8 channel differential audio mixer

[benbiles]

Hi,

I have a DAC wired up outputing 8 channels of differential audio ( balanced audio )

I would like to combine all the 8 channels into one differntial audio output introducing the least noise possible
and keeping the gain 1:1

I would have liked to do the mix in the DAC digatally but it seams that would need to be done in DSP or FPGA.

The volume contols are controlled via I2C from the microcontoller so I'm just looking to combine the differntial channels in the best way.

Would I use 2 op amps 1 non-inverting & 1 inverting in parralel ? if so is there a high quality dual op amp that would do the job well ?

Also does anyone know of an audio proccessor / dac / ??? IC that can digitally mix 8 or 16 tracks down to 2 channels ( stereo output ) whilst having unlimited mix routing control and or stereo balance control for each channel? I searched high and low but could only find TAS5558 that has mix routing but its outputs are PWM !!! and i'm not even sure if the mix routing allows combining of channels ( unclear in the datasheet )

If I can mix 8 - 1 in a clean way without loosing quality / dynamic range then I can use a second 8 channel dac for the 2nd channel output to form an 8 - 2 partly digital STEREO mixer

any ideas would be much welcome.. I will at some point order an FPGA board and try and learn DSP etc but for now I'm still trying to learn the basics !!

 

[benbiles]

[Moved]Re: audio mixer ic

wow, 10 years later and I am asking the same question !!! I don't want to have to use DSP or FPGA for simple 8-2 mixing of audio !! surely its possible in CMOS in a DAC or ADC to have a full crossbar mixer implimentation?
I get the idea theres more money to be made in selling FPGA and DSP kits !!!

Hello all,

I really don't want to have to use a 8 ch ADC, a DSP and a DAC just to produce a simple 8 to 2 (4 stereo) audio mixer.

I want this project to be as simple as possible. Ideally I want a digitally controlled analog mixer IC. Analog Devices and a few other companies made some very capable ICs for this purpose. Now that digital is fasionable they're all discontinued.

Suggestions please? Who still makes these things?
If there are no suppliers I'm gonig to have to go for a design with many more components, or change my control circuit.

Thanks,
Alan G

Added after 43 minutes:

Looks like I'm going to daisy chain 4 LM1972 2 chanel audio attenuators onto a 3 wire bus and then route the analog outputs to 2 summing amplifiers.

not quite as neat as a single IC solution, but it will have to do.

I'm still open to suggestions for 8+ input parts.

I basically want the smallest, cheapest 8 input, 2 output (4 stereo) digitally controlled mixer for line level signals possible.

This is for a box to sit between DVD players, CD players, etc and an amplifier. Pretty simple once I've chosen the parts. The hardest bit will be on the software side for the uC.

Thanks,
Alan

 

[BradtheRad]

An op amp easily performs the role of mixing channels. Or were you looking for something different?

Three channels are shown to illustrate the principle. This assumes each channel's volume has been adjusted upstream.

The potentiometer adjusts gain.

 

[benbiles]

8 channel mixer output op amp type for differential ( balanced ) low noise output

hi, started thread again, could not reply to comment on last thread.. appeard to be closed..

I am trying to mix down 8 differntial (balanced) audio channels to one differential ( balanced ) audio channel.

I'd like to achieve 0dbu differential ( balanced ) line level audio output on mono XLR

If using a differential OP Amp is the best way forward, how about one of these ?

OPA1632 differential input / output op amp

or a Bipola opamp IC ?

THAT1606 - looks like it was made for this job ?

I also found TI DRV135 & AD SSM2142 but both of these are
single ended input only..

Both options require +/-10v approx so I would need a low noise bipola power supply to achieve this.

how about the LTC3260 ? it has 100ma power output ( 50ma on each rail )

or is there a cheaper simpler solution that will produce high quality ?

in the OPamp examples I have seen the mix happens by simply joining the outputs to the same line with resistors prosumably to devide down the voltage and then the OPAMP boost the signal back up again.

can I do this with differential outputs in the same way ? or should I convert to single ended first ?

would love to know what is considered the best way to do this.

 

[FvM]

The thread was closed by accident. Please continue the discussion here.

- - - Updated - - -

I don't see a particular advantage of making the audio signalpath differential inside the instrument. But it's of course possible.

Differential signaling does increase the common mode rejection, e.g. regarding hum interference and crosstalk.

A standard audio mixer circuit uses an inverting OP amplifier topology with n inputs. You can do the same with fully differential OPs.

 

[benbiles]

differential audio reasons,

1, 8 channel DAC(s) output differential audio.
2, there will be a fair amount of UHF RF around the mixer and possiblly in the mixer so I guess it might help although lines are not very long.
3, Main reason ! changing balanced outputs to single ended requires another OPamp?

guess Its time to order 1 x bipoler power supply +/-15v & OPAMP..

do I just add resistors to the + and - audio outputs and hook up to differential op amp ?

The thread was closed by accident. Please continue the discussion here.

- - - Updated - - -

I don't see a particular advantage of making the audio signalpath differential inside the instrument. But it's of course possible.

Differential signaling does increase the common mode rejection, e.g. regarding hum interference and crosstalk.

A standard audio mixer circuit uses an inverting OP amplifier topology with n inputs. You can do the same with fully differential OPs.

 

[BradtheRad]

I see I had the wrong idea when I made post #3. I have not worked with differential *output* op amps, nor differential audio signals. About all I can picture is the output of an audio transformer. I guess you would need a transformer with 8 primaries, and 1 secondary.

 

[benbiles]

Hi BradtheRad, thanks for the comments / ideas.

project is a portable design and should be lightweight. If I had a transformer like that I would have tried it !! probebly would sound great !!

whats anoying with using an Op amp for the balanced audio is the Opamps that work with differntial audio require double rail power supply +/-15v or so. and my design so far gets away with just 2.5v & 5v. I think there is a way to bias the Opamp to accomadate the audios large +- swing in the positive but I'm not sure how to do this. I read this causes DC offset problems and I'm guessing I'd have to use large CAPs to RC filter the massive DC offset.. not sure that would be a good idea.

converting the 8 channels into single ended outputs then mixing would require 8 OPamps + 1 single ended - differential OPamp on output.

so back to the bipoler +/-15v power supply and low noise Opamp that can drive the 0dbu output

The only positive side is that if I make a stereo Panable version of this mixer I can use the one power supply to power both output opamps.

OK, will order the bits and post back here if I get this working..

any other ideas much welcome! The ten year old post pointed out some programable attenuation IC's I did'nt know about so they might come in useful. maybe as master mix volume.. I know I can use a varriable resistor on the OPamp for that but i'm trying to make the mixer digital wherever possible.

I see I had the wrong idea when I made post #3. I have not worked with differential *output* op amps, nor differential audio signals. About all I can picture is the output of an audio transformer. I guess you would need a transformer with 8 primaries, and 1 secondary.

 

[Audioguru]

Opamps do not need a dual polarity supply if they are biased at half the supply voltage and use coupling capacitors. Many modern opamps work with a 2.5V or 5V supply but for audio you need some headroom for unexpected loudness increases.

 

[FvM]

There's a reason why classical mixing consoles are using +/- 15 V (or even slightly higher) supply for the signal path. Modern OPs have low voltage noise, but you need also a considerably lower resistance level to take advantage of it. Otherwise resistor noise can out-compete the amplifiers.

Just a matter of calculation.

 

[benbiles]

going for THAT1280 Balanced Line Receiver to mix the 8 channels into one single ended.

then PGA2310 for master digitally contolled master output volume / attenuation.

then single ended output goes to the THAT1606 OPAMP to drive the balanced XLR output

trying to get LTC3265 or LTC3260 for low noise +/-15v for both OPAMPS for enough headroom.

maybe overkill but i'm trying to make a high quality 8 - 2 stereo mixer, so clicks and pops are not an
option.

would have liked to mix all digitally and just output 2 channels wirh the THAT1606 OPAMP

shame to have to go from balanced to single ended and back again just to make the master volume control and headphone output.. ah well

Opamps do not need a dual polarity supply if they are biased at half the supply voltage and use coupling capacitors. Many modern opamps work with a 2.5V or 5V supply but for audio you need some headroom for unexpected loudness increases.

- - - Updated - - -

so can anyone tell me how I calculate the 8 channel divider resistors?

am i trying to divide peek voltage into 8 and so any high value resistor will work as long they are equall ?

Is there a proper way to calculate this ?

 

[BradtheRad]

The act of mixing does not necessarily need an amplifier. Send each signal through a weighting resistor (or potentiometer). Add one final resistor to ground. Then all the resistors form a divider network. The signal amplitudes are summed to become the output amplitude. It may need attenuation, to avoid overloading the input of your headphone amplifier.

----------------------------------------

In case you are interested in a one-transistor mixer (single-ended 5V supply):



You can add channels at the same node where the two channels join.

Notice the flexibility in regard to whether there is DC component (or none) in the signal.

 

[chuckey]

Studio line normally run at 0 dbM nominal, but should handle +10 dbM without any added distortion, which I would say needs at least 3dBs head room. So I think you should be aiming for a 5 V output, or 14V P-P. This then tells you the minimum Vcc that will be required, or a transformer. The output should have a low impedance (> 1 ohm?), so jacking in extra leads don't cause any change in level. I can't remember the exact circuit of the DAs we used, but they ran from +24V with a TO3 class A power transistor and a 2" cube output transformer. The transformer made them bomb proof from spurious voltages being fed from long input lines into the DAs output.
This is where the design is, the actual mixer is simple, the only thing you have to watch are the differential inputs, are to make them bomb proof and be able to reject an enormous amount of common mode hum.
Frank

 

[benbiles]

done further reading on this.

It turns out nobody made mixing desks with just passive mixing and OPAMP recovery. you go from all 8 channels pumping into the mix , pull down channels 2 -7 to nil and suddenly channel 1 gets loader.

That's unless you have the virtual earth design. negative feedback into the earth of the opamp i think?

it looks like I'm looking for an 8 channel balanced mix bus design. I'm just unclear how the virtual earth ( negative feedback ) concept would work in the +/-15v differential Opamp recovery. 2 opamps one with virtual earth and other with virtual positive??

source here ... **broken link removed**

Studio line normally run at 0 dbM nominal, but should handle +10 dbM without any added distortion, which I would say needs at least 3dBs head room. So I think you should be aiming for a 5 V output, or 14V P-P. This then tells you the minimum Vcc that will be required, or a transformer. The output should have a low impedance (> 1 ohm?), so jacking in extra leads don't cause any change in level. I can't remember the exact circuit of the DAs we used, but they ran from +24V with a TO3 class A power transistor and a 2" cube output transformer. The transformer made them bomb proof from spurious voltages being fed from long input lines into the DAs output.
This is where the design is, the actual mixer is simple, the only thing you have to watch are the differential inputs, are to make them bomb proof and be able to reject an enormous amount of common mode hum.
Frank

 

[chuckey]

The circuits do not have balanced inputs. I think the input stages should be a unity gain buffer with the inputs optimised for CMRR with diode clamps to stop gross overloads. This will then give you single ended outputs to feed one of the above circuits.
Frank

 

[benbiles]

Hi Frank, CMRR? common mode xsomethingx regection ?

I was going to try and use the THAT1280 Balanced Line Receiver to mix 8 balanced into single ended, then send that to another digitally controlled attenuator for master mix controll for that channel.

can I use the THAT1280 as a unity gain buffer? or would I need 8 ( or 4 since there 2 channel ) of them to convert diferential channels to single ended first?

is there no such thing as a 'balanced mix bus' ?

The circuits do not have balanced inputs. I think the input stages should be a unity gain buffer with the inputs optimised for CMRR with diode clamps to stop gross overloads. This will then give you single ended outputs to feed one of the above circuits.
Frank

- - - Updated - - -

I found this on another website,

a balanced audio summing 2 opamp design? apparently the polaritys are the wrong way around in the drawing.. maybe this is what i'm looking for..

then maybe I would have a mono 8 - 1 end to end differentail audio mixer.

Hi Frank, CMRR? common mode xsomethingx regection ?

I was going to try and use the THAT1280 Balanced Line Receiver to mix 8 balanced into single ended, then send that to another digitally controlled attenuator for master mix controll for that channel.

can I use the THAT1280 as a unity gain buffer? or would I need 8 ( or 4 since there 2 channel ) of them to convert diferential channels to single ended first?

is there no such thing as a 'balanced mix bus' ?

 

[D.A.(Tony)Stewart]

• RF Common mode chokes will be necessary to balance the impedance (BALUN) and prevent AM detection of RF into audio noise.
  
• interleaved grounds in layout is useful to reduce crosstalk in high impedance lines.
• use quality caps for low level signals to avoid microphonics (ceramic piezo) and high ripple current electrolytics
• use SMPS to pre-regulate <0.5V LDO's if PSRR is not sufficient.

 

[benbiles]

sounds good, do you have any example schematics of balanced mix bus / summing opamp design with RF rejection ?

• RF Common mode chokes will be necessary to balance the impedance (BALUN) and prevent AM detection of RF into audio noise.
  
• interleaved grounds in layout is useful to reduce crosstalk in high impedance lines.
• use quality caps for low level signals to avoid microphonics (ceramic piezo) and high ripple current electrolytics
• use SMPS to pre-regulate <0.5V LDO's if PSRR is not sufficient.

 

[chuckey]

The THAT 1280 is made for this job. One reason differential amplification is not used, is that you need two faders per channel. Which also need to be accurately matched.
Frank

 

[benbiles]

Hi Frank , thanks for your comment makes total sence you don't want to have two faders per channel !! did'nt think of that one at all !!!

however, since i'm doing the fading digitaly in the dac I wont have that issue..

so, do I need to use 4 ic's (That 1280's) for the 8 channels to mix down to one single ended, then 8 resistor dividers + another OPAMP to recover level? or just the one IC with 16 devider resistors and a balanced output something like schematic i posted earlier?

I'm thinking, keep the signal balanced all the way through, becasue I can !! there will be a lot of other noisey stuff going on in the project so I'm hoping it might help with noise / interfearnce in the design to avoid single ended if I can..

I just ordered some OPA1632's for mic preamps for 8 channel differential ADC chip CS5368. the ADC needs 5Vpp input diferential, a bit less than the That 1510 / 1570 pre amps output. I guess I could have trimmed 1510 or 1570's back to 5Vpp but the OPA1632 just seamed to be made for the job.

well , i'll wire them up and hope i dont see a puff of smoke I'm hoping I can use my volt meter to see the Vpp outputs +/- peek with music or something to check before I blow anything up.. dont have either a sinal gen. or an oscillioscope

ok, i'm off to learn how to make 48v phantom power and preamp cuircit protection / rf choke on the inputs etc..ahhh!!! :bsdetector:

The THAT 1280 is made for this job. One reason differential amplification is not used, is that you need two faders per channel. Which also need to be accurately matched.
Frank