Ampeg Sub Blaster is kicking my butt

[bean]

I'm getting some bleed-through from the rectifiers and low level motorboating from the octave down mixed into the Sallen Key filter before the Sub pot. I just cannot figure out why. I've modded the stage before the rectifier to reduce the gain from it (there was significant oscillation at its output with the stock values) and that seemed to help some but not enough. All voltages good, tried different JFET subs, etc. This thing sounds excellent but the noise problem is a real deal breaker! I also tried an extra 220uF on the VB rail to see if it helped and nope.

Here is the reference schem and my redraw. I have a number of mods but the problem exists built stock or modded. Any thoughts are welcome.

[Rob Strand]

It seems anywhere you have gain there are issues.   Maybe the sign of some bigger problem or something overlooked, like a layout/build bug.

The 47k & 470k resistors on the phase inverter are another source of gain.   They will amplify any noise in the area of the JFET.

You might be able to reduce the damage by connecting C2 to ground instead of Vcc/2.   That will prevent some of the noise on Vcc/2 getting amplified.   

Some other things to try would cap small caps across the feedback resistor on IC1_A, perhaps even across R12 (although that doesn't quite help in the non-inverting state.)   The idea being the issue is possibly high frequency related.  You might be able to rule that out with a scope but it's easy to add the caps and see.

[m4268588]

R28 is the impedance shared by the two circuits to be separated.

I don't know if this will be solved

[Fancy Lime]

My 2 cents: almost everything in this circuit, including the flip flops is DC connected to Vb, one way or another. That's asking for low frequency noise trouble in my experience. I think you may be getting bleedthrough via Vb. Maybe try to split the circuit into AC coupled "audio" and "sub generator" parts with a suitable cap before or after the JFET and give each their own Vb. Possibly buffer both Vbs with an opamp.

HTH
Andy

[Elektrojänis]

I don't see anything that would provide any kind of volume envelope to the sub octave signal. Am I missing something?

If the sub octave is always full on and and the triggering of the frequency divider is sensitive, even tiny amount of noise in the input could cause somewhat random triggering and cause some sort of low frequency disturbance. It could just be the nature of the circuit.

I it is this, it needs some sort of imbalance or hystresis on the triggering. Some sort of threshold the signal must exceed before the divider works.

[antonis]

almost everything in this circuit, including the flip flops is DC connected to Vb, one way or another.

I shouldn't also be suprised for audible crackling pots due to DC offset between op-amps outputs and VREF1..

[amptramp]

The noise from the TL062's themselves usually means that they cannot be used in a signal path and I have only seen them used in tremolo oscillators or somewhere like that where the noise can be filtered into oblivion.  A TL072 might be a better choice and if you have the IC's socketed, it would be simple to try this out.

The TL072 has a noise of 18 nV/SQRT(Hz)

The TL062 has a noise of 42 nV/SQRT(Hz)

If noise is flipping your CD4013's the lower noise op amp may alleviate some of this at the cost of more current drain.

[ElectricDruid]

My 2 cents: almost everything in this circuit, including the flip flops is DC connected to Vb, one way or another. That's asking for low frequency noise trouble in my experience.

+1 agree. This has been my experience too.

What I'd do:

1) Move anything that doesn't have to be on Vb to Ground (C2, C4 and C6). I'd think about adding DC blocking caps in front of the pots, and then taking them to ground too. That reduces the ways that noise on Vb can get into the audio.

2) Use a separate Vb for the remaining Vb connection at R3. Then there's *no* route from the Vb used by the "digital" side (rectifiers and flipflops) to the audio.

3) Route the ground for the digital part entirely separately from the main audio ground path.

[duck_arse]

anytime I've used 4013 to divide by two, I get stupid micropulses instead of square waves, from bad triggering. perhaps a small resistor between D and barQ with a small slowdown cap to ground.

[ElectricDruid]

anytime I've used 4013 to divide by two, I get stupid micropulses instead of square waves, from bad triggering. perhaps a small resistor between D and barQ with a small slowdown cap to ground.

Hasn't this come up before? Old 4013's don't do it, but the modern ones are faster and have more problems.

Incidentally, practically the first bit of "sound synthesis" I ever did, aged about 11 or 12, was to feed a yamaha portasound keyboard through a 4013! That raw square sub-octave gave it a bit more "welly"!

[bean]

Super, lots of excellent ideas here. Thanks guys. I'm going to put it on the breadboard and ground everything I can, add coupling caps, etc.