Trouble with active HPF (high pass filter)

[fezder]

Hehey ferrari, yeah i use filterlab time to time, only bandpass section is quite annoying/hard to use

 

[atferrari]

You need to generate a split supply for it then, there are examples in the opamp 'sticky' at the top of this forum.

https://www.electro-tech-online.com/threads/basic-opamp-circuits.35258/

Probably the best sticky in the forums I frequent.

 

[Nigel Goodwin]

Probably the best sticky in the forums I frequent.

Yeah, that's why I created it - from AudioGuru's excellent posts

 

[audioguru]

For an opamp to be linear, its (+) input must be near half the supply voltage. Its (+) input is at 0VDC when the supply is positive and negative which is "half the total supply voltage".
Then the input can swing up and down allowing the output to also swing up and down.

 

[fezder]

now then, another problem, not with filters but vu section: this is the output from bandpass filter i made, but the next thing i can't figure out is how do i feed signal correclty to my circuits vu-section? it's made of comparators, lm393, wich are connected to single supply. if i hook it like this, some of leds are never off, and that looks quite dumb when no music is going and still it shows bars....and then if i feed signal via bypass capacitor, then it fixes that problem, but then signal is AC and well according to datasheet, lm393 won't like ac fed in input pins. Should i ask this in another/new thread?

 

[fezder]

hmm, unless i change that resistor ladder so there is positive offset? should do the trick

 

[fezder]

yup, that did the trick, thanks for help again!

 

[audioguru]

Why are you trying to make your own LM3915 LED dot/bar driver IC that has 10 comparators, an input that accepts an AC signal averaging at 0V, a built in adjustable voltage reference, simple adjustable brightness, outputs with current regulation and works from a single supply voltage? Each of its steps is 3dB, yours is linear.

 

[fezder]

hehe, yeah i saw that there are ready-made ic's that do what i'm trying to do, but at least this way i learn something, besides it's close. but in future i make this with that simple ic

 

[fezder]

and it's not only vu-meter, idea is that there are 6-8 frequency bands that are checked/sweepd one after another and then each drawn in their own (multiplexed) columns. Here's raw block diagram, there is small difference, linear-log converter right after input amp, basic opamp-diode-thingy. And, i know there is ready-made chip (eq-something? around 3-5$) that does all this, excpect that led-matrix in dil-8 packacke. only thing that needs doing now is that filter section, been doing that currently so hpf (2nd order, 2x1st order HPF->buffer->LPF 2x2st order) But, i gladly hear any stuff you have in your mind that might be useful!

 

[audioguru]

You are making what is called an Audio Spectrum Analyser Circuit. There are many of its circuits in Google Images. Some of them work and others don't.
Many circuits use a Multiple Feedback Bandpass Filter for each frequency band.

 

[fezder]

Audio Spectrum Analyser

well, that's precisely what i was after!

Multiple Feedback Bandpass

i understand some of this, gotta research more of this meaning
hmm, yeah that looks familiar, sallen key or similar?

 

[jjw]

Are you familiar with microcontrollers?
With an 8bit mcu ( Pic, Avr, ... ) running FFT you can make at least 30 band single chip real time spectrum analyzer.
+ some driver electronics for the Led matrix.

 

[fezder]

yeah, buddy made recenty OLED-based spectrum, and i'm somewhat familiar with arduino, suppose that counts

 

[jjw]

There is FFT library for Arduino.
http://wiki.openmusiclabs.com/wiki/ArduinoFFT

 

[audioguru]

A Sallen Key filter makes a good highpass or lowpass but when you add a highpass and a lowpass together they make a poor bandpass. Look in Google to see how a Multiple Feedback Bandpass filter makes a pretty good bandpass filter but it is narrow.
When you multiplex rows of LEDs then since each row is turned off for most of the time they appear to be dimmed. Then you need to blast the current to make them appear bright enough.

 

[fezder]

if i'm not terrbily mistaken, q-factor determines narrowness, right? there are indeed many variations, but i suppose i really need to ditch 358 due the noise, i tested tl071 and it seemed to have lower noise and better overall
about leds dimming during multiplexing, could improving pwm ratio (currently it's 50% around) improve brightness? it shouldn't affect sweep time

There is FFT library for Arduino.

thanks, i'll look and give that more thought, looks handy, and takes a lot less space than this ''bomb''

 

[audioguru]

If you have 5 rows of LEDs where each row is a bandpass frequency, and the rows are multiplexed, then each row is turned on 20% of the time and is turned off 80% of the time so the LEDs will appear to have only 20% of the actual current that they have. You need to increase their current 5 times for them to appear as bright as if they were turned on continuously.

The lousy old LM358 is one of the first low power dual opamps ever made. Therefore it is very slow (trouble above only 2kHz), is noisy and has crossover distortion. The TL07x is made for audio with low noise, very low distortion and good response to 100kHz.

 

[crutschow]

Gyrator bandpass filters are sometimes used in audio circuits since they require fewer parts.

 

[fezder]

hmm, i can see how multiplexing can be quite challenging, when there are say 100 rows or whatever for bigger displays, unless they use charlieplexing.....but iirc, charlieplexing dims even more than multiplexing and is more complex? sorry, bit off-topic