Resonant filterbank idea (schem inside)

[Zero the hero]

I want to share a circuit I breadboarded this evening:

http://topopiccione.atspace.com/temp/uno150.gif

I stumbled against patent number 4117413 by Bob Moog and I tried to mimic this resonant filter bank for making a stompbox version (original patent: http://www.freepatentsonline.com/4117413.html ).
I don't know in which amplifier has been used, so I'm not aware of actual component ratings with exception of the filter section since they are stated in the patent.
The circuit has been breadboarded and it works well, though it cuts some low content of the guitar signal. Middle and high band is preserved and amplified accordingly to the frequency of the resonant filters. 9V DC regulated power supply has been used, and there's a slight distortion in the output; I don't know if the circuit has been designed for higher DC voltage.
I'm too tired and lazy for recording some samples now!

[jrc4558]

Curious. Why an integrator at the output? Something to do with waveshape transformation of the signal? I'm talking about ic2d...

[caress]

link is down...

sounds interesting though, after reading the moog patent.

edit:  ok - you can download the image, though.

[slacker]

you have to copy and paste the link for it to work.

Curious. Why an integrator at the output? Something to do with waveshape transformation of the signal? I'm talking about ic2d...

IC2D is a summing amplifer to mix the signals from the other stages.
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/opampvar5.html

[Meanderthal]

I don't know if the circuit has been designed for higher DC voltage.

I think most Moog circuits are usually +/- 15v. I bet it would clean up some, but 9v(+/- 4.5v) sure is convenient for stompboxes.

[Zero the hero]

Curious. Why an integrator at the output? Something to do with waveshape transformation of the signal? I'm talking about ic2d...

Because I'm naive!
I just need to mix those signal, should C14 be removed?

[Meanderthal]

 No,it might block DC offset. However, shouldn't c14 connect to pin 12(and r22) rather than 13?

[Zero the hero]

Thanks!
The sound seems more brilliant and defined now.

[ambulancevoice]

anyone seen the sherman filterbank?

[Meanderthal]

slight distortion in the output

Just curious... is that gone now?

Also, do ya plan on trying any controls, such as replacing R2, etc... with pots? Or fiddling with cap values in those feedback loops?  Might be pretty interesting...

[jrc4558]

Curious. Why an integrator at the output? Something to do with waveshape transformation of the signal? I'm talking about ic2d...

Because I'm naive!
I just need to mix those signal, should C14 be removed?

I guess C14 is not necessary there... A DC-coupled summing amplifier should be ok, and will allow an all-pass stage for AC signals too... Also no need for biasing resistors.

By the way, always wanted to say, but never had an appropriate chance. I adore your website! Excellent and rare schematics, and photographs are cool too!

[Zero the hero]

slight distortion in the output

Just curious... is that gone now?

Also, do ya plan on trying any controls, such as replacing R2, etc... with pots? Or fiddling with cap values in those feedback loops?  Might be pretty interesting...

Still present, but lower in volume. I'll check with a scope where the distortion is generated, I believe in one or some of the filter stage.
Once I tweak the frequency of the resonant filters, I was thinking of doing a pot-less stompbox (a sort of pedal with just the bypass switch) or adding this filterbank in series with the standard big muff tone control, but I have not much time for experimenting!
Also adding notch filters which cut-off frequency is between the frequency of the resonance is something I would try.

[Zero the hero]

Curious. Why an integrator at the output? Something to do with waveshape transformation of the signal? I'm talking about ic2d...

Because I'm naive!
I just need to mix those signal, should C14 be removed?

I guess C14 is not necessary there... A DC-coupled summing amplifier should be ok, and will allow an all-pass stage for AC signals too... Also no need for biasing resistors.

By the way, always wanted to say, but never had an appropriate chance. I adore your website! Excellent and rare schematics, and photographs are cool too!

THANKS! I'm not very satisfied by some layouts of mine, they could have been better.
I'll remove the cap and the bias resistor, then I'll put everything into a stompbox...

[grapefruit]

R24 and C14 are forming a high pass filter with a -3dB point of 154Hz. I'd increase C3, as omitting it could give you problems by amplifying offset voltage. The wiring of the schematic looks correct. C14 is going to the correct place, the inverting input.

I'd say the filter sections are distorting. You have a gain of 1 at he beginning and a gain of 0.1 at the end (for the filter sections). You'd be best off padding the input, then having a gain of unity or more (or whatever you need) at the output. As the circuit is currently inverting overall, I'd be tempted to keep the non inverting buffer, then have an inverting stage with gain less than 1 (maybe 0.1) before the filter stages. This will give more headroom in the filter stages, an overall non inverted output, and also give you the option of phase inversion for whatever tone control you put in parallel  with the filters. This will work if you control the output gain of the output amp and can still get enough level out of it, otherwise you need a higher supply voltage or rail-rail op amps.

I'd also put a volume control in the filter bypass (wide range tone control) section.

R23 seems redundant to me...
The value of R20 and R21 are around the wrong way, and R21 should be on the other side of C15.

Thanks for sharing the circuit.
Cheers,
Stew.

[StephenGiles]

I got some spamming at that link - probably not good.

[Simon Owen]

I have a half finished Moog Murf alike circuit design based on this circuit.
Expanded to 8 filters - Simulated to roughly match Murf frequency peak values.
PIC sequencer - finished - used a 4 bit dip switch (binary) to select different patterns copied from Murf data sheet.
Can't afford to buy this pedal but I should get a rough aproximation when I finally pull it all together.

[markusw]

Hey Zero,

thanks a lot for sharing your work!!! 

I simulated the circuit in LTSpice (otherwise I can't really figure out what's happening) and I come to exactly the same conclusions as grapefruit.

The filter bands give a massive gain (about 50 dB  ).
Lowering gain to 0.1 before entering the filters seems to be a very good idea. Maybe you need to go even lower.
Filter bands according to the sim are roughly 727, 966, 1370, 1842, 2500, 3500. The frequency response curve looks pretty much like in the patent. So there is definitely a bass loss.
Don't know what output a guitar has at the filter center frequencies. However, running it from +/-15V would help a lot.


The schem below includes the suggestions from grapefruit. According to the sims you get clipping at various bands already at slightly above 1V p/p. Sure that's already quite a level but the input stage as shown reduces the gain to 1/40th.



Hope this helps.

Markus

Edit: Here is a pic for the simulated freq response.

[markusw]

I have a half finished Moog Murf alike circuit design based on this circuit.
Expanded to 8 filters - Simulated to roughly match Murf frequency peak values.
PIC sequencer - finished - used a 4 bit dip switch (binary) to select different patterns copied from Murf data sheet.
Can't afford to buy this pedal but I should get a rough aproximation when I finally pull it all together.

Wow!! 
For me it's hard even to understand the manual without investing a couple of hours 

[grapefruit]

You'd have to check the datasheet for the op amp you're using to make sure you don't draw too much current, but maybe a MAX1044 to get -9V would give you enough headroom. According to Markusw's simulation there's still 12dB of gain in the passbands with 1/40th gain on the front end and 1/2 gain on the output.

I think you need a bass/treble and volume control in parallel with the filters. You may not want to pad the input to this as much, so you could use a voltage divider between the input section and the filters to reduce the level going to the filters only.

The volume control may be slightly more useful if it was on the filters, but unless you want one on each you can't do it and still have a proper virtual earth mixer. If you put it on the bypassed signal put it before the summing resistor. I'd still increase that 470n cap to say 4u7.

Stew.

[grapefruit]

Maybe rather than a normal bass/treble control a parallel 1 or 2 pole LPF in parallel with a 1 or 2 pole HPF with their frequencies set so it balances out the overall frequency response would be better. I can't see you needing to CUT the low or high frequencies.

I'm thinking 2 pole filters near the cutoff frequencies of the lower and upper filters would sound nice.
LPF around  200 or 300Hz.
HPF around 6 or 7 kHz.

Stew.