Gain makeup need from summing two channels?

[Cojo]

Hi all.

I need to sum the outputs from my to stereo busses on my mixer. IOW I need to make two balanced summing resistor network to sum only two channels each.

Now to my question: do I need to make up the gain loss or does the outputs from my mixer manage to provide the nessasary level that's needed?

/Cojo

[ulysses]

It depends. Mostly it depends on whether you're also going to be using those two signals independently at the same time you're using a sum of them. The reason this is important is that it determines whether or not you need to take steps to avoid crosstalk. If you're only using the sum, then the question of crosstalk becomes irrelevant - you're trying for crosstalk. But if you also need to send each of the two signals elsewhere and don't want them polluted with each other, then you need to use an arrangement that minimizes crosstalk. That's the one big sticking point to passive summing - you have to choose between crosstalk and insertion loss. If you don't care about crosstalk, you just use the smallest resistor value that doesn't piss off each output. Put that resistance in series with each channel (or each leg of each channel, if it's balanced,) and then join them after the resistors. In order to reduce crosstalk, you can increase those resistor values or you can decrease the load resistance on its output by adding a shunt resistor. Higher series resistance and lower shunt resistance will both serve to decrease crosstalk and also will increase the insertion loss (attenuation). Rather than go through some examples, let me just direct you to this tech note from Rane Corp that explains it thoroughly. Their website is very good and everybody should read it all.

[Cojo]

Thanks Justin.

So you're saying that if I use low value resistors I don't need to make up for the gain loss but I get higher crosstalk and if I use higher values I get lower crosstalk but also need to make the gain.

How much gain makeup are we talking about if I want as low crosstalk as I had initially?

/Cojo

[David Kulka]

For a passive resistor summing network that is wired correctly, expect 6db of loss, ie 6 db of make up gain needed afterwards.

[Cojo]

Originally Posted by David Kulka

For a passive resistor summing network that is wired correctly, expect 6db of loss, ie 6 db of make up gain needed afterwards.

Ok. Thank you very much!

/Cojo

[Geoff_T]

Originally Posted by David Kulka

For a passive resistor summing network that is wired correctly, expect 6db of loss, ie 6 db of make up gain needed afterwards.

Hi

Mathematically that's true, but, depending on the phasing and summing action of the two signals being mixed, the actual level at the mix point may be higher and need less gain restoration.

[Cojo]

Originally Posted by Geoff_T

Hi

Mathematically that's true, but, depending on the phasing and summing action of the two signals being mixed, the actual level at the mix point may be higher and need less gain restoration.

Hi Geoff.

That's even better.

What I want to do is to connect some outboard after my mixers busses and then sum those signals to one stereo output. Since I do the summing after the outboard I beleive I don't need to worry about neither the crosstalk or the gain. I belive the outboard can provide 6dB gain and the crosstalk can't move trough the outboard back to my mixer... can it?

/Cojo

[Matt Syson]

Hi
Further to Geoff's comments, if the signal is predominantly 'coherent' (the same level and phase on left and right) then there will be practically no loss if the resistors are feeding a high impedance. You have to try it. If the equipment is quoted as being able to feed 600 ohm minimum load then I would suggest using 1K2 resistors minimum as an output feeding full positive on left while receiving full negative on right, (signal common but antiphase) may run into current limiting (distortion).
Matt S

[Cojo]

Originally Posted by Matt Syson

Hi
Further to Geoff's comments, if the signal is predominantly 'coherent' (the same level and phase on left and right) then there will be practically no loss if the resistors are feeding a high impedance. You have to try it. If the equipment is quoted as being able to feed 600 ohm minimum load then I would suggest using 1K2 resistors minimum as an output feeding full positive on left while receiving full negative on right, (signal common but antiphase) may run into current limiting (distortion).
Matt S

Hi Matt.

I understand what you mean but then again not...

You talk about left and right? I'm not gonna connect left and right together!? Do you men hot and cold?

/Cojo

[David Kulka]

Actually, I disagree with Matt and Geoff's comments, though part of it may just be semantics.

The idea that the loss in a passive summer depends on the phase relationship of the signals being summed is a little misleading.

Imagine two identical signals coming up on two faders of a console. If each signal is +4 and the faders are set to unity we'll see +4 on the active summing buss of the console when only one channel is on, but +10 if both channels are on.

With a properly wired passive resistive summer we'll have -2 when only 1 channel is on, and +4 when both channels are on.

So even though the input signals have coherant phase, there is still 6 db of loss due to the passive summing resistors, and 6 db that needs to be made up if you want to maintain unity gain.

Of course, the amount of "build up" when you add the second channel depends of the phase relationship of the two signals, but the passive 2-input summer will always have 6 db of loss compared to an active, unity system. (Again, this assumes that the pasive summer is properly wired. Fortunately, doing so is usually quite easy.)

In the end, the amount of make up gain needed will depend on the type of work you're doing. If for example you were using a passive summer to combine radio commercials for an on air feed, we know that the commercials should never overlap, and we would definitely want 6 db of make up gain. On the other hand if you were using the passive summer to combine two vocals for a duet, the additional voice will cause a ~ 2 or 3 db increase in total level, that you might want to compensate for by reducing make up gain. However, you might choose to compensate for that build up by reducing input levels -- after all, we generally maintain levels in a mix with by keeping the master near unity and backing down the individual input levels, rather than the other way around.

BTW, I'm concerned that the 600 ohm comments may introduce confusion. I agree with Matt that 1.2k summing resistors are probably a good choice. If one does that, there's at least a 99% chance that the everything will work fine, as nearly all pro audio gear has low impedance outputs. (Edited.)

[Geoff_T]

Originally Posted by David Kulka

Actually, I disagree with Matt and Geoff's comments, though part of it may just be semantics.

The idea that the loss in a passive summer depends on the phase relationship of the signals being summed is a little misleading.

Imagine two identical signals coming up on two faders of a console. If each signal is +4 and the faders are set to unity we'll see +4 on the active summing buss of the console when only 1 channel is on, but +10 if both channels are on.

With a properly wired passive resistive summer we'll have -2 when only 1 channel is on, and +4 when both channels are on.

So even though the input signals have coherent phase, there is still 6 db of loss due to the passive summing resistors, and 6 db that needs to be made up if you want to maintain unity gain.

Of course, the amount of "build up" when you add the second channel depends of the phase relationship of the two signals, but the total gain will always be 6 db less than it would be in an active, unity system.

(Again, this assumes that the passive summer is properly wired. Fortunately, it's usually quite easy to do that.)

I'm concerned that the 600 ohm input impedance comments may introduce confusion here. At Matt says, 1.2k summing resistors are probably a good choice. If one does that, there's at least a 99% chance that the input impedance of the "next" stage will be be a total non-issue, as nearly all pro audio gear these days is bridging (10k or more) in.

Hi

I think you threw a red-herring in there with the mention of a feed from one side only.... my original comment was that there must be a 6dB loss because of each resistor looking up the rear of t'other but, big but, two similar sized, in phase signals into the mix would sum together and produce the "pre-mix" level.

Caveats being that it had to be two similar signals, in phase and equal resistors, etc.

This situation only exists with a mix of two sources as the drop increases as more mix paths are added and the summing effect, even if they were all identical and in phase, would not be linear either.

[Matt Syson]

Hi
Re reading the original post, I get the impression that the desk has 2 (two) stereo outputs and the intention is to sum the two left and then the two right together. In this case you you would need about 6dB makeup gain assuming the 2 stereo busses are not being fed from the same source channel.
Solder some 1K2 resistors together and you will be sorted.
Matt S

[Cojo]

Originally Posted by Matt Syson

Hi
Re reading the original post, I get the impression that the desk has 2 (two) stereo outputs and the intention is to sum the two left and then the two right together. In this case you you would need about 6dB makeup gain assuming the 2 stereo busses are not being fed from the same source channel.
Solder some 1K2 resistors together and you will be sorted.
Matt S

Hi agian!

Yes, this is exactly what I want to do.

One thing that bugs me is when I'm looking at the example I got from ulysses (Rane Corp) is that I can't figure out why they have a 20k resistor across the output? What different would it make to put such a high resitor there?

Ohh... and just one last thing to make sure - I reread the technical specification of the mixer and it says that the outputs are terminated with 600 ohm. What does this mean and does it change anything?

Thanks for all help so far! I really appreciate it! thumbsup

/Cojo

[Matt Syson]

Hi Cojo

One thing that bugs me is when I'm looking at the example I got from ulysses (Rane Corp) is that I can't figure out why they have a 20k resistor across the output? What different would it make to put such a high resitor there?

I have not looked at the Rane site but it will be related to the resistors used for the mixing and gived a definite load. Just ignore it.

Ohh... and just one last thing to make sure - I reread the technical specification of the mixer and it says that the outputs are terminated with 600 ohm. What does this mean and does it change anything?
Without you saying which desk you have it is a bit dificult to judge what they are meaning. If it has transformer outputs and EXPECTS to see a 600 ohm load, then they may helpfully provide a 600 ohm resistor which can be switched in so that it is correctly loaded if you are using it into a high impedance input. Transformers like to have a defined load otherwise they can 'ring' or other frequency response errors. Normally other types of outputs are happy with a load from say 600 ohms to 100K or more so don't really need a 'termination' resistor. Actually, not wishing to confuse you more but if you have transformer balanced outputs, then it would be possible to wire the outputs in SERIES. This is a perfectly reasonable thing to do but may 'do your head in' if you change the system or it goes through a patchbay. In this situation the gain makeup required would be minimal if you are feeding a high impedance input. (if the input is 10K then you would need about 0.5dB gain).
Stick with 4 * 1K2 resistors and you will be OK.
Which desk is it?
Matt S

[Cojo]

Originally Posted by Matt Syson

Actually, not wishing to confuse you more but...

How did you know you were?

Originally Posted by Matt Syson

Which desk is it?

It's a Studer 962 and I believe it got transformers on both outputs and inputs.

/Cojo

[Matt Syson]

Hi
The concept of putting the two transformer outputs in series is only the same as putting two batteries in series (two AA cells and a bulb for example) but having the wiring in some connectors and remembering what is happening 'in the heat of the moment' it is probably better to go with a handful of resistors, which could be installed in a little box or something.
Matt S

[Cojo]

Originally Posted by Matt Syson

Hi
The concept of putting the two transformer outputs in series is only the same as putting two batteries in series (two AA cells and a bulb for example) but having the wiring in some connectors and remembering what is happening 'in the heat of the moment' it is probably better to go with a handful of resistors, which could be installed in a little box or something.
Matt S

Thnks Matt!

I have found a small aluminum box and I'm going to order the resistors today.

When I'm at it, I'll going to "through in" a third stereo input to the box. It will be used as a third buss. If I change the resistor values to 1.6k instead of 1.2k the load of the network would be the same I suppose - just some dB more gain to make up and I think I can live with that!

Thanks everybody for contibuting to all the usefull information.

/Cojo