Diffuser PJ Mapping Screen Built!

It is so easy to underestimate how quickly time flies! You may remember two weeks ago I posted about having chopped all the wood and I just needed to get glue, well yesterday I finally got it, and was up until three this morning getting as much as I could done. I ran out of No More Nails about 3/4 of the way through, so I got up first thing this morning and went to get another tube.

Below is a photo gallery of the build from start to finish for your amusement. I have to say, considering the issues I had with the original cuts (breaking a circular saw and having to hand cut half the posts, I am really happy with the results, it looks really beautiful. All that remains now is to let the glue set hard and then get it sprayed matt white, as this will give the best surface finish for the Projections.

1) So as I have touched on before, QRDude is the software I used to create the design for my diffuser. This is a 2D QRD design, with a Scatter characteristic of 333hz, a diffusion characteristic of 666hz and a high frequency cutoff at 3750hz.

2) Here we have all the posts for my build stacked on a secondary baseplate. I intend to build another three of these after university, so I have bought extra MDF for baseplates. The wood used for the posts is Planed treated Red Wood Pine,  44mm x 44mm. The "well" depths vary from 0mm upto 24mm.

3) Here is another baseplate cut and ready to be filled.The well spread is 13, giving the baseplate a width and length of 572mm

4) I decided to go from the two axis first and then build out from there. This isn't really standard practice, you would normally just go row by row, however as I said the circular saw broke, so from 14mm posts down, the hand cut is slightly uneven on a lot of posts. To eliminate the chance of an uneven design, this way lets you glue two sides of every post you place after the axis are done, resulting in a more stable build as you go along.

5) Here you see it at just after the half way point, around one thirty this morning. I knew by this point that I had made the right choice as the lines and block placements were still going smoothly and perfectly.

6) This was just a couple of arty angles before I went to bed.

7) The first of these three was a quick one before I started again today with the fresh No More Nails. This will be painted white as I said, and then placed behind the speakers, central to my mix position. Happy Days!

Advanced colour theory and Digital LED's vs Analogue LED's

Having now been with the system for over a week, and demonstrated it to a number of people, I definitely feel that the idea has legs beyond this project.

As with anything, I am now seeing how it could be advanced and improved upon, though currently I dont have the know how or the time before hand in date. Yesterday I stumbled upon a very impressive system that uses Digital LED's rather than their fully analogue counterparts that I am using. Now, undoubtedly this is a more complex system, bit it requires a laptop for operation and is a lot more expensive, so it would really be for a different market - my system still has validity (phew!). Also, in real terms this is not a true colour organ, where my system arguably is.

These guys have used FFT processing to calculate the dominant frequency band at any given time and display a corresponding colour, it is rather cool! Rather than blending RGB colours together to generate the rainbow. A set number of 12 colour blends, from Red through to Purple, have been mapped across each and every octave. Should the FFT device pick up more than one note, it will send each further harmonic above the fundamental to the LED striip, with its corresponding key mapped colour. The joy of digital control!

Looking into the future, I would definitely like to continue with this side of LED programming. The power of this system really is limited only by one's grasp of maths... You could have many hundreds of different maps for the digital LED strip, for everything from phase correlation to RMS vs Peak metering systems, random pattern gens, all sorts of crazy stuff.

I have watched this video a fair few times now, along with the previous video found on his YT page. Without doubt his system is more powerful and visually "tight" than mine - he specifically said that he wanted to develop a system beyond the standard  colour organ.

Despite this, it makes for an excellent comparison to my current system and efficacy of mapping colour across one octave vs the whole audio spectrum. There are inherent limitations when mapping all light colour across one octave.Going back to colour theory here for a second, if we mix RGB light in equal amounts, we get white light. When you are playing a fully mixed piece of music in my system, the blend of RGB is calculated by the arduino every 25ms, and displayed as a solid amount at that time. With the system seen above, this would not be possible in its current config. With a full piece of music, it would be almost inevitable for the resulting light output to be anything other than white. Even playing a seventh chord in any octave would cause this to happen. The system seen above has found a very clever and attractive way around this, but it simply wouldnt be possible with non-digital LED strips, which are a lot cheaper than their digital counterparts.

Being honest, though it could be easily be modified to be a properly effective audio metering system, in its current state, I feel like my system better represents the frenetic nature of analogue audio signals. It also better contextualises songs and their phrasing, along with mastering qualities.

Initial Observations if the System

OK, so I was supposed to go and get the wood glue to complete the diffuser, but it didn't happen today. In part through written work on the dis, and in part because I was up so late using the colour organ.

I consider myself and electronic musician, sound designer and mix engineer in practice - after uni this is where my main passion lies and I would love to make this my career. I really wanted to develop a means of monitoring music to meet my personal demands, of which there are a few.

Now, we have meters. 1000's of them, far more accurate than anything that I could achieve, however at the same time none of them conform to a few simple wants of mine. Firstly, I'm sick of looking at a screen for the information. Until I get a third monitor in my studio, there is always going to be at least one click involved. So it needed to be outboard. Secondly, I find the whole process very, very dry. Any normal FFT based metering system will look roughly the same and determine the same information. Some have clever features of routing individual tracks to see how each element of the mix sits etc.... But like I say they exist. I have been concerned with something that you can look at that gives you an insight into the sound that inspires you to create more. So it has to be accurate to the point of confidence in what information it is displaying. Thirdly, it had to be relatively cheap, as I am a student and like many others, I am a "bedroom" producer at this stage. I don't know why I used inverted commas there, I actually am.

So after tweaking it to a far better state of reactivity the night before, and just purely listening, I can certainly say that it has changed the whole experience of monitoring my audio! And actually, it does more than I ever expected it would. It really does make it obvious when the loudness wars are in full effect.

With nice dynamic music, the system is an absolute treat, with obvious and consistent colour shifting and blending. Each very well mixed tune I have listened to through it has an obvious colour signature through each phrase. Little fills at the end of bars become a form of visual entertainment as I wonder what the blend is going to come out like.

If the audio material is squashed to much and over limited, it simply results in kinda tinted white light. When you push a mix too hard, effectively all frequency bands become (at least) as loud as they ever possibly could have been in terms of digital full scale. Because this is a three band eq split for the Low, Mid and High frequency bands, which then feeds Red, Green and Blue LED's within a strip, when you push everything to hard, you get extremely similar values for all three colours. If you have been reading through my work, you will know that when we blend all three colours in equal amounts, the result is white.

So in calibration of the system, I took the same approach as the guide to my recently purchased Subpac. It's a seatback subwoofer, look it up! (And buy one, they are the dogs balls) Anyways, they advise you to go through your most well known tunes, and find the level at which the balance of the subpac feels right by comaparison to the level in your cans (or speakers depending on how you choose to set it up).

So I did that last night. I went through my collection, all my Flac files to ensure quality didn't effect my listening. My soundcard has too very high quality headphone circuits built in, and funnily the system reacts best at the level I use on the other output for my Subpac and Headphones... Strange, as it is a pretty driven output and the MSGEQ7 is such a small chip!

Another intriguing property of the system is that it detects low level audio really well too. Remember a few posts ago now, I was talking about using HSL values instead of RGB as a way to improve the stepping sometimes encountered with weak RGB code? Well blending RGB values in this manner seems to be a far more appealing way still. Perhaps it was just because of the shitty chip in the unit the LED's came with for changing colour, but yeah, this is awesomely reactive! You really do see all the colours of the rainbow throughout the course of a tune. AND so far, I have been nothing sort of addicted to either listening or making music through it. AAAAND it only cost me aroun 60 so far. All in all, I'm pretty damn happy. It most certainly works.

The next question for this part of the project is how I am going to package this all up, as right now as you have seen, it's all wires and breadboards and certainly not what I can take up to the uni. Im thinking a frosted glass vase, Im going to TK Maxx tomorrow to see whats up with that. Im also gonna keep an eye out for a nice trinket box for housing my electronics.

I also I have a little suprise in mind, but I'd rather save that for the expo. Love it when a plan comes together ;) till next time.

Objectives Complete!

as I touched on at the end of my last post, I have manged to get my electronics working properly, so now I have my very own colour organ in my room. Advancing on the last post I made, after I got the desired signal from the MSGEQ7 chip, I set about contructing the LED circuit. Up until now, I have been trying to get everything working on one breadboard. This time I decided to split the two breadboards, as I have two... See my earlier posts for how that happened!

If you enlarge this image, you will see that the right hand BB is looking after the EQ circuit, and the left is the LED's. I have also routed the Resistors and Capacitors straight to the power rails and swapped the pin arrangement back to the original advised config from J.Skoba.

In testing it I have also discovered some advancements in the code that could be made, mostly to do with clearing low level distortion from the MSGEQ7 and improving the refresh rate in order to make the colour definition more obvious. These haven't been massively technical or difficult, but they have made a obvious difference to the performance.

I wonder if it is ecause when this code was first written the power of the Arduino Duende wasn't that of the Duo. Basically though, the refresh rate for the MSGEQ7, which is controlled by the strobe and reset pin via the arduino, was set to 80 milliseconds. This was ok, however lowering this to below 30 milliseconds has vastly improved it. this allows for a refresh rate of around 31 frames (equiv) a second. I have just been very happy with its performance since then, however I will continue to lower this amount. Hopefully the modern arduino can drive the MSGEQ7 at a refresh rate of > 10 milliseconds.

The original code used a filter set at a value of 80 to filter out MSGEQ7 distortions, which range roughly between 50-80. I did however notice when monitoring this through the Serial COM that it sometimes creeps upto as high as 86-87... I have found you can nearly completely elliminate these using a filter value of 90 instead. Occasionally I am still seeing some blue flashes on the LED at a really low level, but I am going to put this down to a ghost in the machine; when I monitor the Serial COM, there are no numerical values above 90 happening that correspond to the slight flashing. And when I say slight, when it is placed behind something, it isn't enough light to even notice, even in pitch black. Its only if you actually look directly at the LED strip you notice it.

Another note for anyone building this with no knowledge of electronics, N Channel MOSFETs are tempremental. At first, I thought there was a massive error with what I was doing, I was getting extreme flashing on the blue channel, even with no five volt feed. With the transistors in, this should mean even if your 12volt is connected, you will get nothing through. I swapped out the MOSFET and changed nothing else, everything was fine. I had read elsewhere that they can often be faulty. Buy more than you need, Id say 10 is a good number. Once you have one of these systems, you will want more anyways... ;)

Once the light goes down this evening, I am going to take a video of a few tunes to show it in full effect! In that post I will also include my idea for how I am actually going to implement this. Funnily, we will be going back to diffusion, but this time light diffusion. Considerably easier to implement the its acoustic cousin!

After that, we will be going back to the acoustic diffuser build, as I can now afford the Gorilla Glue I need to get it done. then I will be able to get my analysis of the final system done and complete my dissertation.

A breakthrough! (Finally)

So I decided to take things back to basics. As you know I have been having troubles with the circuits that I have been trying to create from a couple of peoples online blogs and instructables pages.

Both of the circuits I am trying to recreate come from another user called J.Skoba, who is actually credited on the Arduino site as coming up with the guide for how to get the MSGEQ7 and Arduino talking together.

Here is a little sketch he made and uploaded of the circuit diagram. So following this guide I have now managed to get numeric values coming into the Arduino from the MSGEQ7, which is taking a feed from a Mono audio jack this time. I decided to ditch the stereo one, because it is tiny and difficult for my fat finger to soldier the jumpers to. Also, it is a much sturdier and professional looking component.

Apart from this, I swapped out all components for fresh ones, including the MSGEQ7. Being honest, I couldn't tell you if it was a component that was the problem or not, but I will explain the differences between the two, cause there are a few main ones.

First off the in the new circuit, the reset pin is going to digital pin 3 of the arduino, rather than digital pin 4 in both of the other projects I have followed so far. This shouldnt matter in theory, as I have checked through the code and the appropriate lines have been changed to make this work. The main difference in the way I have done personally done the circuit this time, is like J.Skoba, I made the Pin 8 oscillator circuit with the resistor and capacitor attached directly to the positive and negative power rails. I have also taken the capacitor straight from Pin 6 to ground rail. I am just trying to avoid as many extra connections as possible to eliminate problem components at this point and it seems to have worked.

So now I have changed the values according to J.Skoba in the other Arduino code that I have contructed from David Wang and Russells blog posts. I have to stress though, neither of their projects work straight out of the box for me.

EDIT: During writing this post i got distracted playing about with the circuit board and managed to get everything working!! I now have a 5M audio reactive LED strip in my room, which is pretty great! My next post shall be dedicated to that circuit, its differences to the others and how I have changed the code to better suit the purposes.

Problems with the Circuit Part 2

So to continue from my last post, here is a photo diary of me connecting the arduino to the breadboard.

1) Here you see Digital Pins 9, 10 and 11 from the arduino connected to the Gate Pins of each of the N Channel MOSFETs. You also see Pin 7 occupied, it is connected to the Reset Pin 7 of the MSGEQ7.

  

2) Next digital Pin 4 of the arduino is connected to Strobe Pin 4 of the MSGEQ7 (orange wire). Digital Pin 2 of the arduino is connected just underneath the 10K resistor (blue wire).

 3) Here we are connecting the power connections from the arduino to the board. Take the 5v output of the Arduino and connect it to the bottom positive power rail. The ground pin next to the 5v and connect it to the bottom ground rail. Then take the Vin pin on the arduino and connect it to the top positive power rail.

4) Here are the red and white cables seen above connected to the breadboard.

5) Next we have the yellow Vin connected to the top positive rail.

6) Now we look at the analogue inputs of the arduino. Connect analogue pin 1 (A0) to the central pin of the 1K potentiometer.

7) Next, analogue in pin 4 (A3) is connected to the output pin 3 of the MSGEQ7.

8) Finally, I have jumper inserted into the RGB and 12v inputs of the LED strip ready for connection. The RGB lines will be connected to the Drain pins of the N Channel MOSFETs from left to right respectively, and 12v is fed from the 12v power line.

So the arduino code is loaded, the Breadboard and circuitry is complete to the fritzing guide. Hopefully this rebuild may yield some results! Fingers crossed :) One final point for Russell, I decided it would be easier to follow his method of powering it all, and found a similar power brick to the one that came with the LEDs (except I havent stripped the connection off this one). I see no reason why it wont work, it's the same voltage with a slightly lower amperage. Hopefully this should be OK.

The Left one is the original LED PSU, the right one is the one im using.

Problems with my Audio Reactive LED circuit

So I finally got all the last parts for the project and set about building the circuit last night with no success, so I have decided to give a photo blog of the build stage by stage for the benefit of Russell who can hopefully point out to me where I have gone wrong.

This is going to be snappy as I have lots to do and don't want to bore Russell. So from the top then!

1) Insert MSGEQ7

2) Connect Caps (Left 33pf, right 100nf). The left wire of thee right cap goes to Ground Pin 6 of the MSGEQ7. The 33pf's right wire connects to Oscillator Pin 8 (connection will be broken with resistor)

3) Connect 3rd Cap, which sits "over" the second cap. The left wire goes to Input Pin 5

4) Insert 220K resistor between 33pf and Pin 8

 5) Insert second 100nf cap. Pin 1 Positive and Pin 2 Negative

 6) Connect Positive and Negative Pins to the power rail via Cap.

7) Next is soldiering two jump wires the positive and ground connection on this little Stereo Jack Input.

8) You see 5 Pins, Ground is the centre pin, closest to the input. The two at the top here (actually on the right) are Negative, and the bottom two are positive. TRS stands for Tip, Ring, Sleeve. Tip is positive, Ring is Negative and Sleeve is ground.

9) Here is the Input with jumpers soldiered on. (And a slightly burnt desk :/)

10) Here is the TRS taped down to the BB with leccy tape. The positive connects through Cap 2 to Input Pin 5, Ground to the Ground Rail.

11) 3 N Channel MOSFET Transistors are added to the board. N Channels have three pins, Gate, Drain and Source from left to right.

 12) Each source Pin is connected to the ground top rail of the breadboard.

13) The 220k resistor is now connected to the bottom power rail - live to left, ground to right.

14) Here is another angle to show the exact connection, the ground wire connects to the 33pf capacitor left wire, not the resistor.

15) Here we connect the ground of the MSGEQ7 to the bottom ground rail.

16) Next I insert the 1K potentionmeter and 10K resistor. They are not connected.

17) Here, the ground rails are connected together, a button is attached via soldiered jump wires. The live rail is connected to the positive of the button, the negative back to above the 10K resistor. I have also attached the bottom live rail to the right pin of the potentiometer.

18) Here, I attach a jumper to the central pin of the potentiometer for data transfer to the arduino, I have also connected the jumper from the left of the 10K resistor to the bottom ground rail.

19) Here the left pin of the potentiometer is connected to the bottom ground rail.

20) Here are the connections for the arduino attached to the gate pins of the MOSFETs

21) Finally, I attach the live 12v top rail connector which feeds the LED strip power.

So as far as I am aware, that should be the breadboard and circuit good to go, next post is about all the connections from the Arduino to the Breadboard, there will be a tutorial just like this for that too. Hopefully these can go towards helping people do this in the future too! Cheers