Need to pick your brains about making a remote summing mixer

[alexanderbrevig]

Hi all!

PREQUEL: I run a side business as an audio engineer and I the job I most often do is to mix a choir with around 30 hand held mics on stage. Dragging along a huge multi and a complete mixing console only for summing those mics into four vocal groups (bass, tenor, alto and soprano). I want a simple box with 8 XLR in and 2 XLR out. No phantom needed as I'll only use dynamic mics for this.

WHAT I WANT TO DO:
9 XLR inputs, each has a switch for which out it will sum to (except the one witch will be for simply adding a previous box to the sum)
8 digitally controlled preamps/attenuators
2 XLR audio out
2 XLR for RS-485 with max485 (IN/OUT for linking) with switch for inserting termination for last node

I want to use the preamps on my mixing console for the acutal preamping.

The plan is then, that I can have say 5 of these boxes on stage and use a 5 XLR multicable to the stage.
1 for bass, 1 for tenor, 2 for alto (this is where one box will feed into to 'link' input and just add itself to the sum) and 1 for soprano. And lastly, one for RS-485 into a control device I will make.

QUESTIONS:
Should I buffer the signal or simply just use a digital pot to create a remotely controlled LOG attenuation?

IF BUFFER; How does one buffer a balanced signal?
I'm scared if I just use a follower on each line that some distortion artificats (due to mismatches?) will make problems...? Is this a valid concern?

Thanks for your time!

EDIT: found a schematic I did a while back trying to get some of my thoughts down

. Since then I've decided to drop phantom power and use XLR for serial comms (always plenty XLRs during gigs). Probably the only thing that would be the same if I did it again is the digital pot. Terrible layout too, don't give it any attention. Thanks again.

[alexanderbrevig]

I've written down a simple schematic of what my current thinking is (attached).
This is one channel, there will be eight of them, all summing into the SUM2 and SUM3 net.

Should the AC coupling caps be before the summing resistors or not? Maybe I don't need the summing resistors either, as it's kind of built in to the DS1882...?

[Andy Watson]

... only use dynamic mics for this.

I want to use the preamps on my mixing console for the acutal preamping.

Where are you putting the preamps with respect to the rest of the circuitry - your diagrams seem to suggest that you are attempting to do your switching between mic and console? If you want to maintain a reasonable level of signal quality - i.e. noise free - you will have to amplify the mic signal before any subsequent processing.

[Richard Crowley]

There is no particular benefit from attenuating/mixing balanced signals. Almost no modern audio gear does that because the cost/benefit ratio is significantly negative.  Mixing unbalanced signals is that rare combination of easier, better, and cheaper.

For a simper, budget design, I would make mic preamps using the INA217 (or THAT1510) reference design followed by quad digital pots like MCP4461 (which have 4 channels in each package = 50 cents per channel).

For a more high-end approach (and >$$) you could use those combined digitally-controlled mic preamps like TI PGA2500 or THAT 1580/
5171

I have been kicking around a project making INA217/THAT1510 mic preamps mounted right on the back of the XLR jacks. To make projects like this simpler to assemble.  (and line-level input and output and several other handy audio modules also)

While Maxim makes some innovative chips, they have a rather poor reputation for availability and continuity.  I have a growing pile of what would have been cool projects and products if the Maxim chips hadn't been summarily withdrawn.

[felixnavid]

How are you planning to power these buffers? Send phantom power from the mixing console? Or maybe an external power supply?

[alexanderbrevig]

Where are you putting the preamps with respect to the rest of the circuitry

As I've stated I want to use the preamps on my mixing console for the acutal preamping

There is no particular benefit from attenuating/mixing balanced signals.

Well.. As it's a very low voltage signals and long cables (microphones on stage to box, then from box to my mixer) I will need balanced or else it will be nothing left of it once I see it at the console.

For a simper, budget design, I would make mic preamps [...]

As stated above, I want to use the preamps I already have in my console

How are you planning to power these buffers? Send phantom power from the mixing console? Or maybe an external power supply?

An excellent question! I've tried to select ICs with a low current draw. Not sure how much you can pull from a regular phantom power (is there a standard)?
The best would be to have either phantom power or a euro chassis/IEC C13 plug too (but those AC-DC converters are typically expensive).

Thanks for your inputs and questions

[Andy Watson]

Where are you putting the preamps with respect to the rest of the circuitry

As I've stated I want to use the preamps on my mixing console for the acutal preamping
 my console

Have you considered the noise problem?

[Richard Crowley]

Sorry, I didn't even dream that you were trying to combine/mix the mic signals at MIC-LEVEL even before they get to your mic preamps in your mixer.

IMHO, your project is doomed to poor performance because attempting to do that kind of level control, mixing, and combining at MIC LEVEL is simply not practical. But feel free to try it for yourself. Maybe your requirements are modest enough that you can get away with it. But I wouldn't bet on it.   

[alexanderbrevig]

Have you considered the noise problem?

Which of them?

IMHO, your project is doomed to poor performance because attempting to do that kind of level control, mixing, and combining at MIC LEVEL is simply not practical. But feel free to try it for yourself.

I did a test with stereo pots and it seemed to work very well. Why would a digital pot make troubles? Passive XLR summing is not a new thing in live audio... I will give it a shot and report back here just need to get some parts ordered and I'll make a protoboard of it.


This is a typical schematic used in patch bays and the like. The more expensive ones usually use isolation transformers (jensen etc) too, but I don't need it.

Maybe your requirements are modest enough that you can get away with it. But I wouldn't bet on it.   

They probably are. It just needs to be good enough for live audio where the noise floor is already quite high.

Thanks again for your input

[fcb]

AFAIK no one uses remote attenuation without amplification. I can't think of single benefit - and a million reasons it would suck. If your bothered about cost, then make mic amps with NE5532's - and perhaps keep the gain down (no need for 40/60dB gain in the head stage - maybe just go for 20dB, so plenty of headroom so you can use bass-cut in the desk).

[Zero999]

Should the AC coupling caps be before the summing resistors or not? Maybe I don't need the summing resistors either, as it's kind of built in to the DS1882...?

Check whether those electrolytic capacitors will be properly biased.

Alternatively use polypropylene film capacitors which aren't polarised and will have less distortion.

[Andy Watson]

Have you considered the noise problem?

Which of them?

Any of them?
The DS1882 adds up to  2.2\$\mu\$V at its output - presumably twice  (x1.4) for a balanced signal. Then there's the mix of digital control signals to add-in.

[David Hess]

Wow you guys are pessimistic.  Digital noise will only be present when the control signals are being switched.  With a little bit of RC filtering, it might be possible to prevent to much interference even then.

Noise levels will be increased but the big problem will be preventing external EMI from radiated and conducted sources.  With careful design, I think that will be manageable.

[Messtechniker]

Take a look at "Designing a Professional Mixing Console" by Steve Dove.
Downloadable from here: http://84.255.203.119/Steve-Dove-Console-Design.pdf.
Its some 40 Meg. But contains highly valuable information - non only on summing.
As to summing, go to page 96 of Nov 1981.

Yours - Messtechniker

[Richard Crowley]

Wow you guys are pessimistic...

Or maybe "us guys" do professional-level audio where mic-level manipulation is abhorrent because of the signal-to-noise ratio problem. But then our concept of "good" signal-to-noise ratio is probably an order of magnitude (or two) more than casual, amateur use-case.

With a little bit of RC filtering, it might be possible to prevent to much interference even then.

Designing a system that REQUIRES downstream remediation is just poor planning from the professional POV.

[alexanderbrevig]

So this is my current thinking:
If I attenuate the balanced signal at -6dB then the noise present will also be attenuated right?
Because it is balanced, common mode noise will be rejected.
Because the POTs have EEPROM I only induce potential (common mode) noise when changing them? I plan to use internal osx of the yet to be selected micro.

Though, I'm sure if I could make good enough preamps it would be the better way to go if I ever intended to sell this to other audio engineers.
Then I would need phantom power and maybe I want to do I2C TDM over ethernet and I'm stuck for another year with nothing that makes my life simpler.

This is only intended to be a practical way (and maybe it's not) to extend my Soundcraft Si Compact so that I don't have to waste so many channels only to sum down the choir into its bus.
This is the situation I'm trying to get rid of:

(Never mind that extra Behringer X32, it was just for recording purposes)

Thanks again guys.

[fcb]

If you're worried about 'designing good enough' preamps - then don't, just slap down the PGA2500's - they're pretty good.

[alexanderbrevig]

I'm warming up to the PGA2500 suggested by Richard and fcb. Seems easy to implement and more than good enough for my live audio requirements. Maybe I should JUST DO IT and go all in with digital I2S TDM and learn a few many things in the process?

[Andy Watson]

Because it is balanced, common mode noise will be rejected.

Noise that is common, power-supply, switching, ground and external radiant sources, should cancel. However, the noise introduced into each leg of the channel by the active electronics is uncorrelated and will be additive. \$Vn_{total}^2 = Vn_1^2 + Vn_2^2 + \$, etc..

Suck it and see - with one or two channels - I think you will be disappointed. Your original sketch contains most of the components needed - why not go the extra distance and do the job properly?

[David Hess]

Wow you guys are pessimistic...

Or maybe "us guys" do professional-level audio where mic-level manipulation is abhorrent because of the signal-to-noise ratio problem. But then our concept of "good" signal-to-noise ratio is probably an order of magnitude (or two) more than casual, amateur use-case.

But look at the application.  The environment in this case does not demand the lowest noise leaving leeway to sacrifice some noise performance.

With a little bit of RC filtering, it might be possible to prevent to much interference even then.

Designing a system that REQUIRES downstream remediation is just poor planning from the professional POV.

Filtering digital signals to remove fast edges which would otherwise couple into sensitive nodes where mixed signal designs interface is a tried and true technique.  One might as well say that designing a system which relies on common mode rejection to remove external noise sources is just poor planning from a professional point of view since better shielding and ground arrangements could have been used instead of the extra complexity.

Getting back to the design, it is fascinating to consider the old linear techniques for attenuating differential signals under electronic control but I think they suffer from limited control range; I wonder how far they could be pushed though.  Converting to high level single ended signals, processing them, and then converting back to a low level differential signal is straightforward (and I have done this very thing in the past) and would provide a known level of performance.

[alexanderbrevig]

Allright, so I've tinkered a bit with one passive solution, and one using the PGA2500.
I've attached PDFs in case you want to have a look

[Andy Watson]

I think you need to look at the resistance that you are presenting to the attenuators/summing junction. The first diagram does not appear to have any resistance and the second has 330 ohms. If I recall correctly the output of DS1882 is of the order of 20k in mid-position. Also, you might want to buffer these outputs and/or sum into a virtual earth - otherwise the setting of one channel will interact with all the other channels.

[StillTrying]

passive-schematic.pdf
I don't think that passive 1K/1K GND/GGND arrangement is going to work at all. Even just the ground current of the VDD's 7805 will raise the GND/GGND up by 1.5V towards the 2.5V.

[alexanderbrevig]

I don't think that passive 1K/1K GND/GGND arrangement is going to work at all. Even just the ground current of the VDD's 7805 will raise the GND/GGND up by 1.5V towards the 2.5V.

I got overly creative doing that. I would rework the power section to use modern regulators and probably some AC-DC modules so I can use a normal "euro chassis" power plug.

I think you need to look at the resistance that you are presenting to the attenuators/summing junction. [...] otherwise the setting of one channel will interact with all the other channels.

Actually this has me confused. I intended to buffer each channel but then the presence of voltages on the out from other channels will interfere with the feedback?
As you can see, I really am in over my head. I have an intuition with digital electronics but this is my first try at analog.
In my head what I want to do is simple, but when I start thinking about it - it turns out it's not.

Currently I like the preamp solution better. My thinking there is that the output of the preamped signal all go through an 8k and then that sum has a 1k to ground (normal voltage divider, yes?) so the sum of all preamped signals (let's say at 1V) would be 1V at the output as well.

Thanks again guys! I continue to learn a lot (I only started a year ago).

[StillTrying]

My thinking there is that the output of the preamped signal all go through an 8k and then that sum has a 1k to ground (normal voltage divider, yes?) so the sum of all preamped signals (let's say at 1V) would be 1V at the output as well.

A summing inverting amp has near enough zero interaction between it's inputs, because the op amp doesn't let its
-Ve input move from it's virtual earth voltage.
http://www.facstaff.bucknell.edu/mastascu/elessonshtml/opamps/opamp3.html#Summer