AC noise in audio ckt

[rolerbe]

I'm working on a 'Fuzz box' for a bass guitar, ckt diagram below. With fuzz on and gains up, I'm getting a strange (to me) AC artifact -- Oscope images below (AC waveform from isolation xfmr at 10V/div, signal out at 2V/div).

With fuzz on, and the input shorted to ground, the artifact is timed to the AC peaks, and these peaks are flattened. With fuzz off, the AC peaks are 'normal'. The artifact is present (tho much smaller) at the input to the first amp stage, but I think it is somehow related to feedback from the clipping diodes (which make the fuzz).

Any thoughts? Thanks in advance.

[CeaSaR]

At first glance, C8 is backwards. Re-orient so that + is towards V+ from
U2. Also, check the orientation of the 10 uF cap at R5 as it also appears
backwards. If that doesn't work...

What is the power supply? If it is from a wall source, the ripple may be
coming from there. To verify, replace the power supply with the
appropriate battery voltage and look for the ac artifact. If it is gone, you
need to clean up the supply coming into the circuit. Make sure C8 is large
enough and possibly put a small (0.01 uF) bypass cap close to the VCC of
all IC's to keep spurious noise away from them.

CeaSaR

[Bigglez]

With fuzz on, and the input shorted to ground, the artifact is timed to the AC peaks, and these peaks are flattened.

Regarding your schematic, what is U2?
Several of the components (R1,R2,R11, and C8) have
no values.

Where does the power for this project come from?
I don't see a battery or AC adapter. For testing is
this unit operated from an external supply?
What frequency is the unwanted output (when the
input is shorted out)? Does it vary (or stop) when
the controls are changed?

[CeaSaR]

BTW, do you have a guitar cable plugged in when you have this artifact?
If you do, unplug the cable and check again. If the artifact is gone, you
have a grounding problem.

CeaSaR

[rolerbe]

Thanks for the replies. Some clarifications: the box labeled U2 is just the +/- voltage source for the 4053 electronic switch. I eliminated the power supply part of the schematic from the diagram for clarity, as it is just a vanilla 7805 / 7905 dual supply. I do have high freq filter caps on both rails of the pwr supply.

The cap polarity may be reversed in the diagram, only because I don't sweat that too much in the cad program I use. They are properly oriented in the actual circuit, or are non-polar (I did go back and physically check this, so it was a good suggestion).

For these traces, the guitar is not plugged in. To eliminate the input jack as a source of noise, I shorted the tip to ring (ground) on the input jack itself. Output jack is either connected to speaker or not -- no change.

When the gain or drive controls are changed, the magnitude of the signal changes. These images are at max gain and drive for both stages. It appears to be originating in stage 1 input, then just gets dutifully amplified by both stages.

Battery power testing is a good suggestion. Not sure how I'll get to that as I built the power supply on the same PCB. (note to self: Maybe I should put these on separate PCB's for future projects).

[rolerbe]

Not at the project at the moment, but R1 is either 100K or 1Meg, I think, and just added to make sure the Fuzz on LED path in the 4053 switch is not floating with Fuzz off. R2 is somewhere in the range of 470 to 680, to current limit the diode. C8 is a local bypass cap for the 4053, probably 100pf.

[Bigglez]

the box labeled U2 is just the +/- voltage source for the 4053 electronic switch. I eliminated the power supply part of the schematic from the diagram for clarity, as it is just a vanilla 7805 / 7905 dual supply. I do have high freq filter caps on both rails of the pwr supply.

With +5V and -5V supplies the op amps can be referred
to ground. Why are there 10uF coupling caps in each op
amp circuit?

For these traces, the guitar is not plugged in. To eliminate the input jack as a source of noise, I shorted the tip to ring (ground) on the input jack itself. Output jack is either connected to speaker or not -- no change.

What is the timebase rate of the scope PIX?
(To measure the frequency of the oscillations).

When the gain or drive controls are changed, the magnitude of the signal changes. These images are at max gain and drive for both stages. It appears to be originating in stage 1 input, then just gets dutifully amplified by both stages.

None of the controls are labelled. Which is which?
Also, both op amps have variable resistors in the
feedback path, which as drawn could be made zero
(or there abouts). Why are the op amp gains
made vairable but each op amp is followed by
another variable resistor attenuator?

Is this your original design or was it lifted from
another product or source? Has it ever worked?

[rolerbe]

Thanks for the questions.

Actually, except for the AC noise, the box works quite well. high degree of control over the sound quality and degree of distortion. I get good general control over degree of fuzz by the drive pot (10K) output from input Amp A to the diode clippers. But if I want extreme fuzz (very square corners), gain up amp A (1M) to drive that amp close, or into saturation as well. I can gain down and eliminate almost all the buzz, but at the expense of some audio distortion control. And I just want to figure out what's causing it. Haven't done much with op amps, so it's a 'real world' learning exercise.

Gain for amp B (10K || 100K for pseudo audio taper) allows for adjustment for a range of instruments, some with, some without level controls on the instrument itself. Then the amp B drive pot gives finer control over matching audio level between fuzz on/off. perhaps a bit of overkill. The design evolved from an old modern electronics magazine article (circa 1988 or so). I discovered the hard way that virtually no part of the magazine design worked as intended. So, this is essentially a new circuit, loosely derived from that published design. The coupling caps are holdovers from the mag design, but does eliminate any bias in the opamp (which I measured at one point and it was very small).

I don't recall the timebase setting, but the sine-wave shown is 60 Hz taken directly from the AC transformer, which is a wall-wart plug in. So, it must be 2msec/div. This feeds the rectifier bridge to the regulated PS. Output from the regulators looks clean with/without load.

[Bigglez]

Actually, except for the AC noise, the box works quite well.

Fom your scope PIX the "noise" isn't noise,
its 60Hz AC leaking in somewhere. Noise would
look like 'grass' on the scope trace.

degree of fuzz by the drive pot (10K) output from input Amp A to the diode clippers.

So the goal is to clip the input peaks symmetrically?

if I want extreme fuzz (very square corners), gain up amp A (1M) to drive that amp close, or into saturation as well.

So the clipping is hard not soft (as seen in an overdriven
tube/valve amp)?

I can gain down and eliminate almost all the buzz, but at the expense of some audio distortion control.

buzz = fuzz (a typo?) or is buzz another musician term
for AC mains interference?

Gain for amp B (10K || 100K for pseudo audio taper) allows for adjustment for a range of instruments

Wouldn't the different instrument signals
be leveled before the clipper? Once clipped the signal
has a very defined peak to peak value. Wouldn't that
signal be level adjusted to match the input of the amp
or mixing board?

I don't recall the timebase setting, but the sine-wave shown is 60 Hz taken directly from the AC transformer, which is a wall-wart plug in.

So the rogue signal is not only
60Hz, its AC line locked too? You have to find where this
is being injected. The power supply is the obvious
first choice, but any inadvertent ground loop would
do the same. Placing pots in Op Amp feedback is
unwise, and if these are off-PCB there is your antennae
that can pick up crap from any nearby source.

[rolerbe]

From your scope PIX the "noise" isn't noise,
its 60Hz AC leaking in somewhere.

True.

So the goal is to clip the input peaks symmetrically?

Yes. This is a necessary condition or it sounds harsh and 'unmusical'.

So the clipping is hard not soft (as seen in an overdriven
tube/valve amp)?

Not sure I know the difference. The diodes do hard clip the inputs, unless the signal is small. Depending on the gain, you can get rounded shoulders with flat tops, or true squares. this is the 'degree of fuzziness'.

buzz = fuzz (a typo?) or is buzz another musician term
for AC mains interference?

musician term for the interference. That's exactly how it sounds.

Wouldn't the different instrument signals
be leveled before the clipper? Once clipped the signal
has a very defined peak to peak value. Wouldn't that
signal be level adjusted to match the input of the amp
or mixing board?

Yes, exactly. Gain in stage 1 is only to 'square the wave'. Higher gain in this stage gives sharper rise on the signal, rather than rounded sides. Then clipped to +/- 0.6V by the diodes. Then some gain in stage 2 is necessary to match fuzz on/off volumes.

So the rogue signal is not only
60Hz, its AC line locked too? You have to find where this
is being injected. The power supply is the obvious
first choice, but any inadvertent ground loop would
do the same. Placing pots in Op Amp feedback is
unwise, and if these are off-PCB there is your antennae
that can pick up crap from any nearby source.

Well, ample sources of error here. Right now, the pcb is just sitting on on the desk. Pots with fairly long lead wires (not twisted), to mount as controls in the eventual box. Plastic, not metal box initially contemplated. Who knows about ground loops as I laid out and cut the PCB myself and have very little practical knowledge here. Flourescent lights in the overhead radiating who knows what?

But, to your original point, I see very little 'noise'. This is true signal coming from somewhere. I guess I'll look more at the PS, and try to think of a way to 'shield' the board as a test before final box assembly.

Great input. Thanks!

One more question tho, If not front panel pots to control op-amp gain, what 'should' be done?

Thanks again.

[Bigglez]

So the clipping is hard not soft (as seen in an overdriven
tube/valve amp)?

Not sure I know the difference. The diodes do hard clip the inputs, unless the signal is small. Depending on the gain, you can get rounded shoulders with flat tops, or true squares. this is the 'degree of fuzziness'.

Okay, here's more about electronic musical effects
Hard clipping removes the waveform above a threshold,
but retains the zero crossing. Typically used to protect
the following circuit, such as overload protection to a
moving coil meter.

Soft clipping gently attenuates the waveform above a
threshold, and in proportion to the amplitude. Another
name for soft clipping is amplitude compression. It
limits the signal level without distortion (the exact
opposite of 'fuzz'!). Typically used before a transmitter
to prevent over-modulation without distortion.

A tube amp, if overdriven, shows soft clipping and
is widely thought to give the 'tube sound' to the
music. Solid-state amplifiers often use back to back
diodes to simulate the "tube sound".

Right now, the pcb is just sitting on on the desk. Pots with fairly long lead wires (not twisted), to mount as controls in the eventual box. Plastic, not metal box initially contemplated. Who knows about ground loops as I laid out and cut the PCB myself and have very little practical knowledge here. Flourescent lights in the overhead radiating who knows what?

But, to your original point, I see very little 'noise'. This is true signal coming from somewhere. I guess I'll look more at the PS, and try to think of a way to 'shield' the board as a test before final box assembly.

Ouch! You are abusing the Op amps by extending their
inputs further than a few mm. The -ve (inverting) input
is a virtual summing node, and will suck in the smallest
of rogue signals and amplify them. The trick is to place
any controls in circuits that are less sensitive to strays.

One more question tho, If not front panel pots to control op-amp gain, what 'should' be done?

No simple answer to this one. Various methods have been
tried over the years, with varying degrees of complexity
and cost. Here's some I've seen and/or used:

(1) Mechanical shaft extender. Put the pot near the
circuit and bring the shaft to the front panel.
(2) Photo resistor and lamp/LED. The front panel
controls the lamp brightness, which controls the
light dependent resistor, placed close the circuit.
(3) VGA Variable Gain Amp. A circuit that controls
its own gain of the signal under a separate control
signal.
(4) Multiplier. Circuit that does (3) above.
(5) DigiPotor equivalent

We had a recent thread about this topic here.

[rolerbe]

I guess what I should do (after eliminating the rogue AC input) is set fixed gains on the board for both stages that are high enough, then use the ouput pots to control output levels.

Thanks for all the input. I'm learning a lot and still having fun!

[CeaSaR]

I guess what I should do (after eliminating the rogue AC input) is set fixed gains on the board for both stages that are high enough, then use the ouput pots to control output levels.

Either that or use shielded cable for the offboard pots and shield the
enclosure once you have the circuit finished.

I also see that V- only goes to VEE on the 4053. I am assuming (a
dangerous thing to do! ) that it also feeds the TL072P's. The signal
path has it's own ground (floating). A possibility? Maybe?

CeaSaR

[CeaSaR]

I just had a "Ding fries are done!" moment. The only thing connecting
the power supply to the signal ground is C8. Shouldn't there be another
cap of the same size going from signal ground to V-, oriented with + at
signal ground and - at V-? You have filtering on the + supply but not the
- supply! Can't hurt to check it out.

CeaSaR

[Bigglez]

Either that or use shielded cable for the offboard pots and shield the
enclosure once you have the circuit finished.

Dumb idea. The more capacitance that appears at
an inverting node, the smaller the bandwidth and the
greater the risk of oscillation or ringing.

Datasheets and ap notes for Op Amps go to
great length to describe how to remove ground
planes around these nodes. A shielded cable is a
fairly large capacitor.

I also see that V- only goes to VEE on the 4053.
I am assuming...that it also feeds the TL072P's.

Without a full and accurate schematic we'd be guessing...