Active LRC EQ

[mashaffer]

My apologies, I have searched and the only hits I found were full of dead links.

I understand the basic passive LRC notch filter and could probably work out a passive cut only graphic EQ but I am struggling at trying to figure out how to incorporate it into a FB loop for active boost cut EQ.

Any help would be appreciated.

 

[mashaffer]

OK, Using an op-amp tapping Rf works but when using tubes it works great for boost but almost no cut. certainly it cant be an impedance issue can it? Does it need to be a differential input for some reason?

 

[ruairi]

Check out this useful doc from Fred Forssell, it doesn't seem to be on his website anymore.

 

[mashaffer]

Check out this useful doc from Fred Forssell, it doesn't seem to be on his website anymore.

Thank you. I will check them out but I was looking at active.

Perhaps the follower has insufficient current drive or too high of Zout for the chosen values. A SS follower may help. I will follow up on that possibility later today.

BTW I realize I have not accounted for the paralleled pots of the other bands but I put in a parallel 5k in the feedback loop and it of course reduced overall closed loop gain but the limited cut (about 1dB) remained.

 

[mashaffer]

Ok. This is the unit I am basing this on. In fact I hope to be able to use the filter board directly. I will be honest, these complex SS circuits confuse me.


I figure I can build a differential amp something like this but I am at a loss as how the filters are integrated.

 

[mashaffer]

This seems to work. I couldn't figure out how to step the resistances together so ignore the middle lines. The BW seems to be pretty constant but it takes a log pot to make it work.

 

[mashaffer]

Figured it out...use an expression for the second resistance.

 

[PRR]

I do not understand your plans. Maybe cuz I'm not sure where in and out are, and you should use a proper pot part?

The "complex solid state" plan is obvious, once you get past the very bad drawing. Signal come in through a 3.3K resistor. Output comes back through a same-value 3.3K. Pots are hung across the two opamp inputs. LC networks hang off the pot wipers.

Please find R.G. Keen's page on parametric EQ. The truth is out there.

Steve Dove calls this "Swinging Inputs". A mirror-world version is "swinging outputs". Ampeg used this in some tube amps. Again we have matched resistors (except for some reason fudged to slight mismatch), but on the output. Pots across. LC networks from wipers to ground. One side goes to next stage, other side goes back to input as NFB.

 

[PRR]

And then there is the old-old cathode follower output swinger:

 

[mashaffer]

Thank you. I like the looks of the cathodyne... Nice and simple.

I don't quite understand this one. What is Vb? If the 150k input resistor were tied to the top and in+ was grounded it would look like an inverting amp but the input goes to in+ like a non-inverting amp but there is no corrisponding feedback resistor to ground. I could see the array as part of the input resiator of an inverting amp if it wasn't also connected to in+. I suppose the key lies in my first question about Vb.

 

[mashaffer]

Not as high of Q as I was expecting. I suspect a MOSFET would tighten it up a bit.
Now the big question why does it work? FB from plate to its own cathode?

 

[Miles Prower]

Thank you. I like the looks of the cathodyne... Nice and simple.<snip>]

This schemo is very much like this one. When you see an arrangement like that, with the NFB connected to one side of a pot and the input to the other, you have a boost/cut circuit. At the approximate midpoint of the pot, the RLC suck-out filters drop out as they'll be connected to the neutral point. The three 12AX7 sections along the bottom form the "L" part of the filter as active "gyrators" to avoid coils and the associated problems of shielding against hum pick-up. You could ditch the three 12AX7's and replace with series coils to make passive filter sections. It would work the same.

Moving the center of the pot towards the input side removes those frequencies from the input (cut) and towards the NFB end removes them from the NFB (boost). If you had a feedback resistor to ground, you would have a boost circuit only, with no cut capability.

V_b is the non-invert bias voltage. since the op-amp is connected from 9.0V to ground instead of a balanced +/- supply, the non-invert terminal needs to be biased to 4.5V. The output stage naturally splits the voltage so the invert terminal already sees 4.5V.

 

[mashaffer]

Thank you Miles, that makes sense. I was piddling with a Mu Follower LTP and that might be a nice way to implement that.

So far this is the best I have come up with. The 30 Hz slider is a bit of a problem as both the coupling caps and the cap on the 30 Hz filter have to be made very much larger to get a 30 Hz center. Oddly if the filter cap (C1) is eliminated (shorted) correct center can be reached with only moderately big couplers.

 

[mashaffer]

The solution seems to be quite simple. It turns out that eliminating the cathode coupling cap puts the center frequency spot on and alows the other coupling cap to be reduced. Of course that puts the filter network at about 100V so that some caps may need to be changed out due to voltage limitations but since the four lowest frequency are apparently polarized they would need to be changed out anyway. Fortunately the board is layed out such as to make those changes easy. I suspect that the coils themselves should be OK at that voltage.

 

[mashaffer]

Tightens up the Q nicely too.