Angry Charlie last stages not analyzable on LTSPICE ?

[Vivek]

Looking at the part around the LED to ground and the Active Baxandall (or is it James ?) that follows :

My understanding is

The LEDs are effectively across the lower leg of a voltage divider made up of R11 and the Baxandall

The impedance of the Baxandall can vary from around 2K to 25 K ohms based on frequency and tone control settings

So the point where the LEDs clip depends upon the frequency and the tone control settings

The input signal that the Baxandall gets depends upon the LED clipping ie on the frequency and the tone control settings

The last chip will also clip with slight bass or treble boost

Leading to a complex nonlinear joint-at-the hips stages of LED and Baxandall that cannot be fully analysed on LTSPICE.

Is that correct ?

[Fancy Lime]

Spice needs to know the DC state at each point, I think. If D2 and D3 are modeled without leakage, then DC between C11 and C12 is unknown. Try putting a 10Meg resistor in parallel with the diodes and see how spice likes it then.

Andy

[teemuk]

Those polarized coupling caps will appreciate steady 0V DC reference as well. 

[Rob Strand]

Leading to a complex nonlinear joint-at-the hips stages of LED and Baxandall that cannot be fully analysed on LTSPICE.

Is that correct ?

A lot of effects pedals have complex interactions.  It doesn't stop you analyzing them.   The trick is to set the controls to something sensible and start from there.  I forgot to add, if the controls are near the center then the interaction with the 470 ohm is minimal.

If simulator is barfing, sometimes it's useful to add 100MEG or 1G resistors from some nodes to ground to help the simulator solve the equations.    The added resistors rarely, if ever, affect audio circuits.

[iainpunk]

Looking at the part around the LED to ground and the Active Baxandall (or is it James ?) that follows

its neither a baxandall or james, since they were invented specifically to have the full power of a 3 band eq, but with only 2 controls. the mid can be cut by boosting both bass and treble and the mids can be boosted by decreasing both bass and treble.

i also think that those led's need a hueg parallel resistance to simulate correctly, LED's can have leakage ranging from the 30-ish kOhm region to the 5MOhm region, so a parallel resistor of in between that region should be realistic, i usually use 1M for simplicity.

cheers, Iain

[teemuk]

Circuits presented by Baxandall and James never really featured a mid-range control per se but if you think about it deeper the mid-range control is nothing else but just combining bass & treble circuits.
James: Passive. Needs logarithmic potentiometers (so that attenuation corresponds logarithmic sensing of loudness) and therefore also component values with logarithmic ratio.
Baxandall: Active. Circuit is somewhat identical but relies on "balanced" feedback impedances to produce unity gain. Therefore needs linear or S-curve potentiometers and component values with 1:1 ratio.

[PRR]

> The LEDs are effectively across the lower leg of a voltage divider made up of R11 and the Baxandall
> The impedance of the Baxandall can vary from around 2K to 25 K ohms based on frequency and tone control settings
> So the point where the LEDs clip depends upon the frequency and the tone control settings

The divider is at most 470:2k or a couple dB. Not a real loss.

> The input signal that the Baxandall gets depends upon the LED clipping ie on the frequency and the tone control settings

When the (Bax-like) tone control is at MAX gain the last opamp will clip long before the LEDs.

> Leading to a complex nonlinear joint-at-the hips stages of LED and Baxandall that cannot be fully analysed on LTSPICE.

Are you actually getting an error? Or just trying to be smarter than SPICE? (You can't beat SPICE on straight math. Save your smarts for figuring out why SPICE is giving a correct answer to the wrong question.)