[sdiy] equalizer

[Seb Francis]

I know it's not a maths book, but it's a good one ...

The Art of Electronics by Horowitz and Hill
http://www.artofelectronics.com/

Has an excellent chapter on active filters (chapter 5)

Seb



cheater cheater wrote:

>That's how I understand it as well, "roughly", but I do think that
>it's a naive train of thought and that there's more mathematics to it
>than just that.
>Speaking of which, can someone point me to the lowdown rough
>mathematical explanation of filters?
>Don't worry. I'm a maths student. I'll survive 8)
>That's something I was always and always looking for... PLEASE!
>Note that if it's a book, I most possibly can't get it. Unless it's a
>maths book.. then there's a very slight chance. :P
>
>Anyways...
>I didn't know Q was related to the feedback.
>Hrm. interesting
>You mean the feedback circuit around the single opamp/transistor stuff
>right? (I'm not that great with filter circuits... sorry if this is
>one of those "dumb" questions ;) )
>Putting the whole filter in a feedback loop would probably have a
>slightly different outcome, right? I guess it would still start
>oscillating at the peak...
>
>cheers,
>D.
>8)
>
>On 10/10/05, Seb Francis <seb at burnit.co.uk> wrote:
>  
>
>>this is my rough understanding of this (I'm sure someone will correct me
>>if I'm incorrect)...
>>
>>dB/oct relates to the number of poles (stages) in the filter.  you get
>>6dB per pole, so a 36 dB/oct filter is 6 pole which is quite a
>>complicated circuit.
>>
>>the Q comes from adding feedback to the filter
>>
>>Seb
>>
>>
>>cheater cheater wrote:
>>
>>    
>>
>>>shall keep that in mind 8)
>>>what about filters of higher order?
>>>like 36 dB/oct?
>>>there's more to them than just Q - right?
>>>is there even a "standard" 36 dB/oct filter?
>>>
>>>cheers,
>>>D. 8)
>>>
>>>On 10/10/05, harrybissell <harrybissell at prodigy.net> wrote:
>>>
>>>
>>>      
>>>
>>>>.707 gives a smooth roll-off just like two cascaded RC sections...
>>>>1 gives a sharp corner
>>>>anything above that has a really obvious peak.
>>>>
>>>>I've found that if I want really smooth response, even a Q of 1 might be
>>>>too much...
>>>>
>>>>H^) harry
>>>>
>>>>cheater cheater wrote:
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>>>I'll keep the book in mind,
>>>>>thanks!!
>>>>>
>>>>>How did you come up with "Q between 0.707 and 1"?
>>>>>
>>>>>Cheers,
>>>>>D 8)
>>>>>
>>>>>On 10/9/05, harrybissell <harrybissell at prodigy.net> wrote:
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>cheater cheater wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>Well, what I want is a narrow cutoff (like 24 dB/oct? no idea - hints?)
>>>>>>>Usual equalizers have 6 dB/oct - right? Or is it 3?
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>Most are 12dB, some are 18dB.. I don't know of any common 24dB designs.
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>So at 24 dB/oct:
>>>>>>>1. The BP would be real narrow and resonant and crappy and shtuff
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>24dB could still have low, or high resonance.  Q of .707 to 1 would not have
>>>>>>a pronounced peak. BTW the BPF of the state variable is only 1/2 the cutoff
>>>>>>rate of the HPF and LPF.  The 'state variable' design in most commonly a
>>>>>>12dB slope. You 'can' co a 24dB but it gets complex
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>2. The HP and LP would already give me the whole spectrum. So the BP
>>>>>>>would only add stuff - am I right? Or would HP+LP=everything-BP? :)
>>>>>>>
>>>>>>>And in other news:
>>>>>>>can anyone recommend a filter that:
>>>>>>>1. doesn't have a very resonant sound
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>any one with a Q of .707 to 1...
>>>>>>
>>>>>>There are a number of standard filter designs. For memory I will fvck this up
>>>>>>and my listmates will jump my bones. They are Butterworth, Bessel, Chebychev.
>>>>>>One has better flat gain in the passband, one a better cutoff but small peak at the
>>>>>>corner, one has ripple in the passband but even sharper cutoff. They have different
>>>>>>shape of phase shift as well.  Decisions decisions. Most folk use the Butterworth...
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>2. has a pretty narrow cut-off (24? 12? 18 dB/oct? which ones would you choose?)
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>12dB
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>3. has a pretty uncomplicated circuit (I'm going to have to have a lot
>>>>>>>of these in an equalizer it seems - and even more in a whole mixer -
>>>>>>>woe is me!)
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>There are a lot of circuits. I'd start with one I like, then copy that first to get
>>>>>>experienced..
>>>>>>
>>>>>>4. (a plus but not needed) sweeps don't have the usual cheezy resonant whistle sound
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>This goes for LP and HP. I'm counting on you synth freaks here!
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>If you need to sweep the filters, for sure the higher cutoffs will need more
>>>>>>components. Usually a 12dB needs two tuning elements (dual pot, ota etc)...
>>>>>>24dB needs 4 etc.
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>Regarding 3 - am I right to think that SMD components would make the
>>>>>>>[6x2x5=60 (!!!!) (4 bands)x(stereo)x(5 channels)] SIXTY filters easier
>>>>>>>to tune up? Don't they usually fluctuate less? In any case I guess
>>>>>>>it's a must considering space and such...
>>>>>>>I'm *so* buying metal film resistors... and 2% caps.... *cringes*
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>and good caps are hard to find... especially in SMT
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>Also: anyone got the Electrix EQKiller schematic?
>>>>>>>Or the Vestax DCR-1200 Pro?
>>>>>>>I'm very curious what kind of filters they're using, and what order they are.
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>I'd recommend that you get a copy of the old National Semiconductor "Audio"
>>>>>>or "Audio/Radio Handbook"  (out of print but I think someone sells reprints). The
>>>>>>ICs shown are largely obsolete but the concepts are still quite valid, and it is
>>>>>>written
>>>>>>for the common man (well, common man skilled in the art :^) to understand. It shows a
>>>>>>graphic EQ and a 'room equalizing instrument' that are of interest.
>>>>>>
>>>>>>H^) harry
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>Cheers!
>>>>>>>D.
>>>>>>>8)
>>>>>>>
>>>>>>>On 10/8/05, harrybissell <harrybissell at prodigy.net> wrote:
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>>>Steven Cook wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                
>>>>>>>>
>>>>>>>>>A state-variable filter has three simultaneous outputs: lowpass, bandpass
>>>>>>>>>and highpass. I suspect that mixing all three outputs together would
>>>>>>>>>reconstruct the input signal with reasonable accuracy.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                  
>>>>>>>>>
>>>>>>>>True... but the corner frequencies cannot be (individually) controlled... and are
>>>>>>>>
>>>>>>>>really unlikely to be useful in an EQ application.
>>>>>>>>
>>>>>>>>H^) harry
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                
>>>>>>>>
>>>>>>            
>>>>>>
>>>>        
>>>>
>>>
>>>
>>>
>>>      
>>>
>>    
>>
>
>
>
>  
>