I'm building a large Spectrum Analyzer to go on my wall in my room. It's my first big electronic project, and I don't want to order all the wrong parts, or mess it up in some other way.

Parts list:

• 13x Caps and resistors on Bandpass chart (Below)
• 13x LM358 Dual Op Amp
• 13x 8 Pin DIP Socket
• 150x 5mm Ultra Bright LED
• 13x LM3914 Bar Graph Display Driver
• 13x 18 Pin DIP Socket
• 13x 6 Pin Male Connector
• 13x 6 Pin Female Connector
• 26x 10k potentiometer
• 26x Toggle Dip Switch
• 13x 1uf Cap
• 13x 1k resistor
• 130x Ping Pong Ball Halves
• 10x Each Colour of AWG 22 Wire
• 1x Dual Power Supply
• 2x 3.5mm Stereo Male Jack
• 4x 3.5mm Stereo Female Socket
• 13x 94x53mm Strip board

(http://imgur.com/CTBty)

How it works:
1) Audio is split (using a simple passive splitter) between the SA and an Audio System
2) Audio is sent to 12 bandpass filters
3) Each filter sends their signal to the bar graph driver
4) The bar graph driver lights up LEDs (ping pong ball halves diffuse the light)

I used this PDF to calculate values for each filter.

- Am I missing anything important, or using the wrong parts?

- When I test a bandpass filter using a frequency generator set at 5khz, and oscilloscope, the output is only brought down by about half (at higher frequencies, the output gets to near zero). Is that normal, and will it affect the end result?

- Am I using good frequencies for the bandpass filters? They range between human hearing capabilities, but I have no idea what the distribution should be.

- My power supply is a dual 12v 2a. Each of the LEDs are 20ma, but wont all be lit at the same time. Should I get a higher amperage supply? Also, is 12v too much?

Thanks for any help!

 

You should use the LM3915 or LM3916 instead of the LM3914 because these work in db steps which are better suited for audio metering.

- When I test a bandpass filter using a frequency generator set at 5khz, and oscilloscope, the output is only brought down by about half (at higher frequencies, the output gets to near zero). Is that normal, and will it affect the end result?

A bandpass filter should attenuate frequencies below and above the center frequency. Don't quite follow the phrase highlighted above.

Might help to post schematics.

It would be a very good idea to only build 1 channel of your complete setup with the LED's you intend to use to be sure that the brightness is adequate and your circuits work properly... just a suggestion.

You can buy 1/2 a ping-pong ball ???

 

Thanks for the reply.

What I meant was, the bandpass filter wasn't cutting of the frequency very accurately. It should cut it off at 1khz, but there is still about a half gain at 5khz.

And I don't think you can buy halves, but I'll just dremel full ones along the seems.

 

Check out the section on bandpass filter design at this link.

http://www.ti.com/lit/ml/sloa088/sloa088.pdf

 

Not to discredit your work, but I would recommend you look into a micro controller that can do a fast fourier transform (fft) like the adruino before spending the time and money tuning all of the op amps you are using. You can then use any type of led driver you can find a library for.

Also remember op amps run on plus and negative supplies. I am not sure if that is what you meant by duel supply. Lastly, your led driver looks like it has a max input of 5v so the easiest way to cap your op amps at 5
would to make the power supply of the op amps 5v (Positive and negative).

 

We don't know which one of the many filters in the link you are using.
We don't know how you are driving the LEDs.
We don't know if each filter feeds an important peak detector circuit.
Without you posting a schematic then we should go away from this thread.

 

Not to discredit your work, but I would recommend you look into a micro controller that can do a fast fourier transform (fft) like the adruino before spending the time and money tuning all of the op amps you are using. You can then use any type of led driver you can find a library for.

Also remember op amps run on plus and negative supplies. I am not sure if that is what you meant by duel supply. Lastly, your led driver looks like it has a max input of 5v so the easiest way to cap your op amps at 5
would to make the power supply of the op amps 5v (Positive and negative).

The project is also a school challenge, and micro controllers are not allowed.

We don't know which one of the many filters in the link you are using.
We don't know how you are driving the LEDs.
We don't know if each filter feeds an important peak detector circuit.
Without you posting a schematic then we should go away from this thread.

- It's the Wide bandpass filter
- I specified the LEDs are being driven by the 3914 (now a 3915)
- Here is the schematic for the 3915, the only thing in the circuit besides the filters and power supply.

I think I've found answers to all my questions but one:
What should the ranges of the 6 bandpass frequencies be? Obviously they should each be between 20hz and 20khz, but I don't know where. You can see my guesses on the image in the first post.

 

Are you supposed to simply COPY the design from somebody else? Then you learning how to cheat and are not learning about electronics.
Usually you are expected to DESIGN a circuit, not just copy it.

The circuit you copied shows the 6 bandpasses and the resistors and capacitors that make them but no schematic.

 

Are you supposed to simply COPY the design from somebody else? Then you learning how to cheat and are not learning about electronics.
Usually you are expected to DESIGN a circuit, not just copy it.

The circuit you copied shows the 6 bandpasses and the resistors and capacitors that make them but no schematic.

It's only a grade 10 electronics project, it doesn't have to be completely designed by me. As long as I can build it and get it working, it's good enough.

And really, is a schematic even necessary now? My remaining questions only involve a simple bandpass circuit.

 

Google "music frequency range".

Edit: You might want to experiment using filters with a higher Q for sharper separation of bands.
Looks like a fun project !

 

How it works:
1) Audio is split (using a simple passive splitter) between the SA and an Audio System
2) Audio is sent to 12 bandpass filters
3) Each filter sends their signal to the bar graph driver
4) The bar graph driver lights up LEDs (ping pong ball halves diffuse the light)

Do not forget to use buffers as well to separate the frequency specified, otherwise, the spectrum analyzer will not work correctly.

Secondly, one red LED (3mm diameter) can be well lit with a current 3 .. 5mA. So, you can think of an LED power consumption ~ 5mA and not 20mA.

After that, it would be good to show the scheme or diagram, because LM3914 datasheet schematic has "small bug" that make it unreliable. Here, ie electronic scheme, would have made ​​some corrections.

 

Grade 10 was a long time ago for me.
Then I made rockets that blew up and I made bombs that went up like rockets, but not for sckool.

I had a soldering GUN that got way too hot but I managed to assemble a few kits with it.

About 43 years ago I made a VU meter with some National Semi LM3915 modules.
The module was a small pcb with the LM3915 as a chip-in-a-black blob and had a display with 10 LEDs inside.

 

Do not forget to use buffers as well to separate the frequency specified, otherwise, the spectrum analyzer will not work correctly.

Would that just be an op-amp with zero gain between each filter and the LED drivers? I don't understand why that's necessary.

 

Here is an audio spectrum analyser project that multiplexes the display so it uses only a single LM391x IC. You can use 10 LM3915 ICs to make the circuit simpler and the LEDs look brighter: