Hi all! I'm designing a headphone mixer and amplifier for myself, but probably not for sale as I think it will not be very producible.

This was born out of my own need as a musician; no product on the market has all the inputs I want, and is also small enough to fit on my guitar pedalboard. I've already more or less sorted out the physical design; connectors and knobs and PCB will live in a tall-ish aluminum hammond box, with the circuit board below them. Here are a couple of pictures of the assembly:



As you can see it's pretty tight in there, not much room. The connectors are mms away from the bottom of some of the potentiometers. But for audio frequencies, I'm not too worried. I hope!)

The schematic is below. Sorry for the large images, but it's a bit unreadable it I make them smaller! There's nothing revolutionary about this, although I am new-ish to multiple channel audio circuit designs. (I have experience mainly in ADC, analog filtering MCUs, etc.)

The unit will run on 12VDC which is regulated by LDO to 9V for the main supply. There is also a 4.5V reference IC to provide a midscale voltage, since all inputs will be AC coupled. (I don't want to do dual supplies, - although if someone knows of a simple, single IC solution to get like ±9 or ±5V, I'll listen.)


There are two unbuffered inputs, one mono and one stereo, which will come from my guitar pedalboard. (main guitar signal + click). These are relatively high output level so I thought I might not need to buffer them.

The other inputs are:
- microphone in/through (differential, low signal level so it needs gain)
- Monitor in (stereo, also *could* be low level so I added a gain stage, but could turn it down for higher level inputs.)
- Aux instrumental in (just an extra input since I had an extra half opamp, no immediate use for it. :-D )

You can see in one of the pictures that I've already built the prototype PCBA, and started wiring to the connectors and pots. It more or less matches the above schematic.

My main issues so far are getting everything to work properly, which I think I'll be able to do, but I do have some general audio design questions for mixers and headphone drivers.

1. Is there a limit to the number of stages that can be mixed into a single opamp like this? I have 5 for both left and right sides.
2. Should I be doing the mixing stage separately from the headphone driver stage? You can see that they are currently combined into a single opamp.
3. Should I be unity gain buffering the higher level input stages?
4. If I want to implement a master volume control (shown redlined x-d out in the main schematic), what's the best way to do it? obv the 10k pot in series with the 100k for feedback on the driver stage won't get me very much volume change, but playing with component values is easy once a PCB is built.)

Once I sort out the issues in this initial design I will almost surely spin the PCB and improve it. It's not too pricey, I got the qty 3 of the prototype built for around $100. (Double-sided, single drill, singe side silkscreen.)

Thanks for any thoughts! Like I said I'm experienced in electronics, but not in audio design specifically.

Geoff

 

1. Is there a limit to the number of stages that can be mixed into a single opamp like this? I have 5 for both left and right sides.

No particular limit, but the op amp intrinsic noise is amplified by the op amp feedback resistor divided by the parallel value of the input resistors, so the more inputs, the higher the noise gain.
It also amplifies the input DC offset by the same amount.
Using a low-noise audio op amp, that may not be a problem.

2. Should I be doing the mixing stage separately from the headphone driver stage? You can see that they are currently combined into a single opamp.

Combining them in one op amp should be okay.

3. Should I be unity gain buffering the higher level input stages?

Don't see why.
Do you have a reason to think that it's needed?

4. If I want to implement a master volume control (shown redlined x-d out in the main schematic), what's the best way to do it? obv the 10k pot in series with the 100k for feedback on the driver stage won't get me very much volume change, but playing with component values is easy once a PCB is built.)

Likely would be best to add a couple non-inverting buffer amps with a pot at each input with one connection to ground and the wiper to the buffer non-inverting input.

Note: You show the Headphone out with a mono phone jack shorting the two op amp outputs together, which should not be done.
You need a stereo jack with no shorting (below).

 

No particular limit, but the op amp intrinsic noise is amplified by the op amp feedback resistor divided by the parallel value of the input resistors, so the more inputs, the higher the noise gain.
It also amplifies the input DC offset by the same amount.
Using a low-noise audio op amp, that may not be a problem.
Combining them in one op amp should be okay.

Great, I'll keep as is then. I admit to not doing any noise analysis on this beforehand!

Don't see why.
Do you have a reason to think that it's needed?

Not necessarily, just checking - perhaps the low impedance drive after a volume control would give some improvement. No need to add if it's unneeded though!

Likely would be best to add a couple non-inverting buffer amps with a pot at each input with one connection to ground and the wiper to the buffer non-inverting input.

ok, will do in the next rev - suppose I could do that in the same stage as the mixing?

Note: You show the Headphone out with a mono phone jack shorting the two op amp outputs together, which should not be done.
You need a stereo jack with no shorting (below).

View attachment 306897

Thanks crusts!

Yeah the connectors are off board, so they're not actually 100% represented properly in the schematic - I definitely don't short the L/R on the output; I just was lazy about redrawing a new connector for the schematic.

One additional question, if I'm going to go ahead and spin the board at some point.
Any improvements to the above design beyond what you mentioned? (Like maybe including the 10ohm series out of the driver within the feedback loop?)

Also, are there any recommended headphone driver op amps (or specific ICs) that are recommended? with single ended inputs and 9V supply tolerance? I found what looks like a few nice ones from TI, but one was designed for low voltage operation <5V) and the other required differential inputs.

Thanks!
Geoff

 

Likely would be best to add a couple non-inverting buffer amps with a pot at each input with one connection to ground and the wiper to the buffer non-inverting input.

Sorry, I just re-read this bit - do you mean buffer the VBIAS input into U3-3 and U3-5?

 

Why two different opamp types (plus a mention of a third)? The 4580 looks good enough for all of the functions, but I would go with the OP275 for everything.

ak

 

Sorry, I just re-read this bit - do you mean buffer the VBIAS input into U3-3 and U3-5?

No.
I meant add the pot and buffer amps at the output of U3, to feed the headphone output.

But thinking about it, it should also be possible to change P6A to a 100kΩ dual pot (audio taper?) and eliminate R29 and R30.
So when the pot goes to zero ohms (100% negative feedback), the output will be essentially zero and 100kΩ will give maximum gain.

 

Sorry, I just re-read this bit - do you mean buffer the VBIAS input into U3-3 and U3-5?
[/QUOTE

Vbias should be decoupled to gnd with a suitable cap this port is sensitive to stray positive feedback (oscillations) that you found with the pot , now removed.

I assume your headphones are high Z types > 200 ohms? otherwise bass rolls off at 180 Hz with 8 Ohm.

The 10 ohm Rs could be put inside the loop to lower the Zout and distortion..

 

No.
I meant add the pot and buffer amps at the output of U3, to feed the headphone output.

But thinking about it, it should also be possible to change P6A to a 100kΩ dual pot (audio taper?) and eliminate R29 and R30.
So when the pot goes to zero ohms (100% negative feedback), the output will be essentially zero and 100kΩ will give maximum gain.

Yeah, that's probably the easiest way!

 

yah, probably good practice to decouple near the U3 inputs, but there's a lot of capacitance on the reference chip that generates VBIAS too....

Next rev the 10Rs will go inside the feedback loop!

 

Why two different opamp types (plus a mention of a third)? The 4580 looks good enough for all of the functions, but I would go with the OP275 for everything.

ak

No super good reason, and changing opamps is easy enough! (even tho these are SOIC-8s)

 

I did not see the important headphones impedance and power supply voltage (VCC) in this thread.
Most opamps are spec'd with minimum2000 ohms loads (some drive 600 ohms with higher distortion) and 30V supplies. The 47uF output capacitors indicate minimum 300 ohms per ear for good bass.

 

Look into the OP275. It is built to drive 600 ohms with very little distortion increase. Impressive part.

ak

 

Many headphones are only 32 ohms.
The OP275 opamp can drive 600 ohms loudly and with low distortion only with a high power supply voltage.

We do not know the headphone impedance and power supply voltage for this thread.

 

This will be for driving in-ear monitors primarily. (If you're not familiar with these, they are headphones that musicians use that are often molded to their ears.)


A few minutes of googling reveals that IEM impedance can be low, like maybe 32-300ohms. But I suspect they must also not require much power, with the drivers being so close in the ears, coupled with the isolating form factor.

So I imagine they wouldn't need too too much power to drive well. In practice, I haven't noticed much difference in setting volume levels on my own IEMs vs traditional headphones.

 

Many headphones are only 32 ohms.
The OP275 opamp can drive 600 ohms loudly and with low distortion only with a high power supply voltage.

We do not know the headphone impedance and power supply voltage for this thread.

The pads for those output caps are big. I could pretty easily replace them with 100uF or 220uF or maybe even larger caps.

 

The Shure SE846 In-Ear Monitor has very good ratings and has a sensitivity of 114dB/mW. Each earphone is 9 ohms.
One mW in 9 ohms is only 0.095V so my calculator says the power for it to be very loud is low.
I think low noise from the driver might be heard.

 

I'm working through my prototype design, and having some issues I might solicit advice for.

First off, just listening, the headphone output is pretty hissy. The one thing that seemed to clean it up was disconnecting the REF5045 pin 6 from the VBIAS net and injecting an external voltage (2xAAA batteries for ~3.1V) With the battery supplying the reference, the output is not hissy at all, but with the ref5045 supplying VBIAS, it's quite hissy.

Substituting a different reference chip, a max6017 (a series 4.5V 3 pin reference) changed the characteristic of the hiss; it was lower frequency. I can also report that playing with the bypassing of the REF5045 slightly changed the character of the output hiss as well, but not much. As you can see in the schematic, there's a good smattering of different value caps on the VBIAS net: 47uF, a couple of 1uFs, and a couple of 0.1uFs, hopefully this would give me a nice good wideband capacitance across VBIAS. THe datasheet says you need a 1-50uF cap on the output, so what I have seems like it ought to be sufficient. I also have the datasheet recommended 1uF cap on the NR pin to reduce noise as well.

Any thoughts on the hiss? obv a battery isn't a real solution :-D, but good for troubleshooting!

 

Any thoughts on the hiss?

Try adding a small resistor (e.g 1kΩ) in series with the reference output before the filter capacitors

 

Why does the AC audio circuit have a very accurate DC reference IC that produces noise?
The accurate DC reference is used in a voltmeter or thermometer circuit, not in an audio circuit.

The - input of an opamp is biased by the negative feedback resistor.
The + input of the opamp is biased by a bypassed resistors voltage divider.

 

Audioguru beat me to the answer.
Unless they are specifically rated for low noise, don’t use a reference for the mid-bias but rather use the resistor divider as shown.
For a very stiff “virtual ground” that can drive several opamps, try the TLE2426.
https://www.ti.com/product/TLE2426