DC blocking in headphone amp with virtual ground

Hello, everyone,

First, let me preface the following with the understanding that I am not looking to troubleshoot someone else's design but trying to gain knowledge for the future.

I have a project that I've recently finished up and one of the things that I added on was a headphone amp which is powered via a rail splitter and is based off of an OPA2162. There are also film caps for on the IC power rail connections to ground as well as a film cap on the input and output audio connections. The headphone amp works, has plenty of gain (probably too much) and sounds as expected. However, when I turn the volume knob with no program material passing through, there is quite a bit of noise that sounds like a noisy pot or leaky caps or both. On the output of the headphone jack itself I can confirm DC spikes between the different connection points on the TRS jack when turning the knob. I tried placing electrolytic caps and 100k resistors in various combinations of places to see if I could block or bleed off the DC but I haven't been able to find such a configuration. I confirmed that there is no DC being transmitted from the source feeding the headphone amp input.

If I am able to move forward with this project, it is possible that the headphone amp will be a virtual ground design, so my question is what could be the source of the DC spikes when the headphone amp is being adjusted? The follow up is how can this be mitigated? Like I said, I messed around with caps and resistors and I think that is still the answer, I just don't know where to connect them and at what values.

Thanks!

Paul

Is the headphone return current going into the virtual ground? That is not recommended because the virtual ground would have to absorb that current. It should be possible to make something work from a single supply but it's not a trivial exercise.

Post a schematic. You'll always get much better answers if you post a schematic. There might be a way to tweak your design and make it work (such as by just bypassing the Vg with a huge cap to real ground).

It was a cheap CMOY based circuit kit using an OPA2162 and a 2x50k pot (log or linear if doesn't say) which didn't come with a schematic, but if needed I could hand draw one up if needed. Again, I don't necessarily care about fixing this circuit as it it what it is and serves the purpose of testing a build with an integrated headphone amp, though I am interested in finding out how the pot is the source of generating the DC when circuit overall is clean and working properly.

You mentioned a concern that I had with absorbing current with the virtual ground as that is what I was thinking about as well. One of the things I was trying to do is connect areas where DC spikes occurred to various resistors and caps and take the opposite ends to the actual earth but with no luck. So I'm mostly trying to figure out for down the road if I am going to incorporate a headphone amp in builds the approach to use if using a virtual earth/voltage divider scheme is going to inherently cause issues. Figuring out the ideal design approach for this part of the circuit will have implications on how the PSU is designed.

I have a second kit that I can use for testing ideas as the first one is in a build that is completed and I have no desire to open it back up now that the rest of that project works.

Thanks!

Paul

The Cmoy does not play well with other devices on the same power supply. It was made to take signal frm a WalkMan and have its own battery. Any other uses, you probably really want a LM386.

PRR said:

Any other uses, you probably really want a LM386.

Click to expand...

Nobody should be using an LM386 for anything (except maybe to give you a nostalgic feeling?).

I realize people want to build something elaborate because this is DIY but there are modern parts that are purpose build for a lot of scenarios. I would just buy a single supply headphone IC. Just go to the "audio amplifiers" category on Mouser and use "search within results" for "headphone" [1]. Yeah, most of them are low voltage but that's GOOD because it acts as a power limiter (5V/2 driving 32ohms is 200mWp which is crazy loud).

[1] BH3548F-E2, SSM6322 (fancy but could probably be made to work with single supply), PAM8019 (class D but has less distortion where it matters than many of the other non-class D chips driving 4 ohms to 1.5W!), MAX97220 (builtin charge pump eliminates output caps), ...

Some early consumer audio power amplifiers used single supply. Their remedy was to use a large coupling capacitor in series with the output, and return the load to ground.

JR

> Nobody should be using an LM386 for anything

Thank you for the strong opinion!

squarewave said:

Nobody should be using an LM386 for anything (except maybe to give you a nostalgic feeling?).

Click to expand...

It's true the LM386 is long in the tooth. I think I first used one (or a close relative) for the headphone output in a DJ mixer kit I sold in 1978 (over 40 years ago). That doesn't mean it won't work.

I realize people want to build something elaborate because this is DIY but there are modern parts that are purpose build for a lot of scenarios. I would just buy a single supply headphone IC. Just go to the "audio amplifiers" category on Mouser and use "search within results" for "headphone" [1]. Yeah, most of them are low voltage but that's GOOD because it acts as a power limiter (5V/2 driving 32ohms is 200mWp which is crazy loud).

[1] BH3548F-E2, SSM6322 (fancy but could probably be made to work with single supply), PAM8019 (class D but has less distortion where it matters than many of the other non-class D chips driving 4 ohms to 1.5W!), MAX97220 (builtin charge pump eliminates output caps), ...

Click to expand...

There is some merit in KISS.... but indeed there are multiple more modern solutions.

JR

PS: I used a cute class D IC to drive speakers in my drum tuner efficiently from batteries. That low voltage class D IC might not be appropriate for higher impedance headphones that need more voltage swing to melt your ear wax.

JohnRoberts said:

PS: I used a cute class D IC to drive speakers in my drum tuner efficiently from batteries. That low voltage class D IC might not be appropriate for higher impedance headphones that need more voltage swing to melt your ear wax.

Click to expand...

Funny I was just looking at the datasheet for that. It has a separate AB output for headphones. The class D is for driving speakers. But the interesting thing is the DC volume control.  You can use one cheap pot for good tracking. It's pretty much optimized for a small portable speaker with headphone option. Of course one has to look carefully at each chip and find something that really fit's their need. The chips are so specialized now, there's something for every scenario.

squarewave said:

Funny I was just looking at the datasheet for that. It has a separate AB output for headphones. The class D is for driving speakers. But the interesting thing is the DC volume control.  You can use one cheap pot for good tracking. It's pretty much optimized for a small portable speaker with headphone option. Of course one has to look carefully at each chip and find something that really fit's their need. The chips are so specialized now, there's something for every scenario.

Click to expand...

I used a different part, but that one has the same low voltage issue. My tuner is battery powered and 4 fresh AA batteries in series could over-voltage that technology. I had to pre-regulate my unregulated supply to not kill the class D chip.

This is surely too much information for the OP.

Sorry about the veer.

JR

JohnRoberts said:

I used a different part, but that one has the same low voltage issue. My tuner is battery powered and 4 fresh AA batteries in series could over-voltage that technology. I had to pre-regulate my unregulated supply to not kill the class D chip.

Click to expand...

What if you use 3 cells for 4.5V and then bridge the outputs into a 4 ohm speaker? Says it works down to 2.8V.

squarewave said:

What if you use 3 cells for 4.5V and then bridge the outputs into a 4 ohm speaker? Says it works down to 2.8V.

Click to expand...

I am driving two 8 Ohm speakers in series, and they get plenty loud enough to excite drumhead vibration modes. The class D amp driving 16 ohms sips current. 

I am not about to redesign my drum tuner now. Battery life was an important design concern.  My first generation tuner used a 9V battery and class A/B amp... not very good battery life. 

The current 3.3V processor is powered through a LDO Vreg, so 4xAA cells will work for a pretty long life (something like 0.9V per cell drop out voltage).  The processor runs lower than 3.3V and LDO regulator has an under voltage flag, so I warn with blinking power LED if/when the batteries are low...  (unit still works down there in a pinch, but tuning results are not guaranteed). 8) 8)

The pre-regulator to scrub off the excess voltage from fresh batteries is a bit of extra complexity but modern mosfets made that relatively cheap and easy. It is not tightly voltage regulated but pre-regulator is LDO too....  Pass mosfet turns hard on below something like 5V.

JR

All righty, I managed to get back to looking at the headphone amp that spurred another quest for knowledge. I sketched out the schematic from the board and is attached (op amp is OPA2162). The thing I left off is the virtual earth power scheme used to supply VDC to the op amp. What I found is that when using an input from a battery powered personal music player, there is no DC discharge to the headphones when the ground reference is coming from said player. However since I am using this in a mixer design, the audio input ground and 0V tie to the star ground at some point, so using either for headphone ground creates the same issue at hand.

Again, this is in a prototype and any future designs would accommodate a +/-VDC PSU for the headphone amp section.

One thing that I do need to change is overall gain as I can only turn up the volume to about 9 o'clock when using program material at unity, so changing the feedback resistor to suit my needs should hopefully lower the scratching noise which for this build will be an acceptable compromise.

Thanks!

Paul

Wrapped up the final adjustments and using 10R for the feedback resistor gives me the reduction in overall gain that I am looking for and doing so minimizes the scratching noise. What I've come to believe is this design was intended to be used with crappy earbuds as it is sold as a kit to work with a 9V battery, which the OPA2134 will see as +/- 4.5V. Using it in the stock configuration would not sufficiently drive a real set of headphones. Running it with a 24VDC PSU as I am doing in build where this circuit is being used allows headphones to be driven without distortion, however this requires adjustment to resistor values. The value of the pot is 50k but I think in a previous version 10k was being used, which may have been better suited for my use. Since I didn't have a dual gang pot in this value, changing the feedback resistor was the next option. I know there is probably a penalty to be paid for changing this resistor and not the others, but I am very happy with how it sounds and program material seems to retain it's full frequency spectrum and the turning range of the potentiometer is much more usable.

At any rate, thanks to everyone here I learned somethings for next time and figured out how to make the circuit on hand work for time being, so I'll put that in the win column.

Thanks!

Paul