# Parametric Equalizer with State Variable Filter

I'm new to this forum but hope to get some answers to a project I'm working on. I'm trying to build an parametric equalizer with several bands (5 Bass-Bands, 3 Middle-Bands, 3 High-Bands). In the first picture below you can see the Middle-Band. I use a state variable filter design. It's based on the Urei 545 parametric equalizer. I would like to connect all the bands parallel to each other. I tried but in the simulation I get a loss of cut and boost, almost 5dB. Is this the right way to connect those two filters parallel together (see second picture)? Why do I loose the cut and boost? Thanks in advance for your help. And in general I think my impedance level is to high. I think I need to lower the values of the resistors in order to reduce noise, is that right?

Ian

Here I'm connecting the two middle-bands parallel.

Parallel circuits are loads to each other. Such attempt should be based on math, not on guessing.

If you want to continue on the experimental route, at least have a proper summing amp for the parallel output signals. Otherwise the outputs fight against each other. They are designed to see only an opamp input.

ADD: Fast transfer function calculations showed that your idea - altough realizable with proper summing, leads easily to stability problems without precise component values and math based design. You want a filter, not an oscillator. In addition the adjustments of the bands would surely interact, you want independent bands.

If you can understand the following formula Uout=-Uin(1+2(1-A)F)/(1-2AF) where F is the complex transfer function of a 3 opamp state variable section and A is a real number between 0 and 1 (=cut/boost position parameter) you should see that the denominator should not be zero. How do you can be sure of it? Answer: Only with math based design or by having some incredible luck.

See TimWescott's answer. He's right. Two band filtering is achieved by feeding the same signal through two cascaded filters. In decibels their effect is summed. The cascaded filters are designed to have 0dB gain when the frequency is not near the adjusted center freq.

BTW. Multiband graphic equalizers are not cascaded filters, but resemble a little your idea. Too many cascaded sections detoriate the signal useless due the non-idealty of opamps. I haven't seen multiband parametric EQs which are like those fixed band graphic equalizers (=have parallel circuits), only cascaded ones.

• Thanks for you answer. I could really need some help on that topic, I thought the U4 OpAmp acts as a summing amp. How could I connect the two bands otherwise parallel? I'm really trying to understand analog circuits but I find it quite difficult sometimes. I'd really appreciate some help. – Ian Oct 26 '18 at 23:30
• Thank you very much for your answer. I do see that the schematic was designed to cascade those filters. However, what about a high-end equalizer where you might need 11 Bands to have full control, plus shelving filters and high- and low-pass filters? As you said, cascading to many of those filters will detoriate the signal. But how many is to many? That's why I looked at the concept of grapic equalizers where many bands are put together parallel. I thought I could adapt this topology to this parametric equalizer. How else would you design a parametric euqalizer with 11 bands? – Ian Oct 27 '18 at 12:50
• @Ian Good enough (=acceptable for high end studios) design probably is out of my range. Sorry. I would try to do it in digital domain. There's no limit if one uses enough calculation power for high resolution math. But that's computing, not an opamp circuit. Programming it to have an usable user interface and real time speed is also out of my range. – user287001 Oct 27 '18 at 13:08
• @Ian Check this. If you can understand the math, you have a good possiblity to have some progress. mysite.mweb.co.za/residents/cyb00746/audiodocs/Analog_EQ.htm – user287001 Oct 28 '18 at 0:13

If you look at the schematic for the Urei equalizer, you'll see that the equalizer sections are cascaded -- not run in parallel. Filters change the phase of signals, and adding signals of different phases at the same frequencies does odd things.