the legendary tg12412

[etheory]

Low/High pass of the TG12414:

Clear["Global`*"];

(*LPF*)
(*5.6*^-9*)
c1a=c1;
c1b=c1;
c1c=c1;

(*HPF*)
(*1*^-6*)
c2a=c2;
c2b=c2;
c2c=c2;

r28=6.2*^3;
r29=6.9821*^3;
r32=3.3*^3;
r33=12*^3;
r35=1.3*^3;
r36=30*^3;

ic1a=s*c1a*(v4-v2);
ic1b=s*c1b*(v6-v4);
ic1c=s*c1c*(0-v6);
ic2a=s*c2a*(x-v1);
ic2b=s*c2b*(v2-v3);
ic2c=s*c2c*(v4-v5);
ir36=(v4-v1)/r36;
ir35=(v1-v2)/r35;
ir33=(v6-v3)/r33;
ir32=(v3-v4)/r32;
ir29=(0-v5)/r29;
ir28=(v5-v6)/r28;

eqv1=-ic2a-ir36+ir35==0;
eqv2=-ir35-ic1a==0;
eqv3=-ic2b-ir33+ir32==0;
eqv4=-ir32-ic1b==0;
eqv5=-ic2c-ir29+ir28==0;
eqv6=-ir28-ic1c==0;

(*System*)
r1[c1_,c2_]=Solve[{eqv1,eqv2,eqv3,eqv4,eqv5,eqv6},{v6,v5,v4,v3,v2,v1}][[1]][[1,2]]/x//FullSimplify
BodePlot[TransferFunctionModel[{r1[5.6*^-9,1*^-6],r1[4.7*^-9,330*^-9],r1[3.9*^-9,220*^-9],r1[2.7*^-9,150*^-9],r1[2.2*^-9,100*^-9],r1[1*^-9,100*^-9]},s]]

 

[turnerk6]

Okay that makes more sense. Thanks for the help! I think the 30k and 12k feedback(?) resistors are confusing me in that section. What would you call the topology of the inductor-based filter sections? The closest I can figure is a DC bootstrapped Sziklai pair? I'm trying to understand how the resonant bell is formed and controlled by those stages.

 

[turnerk6]

I just whacked the whole schematic into LTSpice, and it's a really cool circuit.
Simulates perfectly - as is.

I tell you what though, it would be pretty expensive (in terms of what I can afford for DIY anyway, especially with those 5P5T and 1P21T switches and inductors) to put together as is, but a cut-down version, where you choose which of the LOW, BL, MED, SH and HIGH curve types you want per band (i.e. fix it to one option for each of the 4 bands), could be quite affordable, especially considering the lack of iron and simple transistors that are still available (I would probably sub BC177 for BCY71) and it would be an amazing thing to have around....

Very very very cool - this is going on my long term list of crazy projects to build.... Thanks so much for posting the original info, I LOVE the curves of the plug-in, but to have a physical version would be even better. I might even have to build a version with just that high band - it's such a nice curve shape and set of frequencies for adding sparkle.

Hey etheory, what transistors did you use for LTSpice simulation? I'm getting weird results with the transistors I'm selecting.

 

[etheory]

Hey etheory, what transistors did you use for LTSpice simulation? I'm getting weird results with the transistors I'm selecting.

It's a very robust circuit. Literally any transistor should work. BC550C and BC560C are the logical modern choices for reasonable audio transistors.
But you can still get BC177/BC179/BC109 and get the original sound if you know where to look.

 

[turnerk6]

Can anyone explain why this circuit attenuates the input signal so much at the first stage? In LT Spice I see around 10dB drop through the first stage.

 

[kpearsall]

Can anyone explain why this circuit attenuates the input signal so much at the first stage? In LT Spice I see around 10dB drop through the first stage.

Headroom before clipping I believe, it’s very common with equalizer circuits

 

[abbey road d enfer]

Can anyone explain why this circuit attenuates the input signal so much at the first stage? In LT Spice I see around 10dB drop through the first stage.

All the active stages are Virtual Earth. Gain is defined by the ratio of the feedback resistor to the input resistor.
The pin labelled In has a gain of zero dB (unity gain).

 

[turnerk6]

All the active stages are Virtual Earth. Gain is defined by the ratio of the feedback resistor to the input resistor.
The pin labelled In has a gain of zero dB (unity gain).

Can you explain this a little for me? Or what is the name of the topology that I can go through an analysis. I've stared at this circuit for hours and still can't understand it any better ha

Which resistors are you referring to as the feedback resistor and input resistor? R29 and R26?

 

[abbey road d enfer]

Input resistor is R25+R26, feeback resistor is R27, so gain at VT2's collector is 12/(12+27) or -10.2dB below input. Actually, it('s a litle less, because the open-loop gain of the stage is not infinite.
The stage is a very basic common-emitter stage (VT1) with an added common-collector stage (VT2), also known as voltage follower.
The open-loop voltage gain comes solely from VT1, and VT2 provides current gain.
Since the operating level is about 10dB below nominal, gain has to be made up in the last stage, which gain is R53/(R51+R52) = 3.16 or 10dB.

 

[turnerk6]

Input resistor is R25+R26, feeback resistor is R27, so gain at VT2's collector is 12/(12+27) or -10.2dB below input. Actually, it('s a litle less, because the open-loop gain of the stage is not infinite.
The stage is a very basic common-emitter stage (VT1) with an added common-collector stage (VT2), also known as voltage follower.
The open-loop voltage gain comes solely from VT1, and VT2 provides current gain.
Since the operating level is about 10dB below nominal, gain has to be made up in the last stage, which gain is R53/(R51+R52) = 3.16 or 10dB.

Thank you for explaining. I guess it's a little bit simpler than I thought it was. I will try to get that bit to sink in. Can you tell me what node acts as the virtual earth?

 

[abbey road d enfer]

Can you tell me what node acts as the virtual earth?

The base of VT1.

 

[turnerk6]

The base of VT1.

Sorry to pester you more but what exactly is the purpose of VT15, the BCY71 at the end of the schematic?

 

[abbey road d enfer]

VT11's collector drives VT12's base. Vt12's collector drives VT13's base. This creates a very high gain inverting stage.
Stabilized by feedback resistor R53.

 

[turnerk6]

VT11's collector drives VT12's base. Vt12's collector drives VT13's base. This creates a very high gain inverting stage.
Stabilized by feedback resistor R53.

Thanks. What about the final BCY71 on the right? Is it clamping the 24Vdc supply down?

 

[abbey road d enfer]

Thanks. What about the final BCY71 on the right? Is it clamping the 24Vdc supply down?

Aah, OK, this one...
It's a capacitance multiplier. It filters out the incoming voltage variations.
Not very efficient IMO.

 

[warpie]

Just curiosity but how much current would such circuit draw? Is it possible to calculate only from the schematic and if so, what would be the way to do so?

 

[turnerk6]

Just curiosity but how much current would such circuit draw? Is it possible to calculate only from the schematic and if so, what would be the way to do so?

My LTspice model has the entire TG12412 circuit drawing about 12mA from the negative rail and about 130mA from the positive rail.

 

[turnerk6]

Aah, OK, this one...
It's a capacitance multiplier. It filters out the incoming voltage variations.
Not very efficient IMO.

Is something like this reasonable/worth doing:
1. Remove the capacitance multiplier
2. Remove R62, R64, and the two zeners that make up the +23Vdc
3. Add the attached triple rail supply for the +24V, +23V, and -28V

OR would you remove the capacitance multiplier, keep R62, R64, and zeners in place, and use a dual rail +/-28V LM317/LM337 based supply?

OR is the circuit perfectly fine as is and it shouldn't be messed with?

 

[abbey road d enfer]

You don't need separate 23 and 24V rails.
I believe the 28V rail is not necessary, since it's not used directly.
Just a bipolar 24V supply is enough. You may have to adjust some resistor values, particularly those that go from the -20V rail to the bases of the transistors.
Anyway modern transistors have a different Hfe than period ones so you'll have to adjust the operating point of each stage.
Remember that monolithic regulators did not exist at the time, which tended to result in complicated designs.

 

[abbey road d enfer]

Just curiosity but how much current would such circuit draw? Is it possible to calculate only from the schematic and if so, what would be the way to do so?

Whenever such a question arises, I always recommand simulating the circuit in LTspice or similar. You get a better answer than any calculation, as long as you have accurate models for the active components.