Bandpass for VU.

[neo]

First off, accuracy is not the goal. Pretty lights are the goal, and with pretty lights in mind would the below circuit in conjunction with an lm3916 work in a relative sense?

[danadak]

You will have poor frequency band resolution. And of course signal loss.

Attached a sim, you can play with values to control the response.


There are 4 channel switched capacitor filters -

http://www.newark.com/c/semiconductors-ics/filters-active/switched-capacitor-filters?DM_PersistentCookieCreated=true


You could do it via FFT -

http://mikami.a.la9.jp/meiji/b3/AN42877_12.pdf

https://community.cypress.com/thread/28886



Regards, Dana.

[neo]

Thank you but the math makes my head spin.

[drussell]

You should at least be changing the capacitor values in each band so you don't end up with silly values like 7.6 meg.   

If you get the outputs from each band to be reasonably close, you should be able to trim it so each LM3916 shows the proper level with a (same amplitude, different frequency) test signal in each band and make a reasonable looking "spectral analyzer". 

Personally, I would do it with active filters for the filters.  It would only take a few op-amps to be able to make the filters much, much, much better....

[neo]

You should at least be changing the capacitor values in each band so you don't end up with silly values like 7.6 meg.   

If you get the outputs from each band to be reasonably close, you should be able to trim it so each LM3916 shows the proper level with a (same amplitude, different frequency) test signal in each band and make a reasonable looking "spectral analyzer". 

Personally, I would do it with active filters for the filters.  It would only take a few op-amps to be able to make the filters much, much, much better....

I got a better idea, use the MSGEQ7 with a different display, if i could get it to work.  Also those values were spit out by calculator,  as i said i cannot get my head around the math.

[Audioguru]

Your filters are very poor, an RC as a highpass and an RC as a lowpass that affect each other and have very gradual slopes.
You said "an LM3916" but you need four LM3916 ICs with one bandpass filter at each IC's input.

[neo]

Your filters are very poor, an RC as a highpass and an RC as a lowpass that affect each other and have very gradual slopes.
You said "an LM3916" but you need four LM3916 ICs with one bandpass filter at each IC's input.

I meant an lm3916 per filter, thank you for noticing and they are poor, their also as far as my math skills will take me in the world of bandpass filters.