QRe: [sdiy] Reprinting Electronotes

[mark verbos]

The Buchla 296 has 16 bands  made of  3 staggered serial bandpasses 
centered at....
lopass at 100
150
250
350
500
630
800
1k
1.3k
1.6k
2k
2.6k
3.5k
5k
8k
13.2k (labelled >10k)

the Buchla 295 has bands made of a single bandpass centered at...

lopass at 100
200
350
500
700
1k
1.4k
2k
3.5k
hipass at 7k

these claim to be spaced to "compliment the discrimination curve of the 
human ear". Could be a good choice, since that's what you and the 
majority of your audience will have. ;)

the Serge resonant EQ has bands, I assume made of single bandpasses, 
centered at...

29
61
115
218
411
77
1.5k
2.8k
5.2k
11k

this one is regarded as "great sounding". It's bands are not spaced by 
octaves so it doesn't accentuate any particular note in a scale. It also 
claims to be resonant past a certain level, I don't know how this is 
implimented.  Maybe it is global feedback? The panel makes it look like 
it is actually an equalizer that has no gain in the middle of the knob 
and can boost or cut, not like a filter bank where you have no gain at 
the top of the control and and full attenuation at the bottom.

good luck.

mark



 

jbv wrote:

>
>Ian ,
>
>>But if someone wants to make a filter bank, well, jeeze, that's a really
>>simple thing to figure out how to do.  Most of the relevant non-obvious
>>information has been posted (for free!!!) on this thread.
>>
>>Here's what you do:
>>
>>1.  Pick out a good low noise opamp.
>>
>>2.  Search the web or some textbook for opamp bandpass circuits.  There are
>>zillions around.
>>
>>3.  Pick a circuit and use the textbook equations to design a set of
>>filters with Q of 25 and frequencies spaced at a ratio of 2.1^(1/5).
>>
>>4.  Design a mixer (or just a summer) circuit to combine all the outputs.
>>
>>5.  Add some feedback to vary the Q.  (Either locally, in sub-banks or
>>globally. You get to choose!)
>>
>>6.  Build it and tweak up all the frequencies.
>>
>
>Great, you're answering my questions before I find the time to ask them...
>
>As for "frequencies spaced at a ratio of 2.1^(1/5)", how many bands per
>octave does that mean ? Three ? Five ?
>And what are the practical frequency limits (both high & low ends) for
>such a module ? 100 Hz to 10 KHz ?
>
>Thanks,
>JB
>
>
>