Circuit Review: Audio Routing

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
Hello! I've been designing a new feature for my animatronics. First, the audio is being sent (via Bluetooth) to a pair of highly-directional speaker/amplifier systems. I plan on keeping the audio local to each character, as it is used as input to the servo-control subsystem for synchronized speaking and audio.

However, the other channel will be amplified locally for an internal speaker and sent to the DSA systems. One line level output will feed an amplifier and two mixers... It's not practical to split out the internal amplifier feed, but I plan on using an active splitter to feed two mixers... Each mixer will mix/balance up to four sources and transmit the audio signal to the DSA via Bluetooth. Each mixer input will come from one side of the splitter, balanced with a 10K Linear Pot.

I've researched both active splittters and mixers. I found an article on Learning Electronics for a three channel splitter. I modified the output resistor values to make it a two-channel splitter. Here is a closeup of my splitter model in LTSpice:
AudioSplitter.JPG


AAC has a thread on a mixer, but I used another circuit which is very similar as it was easier (IMHO) to modify to include the splitter. The mixer circuit is attached to this thread.
Multi-channel Mixer.JPG

So, my circuit combining the two functions is shown below. This image is pretty tiny, so there is a .pdf and .asc file as well.


SceneMixer.JPG

Basically, I have a couple of uncertainties.

  1. The line input is coming from a VS1053 codec. It's datasheet states clearly "do not connect to ground." But I wonder if that means ground of the VS1053 (as the audio ground is created by the chip). But if I'm connecting to another circuit, whose ground is not connected to the VS1053 supply ground, I should be ok. As long as digital ground is separate from analog ground. (I'm struggling with this concept)!
  2. I've adjusted the splitter resistor network. The original circuit had three outputs, each fed with an 330Ω resistor. The ratio works out to be an even 33%. But with two outputs, I changed it to 470Ω. The total resistance is close to 1KΩ in both cases and the ratios are commensurate with the number of outputs (1/3 for three outputs; 1/2 for two outputs)
  3. I've not figured out what to do if I only need 2 or 3 sources... Normally, I'd explicitly wire up an unused op amp. But in this case, it might be in use later, just not now. Looking at the circuit, I may be set, as there are no floating inputs. But I'm also not sure.
  4. And to close out, any other comments are welcome! I'm more comfortable writing code for a microprocessor than analog electronics. But this is a particularly critical project that I can't screw up.
So on that note, thanking you in advance,
Don dj Lutz
dj's fantasi
 

Attachments

crutschow

Joined Mar 14, 2008
24,077
Here is a closeup of my splitter model in LTSpice:
How much gain do you want for the mixer circuit?
Note that Spice interprets a 1m resistor as 1 milliohm.
You need to use 1meg if you want a 1 megohm resistor (but that will give a lot of mixer gain).
The line input is coming from a VS1053 codec. It's datasheet states clearly "do not connect to ground."
There is a line in the data sheet that says the common buffer GBUF must not be connected to ground.
Is that what you are referring to?
If so just follow the data sheet recommendations.
But if I'm connecting to another circuit, whose ground is not connected to the VS1053 supply ground, I should be ok.
Not ok.
All circuit grounds must be connected together at some point, otherwise the signal return current from the codec has no place to go between circuits.
I'm sure you know that all electrical signals always need a complete (return) path. :)
I've not figured out what to do if I only need 2 or 3 sources..
Just ignore the unused inputs and turn their input gain to zero to minimize noise pickup.
 

Audioguru

Joined Dec 20, 2007
11,251
You are using low power but normal LT1012 opamps in an circuit from CircuitsToday in India that was designed for an old and odd Norton LM3900 quad opamp. The voltage gain settings of a Norton opamp is completely different to a normal opamp.

Also, the 470 ohms output resistor values are too low for an ordinary opamp and are much too low for your low power opamps.

Your 0.01uF output capacitor values are so low that they might feed only ultrasonic "sounds" to an amplifier.
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
Thanks for the link. I’ll look at in more detail later.

My circuit requirements are:
  1. To allow line out from a VS1053 to be input to an amplifier and two Bluetooth transmitters (or a mixing circuit)
  2. To be able to “pan” the sound level between the 2 BT xmitters.
  3. To be able to mix up to four sources into the two BT xmitters.
@crutschow - I don’t think I’ll need any more than a unity gain from the mixer. Input levels are a pre-set (100K log pot).

I definitely have some confusion about the grounds. The audio from the VS1053 comes from the codec channels and the current returns via GBUF. You can’t ground GBUF because it’s around 1.2V w.r.t. the codec’s supply. In that case, grounding would cause a short. But if I had an external amplifier or other circuit (like Bertus’ pan circuit, the input jack would need to be grounded. And in the circuit, it is. There is no path for GBUF to short. It would be 0V w.r.t. the second circuit. Since it’s external, there is no path to the codec ground; it’s all relative to the second circuit.

I’m sure you know that voltage measurements are always between two points Otherwise, you could never use line out on the VS1053! What am I missing?

@Audioguru - I chose CircuitsToday as kan example because AllAboutCircuits mixer looked similar but had less detail.

You always catch my most egregious errors. However, your advice is difficult to use. Obviously, I’m misunderstanding something. Like “your 0.01uf output capacitors are so low that they might feed only ultrasonic “sounds” to the amplifier.” I’m glad to know that but a more helpful reply might include “... try a 470f * cap.”

A big boatload of thanks for your replies.
 

Wolframore

Joined Jan 21, 2019
1,530
Hey dj, the capacitance and the impedance form a high pass circuit... so your 0.01uF would pass ultrasonic... usually the coupling caps are about 1uF-10uF... depends on your circuit and what you want to cut off... the formula is C=1/(2 * pi * F * R).
 

crutschow

Joined Mar 14, 2008
24,077
I definitely have some confusion about the grounds. The audio from the VS1053 comes from the codec channels and the current returns via GBUF. You can’t ground GBUF because it’s around 1.2V w.r.t. the codec’s supply.
The data sheet says:

"If GBUF is not used, LEFT and RIGHT must be provided with coupling capacitors. To keep GBUF stable, you should always have the resistor and capacitor even when GBUF is not used."

The typical circuit application shows a 10Ω resistor in series with a 47nF capacitor to ground for that.
That should provide an AC return path for the signal.
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
Bertus, after reading your link in detail, I'm not sure that a pan circuit is what I need. I’m using a highly Directional Speaker/ Amplifier (DSA) system. That is, guests on the left of the “stage” cannot hear the right speaker; guests on the right of the stage cannot hear the left speaker.

Upon writing this above, I think that perhaps a pan system will work! A source on one side of the “stage” (for discussion, let’s assume it is on the right) will be “panned” so that it is louder to the guests on the right. It will have to still be heard, albeit at a lower level, to guests on the left.

That will work with the circuit in the link you provided. I might build it with three pots in series... two trim pots, on each side of the central pot, will set a minimum level so that neither side can accidentally be turned off.

In a normal situation, this is a desirable feature because each speaker, left and right, can be heard throughout the audience. With a DSA, if no signal is supplied to one speaker, the guests on that side will hear nothing - an undesirable result - they cannot hear the other speaker!

I would assume that the outputs of four of these pan circuits could be fed through a set of parallel output resistors - apparently something larger than 470Ω - to blend the output of the four sources. And apparently I need an output capacitor larger than 0.01μf (I’ll try 10μf) (I tried to calculate a value with the formula @Wolframore provided, but I’ve been unable so far to determine the impedance if a typical Bluetooth transmitter).

Am I getting closer?
 
Last edited:

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
The data sheet says:

"If GBUF is not used, LEFT and RIGHT must be provided with coupling capacitors. To keep GBUF stable, you should always have the resistor and capacitor even when GBUF is not used."

The typical circuit application shows a 10Ω resistor in series with a 47nF capacitor to ground for that.
That should provide an AC return path for the signal.
Thanks... but GBUF will be used. One channel (let’s say the right channel for this discussion) will be sent to a splitter/ mixer/ pan circuit for processing before being sent to a Bluetooth transmitter (or subsequently an amplifier)

The circuit I originally proposed uses bipolar op amps. Where I am stuck is that the audio source and splitter/ mixer/ pan circuit are powered by independent power supplies. The analog source signal is measured w.r.t. GBUF. the analog signal in the processing circuitry is measured w.r.t. ground of a ±12VDC supply. I imagine this ground and GBUF connected do that.

Another way of stating my confusion is how in the world does an audio signal from the VS1053 get processed if GBUF cannot be connected to ground.

I haven’t yet grokked the wiring of the pan circuit.
 

crutschow

Joined Mar 14, 2008
24,077
how in the world does an audio signal from the VS1053 get processed if GBUF cannot be connected to ground.
I believe the problem is that the GBUF AC signal has a DC bias.
So it can be connected to ground (for the AC signal) with a capacitor to block the DC.
(Makes sense, right?)
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
I believe the problem is that the GBUF AC signal has a DC bias.
So it can be connected to ground (for the AC signal) with a capacitor to block the DC.
(Makes sense, right?)
Yes, you know, it makes perfect sense! So do you think that what I’ve been missing all along, is to connect GBUF (analog ground) to GND (circuit ground) with a 10Ω resistor in series with. 47nf cap?

Thanks for hanging in there...

I figure I can breadboard your recommended pan circuit with a couple of trim pots, to determine the limiting resistors needed for my custom mixer.

Then all I need is to determine the value for the output resistors and an output capacitor value.

Awesome
 
Last edited:

Wolframore

Joined Jan 21, 2019
1,530
Can’t wait to see what you think about the VS1053 module.

Dj, you have a pretty ambitious project here. I have many mixers but would never have thought about building my own.

Mic/line in then a pad
Preamp (eq?)
Pan circuit

Summing to another stage or main output.

That’s very cool.
 

Sensacell

Joined Jun 19, 2012
2,544
Could you not eliminate a lot of parts by just having two output amps, with summing resistors connected to each input signal?

This would eliminate 6 X LT1012 amps?


I have never designed an audio mixer, but my gut tells me it could be greatly simplified.
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
I believe the problem is that the GBUF AC signal has a DC bias.
So it can be connected to ground (for the AC signal) with a capacitor to block the DC.
(Makes sense, right?)
Yes, you know, it makes perfect sense! So do you think that what I’ve been missing all along, is to connect
Hey dj, the capacitance and the impedance form a high pass circuit... so your 0.01uF would pass ultrasonic... usually the coupling caps are about 1uF-10uF... depends on your circuit and what you want to cut off... the formula is C=1/(2 * pi * F * R).
Ok, while I never was able to determine the Bluetooth transmitter input impedance, I tried some numbers and assumed the frequency was 20,000Hz (upper limit of the audio range). And I assumed the input impedance was 16Ω...

2*3.1415928*20000*16 =
2,010,619

1 / 2,010,619 =
0.000 000 5 =
0.5μF​

Or assuming an 8Ω input impedance

1 / 0.000 000 1 =
0.1μF​

Both are far below the 1μF absolute minimum @Wolframore suggested.

So what’s a reasonable value? So I can figure out the proper way to calculate the cap value.
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
Could you not eliminate a lot of parts by just having two output amps, with summing resistors connected to each input signal?

This would eliminate 6 X LT1012 amps?


I have never designed an audio mixer, but my gut tells me it could be greatly simplified.
No, because I effectively have four inputs that need to be sent to eight outputs.

Each of the four inputs need to have a unique and relative split of their signal strength to two different amps.

There are four such modules. Four of the split inputs needs to be summed and sent to one Bluetooth transmitter. The remaining four split inputs needs to be summed and sent to the other Bluetooth transmitter.

This is a minimum of eight op amps.
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
I might expand this module to eight sources during manufacturing. You know as soon as I make a four channel module, I’ll need fifth, sixth, etc... inputs! I already can foresee five inputs!
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,923
The module would have a faceplate that can accept 8 - 3.5mm stereo audio cables, and a total of 16 pots. 8 log pots as the audio level preset and 8 additional pan pots. On the reverse of the enclosure are two - 3.5mm stereo jacks, into which two Bluetooth transmitters are plugged.

I’ll have a lot of Bluetooth transmitters in the area. 20-40 as a rough estimate. There'll be 20-40 receivers in the same area (~5,000 sq.ft.)
 

Audioguru

Joined Dec 20, 2007
11,251
A Bluetooth transmitter will not have an audio input impedance of 8 ohms or 16 ohms which is a speaker impedance. Instead its input impedance will be 10k ohms or more.
An audio coupling capacitor value is not calculated to pass frequencies higher than you can hear. Instead it is calculated to block DC and very low frequencies that you cannot hear. You want the coupling capacitor to pass audio frequencies.
I assume your audio is speech which goes as low as about 50Hz. Then a coupling capacitor value to pass audio frequencies above 50Hz into 10k ohms is 1 divided by (2 x pi x 10k ohms x 50Hz)= 0.32uF. Use a 330nF film capacitor.
 
Last edited:

Wolframore

Joined Jan 21, 2019
1,530
Yes, you know, it makes perfect sense! So do you think that what I’ve been missing all along, is to connect


Ok, while I never was able to determine the Bluetooth transmitter input impedance, I tried some numbers and assumed the frequency was 20,000Hz (upper limit of the audio range). And I assumed the input impedance was 16Ω...

2*3.1415928*20000*16 =
2,010,619

1 / 2,010,619 =
0.000 000 5 =
0.5μF​

Or assuming an 8Ω input impedance

1 / 0.000 000 1 =
0.1μF​

Both are far below the 1μF absolute minimum @Wolframore suggested.

So what’s a reasonable value? So I can figure out the proper way to calculate the cap value.
Hey dj it’s a high pass you don’t want to cut off everything below 20k

That frequency is the -3dB point

Again it depends on where you need the cutoff.

edit- I agree with audioguru 16 ohms would be too low impedance, which could potentially pass too much current... do you have the datasheet?
 
Last edited:
Top