Capturing the Sound of a Space

I’ve been interested in the use of convolution reverbs for a while, and was particularly inspired listening to this interview with Nikolay Georgiev by Lij Shaw for his brilliant Recording Studio Rockstars podcast series. During the interview Nikolay explains how he has fine-tuned his process of capturing the acoustics of a space using a mobile recording rig and various means of generating an impulse (including sine sweeps, a starter pistol and bursting inflated condoms!)

The basic concept is simple enough, excite the space with a burst of acoustic energy and  record the resulting response. The recorded audio is an acoustic  signature of that space which can be applied to any other sound through the process of convolution (literally ‘folding’ the sounds together).

There are two ways of performing this process, the first is the method I will explore in this blog post, using a balloon popping or other short loud burst of sound to approximate an impulse. This is the easiest method as the recorded file can be used directly in the convolution plug-in. The other method involves recording a sine sweep played back through a speaker into the space. The resultant recording needs to be de-convolved to create the impulse response. Although both methods are widely used, the sine sweep is considered better and there are some very good reasons why, but in practical terms you can still achieve great sounding results using a bursting balloon although it may not offer the most accurate representation of the space.

It was my sons 10th birthday the other day, and whilst clearing up the aftermath I found myself with a whole load of inflated balloons that needed disposing of. Perfect, I remembered I wanted to try recording some impulse responses, and this would be a great way to give it a go. I thus equipped myself with the following items:

  • Inflated party balloons (the round ones, not the sausage-type)
  • a pin
  • a Tascam DR-70D

The Tascam is a handy portable recorder that can capture up to 4 simultaneous tracks in uncompressed high quality format, with the added bonus of having 2 built-in small diaphragm condensor mics . You could equally use a computer soundcard and one or two mics.

One of the spaces I was going to capture was a small stairway that runs along the back wall of our house. I have previously had some good results using the corridor for re-amping drums , placing a mic there and playing back sampled drums into the space to add ambience.

The lower part of the stairway has a sloping ceiling which gives a bit of a flutter echo.

IMG_2142

The upper stairs has dense, bassy sound to it.

IMG_2143

I also wanted to try capturing the bathroom downstairs, it’s about 2m x 2m square, completely tiled floor and walls with a textured ceiling, and is quite reverberant.

IMG_2146

The recorder was mounted on a mini-tripod and sat about 6 inches off the floor for the stairway capture, and on the window ledge for the tiled bathroom. One of the pitfalls I encountered in capturing the impulse response was that the initial pop of the balloon is very loud compared to the resulting echos and can result in clipping in the recorder. It took a few goes to get the levels just right, but once set the actual process was very simple: hit record, hold the balloon up and prick it with a pin. I didn’t experiment a great deal with the effects of location, but I generally had the balloon above the recorder when I burst it. 

Post-processing was very simple. I trimmed the impulses such that they started just before the initial transient and then faded them out shortly afterwards. I used Audacity but any DAW or editor will do the job. Once edited, just export them as wav or aiff files ready to load into the convolution plugin of your choice.

Screen Shot 2019-03-11 at 21.30.53

I used the Convolution Reverb Pro device in Ableton Live Suite (it’s a Max for Live device), I like this plug-in as you can just drag and drop the IR file from finder to the device, but it’s pretty simple to do this with Space designer in Logic Pro or Waves IR-1 for examples of other convolution plug-ins.

Screen Shot 2019-03-11 at 16.03.20

Once loaded into the convolution plugin, simply pass audio through it and hear how it sounds.

To demonstrate I’ll use a simple drum loop, here it is dry:

Now 100% wet with the ‘Stairway 1’ IR:

I like the nice low-end thump to that particular reverb

With 100% wet  ‘Stairway 2’ IR:

This one is a bit more present and in your face.

With 100% wet  ‘Bathroom 1’ IR:

This one has a splashy, hard character.

Notice how the character of the space is imparted on the loop. By adjusting the wet dry blend you can easily dial in as much or as little of the ambience as you like.

Another thing you can do is mess with the actual impulse response to change the character of the reverb. For example, here is the ‘Bathroom 1’ IR where I’ve altered the envelope of the sound using volume curves in Audacity:

With 100% wet  ‘Bathroom 1 Processed’ IR:


It gives it a more non-linear type of sound.

You can also apply other processing to the IR’s to create more complex and interesting effects, here is an example where the same IR has been passed through a phaser, envelope filter and finally high passed:

Although there are lots of ways you can refine the process, and there are further subtleties involved that I haven’t explored here, I was surprised how easy it was to capture the reverbs and use them pretty much straight away. I’m looking forward to getting out and about and sampling some more interesting spaces, there’s an old railway bridge near where I live that has a cool echo that I’d like to capture. So if you see a bloke with a balloon hanging out around tunnel entrances, it’s probably just an audio geek capturing impulse responses…

Modifying the Rode K2 mic Part 1: Tube Change

I originally intended to replace both tube and capsule on this mic, but after experimenting and comparing I ended retaining the original capsule but with a replacement tube. In this post I’ll describe the simple tube change and give some audio examples. In the next post I’ll go over the capsule change. To be clear, the capsule upgrade made the mic sound different, rather than better, and based on what I already had in my mic locker I chose to return to the original capsule.This highlighted the importance of following your ears rather than what you read on the internet!

Okay, so here’s what you need:

A Rode K2 valve microphone ( I got mine second hand off EBay for £300 which is a fair price). These are decent mics and are very useable straight out the box. They are multi-pattern, very well made with a good quality power supply, cable and shockmount (unlike the Chinese valve mics often used for mods such as the Apex 460). The case is a bit tacky but you can’t have it all.

A NOS (New Old Stock) Telefunken E88CC (6922) valve or equivalent

20170203_180548708_ios

I learned a lot about the vintage tube market whilst researching these. Firstly, they are not cheap! You have to be careful as there are a lot of rip-offs out there, people trying to pass off newer tubes as vintage ones but fortunately there is a wealth of info out there to help as well as several reputable dealers that get recommended by many different people. I paid approx £125 for one from NOS Tube Store , their service was great and the tube was as described. This particular one is a sought-after Telefunken made in West Germany in the 1950’s. There are many other fine vintage tube options but I didn’t have the time and money to experiment, and this tube was highly recommended by a lot of people. If you want to know more and have time to burn, just check out the forums. Different tubes will impart a different sonic character to the mic so the ultimate test is to use your ears!

How to do it

This is very easy and I would recommend you try this first and see if you like the way it sounds. I recommend you record some different sources with the stock tube before you proceed, so you can compare afterwards and understand for yourself how the mics character has changed.

20170203_180200307_ios

 

Firstly, undo the metal retainer ring at the base of the mic and remove it. Then undo the metal sleeve that forms the main body of the mic (hold the mic by the grill end and unscrew). It should then slide off to reveal the tube and circuitry of the mic as shown in the picture.

20170203_180255412_ios

The tube is secured in position by a plastic clip which sits on top of the tube (pointy end) and the clip is held in place by 2 screws. Using a Phillips screwdriver, loosen both these screws and fully undo 1 of them to flip the plastic clip up and out of the way.

20170203_180402706_ios

The tube can now be pulled carefully from its socket, a slight amount of wiggling (of the tube, not you) may help!

20170203_180229656_ios

Carefully insert your new tube into the socket, it only fits in one orientation, make sure it is fully engaged. Replace the plastic retaining clip over the top of the tube and tighten the screws.

To my ears the NOS Telefunken tube makes the mic more mid-focused, the low frequencies are still clear, warm and big, the highs are a little less bright and sharp but the mid presence makes the mic cut through the mix more. I also find it more pleasant to listen to than with the stock tube, but it may not suit every application. I’ve presented some comparisons here to give you an idea of the difference. All clips were recorded via the preamps on my UA Apollo interface, I’ve endeavoured to maintain the same conditions and distances between mic and source for each comparison, but there may be small variations.

Firstly, male voice (don’t worry, I’m not singing)

Stock Tube

NOS Telefunken E88CC

Female voice

Stock Tube

NOS Telefunken E88CC

Double Bass

Stock Tube

NOS Telefunken E88CC

Judge for yourself and decide if this mod is worth the money, for me it’s a yes, but I will qualify it by saying I have other flavours of mic tonality in my locker, this adds something different, and it does it rather well.

In the next post I’ll describe changing the capsule and post some examples of the difference that makes.

 

 

 

 

AKG C414B Hum problems

I thought I’d write a quick post on this in case some of you experience a similar problem, it may save you hassle and expense.

I have a pair of AKG C414B mics, they are real workhorses and as the cliche goes, the studio Swiss army knife. On top of that they are solidly built and very reliable. But sometimes things go wrong. I recently did a bit of drum recording for myself and was listening back when I noticed a faint but distinct mains frequency hum on the overheads (my 414s). I had them running through my home made Neve preamps and then into a pair of line-level inputs on my UA Apollo interface. I automatically assumed the hum was coming from my home-made preamps rather than the solid, Austrian-engineered AKG mics. I duly set out to test, swapping cables, microphones and trying various configs of phantom power on / off on each channel, until it was clear that the problem lie with one of the 414s.

Faced with an expensive repair, I thought, well, let’s open it up and see if anything obvious was up with it.

Hmmm, lots of surface mount components and no obvious signs of trouble. A quick Google search led me to this post. Not exactly the same problem but similar, so I read on. Turns out that the grill / mesh of the mic connected to ground via pin 1 of the XLR forms a Faraday Cage, which helps shield the high-impedance capsule from EMI noise in the room. The mesh makes contact mechanically and if this connection is a little dodgy, it won’t work and there will be noise. I tested the resistance between pin 1 on the XLR and the mesh, it was a variable which immediately suggests a problem. I went about gently squeezing the base of the mesh in an attempt to improve the mechanical connection. Checking again, the resistance was now consistently minimal, so time to test. I put the body back on and plugged it in, powered it up and hey presto, no noise. Phew!…. So if you’re having this kind of issue, try this first, the mic is easy enough to open , just remove the 2 screws on the base (star driver or a flat head screw driver will do) and the smaller cross head screw in the XLR connector base and then slide the body off.

“Boutique” DI box on a budget

I’ve been meaning to build a passive DI box for sometime, I have a couple of active ones and it’s nice to have an alternative. Also, I recently bought a Korg Minilogue and figured a good quality passive DI would be a good match.

Passive DI’s are very simple, they are essentially a step down transformer used to bridge the high impedance output of an instrument to the lower impedance input of a mic preamp, as well as to convert from an unbalanced to balanced signal. As they are based on a transformer, the performance of the DI box is pretty much uniquely governed by the transformer, and good quality transformers aren’t cheap. There are a number of high quality DI boxes out there, such as the Radial JDI, but being a DIY fiend and a slight cheapskate, I thought I could have a go at making one, how difficult (and expensive) could it be?

There are a few great posts giving instructions on building a DI box from scratch, but what I noticed was that as you get all the parts together, the bill of materials creeps up and the cost benefit of doing it yourself  becomes less attractive. Now, cost is not the only reason why I build things, in fact, those EZ1290 preamps ended up costing me way more than I expected (they were a labour of love), but frankly, if I can build 2 good quality DI boxes for the price of buying one, then that’s worth the effort for me.

Whilst doing my research I noticed there were a number of cheap passive DI boxes out there, and I’m guessing the reason they were so cheap was because they used a poor quality transformer, so rather than build a box from scratch, why not just upgrade a cheap one? I found this Cobra DI box on EBay, it only cost £9.95 (+£2.95 p&p) but comes in a metal case with pad and ground lift circuitry and a thru jack. Perfect as a donor, all I needed was a transformer.

20160703_152741

I went for this Cinemag CM DBX (EBay £64.99)

s-l300

Good alternatives include:

Jensen JT-DB-E/EPC (as used in the Radial JDI)

Lundahl LL-1530, LL-1576/77/78

Sowter 4243, 8044

The Cinemag website includes schematics for a DI circuit using their transformer, comparing this to the COBRA DI box there were a few components missing but I thought I’d wire it up as a direct replacement for the existing transformer first and then see if there were any issues.

Firstly, open up the COBRA DI box by unscrewing the 4 retaining screws on the base.

20160703_152809

With the cover removed you can see the simple and robust construction, the original transformer is through-hole mounted on the PCB.

20160703_152827

Flipping the box over and working with a solder sucker, remove the original transformer.

20160703_153041

You can see the obvious differences between the original and upgrade transformer, the mu-metal can for shielding, but the Cinemag also has individual shielding for each winding and will use better quality wires and laminations.

20160703_153420

I soldered the Cinemag wires into the existing PCB holes following this wiring scheme:

  • Yellow – Input Tip
  • Orange, Black, White – Input Sleeve (PCB ground)
  • Gray – XLR 1
  • Red – XLR 2
  • Brown – XLR 3

I tested this configuration and it worked fine, but I noticed that in the Cinemag schematics they wire only the white cable to case ground (not the black and orange too), so I modified the wiring slightly such that black went to the PCB ground and white connected to the case via the existing green cable.

I will admit that I was feeling lazy and didn’t fancy drilling the case or the PCB to mount the transformer, so instead I used cut up neoprene pipe insulation to snuggly hold the transformer in place between the case and PCB. Once the case is on the transformer isn’t going to move anywhere and there is no strain on the connectors.

20160703_16182920160703_161847

And that’s it, job done, a very simple and quick route to getting a high quality passive DI box on a budget.

 

 

 

 

Piano Re-Amping

I recently watched a brilliant online recording masterclass with Sylvia Massey courtesy of Creative Live, very inspiring and it reminded me of something I’d lost sight of, the spirit of experimentation. During the class she records a band live and demonstrates some of the techniques she uses as well as her creative process. What struck me was how playful it was. Among the many things she did (including vocal feedback delay using 2 mobile phones a mic and a monitor speaker,  and passing a guitar signal through a power tool), she re-amped a snare by passing the recorded track back through a speaker with a snare strapped to it. The recorded sound of the re-amp was blended with the original snare to get fatter drum sound. This got me thinking, what else could you apply that too?

I’ve been doing a few tracks recently where we’ve used midi piano, I’ve been using the sampled grand piano in Ableton which isn’t bad at all (to my ears at least), however, it’s still a sampled piano and lacks a little dimension. I had the idea of re-amping the piano by passing the sampled piano track back through a speaker playing into the body of the upright piano I have at home, then micing the piano to get a new piano track. The idea being that this new track contains the string and body resonances of the real piano. The final piano track would then be a blend of the 2.

I set about doing this, it was relatively simple. A spaced pair of AKG C414s aiming inside the top of the piano, close in on the strings, and then a speaker placed at the bottom of the piano aiming into the body (I used my Avantone Mix Cube for the job).I removed the lower panel of the piano so the speaker could play into the body (see photo)

Using a real piano to enhance the sound of a sampled piano
Using a real piano to enhance the sound of a sampled piano

Once all this was set up, it’s just a case of pressing play and record. In the mix, just blend the original with the re-amp track according to your taste.

Here’s the original sampled piano track:

 

Here’s the re-amped track:

 

Here’s a blend of the 2:

 

You can hear how it adds a depth and dimension to the sampled piano without recourse to loads of plugins.

I’ve also tried passing other signals through it, including vocals and guitar tracks,it lends an interesting colour.

I’m liking the natural colouration you get with re-amping, it’s something I’m definitely going to explore more.

 

 

DIY preamps continued….

As promised in the previous post, here are some sound samples to demonstrate the sonic differences between the DIY EZ1290 Neve-style Mic preamps and the stock preamps on the Universal Audio Apollo firewire interface.

The setup for this was pretty simple, a remote drum recording session where the drummer (the fabulous Mr Mark Whitlam) was given a score and a backing track to record the drum part to. The session was carried out in Marks garden studio cabin, a compact but decent sounding purpose-built space that has been acoustically treated.

Listening to the stye of the track and how the drums sounded in the space I decided to go for a simple 4 mic setup: 2 overheads placed more out in front of the drums to get a more balanced picture of the kit as a whole, supplemented with kick and snare close mics.

The drums themselves were:

Vintage 1960s Zildjian A 14″ hi hats, 22″ Istanbul Agop Azure ride (next to hats), Bosphorus 21″ medium thin ride and an 18″ bosphorus thin crash. Drums …. Snare: Canopus Zelkova, 1960’s premier Olympic 20″ bass drum and 12″ Tom, modern premier 14″  Tom

Of course, all properly tuned…..

On the overheads I used 2 Oktava MK-012 modified by Micheal Joly at Oktavamod, on the kick an Electrovoice RE20 and on the snare a cheap and surprisingly cheerful Audio Technica AT2020 (high SPL handling, nice response, good rejection).

On the following audio examples you are listening to just the overheads so you can hear more clearly the differences between preamps. Note that these are different takes, although the drummer is incredibly consistent, so the comparison is not entirely precise.

Firstly, through with the UA Apollo preamps

 

And then with the EZ1290 preamps

 

You can hear the subtle differences, especially when you listen to that ride cymbal from about 10 seconds in. The Apollo preamps are very good, very clear and crisp, but the EZ1290 has a smoother sound, more open and somehow with a better sense of space.  What do you think?

This is the only direct comparison I’ve done but I’ve been working a lot with these preamps on voice and guitar and really like the sound I’m getting, I’m finding I’m using less processing further down the line to shape the sound. I’ve particularly enjoyed the combination of this preamp and the modified Apex 460 valve mic mentioned in a previous article.

 

The DIY preamp saga

There comes a point when many a recording engineer starts to consider the finer points of the recording chain, and inevitably finds themselves in the esoteric world of mic preamps. These days there are so many choices both off the shelf and DIY, so many opinions, and so many fuzzy descriptions of ‘warmth’, ‘open-ness’, ‘transparency’ etc…It’s hard to know where to begin.

I began by accident. I bought a job lot of electronic components off EBay, the seller claimed that all the components required to make a Neve preamp were included. Well, that was optimistic, I think there were about 4 capacitors and a handful of resistors that could be used, but anyway it piqued my interest. I found out the project board he was intending to use which led me to the fabulous www.groupdiy.com website, where Martin Adriaanse has created a complete guide to building the EZ1290 Neve preamp clone, including a bill of materials, circuit schematics, build guide and some basic troubleshooting points. You can purchase the PCBs directly from him (they are not available anywhere else).

My first decision was whether to go the complete DIY route (where you have to source all the parts yourself) or go for one of the many kits provided by the likes of Seventh Circle Audio or JLM Audio . Being the masochist that I am, I opted for buying the PCBs and sourcing components myself using mainly Mouser in the UK, how hard could it be? Of course, some of those nice polystyrene capacitors weren’t available on Mouser, nor were the BC184 transistors and a few other little things, so I ended up on EBay where I found pretty much everything else I needed.

The Neve sound is all about transformers, and not just any old ones; properly designed, carefully manufactured transformers made for the specific task of interfacing microphones and line level signals. You need 2 per channel, an input mic transformer and an output line transformer. These aren’t cheap, I got mine from Audio Maintenance Limited who also sell the Grayhill rotary switches used for the gain selector. All in, 2 channels worth of xformers and switches cost the best part of £200 once delivery was factored in.

Once the components were in, soldering the PCBs was a relatively straightforward task, pay attention to put things in the right place and use a heatsink on those capacitors. Use a fine tip soldering iron (especially for that rotary switch, the pins are tiny and close together) and audio grade solder.

Soldering the components onto the ez1290 pcb
Soldering the components onto the ez1290 pcb

I started with resistors first, then smaller caps, then transistors and larger caps.

20150304_105822
Almost complete ez1290 pcb

Note the overkill polystyrene capacitor on the right, couldn’t find the right voltage spec one for that capacitance so had to go for a 400V spec, hence the size. Still fits and works!

Building the boards was the easy bit to be honest. Once they were done the fun of trying to fit everything into a case and route the wiring began. I got a 1U rack case off EBay again, it was from a network router so needed a bit of tidying up, and I also ordered a plain aluminium front panel which I was going to attach on top of the existing front plate. I’ll admit that the rack case was probably my biggest cock-up. I thought it was full depth but when it arrived it was a shallow one, so stuffing all the PCBs, transformers and connectors in was going to be a challenge.

20150306_150725

20150306_150739

I put the transformers for each channel down the sides and the PCBs in the middle. This just about worked, I had to leave space for input, output and power supply sockets too.

I wanted to make use of the input transformers multi-taps, so I could make a switchable impedance for the mic inputs. Although not strictly necessary these days as most mics are designed to work with the typical high impedance of most mic preamps, it can help when interfacing with older mics and it also offers another tonal option for a given mic. The taps on the primary side of the transformer allow the two halves of the primary to be wired in series (1200Ω) or parallel (300Ω), I wired in a switch to change the configuration via the front panel.

Nearing completion, fitting the front panel
Nearing completion, fitting the front panel

The only other switch I fitted was a +48V phantom power switch. Phantom power is introduced to the hot and cold connectors on the input XLR via 6.81K resistors. I left out phase and pad switches for now. I also opted not to fit the output trimmer. You ideally need to use a high quality pot, I think I may add this for the next build as it allows you to drive the input stages harder for more grit if you want it. Some people say that the trimmer degrades the sound quality slightly, that’s a better reason than not being arsed to fit it.

The other important piece of the puzzle is the power supply. These boards need a +24V supply as well as +48V  for phantom. I had no choice but to locate all of this in a separate enclosure, though with careful planning you could fit all this into a larger rack unit (watch out for induced mains hum from the power supply transformer). JLM do a suitable power supply, I already had a toroidal transformer so found a PSU kit from a Russian supplier on EBay. These do the job, but I blew the first one with an accidental short circuit.

Once everything is connected, you need to bias the output transistor, for this you ideally need an oscilloscope and a signal generator. I used a 1kHz test tone from my sound card and a borrowed scope from the local university, very useful diagnostic tool and I might have to invest in one if I’m going to do this more often. Biasing is explained in the build guide and is a pretty straightforward procedure, just adjust the trimpot till you get symmetrical clipping and your done. Pushing it too far in the wrong direction can make things get very hot.

My first board wouldn’t bias properly, I ended up swapping all the transistors in the output driver stage and it all worked ok after that, took a bit of head scratching though, the build guide gives a few tips such as the typical DC voltages at each of the transistor pins but it wasn’t obvious which was the culprit.

I had some intermittent hum issues, I was using the case as a ground so I implemented a star grounding scheme, seems to have worked.

I used a powder coated metal enclosure for the power supply, this is a pain as the individual panels don’t necessarily conduct with each other so the safety ground had to be wired to each panel to ensure proper earthing.

I blew the 1W resistors on the + side of the board supply rails by letting the underside of the board accidentally touch the case (which is earthed) when testing, won’t be doing that again.

The finished product (Roof off)
The finished product (Roof off)

So here is the finished version. Sure, it’s a little chaotic and Heath Robinson in there, I used whatever material I had to hand to make things like cable guides and mountings for the transformers. But this is my first build.

I’m glad I chose to do it this way, you learn so much more by going through the stages yourself, figuring out details like how to package and connect up the system and then troubleshooting it, making mistakes, blowing components, ordering the wrong thing etc….I know what to do next time. Speaking of which, I’m considering this project as a prototype; I’m planning to get another 2 boards and build a 4 channel unit into a 2U (full depth) case, and add pad and phase switching for each channel, and maybe a switched mode power supply instead of that bulky and noisy toroidal transformer.

All in all it has been a lot of work (and expense) to get these mic-pres up and running, however, they work beautifully and I’ve been getting some great results on guitars, vocals and drums. I’m even starting to use words like ‘warm’, ‘open-sounding’ and ‘big’ when I describe the recordings. In the next post I will put up some examples to illustrate why I think they are worth the effort.