Ambi-Alice a First Order Ambisonic Microphone

by DJJules in Circuits > Audio

19805 Views, 115 Favorites, 0 Comments

Ambi-Alice a First Order Ambisonic Microphone

AmbiBuildCover-3 copy.jpg
Ambi-Alice-1.jpg

An Ambisonic Microphone records audio in multiple directions allowing you to change the direction of where you are listening after recording, sometimes while you are listening. There are several of these on the market but we are going to build our own. Construction uses a 3D printed housing, very good mic capsules and deceptively simple electronics. You can choose two mounting methods, one will mount on a traditional mic strand using a mic clip. The second integrates a ¼-20 bolt into the housing allowing you to use alternative mountings such as GoPro mounts and Cold Shoe camera mounts. Both have a great eco system of options. You could use a ¼-20 nut as well, but I do not recommend it. The electronics, which use 48V phantom power are just two components, both of which will fit into the XLR connector shells. Don't let the simplicity fool you: This mic is better than some that cost a lot more. It has low self noise and sounds great. I have used it for very quiet ambience recording, a choir and loud traffic noise. Listen to them all in the final video at the end.

Supplies

Project Files are: here.

All the rest of the supplies and materials are in each section as they apply

A Brief History of Ambisonics:

TetraPatent.png

The original concept was developed in the 1970’s in the UK. The first one was the Sound Field microphone. It was invented by Michael Gerzon and Peter Craven and used a tetrahedral array. This generates what is referred to as “A” format. That means it is raw four channels of audio that needs to be further processed when mixed down. The other format used is “B” format which for first order ambisonics, is encoded into XYZ and W. W is the equivalent of an omni microphone signal. X,Y and Z are figure 8 pattern responses facing, front/back, left/right, and up/down. The tricky part is sorting all of this out when you bring it into your DAW. There are two standards for B format and the only difference is in the ordering of the channels. I am using Reaper and this instructable includes templates and links to all the plugins required. Way more in the “How to use this thing” section.

“A” format: Straight unprocessed audio from the individual capsules -- Raw audio from the mic

“B” format: Encoded into “X”, “Y”, "Z”, and “W” -- Start of the Magic

The original patent for this was filed 45 years ago and what was missing, in my mind, was the ease of use part. Now we have really good portable digital recorders and audio workstations. The recent improvements have all been with processing and plug-ins on the software side. There are also mics with more than four capsules for higher order ambisonics. We are going to focus on and build a first order ambisonic microphone, the Ambi-Alice. So named in recognition of Scott Helmke, and his first truly amazing Alice DIY mic. The name fits as well here, as Ambisonics is truly a rabbit hole, that once you descend into, keeps pulling you further into the depths.

Let's Build It!

AmbiBuildDay2-6.jpg
AmbiBuildDay2-9.jpg
AmbiBuildDay2-3.jpg
AmbiBuildDay2-1.jpg
AmbiBuildDay2-2.jpg
AmbiBuildDay2-10.jpg

Here is what you will need: Four Mic Capsules. We are using TSB2590’s that are really good. They are 1” diaphragm electret condensers that also contain EMI/RFI suppression capacitors. Meaning we don't have to enclose them in a metal screen. TSB2590

Mogami Wire W2697. This is critical as it is matched mechanically to the 3D printed mic body. You will need 4 times the overall length of the mic cable you want. The beauty of using the mic close to a recorder means that they don't have to be too long. I built several mics with 4-5 foot wires along with a few that have 10-12 feet of cable.

The Simple P-48 Circuit parts:

Male XLR Connectors (4)

Colored Boots for the XLR Connectors (Red, Yellow, Green, Blue)

And the most important! The 3D printed Body. There are two options:

For the “Normal”or “Stick” mic build you will need a 6-8 inch piece of ½” PVC pipe.

Rubber Drone "Fright Controller Gimbal" :-) mounts

Multicolor sleeving (Techflex PTN0.13)

Misc Parts:

M3 Screw and Nut assortment

Supplies:

Multi Colored electrical tape

Paper clips (Binder clips)

E6000 glue (a must! It allows repairs)

Basic Soldering and electronics tools

Model paint (for color coding)

The Build

AmbiBuildDay2-13.jpg
AmbiBuildDay2-17.jpg
AmbiBuildDay2-5.jpg
Ambi-BuildCapsule-1.jpg
AmbiBuildDay2-19.jpg
AmbiBuildDay2-20.jpg
Ambi-BuildCapsule-2.jpg
Ambi-BuildCapsule-5.jpg
AmbiBuildDay2-18.jpg
Ambi-BuildCapsule-7.jpg
Ambi-BuildCapsule-4.jpg
Ambi-BuildCapsule-6.jpg
AmbiBuildDay2-22.jpg
AmbiBuildDay2-25.jpg
AmbiBuildDay2-26.jpg
AmbiBuildDay2-23.jpg
AmbiBuildDay2-24.jpg
AmbiBuildDay2-27.jpg

Here is a full assembly


Here are the steps:

Prep the wires for the Capsule end

Solder the capsule wires to the capsule, connecting the Source (“S”) and Ground (“G”) terminals in the process. This makes the internal fet into a common source amplifier.

Note that the “G” is Ground and not Gate. Internally to the capsule the gate is connected to the capsule and there is an internal leakage diode that biases the fet properly. This makes the circuit very easy for us to interface to!

Prep the Capsule Holder by painting the labels.

  • Red = FLU (Front Left Up)
  • Yellow = FRD (Front Right Down)
  • Green = BLD (Back Left Down)
  • Blue = BRU (Back Right Up)

Glue each Capsule into place in the holder

Label the end of the wire with a piece of colored tape to match the wire to its capsule

Put in the Silicone Rubber Bumpers for vibration dampening

Fed the wires through the Head Basket Bottom

You need to take the tape off to feed through, but put them back on as soon as they are through. You will not regret this

Settle the Capsule Holder into the Lower Head Basket, aligning the rubber grommets

Glue on the Upper Head Basket

Use a few paper clip/binder clips to hold it in place

Put colored sleeving over the individual wires

Optional but Recommended! Alternatively, you can use one larger sleeve for all four.

There are two paths now…

One: the GoPro Mount version

Two: the Stick Mic version

GoPro Mount

  • Carefully fit the wires into the recessed channel on the Head Basket base half and slide the other half into place.
  • Double check that none of the wires stick out of the channel or you could pinch them when tightening down with the M3 screws and nuts.
  • Using 10mm (12mm works as well) M3 screws and nuts secure the other half of the base.
  • Carefully tighten the screws ensuring no wires get pinched

Stick Mic

  • Insert all four wires through the PVC pipe and slide the PVC pipe all the way up into the Head Basket housing.
  • Glue the PVC pipe into the Head Basket base with E6000 glue.
  • Slide the wires through the strain relief base piece.
  • Glue the base piece into the bottom of the PVC pipe and let dry.
  • Secure the other half of the base strain relief with 10mm M3 screws and nuts.

Both Styles:

Put the XLR boots on the wires before proceeding!

I have forgotten this and it is a pain to un-solder and re-solder the wire to fix this!

Add the Electronics and solder the male XLR connectors

Prep the Resistor and Capacitors

  • Fold the negative lead up and back on the 3.3uF capacitor.
  • Twist the negative lead to one end of the 100K resistor lead.
  • Solder the connection and trim to just below the height of the capacitor body.

Trim the bottom of the resistor and capacitor as shown in the pictures and video.

Tin the Male XLR jack terminals.

  • Solder the bottom of the resistor to Pin 1.
  • Solder the bottom (+ lead) of the capacitor to Pin 2.

Prep the wire as shown in the picture with equal red and white wire lengths, tinned on the end and a shorter sleeve/shield twisted and tinned.

Solder the wire to the XR jack as follows:

  • Red to Pin 3
  • Shield to Pin 1

This is our ground connection for the shield and minimizes EMI/RF interference. It only needs to be connected at one end to do this.

  • White to the junction of the Resistor and Capacitor.
  • Place a small piece of electrical tape over the resistor/capacitor/white wire part.

Assemble the XLR connector

  • Place the plastic insert over the wire and sleeve.
  • Slide the XLR pin and components into the sleeve.
  • Insert the XLR pin assembly into the metal connector housing.
  • Screw on the the boot Repeat for the three remaining connectors.

Congratulations! You just built an Ambisonic Microphone! One at a time, test each channel to ensure they work properly.


Recording: the First Part of Getting It Right…

ZoomAmbi-1.jpg
ZoomAmbi-4.jpg
ZoomAmbi-5.jpg

The microphone is just the starting point. There is way more to Ambisonics than can be covered here. One of my colleagues collaborating with me on this is actively working on his doctoral thesis for ambisonic microphones.

I have mentioned how “keeping track of all of this” is the hardest part. This is why we color coded everything. I did a lot of research on this and there are two ways to do this with 4 capsules in a tetrahedral array, by this I mean there are two possible ways of selecting the positions for the 4 capsules on the cube: Here is a Reference.

FLU-FRD-BLD-BRU (Soundfield and Core-Sound microphones use this) as does the AmbiAlice
FLD-FRU-BLU-BRD (DPA-4 uses this)

Connections: So we plug the four cables into the Zoom F6 (or other recorder) like this:

  • Input 1 = Red = FLU (Front Left Up)
  • Input 2 = Yellow = FRD (Front Right Down)
  • Input 3 = Green = BLD (Back Left Down)
  • Input 4 = Blue = BRU (Back Right Up)

Settings on the ZoomF6:

Without diving into the manual and quoting it (It’s on Pages 104 - 108) Link the first four channels via setting the link mode to Ambisonic Set the Ambisonic mode to Ambisonics A Yes, there are encoders in the Zoom. We are not going to use them.They work but don't know the specifics of the microphone. And, most of all, once you convert to AmbiX or FuMa, there is no going back. In my mind it is like trying to mix a recording live. Yes you can get close, and maybe even a great mix.But, it is so much easier and better to do it post recording. Set the Mic position to whatever position the mic is while recording. I almost always have the mic upright. Don’t forget the turn phantom power on and to 48V for inputs 1-4. I like to set the monitoring to pan channels 1,3 hard left and 2,4 hard right. It makes monitoring while on location easier as it matches the general direction of the capsules.

Wind Protection:

For outdoor use you need wind protection. This was one of our big concerns. Tom was able to make the headbasket just small enough to fit into a windjammer. This is the one we use: Rycote Baseball

Post Recording: Let the Fun Begin!

AmbiFormat.png

Ok, let's briefly investigate the encoding and decoding process here. Ambisonics gives you the ability, in theory, to perfectly encode sound around you and then perfectly recreate it while focusing in whatever direction you like. Real life is a bit different. Just like there isn’t one perfect microphone for every use case, or one perfect position for stereo mic placement, ambisonic microphones are one more tool to add to your audio arsenal. Let's look at the math and then briefly, discuss.

The simplest encoding technique is this:

  • W (all) = FLU+FRD+BLD+BRU (Omni: Hey, just sum them all together!)
  • X (front/back) = FLU+FRD-BLD-BRU (Kinda like MS: Sum the “Front”, subtract the “Back”)
  • Y (side to side) = FLU-FRD+BLD-BRU (Kinda like MS: Sum the “Left”, subtract the “Right”)
  • Z (up down) = FLU-FRD-BLD+BRU (Kinda like MS: Sum the “Up” subtract the “Down”)

In reality this would work perfectly except all the capsules are not in the exact same space. And crucially, the response of the capsules off axis isn't perfect, and then each capsule is not matched to the other capsules. This is where the real challenges come in. There are ways to measure the microphones and create a file that can be used to calibrate the encoding process further. Those are beyond the scope of the instructable. I have built five of these so far. I sent one to David McGriffy, who among other things, is the developer of VVencode. He calibrated it and sent me the calibration files. In theory those are for the specific microphone that I sent him. My experience is that they work (and do so really well) with the other ones I have built. Tom Benedict, the designer for the housing has also built a few and the calibration files work with his.

To be clear, there are vendors of ambisonic microphones that provide individual calibration files for each microphone and offer recalibration services periodically. While I am not disputing the results they get, there are also other manufacturers of mics (B&K) who claim that their mics will be within +/-1dB decades later. Yes they make measurement microphones. Also there are other vendors that make ambisonic microphones that do not calibrate each one but do provide a plug in specific to their microphone (RØDE). Bear in mind these are observations I have from researching this and, well, drinking from the ambisonic firehose of knowledge. One of the other big observations I have is that the workflow is not standardized. The final one is that the final use case varies as well. I have been mostly decoding mine into binaural to listen on headphones immersively. I discovered a multitude of decoders and they all sound slightly different. As you might have observed, everyone's ears are different and the spacing between them varies as well. There is something you can use to “calibrate” your decoding plug in to your head called an HRTF or “Head Related Transfer Function”.

We are also not going to do that. I have also been recording with this as the center mic with a pair of outriggers for classical recording and ambience with great results. I can steer the image with the ambi mic and keep the levels left right the same. Makes the ultimate Decca Tree! Other use cases are decoding the mic into virtual mics with whatever pattern you would like. The cool part with this is that you can decode it multiple times simultaneously in multiple patterns that you can either mix back into stereo or use as feeds for multiple speaker outputs.

Angelo Farina is, in my opinion, one of the world's experts in both ambisonics and spatial audio. He uses higher order ambisonic microphones and multiple speaker arrays to recreate choral and orchestra performances this way that are amazing. He is based in Italy at the University of Parma.

Len Moskowitz, another expert and owner of Core Sound, focuses on using two second order ambisonic microphones while recording. Then in post, decoding them into a perfect cardioid ORTF pair. Which of course, can be repositioned on the fly while decoding. That methodology goes back to my statement about workflow. Along with the two Core Sound OctoMic™ microphones, you need 16 channels of outboard recording. OK, as you can surmise, this is a quite deep rabbit hole to go down. We are only going to dip our toes but I will show you the paths you can dive into.

One other thing to be aware of -- Standards! There are of course not just one, that would make life easy. "B Format" has a few. Read more here. FuMa and AmbiX are the two main standards with AmbiX becoming prevalent. For first-order signals FuMa uses the channel sequence W-X-Y-Z, while AmbiX uses W-Y-Z-X. If something sounds like you are standing sideways to it... You have selected the wrong one.


The assumptions I am making here are:

  1. You are using Reaper for this. You can download it and trial it for free then to purchase (which I highly recommend, is only $60 for a non commercial license)
  2. This can be encoded with Sparta Array2SH but I most strongly recommend buying VVencode because there are calibration files for the build supplied below.
  3. You know how to add plugins to it.
  4. You haven't done multichannel stuff with it yet… Well, because I had not.
  5. Software we are using:
  6. Reaper This is the DAW that I feel handles Ambisonics really well. A lot of the full tools below were developed to use in Reaper (although they work in other DAW’s)
  7. VV Audio This is the brainchild of David McGriffy. VVencode is what I recommend you use to encode the Ambi-Alice. I am including calibration files that David created for this. You can use SPARTA’s Array2SH as well, but this one is the best in my opinion. David has been doing this for 20 years.
  8. Sparta (Spatial Audio Real-Time Applications) These are another open source plugins from Finland. From their website: SPARTA is a collection of flexible VST audio plug-ins for spatial audio production, reproduction and visualisation, developed primarily by members of the Acoustics Lab at Aalto University, Finland. It includes one called Array2SH or "Array to Spherical Harmonics" described here in and AES paper. This is the second best way to encode the Ambi-Alice.
  9. Ambisonic ToolKit or ATK This is a great set of open source plugins for Reaper that focus more on decoding into various things and encoding of mono or stereo signals.
  10. IEM Plugin suite From their website: The IEM Plug-in Suite is a free and Open-Source audio plugin suite including Ambisonic plug-ins up to 7th order created by staff and students of the Institute of Electronic Music and Acoustics.
  11. Harpex Commercial plug in that are really good. From their website: Harpex is a signal processing algorithm designed to extract the maximum amount of spatial information from sound field recordings.
  12. Blue Ripple Sound Mix of commercial and free. One of their free ones is a great spatial visualizer called Flare.

I went through the basic setup in Reaper and how to save a template. I personally love the binaural decoding with a big caveat. Just like 3D Video and movies are not perfect because everyone's eyes are different i.e. pupil spacings vary and everyone has some astigmatism, 3D Audio is similar. It is actually harder than vision. Our ears all vary even further. each lobe, nook and cranny have filters associated with them. In real life our brains combine all of our sensory inputs to "tell us" where something is. If one is off it messes us up. Which is why 3D Movies can cause motion sickness. Your inner ear tells you brain something is off. One thing I have not played with yet is using a head tracker linked to the binaural decoder to rotate the sound as I move my head. I do have one on order and will do a separate instructable on that in "Advanced use"

Beamformers are another favorite thing to use. I quickly go through how to setup ORTF decoding. And yes, before you comment below, I know that true ORTF spaces the two microphones 17cm apart. We are not capturing that part. However, you can create a virtual mic or mics that point in any direction you like. Which can be very useful. We are just scratching the surface here. I feel like Ambisonics is one of those things that has been around a long time but all the tools were not all easily available. And my building the Ambi-Alice, your cost of entry is low but you are also getting a world class first order Ambisonic Microphone. Exciting times!

Here it is in front of a couple trains, including the Union Pacific 4014 Big Boy Steam Engine


Wrap Up

AmbiBuildCover-3.jpg

There is so much more that I could have added to this. I encourage you to explore Ambisonics and this is a great entry point.

Couple disclaimers and my thoughts:

  • We are not calibrating each individual microphones. There are several thoughts on this. Mine is that the QC process of Transound is pretty up to modern standards, so capsule to capsule variance is low.
  • It is first order so it isn't going to be perfect -- no microphone is.
  • The biggest challenge to existing first order mics is the capsule off axis frequency response. These have really good off axis response so they just sound good.

There is so much more to sound transmission in air than all the encoding and decoding algorithms take into account anyway. Water and Sonar are actually way easier to deal with. I was in the Navy on submarines and I know underwater sound really well.

I am including audio files I recorded with this so you can determine for yourself weather building one is worthwhile.

There is a Choir that was recorded with the prototype in the BM-800 body. It is all one take as it was a UIL submission. You can hear the noise floor of mic (or should I say you can't hear the mic, only they room!)

I short recording in a semi Anechoic chamber. Thank you UNT! Go Green!

There is really quiet ambience form Colorado in July, Front of my house with neighbor mowing lawn and our summer cicadas and finally I went to a very busy off ramp in Dallas at Royal Lane and got some traffic.

A huge Thank You! to:

  • Tom Benedict, he did all the CAD work on the capsule holder mic body.
  • David McGriffy, who wrote VVEncode, is a long time Ambisonic fan and provided valuable insights the whole thing -- along with calibration files!
  • Professor Xinrong Li and The UNT Engineering Department for use of their test facilities.
  • And of course, the people of the Mic Builders Group. A wealth of knowledge for sure!

A request, since you have gotten this far. Please fill out my Ambisonic Survey I found both a wealth of information on Ambisonics researching this over the past year or two and simultaneously few people who were actually doing it.

Enjoy and let us know if you build one!