Input requested for DIY Mixer-Preamp

[Quercus]

Hello - I'm working on an updated version of my mixer-preamp-amp and want input/comments on the mixer-preamp stage.

For background, the application is audio input from two computers plus a 3rd aux input, and output to a couple of 6 ohm monitors. This device allows variable mixing of the three inputs and then amplifies it for high-level output to the monitors. The amplifier stage is a TDA7294 based amp with 100W per channel.

Notes:

• Op amps are LM4562
• Grounded portion of the input is not shown
• Op amps are powered by dual-rail +/- 12V (with decoupling caps), not shown
• Mixer circuit below is repeated twice more, once each for the 2nd and 3rd input channels
• "MIX LVL 1" connects to 20K Audio taper pot for multi-channel level control
• "PA IN L+ 1" combines with the left input of the other two channels and enters the preamp. The three right channels similarly combine for the right channel preamp (not shown)
• Output from the preamp ("PA OUT L") then connects to the TDA7294 amplifier circuit via another set of 0.47 uF caps.
• TDA7294 Amplifier circuit it taken simply from the TDA7294 datasheet: Datasheet
• Full project schematic is attached as a pdf

Mixer:

Preamp:

• This shows the left channel preamp and receives the paralleled input from the three left channels
• Right channel preamp not shown

I welcome input and comments.

Thanks,
-Q

 

[ronsimpson]

What is the point of the 33pF and 100k pot? Volume?

 

[Quercus]

There's supposed to be a connection between 2 and 3 of the 100k pot which in turn is intended to help adjust the gain (volume control takes place between the preamp and the amp stages). As for the 33 pF, my understanding is that it helps prevent high frequency noise. But if I'm off on something please enlighten.
-Q

 

[ronsimpson]

 

[Tony Stewart]

You cannot dive Vin- directly from OA-s without mixing resistors to control max gain with Rfb = 122k , Av= -Rfb/Rin

33pF only offers about 20 dB fixed attenuation in the AM band regardless of pot setting depending on missing Rs resistors normally set to be greater than 22k for some attenuation, like 0.33 or 66k then a max gain of 122k/66k= 2x so +/- 6dB

 

[rjenkinsgb]

The mixing resistors, 22K, are shown as the last component in each of the preceding input buffer stages.

I must admit I do not see any advantage to having input buffers for a fairly low impedance system - eg. PC audio outputs. They could drive the level pots directly.

I'd be more concerned about ground noise, when connecting several separately powered sources; converting the buffers to differential amps could then be useful?

 

[Quercus]

Thanks all - making some corrections from above and tweaking values of C112 and R120.

I'd be more concerned about ground noise, when connecting several separately powered sources; converting the buffers to differential amps could then be useful?

I am very careful about my ground paths throughout the entire circuit and PCB to avoid ground noise. That said, I'm unclear on what the thought is here in converting the buffer into a differential amp - isn't an OA a differential amp to begin with? I've done some looking around trying to understand your intention but short of a pro-audio setup with true differential inputs (which I don't have), I'm not clear on what you're thinking. Can you elaborate please to help me understand the suggestion?

Thanks!

 

[rjenkinsgb]

I am very careful about my ground paths throughout the entire circuit and PCB to avoid ground noise

It's the ground voltage differences with the external equipment that will introduce noise. PCs are usually terrible for it due to the high and rapidly changing internal PSU currents, without counting "ground loops".

Ground loop (electricity) - Wikipedia

en.wikipedia.org

Problems are pretty much inevitable as soon as you start connecting separately, AC -powered equipment to a common device.

I did mean true differential inputs - use the positive input (V2) as the ground at each source, with screened cable and the screen only connected at the amp end, or isolated (non grounded / plastic body, if you are using a metal case) input connectors with the screen from the external source connected to positive (V2) on both channels.

You can add eg. 1K from positive V2 in to circuit ground at the amp input, to stop if being completely floating, if you wish.

The other option is to use inline stereo isolating transformers in the external wiring, to break the grounds - but it seems a waste to do that when you only need a could of extra 47K resistors with the circuit you have!
(I'd probably use 10K at all four points).

 

[Nigel Goodwin]

It's the ground voltage differences with the external equipment that will introduce noise. PCs are usually terrible for it due to the high and rapidly changing internal PSU currents, without counting "ground loops".

Ground loop (electricity) - Wikipedia

en.wikipedia.org

Problems are pretty much inevitable as soon as you start connecting separately, AC -powered equipment to a common device.

I did mean true differential inputs - use the positive input (V2) as the ground at each source, with screened cable and the screen only connected at the amp end, or isolated (non grounded / plastic body, if you are using a metal case) input connectors with the screen from the external source connected to positive (V2) on both channels.

You can add eg. 1K from positive V2 in to circuit ground at the amp input, to stop if being completely floating, if you wish.

The other option is to use inline stereo isolating transformers in the external wiring, to break the grounds - but it seems a waste to do that when you only need a could of extra 47K resistors with the circuit you have!
(I'd probably use 10K at all four points).

I think you're greatly overcomplicating this - and there's no point using balanced inputs if the source isn't balanced, because you then have to strap the balanced inputs as unbalanced.

This project is absolutely trivial, one opamp per channel, three pots, four resistors and a few caps - it's a basic simple virtual earth mixer. I knocked one up on my bed in under an hour years back, as I had a disco to do - only two channels required, and I built it in a tobacco tin (nice screened boxes you can solder to). I powered it from a PP3 9V battery, a with couple of extra resistors and a capacitor for a split supply - it worked absolutely perfectly.

 

[rjenkinsgb]

there's no point using balanced inputs if the source isn't balanced, because you then have to strap the balanced inputs as unbalanced

There is a very good reason:

The reference input connects to ground at the audio source (ie. mixer input socket "ground" is differential reference rather than actual local ground), so the input is purely the wanted signal at that source point and any ground noise or hum between different devices is ignored.

As I said in post #6, I don't see any point of input buffers for line level signals, they could connect directly to their respective volume pots.

The OP is using extremely high spec audio opamps though, and it seems a shame to use those then inject a load of external ground noise. The buffers can be converted to differential inputs by adding two resistors, and he can compare results with them connected as above, or just ignore the positive in so the opamp ref is to local ground.

 

[Nigel Goodwin]

There is a very good reason:

The reference input connects to ground at the audio source (ie. mixer input socket "ground" is differential reference rather than actual local ground), so the input is purely the wanted signal at that source point and any ground noise or hum between different devices is ignored.

Sod all difference mostly theoretical. Try it and see, don't make work for yourself when it may very well not be needed.

As I said in post #6, I don't see any point of input buffers for line level signals, they could connect directly to their respective volume pots.

I agree, hence I said one opamp per channel.

The OP is using extremely high spec audio opamps though, and it seems a shame to use those then inject a load of external ground noise. The buffers can be converted to differential inputs by adding two resistors, and he can compare results with them connected as above, or just ignore the positive in so the opamp ref is to local ground.

Doesn't matter how good the opamps are, line level makes it meaningless - 741's would be perfectly fine (just to annoy AG - and were probably what I used in my emergency tobacco tin mixer, I wonder what happened to it?).

 

[rjenkinsgb]

Sod all difference mostly theoretical. Try it and see

I know that I have to use stereo ground isolation transformers on my existing shared PC amp setup, for exactly that reason - the ground loop noise is bad. The isolating transformers eliminate it.

When I get around to rebuilding the amp setup, I will be using differential inputs.

 

[Quercus]

@ ronsimpson, Tony Stewart, rjenkinsgb & Nigel Goodwin:

Interesting discussion above and thanks for all the input. This helped me a lot. It really did.

Below is the updated design. I'm retaining the buffer stages and I've added the differential inputs (slept on it and agree it's a good change and minimal additional cost/complexity). I've removed the cap from the pre-amp stage. I already have a filter on the buffer stage, and since there's a trim pot on the preamp stage, the cutoff frequency could vary quite a lot. Tweaked several of the values as well.

Help much appreciated!
-Q

 

[rjenkinsgb]

I've added the differential inputs

For a differential amp to work correctly, the two input resistors must be the same (47K, OK) and the two feedback/reference resistors must also be the same - you have 47K and 1K there, so it won't work.

The 1K I mentioned was an option (probably not needed) which would connect from "MIX 1 LG IN" to ground, if you wanted to be able to use either single ended or differential input.

 

[Quercus]

All feedback resistors are 47K. The 1K is similar to the Rg resistor in the differential schematic above, as well as most other differential amp designs I've seen. I could see positioning the 1K (Rg) on either side of the 47K (R2) as I don't see it would make much of a practical difference for this application.

 

[rjenkinsgb]

Rg is part of the differential balance and MUST be the same value as Rf, when using equal R1 & R2, otherwise it's a waste of space.

(Technically, the ratio of R1 to Rf must equal that of R2 to Rg, so with R1 = R2, Rf must = Rg).

Imagine putting eg. +4V on both inputs; the opamp +in will be at 2V and the output must then go to 0V to hold the -in at the same voltage; 0V out as there is zero differential between the input.

 

[Quercus]

Rg is part of the differential balance and MUST be the same value as Rf, when using equal R1 & R2, otherwise it's a waste of space.

(Technically, the ratio of R1 to Rf must equal that of R2 to Rg, so with R1 = R2, Rf must = Rg).

Imagine putting eg. +4V on both inputs; the opamp +in will be at 2V and the output must then go to 0V to hold the -in at the same voltage; 0V out as there is zero differential between the input.

I see that now. Thanks for for helping me see the point. Will be updating the schematic to change the 1K to 47K. Not going to bother putting a 1K from MX 1 LG IN to ground.
-Q