these are and how well they perform. There are two functions that the internal electronics need to perform. The circuit has to have a very high impedance so that it does not load down the capsule. This was originally done via a tube and then later with a fet. Then it needs to be able to drive a long microphone cable while not affecting the signal. This circuit achieves that really well. Interestingly, the Operational Amplifier or "OPA" we are using, the OPA1642, can swing rail to rail so the capsule will distort long before the internal electronics. The noise floor of the circuit makes it less of a factor than the capsule self noise, and overall the electronics have about a 130db dynamic range (if I put on my marketing hat). One small change in components. On the OPA boards we are using wima polyester .1uF capacitors now. I originally spec'ed a MLCC one that can exhibit a “microphonic” phenomenon. This was pointed out to me and while they can, I’ve never noticed it and decided to experiment. I put one in place of the capsule in my OPA Alice build. If you tap it with a small screwdriver, you can get some sound. Real world: You would have to drop the microphone or hit it to have any effect, and that will be swamped by the dropping noise anyway. But, for the sake of argument, I replaced them in the BOM for this build. They are 61 cents each vs 41 cents for the old ones from Mouser. They are also used for filtering and not in the audio path either so I am most certainly not retrofitting anything I have already built. :-)
Step 3: Lets Build Them!
The final thing we will need to build the mics are donor bodies. I wanted something nicer than a BM-800 body for these and found something called a “mini U87” on Aliexpress. I bought a pair based on a recommendation of a friend and was very impressed. The first of two orders were machined very well and made out of some kind of copper/bronze alloy. So took all the paint off and hand polished them giving me a steampunk retro look. I built two of each and then realized I can do something really cool with three of the dual capsule one’s so I went to order another body and… They were discontinued. I found something that looked similar and ordered it. OMG, supply chain at its finest here: The body is great and physically identical to the original orders machining included, except it was made of a silvery metal I am assuming is a zinc based alloy. And it came with a mic mount that looks just like the original Neumann U87… Except it doesn't fit the mic body. No worries for what we are doing. You can use any donor body from the BM-800 to some of the more inexpensive MXL microphones. Enough already, let's build them!
Parts List: All the PCB’s are available to order from PCBWay. They do a fantastic job and are very inexpensive. One note on color selection. In the past I ordered red, blue and green, which are all great with white silk screen. My first batch of the Dual Channel boards I picked yellow. Because I thought it might look cool. It really doesn't, it really isn’t yellow, and it is hard to read the white silkscreen with it. Stick with one of the primary colors. I do recommend a different color for each board type so they are easy to tell apart at a glance. Huge shout out to Homero Leal who laid these out and let me upload them. We are working on fully populated versions if there is interest. And my only request here. If you order the boards please consider tipping us. We make no money off of these and it helps buy the next build.
PCB’s: Hex Inverter Multiplier Hex Inverter Multiplier
Single Channel Dual OPA Circuit (For the TSC-1) Single Channel
Two Channel Dual OPA Circuit Dual Channel
Complete BOM is attached with a separate sheet for each PCB. Mouser part numbers are included.
Update September 16th 2021: All circuit boards are available from JLI Electronics fully populated with surface mount components. https://www.jlielectronics.com/diy-accessories/
You can now build these with minimal soldering!
We will need a five pin XLR insert for these and the easiest solution I have found (as in least expensive) is to buy one of these and disassemble: This
For the dual output mic, a breakout cable is needed:
10 ft of dual mic cable
White and Red XLR boots Select the color when adding to the order
The TSC-1 Capsule (single output Cardioid)
The TSC-2 Capsule (dual output) Donor Body - See Text
K87 saddle, 3D printed
Step 4: Final Construction and Notes
Some final construction notes:
If you have any buzz or handling noise, please ensure that you sand the surfaces of the body parts that come into contact with each other. Usually the offending part is the body cylinder, specifically the ends. If there is paint on the end, it won’t conduct so it wont connect to ground.
The OPA board really needs to be clean, especially around the 1Gig resistor and the OPA itself. I use water based flux when I solder and clean it off with dish soap and water and a small scrub brush like a toothbrush. Then rinse with isopropyl alcohol and thoroughly dry.
If you don’t get 80VDC at the 1M resistor and diode junction at the output, check the diode polarities. I had one backwards on a build and it caused some weirdness. The Hex Inverter board also needs to be really clean as well. One of the downsides to water based flux is when damp, it conducts. So it is affected by humidity.
UPDATE: Added Troubleshooting Guide March 2022
Step 5: Testing and Usage
I powered mine up with headphones and ensure that you get output from both sides of the capsule. With both inputs panned center, turn the preamp gains all the way down and bring up just the front capsule while talking into it. Get close and intimate and you should hear the proximity effect. Now slowly raise the rear capsule level and as the gains become even, the mic becomes a full omni and the proximity effect will go away. Keep raising it and you will start making it cardioid, but with the rear capsule dominant as it will be out of phase with the front based on it facing away from you.
Here are the general patterns that microphones can have:
To make them in post you record both channels and then do this:
For Omni: Equal mix of Front and Rear.
For Subcardioid: Mix in less of the Rear.
For Cardioid: Just the Front. In fact, you can just plug the front XLR in and use it standalone as a Cardioid microphone.
For Figure 8: Equal Mix of the Front and the Rear with the Rear polarity inverted.
For Hypercardioid and Supercardioid:
I made this a pdf that is vector so you can zoom in for detail. Now here is the best part: What if you are using the mic as a figure eight and didn't quite get the null spot for the sound you don't want? Well if you look at the graphs for figure 8 you see the null is exactly at 90 and 270. Of you look at the Hyper Cardioid, the null is about 30 degrees more to the back of the mic. Just by adjusting the level up and down of the rear, you can steer the null spot in post. How cool is that?
Step 6: Wrap Up
Thank you to Homero Leal for the PCB layouts, Tom Benedict for the capsule holder CAD design and Kady Speeks and her voice for being my microphone tester.
I'm not the first person to think about taking both signals out of the mic simultaneously. Here are links to commercial ones and their stories along with some great reading material. To me we are getting to the point where it is harder for a DIY guy to sort out the mechanical switches etc to make the mic pattern selectable than it is to just get it all out of the mic. The cost and quality of multichannel interfaces has dropped dramatically along with the storage and editing power of today's computers. My Microphone Parts RK12 built has an internal switch for Omni and Cardioid. I have to unscrew the mic body to get to it. And then remember that I left it in Omni when I can't figure out why the vocals are not right… If I am doing my math right these builds have about an 8 or 9dB of self noise, truly world class. I’m working on a test setup for this to verify and will update. Just as a Honda and Toyota drove changes in manufacturing quality for cars, that same progress has happened across the boards in most mechanical things. Check out a TSC-1 or 2 capsule. You will be impressed.
Participated in the
Anything Goes Contest