When I connect the DAC audio to the summing circuit, it slightly attenuates the bluetooth audio.

Yes, since the inputs act like a voltage divider, when more than one input is connected.
To avoid that change you can connect the unused inputs to ground.

There are audio connectors that can do that automatically (below):
You connect pin 3 to ground which then connects the pin 2 input to ground when not plugged in.

 

Yes, since the inputs act like a voltage divider, when more than one input is connected.
To avoid that change you can connect the unused inputs to ground.

There are audio connectors that can do that automatically (below):
You connect pin 3 to ground which then connects the pin 2 input to ground when not plugged in.

View attachment 351138

I can’t do that because all inputs are used. The Bluetooth audio will play and so will the voice prompts from the DAC at set altitudes

 

I can’t do that because all inputs are used. The Bluetooth audio will play and so will the voice prompts from the DAC at set altitudes

Then as long as all inputs are connected, the gain won't change.
But for testing purposes, you can ground unused inputs to simulate they are connected (since the outputs of those devices are low impedance and essentially look like AC ground).

 

When I connect the DAC audio to the summing circuit, it slightly attenuates the bluetooth audio.

Which schematic (post #) are you using.

ak

 

Which schematic (post #) are you using.

ak

Post no. 3

Tho the resistor are actually 10k not 20k

 

After all these posts, you’re still using the incorrect circuit instead of the one shown in post #8?

 

After all these posts, you’re still using the incorrect circuit instead of the one shown in post #8?

Why does figure 3 in post #8 look different to figure 3 in the LM4871 datasheet (see below).

So if I use 20k resistors and 0.39uF capacitors for each of the 3 inputs with a feedback resistor of 20k it will work ?

The existing gain is 7 (as shown in post 12). If I use the resistor values in post #8 the gain will be 1, so surely it will be even quieter than it already is ?

Even if I directly connect an audio source (bluetooth left, right or DAC) directly to the input of the LM4871 eval board its very quiet. So how can decreasing the gain from 7 to 1 make it any louder ?

 

► Ti op amps compare ◄

. . . ▄ simplified auto-centering thingy https://tinyurl.com/24pt2ezj (Falstad simulation)

 

This is post 8, modified to fit your requirements. Panic Mode outlined the changes. It should work although the gain may require adjustment.
Now this is post 3, with a biasing scheme that is shorting the audio signals.

 

If efficiency is important, I would work at eliminating the noise from the digital amp.

 

basically the stereo difference is at most cases orders of magnitude less than the mean audio level =
= you can weigh the L+R audio to be comparable to your "digital input"

(( . . . also at #29 [ #3 ]) the resistor is missing from the output of the TLV2372 (fixed to the supply median) to the summing point of the input sources ))

 

Now this is post 3, with a biasing scheme that is shorting the audio signals.

schmitt_trigger is right. virtual ground using buffer OpAmp is fine. but it also means that it will do what it designed to do - prevent changes at that point. basically it would also dampen (short circuit) the other three audio channels that you are trying to amplify.

 

The Fig. 3 circuit is Post #8 and Post #29 has the LM4871 input as a virtual ground summing point due to the negative feedback from Rf.
I see no problem with that.
It is not "shorting" the input signals, it is summing them as currents, as any typical inverting op amp summing circuit does.

 

schmitt_trigger is right. virtual ground using buffer OpAmp is fine. but it also means that it will do what it designed to do - prevent changes at that point. basically it would also dampen (short circuit) the other three audio channels that you are trying to amplify.

View attachment 351223

I do have that resistor fitted, it’s value is 100k

What I find strange is that if I connect any of the s single audio channels (Bluetooth L or R or DAC) directly to the input of the LM4871 eval board input (ie no series resistors or caps) the sound is still very quiet. And the gain of the LM4871 board is 7.

 

I connect any of the s single audio channels (Bluetooth L or R or DAC) directly to the input of the LM4871 eval board input (ie no series resistors or caps) the sound is still very quiet.

What is the output voltage of those signals?
What speaker are you using?

 

Signals from Bluetooth and DAC are shown in scope captures in post #20

I’ve tried 3 different speakers…

https://thepihut.com/products/mono-enclosed-speaker-3w-4-ohm?variant=31955927249

https://thepihut.com/products/20-x-...-ph1-25-connector-8-2w?variant=43764408516803

https://www.torkworld.com/collections/frontpage/products/tork-xpro-element

 

I’ve tried 3 different speakers…

Can you measure the voltage across the speakers?

 

Can you measure the voltage across the speakers?

Would I need a differential probe to do that?

Or use two probes and subtract the difference using a math channel?

 

to avoid audio channels influencing each other, you may add buffer to each audio input before summing, as for measuring output across speaker, yes differential probe is ideal, or using two probes and let the scope do the math Ch1-Ch2... or just measure one channel at a time and post both screenshots. for audio signal use single tone like 1kHz

 

to avoid audio channels influencing each other, you may add buffer to each audio input before summing

There should be very little crosstalk from one input to the other with a virtual ground summing junction, so I see no need for an added buffer.