Combining 2 channels to one

[klitgt]

Hi
I have an active x-over with 2 channels in both high pass and low pass lines. Now I want to combine the 2 channels (left & right) in the low pass line after the L & R output buffers to feed one single poweramp that drives one single subwoofer.
As far as I know this is possible if I connect the outputs through resistors as shown in the diagram attached.
But the values of the resistors is a ??? to me.
Does anybody know which value to use?
And am I right with the principle shown in the diagram?

Thank's for inputs !

[Bill Fitzpatrick]

Try 2200 ohms.

[paulb]

Originally posted by Bill Fitzpatrick
Try 2200 ohms.

I agree with Bill. It's the bottom resistor in your avatar.
But anything in the same range will work too.

[klitgt]

Thank's guys!

Now that I have a dialog with some knowledgeable persons, can you tell me why it is necessary with these resistors, why not just connect the leads from R and L without resistors?

[Sch3mat1c]

Because if R is on a positive peak while L is on a negative peak, the mixed output shows the average, i.e. zero, but between them, they try feeding each other current. If the outputs don't use current-limited op-amps, it's a good smoke recipie. If they do, it probably won't come out right because the two amps are trying to fight each other (due to negative feedback) for control over the output. Using resistors fixes this.

The value of resistance depends on how much crosstalk/loading you want (i.e., current flowing into and out of each terminal) and the attenuation from feeding the load impedance. Ideally you'd use infinite, followed by infinite gain to counter the attenuation. That doesn't work well in practice, but where an input impedance is tolerable (well, always), a "virtual ground" mixer works well.

Tim

[klitgt]

I found an answer by using the inside of my head: The buffers have low output impedance and if I connect the output leads without output resistors, the output of each buffer looks into a low impedance (the other buffer). This impedance is too low, it must be the same high impedance as if the buffers look into a power amp with an impedance of typically 10-20K.

10-20K for the output resistors would be too much, as the resistors form a voltage divider together with the input impedance of the power amp.....and thus attenuatte the signal too much. So the suggested 2200 Ohms for the resistors is what I will go for, thank's again to all

[AndrewT]

Hi,
connecting those two resistors to an inverting opamp is the normal way to SUM 2 inputs.
The inverting pin on an opamp is a virtual earth and the two outputs each see the resistor as a load to earth with no crosstalk.
The opamp reads each input and adds them together to be passed on to the next stage. It also cancels any stereo information. But into a sub-bass there is probably no real stereo information anyway. Modern electronic mixes could easily put different info irrespective of frequency and if the producer did this you would lose some or all of that as well.

If your next stage gave you access to the inverting input you could feed into that instead of the non inverting input.

[klitgt]

Originally posted by AndrewT
Hi,
connecting those two resistors to an inverting opamp is the normal way to SUM 2 inputs.
The inverting pin on an opamp is a virtual earth and the two outputs each see the resistor as a load to earth with no crosstalk.
The opamp reads each input and adds them together to be passed on to the next stage. It also cancels any stereo information. But into a sub-bass there is probably no real stereo information anyway. Modern electronic mixes could easily put different info irrespective of frequency and if the producer did this you would lose some or all of that as well.

If your next stage gave you access to the inverting input you could feed into that instead of the non inverting input.

Do you mean the way showed in the upper example (see attachment)?
Should there be a feedback loop (blue lead) with a resistor and if yes which resistor value? And what about the non-inverting input (pin3 at the opamp)?
What about the lower example. Are input resistors needed here and what about feedback from output to pin2? If needed, should a resistor be added? What value?

In my CDP the differential signals from the I/V converters go into pin 3 (positive ½) and pin 2 (negative ½) of the opamp that combines the signals to single ended in each channel. This has inspired med to draw the lower circuit example attached. In the CDP there is a feedback from output (pin6) to pin2 with a 2,2K resistor in parallel with a 180p cap.

[klitgt]

Originally posted by AndrewT
Hi,
connecting those two resistors to an inverting opamp is the normal way to SUM 2 inputs.
The inverting pin on an opamp is a virtual earth and the two outputs each see the resistor as a load to earth with no crosstalk.
The opamp reads each input and adds them together to be passed on to the next stage. It also cancels any stereo information. But into a sub-bass there is probably no real stereo information anyway. Modern electronic mixes could easily put different info irrespective of frequency and if the producer did this you would lose some or all of that as well.

If your next stage gave you access to the inverting input you could feed into that instead of the non inverting input.

Andrew, now I have read your reply a couple of times and I think I have understood it better.
I found an appication note on NE5532/34 and here the non-inverting input is connected to ground if the inverting input receives the signal.
But I do not find any information about the reason for
"connecting those two resistors to an inverting opamp is the normal way to SUM 2 inputs."
Why not use the non-inverting input and connect the inverting to ground?
Again I would appreciate som guidance as to feed-back from output to input etc.

"If your next stage gave you access to the inverting input you could feed into that instead of the non inverting input."
My next stage is a poweramp with only non-inverting input (I guess if there is only one input on a poweramp it is a non-inverting).

So in conclusion I have to find an opamp and do as you suggest. But what are the benefits of your solution as compared with my first suggestion (see picture in the beginning of this thread) just connecting the two resistors to the input of the poweramp, without using any opamp?

[AndrewT]

Hi,
your top diagram is the summing amp. You can have as many inputs as you have resistors connected to the inverting input.

If feedback value =input value then gain=1. Two inputs will give a total output =2 times a single input.
By changing the ratio of feeback value to input value you can select the gain/attenuation you need.

Since your power amp is conventional you cannot use it in the summing manner and you need the summing opamp at its input.

You can connect the non-inverting input direct to ground but this gives a DC offset at the output.

Instead it is better to connect a resistor between non-inv & ground. The resistor value = // resistance of all the connections to the inverting input. In your example diagram if feedback value =2k2 then ground resistor =2k2/3 =733R. THEN 2* 1v INPUT GIVES 2V OUTPUT.