Passively combining two audio channels

[Signaling]

Hi, what sort of resistor values should I use for passively combining two audio channels from the same source (computer/phone headphone output) to drive a mono chip amp?

I know an op-amp is the best way but for simple mono chip amp projects I think resistors are probably good enough, I'm currently using 1k resistors but can't help think that value might be a bit too low.

I put a switch in there to select between combined channels and just the one channel, and when I combine them the sound is a bit "muddy" sounding although I can't quite put my finger on it. I've done some output frequency sweeps and it all looked pretty even on the scope across the audio range.

Should I use higher value resistors instead of 1k to passively combine audio channels?

Thanks.

 

[tomchr]

If you have both sources plugged in, but only one of them is providing signal, the other will fight to keep its output voltage at 0 V. That's what creates the mud.
You can increase the resistance 5-10 kΩ should be plenty) but depending on the input capacitance of your amp, you might end up bandwidth limiting the circuit with that.

A better option would be to use a summing amp. That requires power, though. An input selector switch would be a passive solution.

Tom

 

[rayma]

I put a switch in there to select between combined channels and just the one channel,
and when I combine them the sound is a bit "muddy" sounding

Try a mono recording and see if it's still muddy. If not, then do use higher values,
though too high will start to lose level.

 

[Signaling]

If you have both sources plugged in, but only one of them is providing signal, the other will fight to keep its output voltage at 0 V. That's what creates the mud.
You can increase the resistance 5-10 kΩ should be plenty) but depending on the input capacitance of your amp, you might end up bandwidth limiting the circuit with that.

Both channels are from a single low output impedance source, does the above still hold true for that?

Try a mono recording and see if it's still muddy. If not, then do use higher values,
though too high will start to lose level.

Thanks, I'll give that a go and see what I get. Would this video be ok for testing if I fed it into the amp from my laptop? YouTube

 

[Galu]

I use the same stereo to mono summing circuit as yours to connect a stereo CD player to a mono valve amp, and it sounds fine in this application.

However I have seen the basic circuit embellished by the connection of a 20K resistor between the point of intersection of the two 1K resistors and Gnd.

I wonder if this addition could possibly clean up your 'muddy' sound.

 

[PRR]

If the stereo output is intended to drive modern headphones, even 33 Ohms is acceptable for mix resistors.

With the low input impedance of some modern amps, 10K is rather a lot.

470 Ohms seems a happy medium. If you have over-stock of 100 Ohms or 2.2K, that will work also. It isn't fussy.

 

[jvhb]

Hi, Rane Note 109 provides some good information and suitable values:
Why Not Wye?

They also suggest 470 ohm, but with a 10K to ground from the summed signals.

 

[Signaling]

Ok I think its easier if I show you how its been set up, all that extra filtering is because it would pick up a strong local radio station if I plugged in a long aux cable in and left it floating (layout is tight and minimal). The ferrite beads are those little through-hole things that look like jumpers with a small 5mm slug of ferrite slipped over, so its not even one turn.

Does anything look off? Also I guess muddy sounding was a bit strong, it just doesn't seem as bright and crisp when the two channels are mixed (it is noticeable when the switch is toggled whilst something is playing). It doesn't sound terrible but there's obviously something going on.

If the stereo output is intended to drive modern headphones, even 33 Ohms is acceptable for mix resistors.

With the low input impedance of some modern amps, 10K is rather a lot.

470 Ohms seems a happy medium. If you have over-stock of 100 Ohms or 2.2K, that will work also. It isn't fussy.

I've come across an "input resistance" for my chip amp (LM386) but it doesn't actually say input impedance so not sure if its the same thing.

I use the same stereo to mono summing circuit as yours to connect a stereo CD player to a mono valve amp, and it sounds fine in this application.

However I have seen the basic circuit embellished by the connection of a 20K resistor between the point of intersection of the two 1K resistors and Gnd.

I wonder if this addition could possibly clean up your 'muddy' sound.

I've got a 1MEG there at the moment to discharge the input filtering cap, is that too high?

Hi, Rane Note 109 provides some good information and suitable values:
Why Not Wye?

They also suggest 470 ohm, but with a 10K to ground from the summed signals.

Thanks, I've read that page a few times but I have to admit I don't really understand the last section.

 

[jvhb]

(...)Thanks, I've read that page a few times but I have to admit I don't really understand the last section.

You mean the "Termites in the Woodpile" section?
I don't think this is relevant in your situation anyhow. It pertains to a PA mixing scenario, where both the two original and the mixed output are all used at the same time.

Basically you are just making a "Stereo-to-Mono Summing Box", with an off/on switch - If I understand correctly.
But I think you should wire the switch so a 22K to ground is switched in, when your are summing the signals. As shown in the Rane note.

 

[Signaling]

You mean the "Termites in the Woodpile" section?
I don't think this is relevant in your situation anyhow. It pertains to a PA mixing scenario, where both the two original and the mixed output are all used at the same time.

Basically you are just making a "Stereo-to-Mono Summing Box", with an off/on switch - If I understand correctly.
But I think you should wire the switch so a 22K to ground is switched in, when your are summing the signals. As shown in the Rane note.

Is this acceptable (its the most practical way for me to add the resistor)? And could the resistor be higher than 22k? As I don't want to drop the signal level too much at the pot wiper when the switch is switched. R2 was just to prevent the right channel from floating when switched out.

All resistors are those little 100mW carbon film types.

 

[jvhb]

I am not an expert in this - I have basically just followed the Rane note, and it has worked fine. But as far as I can see, your mixing circuit will work fine as drawn now:
You switch in 12k in series with 10k to ground when combining L and R, and they both go through a 1K. So it basically follows the Rane recommendation, as well as the posts from experienced users higher in the thread.
I don't think the value of the ground resistor is critical, anything from 10k to 30k should be fine. But a lower value will give more attenuation of the input signal.

 

[Signaling]

Update: 12k resistor added as shown in my last post and I don't know what the reason is but it sounds better when the channels are mixed now with next to no noticeable difference when I switch the switch.

I really can't explain it, maybe that mixed point stays a little more balanced?