Whine on high gain buffered pedal - not when double buffered

[Ksander]

Hi!

After my brother got an Erica Synths plasma drive, I had to DIY something like it. Then I found the "neon-drive"  (https://youtu.be/BmYvxnKo2pA?feature=shared, and built something like it:

• the guitar signal goes into a variable gain op-amp buffer (TL072)
• then into an LM386 for more gain/power
• then through a 1:20 transformer
• and then the signal is clipped using the neon bulb
• scaled down with a big voltage divider
• and finally sent to the amplifier.

The effect works, in that I get a nice overdrive like effect (partly because of the LM386 itself clipping) and a cool visual effect from the bulb. However, a high gain settings (any high-gain combination of the op-amp/LM386 gain) there is a high pitched whine. First I thought it was due to some feedback because of I/O wires crossing, so I used insulated wire from the jacks to the switch and from there to the circuit board instead. That didn't solve it. Now I noticed that putting another buffer in front mitigates the problem. Why? Is there anything I can do to the buffer inside the pedal to get rid of the whine? 

[Ksander]

With a picture:

[MrStab]

So the buffer that resolves it is external?

Which true bypass configuration have you used?

Without the additional buffer, how close does the raw, high-impedance input signal get to the output of any of the gain stages (or the final output) on the board?

Is there a feedback loop capacitor in the TL072 gain stage?

[ElectricDruid]

Have you got a schematic of the pedal please?

[Ksander]

This should be the schematic (i just opened it up and drew it out):


The type of bypass is true bypass w. a 3pdt switch. Signal is switched to the output or to the board and then to the output.

As 'buffers' in front of this pedal, I tried an anderton compressor and an orange squeezer.

I'm not sure about the question on signal levels. The guitar produces 40mV (RMS?) if i recall correctly, and the max input level for the 386 is 0.4v I believe. The neon bulb discharges at 60V, and that level is reached, because I can see it discharge.

EDIT: the cap in the feedback loop is 100pF

[Clint Eastwood]

A power supply to ground decoupling capacitor for the LM386 as close to it as possible might help.

[MrStab]

I meant impedance, not voltage amplitude. Even a crazy-output passive Invader pickup, or similar, with an output swing of several volts peak-to-peak has a high impedance and is susceptible to crosstalk and feedback.

In simple terms, a source capable of delivering more current is interfering with the bare input unless it itself is delivering more current.

This is very likely a layout issue, be it on the board or offboard wiring. Extreme gain is possible to pull off (reading posts on the BSIAB pedal oscillating may help), but sometimes nothing can be done unless the layout is corrected or the gain is reduced. Does shoogling the input & output wires to/from the jacks, while oscillating, affect the pitch at all?

[antonis]

Is there anything I can do to the buffer inside the pedal to get rid of the whine?

Sorry but I can't see any buffer inside pedal..

[Ksander]

Is there anything I can do to the buffer inside the pedal to get rid of the whine?

Sorry but I can't see any buffer inside pedal..

Unless I'm mistaken, it is an inverting op-amp buffer with 100kOhm resistor from the input to the inverting terminal, and a 1MOhm pot and 100pF cap between the inverting input and output. This is how I found inverting buffers on muzique-dot-com and various other websites. Could very well be I did something wrong nonetheless...

Is the 100kOhm perhaps too small?

[Ksander]

I meant impedance, not voltage amplitude. Even a crazy-output passive Invader pickup, or similar, with an output swing of several volts peak-to-peak has a high impedance and is susceptible to crosstalk and feedback.

In simple terms, a source capable of delivering more current is interfering with the bare input unless it itself is delivering more current.

This is very likely a layout issue, be it on the board or offboard wiring. Extreme gain is possible to pull off (reading posts on the BSIAB pedal oscillating may help), but sometimes nothing can be done unless the layout is corrected or the gain is reduced. Does shoogling the input & output wires to/from the jacks, while oscillating, affect the pitch at all?

I don't notice any change when moving the wires around; changing to coaxial wire didn't make a difference either (grounded at one end). It may be on the board.

[MrStab]

Probably not related, but non-inverting buffers are generally preferred when facing high-impedance inputs to avoid signal being in series with thermally-noisy large resistors. That's more of a hiss issue than oscillation, though.

[antonis]

Unless I'm mistaken, it is an inverting op-amp buffer with 100kOhm resistor from the input to the inverting terminal, and a 1MOhm pot and 100pF cap between the inverting input and output.

You're free to call it as you like but, IMHO, "buffers" don't either invert or exhibit voltage gain,,

[PRR]

Audio goes through air (or plastic!!) same as radio waves. Not as well, but you put a transmitter an inch away from a receiver and a gain of thousands between, it's gonna feedback.

You don't need a Vb buffer for just one opamp.

Use the other opamp as a unity gain LOW-impedance input buffer. Use a shorting input jack. 1Meg bias resistor, zero NFB resistor. Keep all this input stuff small and tight so it doesn't pick-up signal. Keep it AWAY from your power output section, the neon and stuff around it, right up to the 10K at the end. Make all this small and tight.

[Ksander]

Audio goes through air (or plastic!!) same as radio waves. Not as well, but you put a transmitter an inch away from a receiver and a gain of thousands between, it's gonna feedback.

That makes a lot of sense of course - thanks for pointing that out!

You don't need a Vb buffer for just one opamp.

Use the other opamp as a unity gain LOW-impedance input buffer. Use a shorting input jack. 1Meg bias resistor, zero NFB resistor. Keep all this input stuff small and tight so it doesn't pick-up signal. Keep it AWAY from your power output section, the neon and stuff around it, right up to the 10K at the end. Make all this small and tight.

Perfect. Thanks, I'll try this ASAP.

[MrStab]

In your defence, 60V is quite hardcore!

[Ksander]

In your defence, 60V is quite hardcore!

Actually, I have another version (which doesn't have the issue; different 'buffer'), which uses an NE555 charge pump to get from 9 to 18V, to power an LM386-4. with max gain of 200 and a 20x transformer, this should theoretically reach 40mV x 200 x 20 = 160V...

[MrStab]

If instrument level to near-tube-amp-level is a necessity, tube amp designs would be a good place to look for inspiration.

[Ksander]

...

You don't need a Vb buffer for just one opamp.

Use the other opamp as a unity gain LOW-impedance input buffer. Use a shorting input jack. 1Meg bias resistor, zero NFB resistor. Keep all this input stuff small and tight so it doesn't pick-up signal. Keep it AWAY from your power output section, the neon and stuff around it, right up to the 10K at the end. Make all this small and tight.

I have now tried the above: input signal to 100nF capacitor; to non-inverting input & biased to 1/2 Vcc (4.5V) using two 1MOhm resistors; a jumper between the inverting input and output (0Ohm NFB) and then a 10uF capacitor to the next stage (exactly as one of the examples on muzique.com). Unfortunately, this has made the feedback issue way worse.

I'm wondering: isn't the input impedance of a TL072 very high? The datasheet doesn't seem mention this term explicitly, but the 'common-mode input resistance' is 1TOhm (1x10^12Ohm) . Should I use some relatively low resistors (100kOhm) in the buffer instead?

[antonis]

Should I use some relatively low resistors (100kOhm) in the buffer instead?

Only if you can afford an input impedance of 50k (100k//100k)..
You see, for non-inverting configuration, the dominant factor of input impedance is the bias configuration..
Paul above proposed you the easiest/quickest (but not "quietest") modification for a 500k input impedance buffer..
Alternatively, you can implement a 10k/10k (or so) voltage divider, with a 100μF cap from their juction to GND and a 1M resistor from their junction to non-inverting input..

[Ksander]

...

Alternatively, you can implement a 10k/10k (or so) voltage divider, with a 100μF cap from their juction to GND and a 1M resistor from their junction to non-inverting input..

I see! I have implemented your suggestion, and it worked like a charm. No more whine whatsoever. Thank you very much!