Giant cylindrical color organ/VU meter for a nightclub

[Tony Stewart]

Linear scale amplitude LED scales are pretty uninteresting. Audio dynamic range is over 100dB but music in bars might only be 50 dB SNR. This is why music with low bass doesn't even register.
To make the visual experience match the audible perception, one needs at least a 40 dB range or 100:1 or about a 7 bit binary scale, 1,2,4,8,16,32,64,128, or 46 dB range would be 8 bit or ...256:1 dynamic range. This can be done with a diode feedback log amp to drive the linear scale LED driver for each band .


But in retrospect Windows Media Player has much more interesting Visualizations, and these are nothing compared to other players.
I would consider a ceiling projector down onto the glass with a lens to focus on a skewed depth, not sure, using exquisite visualizations from media player on a PC.

But if you want to emulate an LCD backlit glass you need a light diffuser of white translucent plastic on one side if you drive the glass edges with LEDs, top and bottom. Perhaps some other method to diffuse the light evenly.

What is the glass thickness? Like a curved glass shower ~10mm?

Even modulated colours are not that interesting

like this is pretty boring
https://goo.gl/jGk7D8

unless it's a whole wall with other special effects with nice saturated colours or you have a multi million dollar budget like American Idol.

 

[mjkelly93]

OK thank you for your explanation. I have a much better understanding of how im going to go about this after all of the advice i found here on these boards.

My parts should be here either over the weekend or early next week and im going to try and put a prototype together. I'll report back after ive tinkered around a little.

 

[mjkelly93]

Tony im not doing this on glass or a wall. Im going to make a 3 foot wide cylinder thats 10 feet high with panels similar to the c0lumns in my club now.

Its going to react similar to this with the way it emanates from the center of the cylinder outwards. It starts at 40S when it embeds the video it removes the time part of the URL

But its going to look like this with these panels except instead of being verticle 12" pieces, they will be horizontal like 12" tall stripes on the cylinder. These are actual columns i built in my other club using a MSGEQ7 and arduino.

Another video to show better change in the colors with the frequency:

So its going to emanate outwards from the center like video number 1 but the panels will react like video number 2 and the more intensity of the audio the farther outwards the panels will light up, like an volume unit meter.

 

[mjkelly93]

OK well today my parts came in from tayda and ive been at it all day. Surprisingly, i was able to get it to work right but it needs ALOT of tweaking to get it right.

I did eleven 1' strands of SMD 5050 to test it out. I followed audio guru's advice on dividing the voltage using resistors. The signal comes out from the arduino to the first RGB set of transistors this goes straight to the first line in the middle. Then, i take that signal and using a resistor pass it on to another set of RGB transistors. This transistor runs the layer of LEDs above and below the middle one. Then i do the same for the next set. That controls one more layer out.

I decided to use 1/2 watt 1M ohm resistors throughout and it definitely proved that the concept can be done. However, i think they need to be stepped a little more. Theres too much resistance from the middle line to the nearby ones so im going to be tinkering with different resistor values for a while to get it dialed in right. Also, it seems that after the 3 middle rows, the outer 6 rows just light up at once rather than stepping down the intensity like its supposed to. Strange. I have to assume this has to do with using the wrong sized resistor to step it down further.

Im thinking maybe like 200k then 400k then 1M then maybe 2M, something like that.

Anyway, heres a crude video of them in action. They get washed out a little but you can get the idea. Plus, they seem to stay dimly lit even when theres no sound so idk wtf is up there. My other ones dont do that. Maybe interference? I dont think so bc theres an interference variable and when you crank it up, they still stay lit.

 

[mjkelly93]

Dialing it in a little. Instead of 1M ohm resistors, im now doing 220k, 110k, 1M, 1M and its looking better but i bet i can dial it in even further. Check it out now.

 

[audioguru]

Yes, I like listening to the latest songs by Ellie Goulding.

The sensitivity to loudness of our hearing is logarithmic or exponential which is why the 10 steps of an LM3915 are not linear but are 3dB apart (1.414 times more level than the next step) then each step is only a little louder. The voltage divider resistor values in the LM3915 show the logarithmic level steps. Since each step is 3dB from the next step then 10 steps are 30dB apart which is from low loudness to fairly loud or from fairly loud to very loud. The total range from quiet to very loud would be maybe 100dB or 120dB.

Years ago in a disco I measured peaks of 120dB and the ambient without music was 70dB so the range was only 50dB.

Your resistor values in your voltage divider seem to be too high for transistors and might be better suited to drive low input current opamps. Please post your schematic.

 

[mjkelly93]

OK ill try and get a schematic together.

I think im going to scale it down a little more yet. Right now i think its 100k, 220k, 270k, 330k, 470k, and 1M ohm.

Im thinking about trying like maybe 50k, 80k, 100, 220k, 270k and on.

This is the newest video with 4 more bars. You can notice how the middle 5 bars react the best so im thinking the resistance is still too high.

 

[audioguru]

The middle 3 bars look like they flash the same.
The other bars do not flash, they stay dimly lighted.

 

[mjkelly93]

Its been corrected except youtube is taking 6 hours now to upload a 1 minute video.

I decided to do 22k, 47k, 100k, 156k, 220k, 270k and they look much smoother.

And, taking video staring directly at LED's misrepresents how they are reacting as well. I take the video and then i look at the video and im like thats not exactly how they look in person.

Here is that schematic though. This only shows the color blue though.

 

[Tony Stewart]

MJ Kelly your schematic doesn't even come close to replicating the video. with the VU color frequency peak ONLY level on with hold effects, rather than peak and lower threshold ON and no HOLD time.

Forget the design and work on the algorithm, first in words then in numbers

Parameters like Spectral band , peak range, hold range, Bank color and intensity decay time.

 

[mjkelly93]

Well tony i have to respectfully disagree since i know the schematic like the back of my hand and im sitting here watching it work right in front of me as i type this.

I looked over the diagram twice before i put it up but maybe i still missed something. Tony what do you see is wrong?

 

[Tony Stewart]

Note the outer level is RED Only, then a few blank levels then 2 GREEN then a blank level then 2 BLUE from centre.

And this is just a snapshot in time. The colours are not gradient with audio levels at all.

 

[mjkelly93]

I know im not trying to build that. I was just using that video to demonstrate how they pulsate from the center outwards.

I couldnt find a video showing what im trying to do. That was the closest thing i could find.

 

[Tony Stewart]

symmetry from centre is easy. doing the rest must follow my suggestion

 

[audioguru]

There is no schematic. Instead it is a photo of a solderless breadboard layout with messy wires all over the place.
The four resistors in series do not make a voltage divider.
Comparators are missing (to turn on the Mosfets).

The LM3915 has comparators that compare the audio input level with a logarithmic voltage divided DC reference voltage. When the audio input level reaches the comparator's reference level then its LED turns on. Then each output turns on at a different audio level.

 

[mjkelly93]

Right... well Tony i didnt follow your suggestion and they came out exactly as i wanted.

And audioguru, i appreciate your criticism, i really do, but its working exactly as i wanted it to. In fact, it looks better than i thought it would. And as for the diagram, i couldnt care less if its messy or not. Its not like i sketch these things out on a daily basis.

I chose not to use the LM3915 because i wouldn't be able to assign different colors to the different frequencies of the signal like i could with the MSGEQ7 and the arduino.

And i apologize for not knowing all the proper terminology throughout this thread. Im a business owner not an electrical engineer.

Its now 6 hours of youtube uploading a 1 minute video. If it ever finishes, ill post it.

 

[Tony Stewart]

Ok then what u showed and wanted are different

try inverting it with outer rings together going towards the middle the peak sound coincides like a handclap of thunder

 

[mjkelly93]

Yes thats what i said earlier. I was just using the video to demonstrate how the signal radiates from the center outwards.

And it is sort of representative of what the columns will look like when they are done. But it will look wayyyyy better than that video.

 

[mjkelly93]

OK so i decided to start uploading a new video and the other finished a minute later. How about that.

I think i dialed it in slightly more after taking this video but this is definitely with the new resistor line up.

And thats an interesting idea inverting it so it comes together in the middle. After they are built, ill definitely give that a go.

 

[audioguru]

An LM3915 can do what you show in your new video. The center LED line is output #1, the next two lines away from the center are output #2, then the next two lines are output #3 etc. You can reverse the order so the lines light from the outside towards the center as the loudness increases. The LM3915 allows a BAR mode or a DOT mode. Your new video shows a BAR mode. If you add a peak detector as shown on the LM3915 datasheet then each maximum level will light the lighted LEDs at the same brightness but on your video the instantaneous levels cause the outer lighted bars to look dimmed.

Each LED color needs its own LM3915. Each LM3915 is fed from its own bandpass filter.