distortion and oscillation in headphones output

[Dimitree]

I'm using this circuit to provide a mixing section and headphone output interface to my a digital synthesizer that I designed (that runs at 5V DC).
I basically copied this circuit from my Korg Volca synthesizers, all those little machines have this circuit on the output, and found it handy since the supply voltage is the same as mine.
The only modification I did was to increase the gain. The original has unity gain achieved by 22K resistors.
I needed more gain so I changed those resistors. But now I got loud oscillation (around 4KHz) and a noticeable distortion when using a mono cable connecting this output to my line input interface. When using the headphones it sounds fine.
I'm not an expert by any means, so I just guess that, by grounding the ring on the headphone jack when using a mono cable, the shorted channels draws too much power from the opamp and make the other channel distort and oscillate. Is that right?
If so, is there any workaround? Or what could be the issue?
I tried to replace L3/L4 with 100R, the oscillation disappeared but the level when using headphones was dramatically reduced.
So I tried to move the 100R inside the feedback path of the opamp, but that didn't fix the oscillation.

The chip that I'm using is the TPA6111A2
Any suggestion? thank you really much, and sorry for any obvious mistake.

 

[ccaudle]

is there any workaround?

The easiest would be to either add separate mono connectors for line-out, or make a custom cable that does not short one of the channels.

 

[Dimitree]

Thanks! And what could be other solutions keeping a single connector/standard cables? One option I tried was to use another opamp before the TPA6111A2, having the gain in the first opamp and unity at the TPA6111A2. That worked but was a little bit noisier of course.

 

[JohnRoberts]

I'm using this circuit to provide a mixing section and headphone output interface to my a digital synthesizer that I designed (that runs at 5V DC).
I basically copied this circuit from my Korg Volca synthesizers, all those little machines have this circuit on the output, and found it handy since the supply voltage is the same as mine.
The only modification I did was to increase the gain. The original has unity gain achieved by 22K resistors.
I needed more gain so I changed those resistors. But now I got loud oscillation (around 4KHz) and a noticeable distortion when using a mono cable connecting this output to my line input interface. When using the headphones it sounds fine.

4kHz is pretty low frequency for oscillation. look for something funny about wiring and grounds when using the mono cable. Is that mono cable shorting left and right together? IC may not like that. If 100 ohms loses too much level try smaller Rs.

JR

I'm not an expert by any means, so I just guess that, by grounding the ring on the headphone jack when using a mono cable, the shorted channels draws too much power from the opamp and make the other channel distort and oscillate. Is that right?
If so, is there any workaround? Or what could be the issue?
I tried to replace L3/L4 with 100R, the oscillation disappeared but the level when using headphones was dramatically reduced.
So I tried to move the 100R inside the feedback path of the opamp, but that didn't fix the oscillation.

The chip that I'm using is the TPA6111A2
Any suggestion? thank you really much, and sorry for any obvious mistake.

 

[ccaudle]

what could be other solutions keeping a single connector/standard cables?

I would think that stereo cables inserted into a stereo output would be considered "standard."
The design goals for a good headphone amplifier and a line amp which will allow shorting channels together are fundamentally in opposition.

One option I tried was to use another opamp before the TPA6111A2, having the gain in the first opamp and unity at the TPA6111A2.

That should reduce the difference in the output signals between the two channels of the TPA device (greater feedback percentage for that circuit). It is a little surprising that the noise would be noticeably different, that could probably be optimized, but you would still have one channel of the TPA driving into a dead short. Have you checked the device temperature when you do that? I'm a little surprised the device did not go into protective shutdown. Or maybe one channel did, the datasheet just says that continous power dissipation is internally limited, it does not clarify if that is a device or channel protective circuit.

 

[merlin]

I tried to replace L3/L4 with 100R, the oscillation disappeared but the level when using headphones was dramatically reduced.

Have you tried simply using a smaller value, like 10 to 33 ohms?

 

[Voyager10]

You could also try removing R68 and R42, and adding more local supply decoupling, e.g. 100uF close to the IC.

Are the pi filters (C48, L3 etc) needed? The TI datasheet doesn't have them.

 

[Dimitree]

For comparison, this is the Korg Volca schematic where I copied the output network.
By the way, what is the purpose of R210/R212?

 

[Voyager10]

R210/R212 provide a DC current path from C149/C160 to ground, so that the -ve ends of those capacitors are at 0V with nothing plugged in. It stops a thump when you plug the headphones in.

The filter may be for EMC filtering, if this came from a commercial design with mixed analogue/digital circuitry.

 

[gyraf]

..isn't that design wrong: R223 feeds (for detecting "headphone attached") some +V into the wrong, minus side of C160?

Note that in your version, to get higher gain, your input impedance to the circuit gets very low (820 Ohms) - which may not be great for the preceding stage, and which kinda messes up the taper/curve of the 10K volume pot

/Jakob E.

 

[Voyager10]

The TPA6111 is a single-supply-rail amp, and the outputs are at Vdd/2 with no signal. C160 will still have its voltage the right way round.

 

[gyraf]

Aah, all is good then..