resistor values for 16 to 2 passive summing box?

[jeffinaustin]

Can anyone provide a simple answer for what to put on the input and what to put on the output? 5k'ish seems to be what I've seen on most on the input, but the output seems to be a little less sure.

[JohnRoberts]

You want a simple answer, or a correct one?

There are probably a number of usable values but the the general criteria are;

1) Resistor values high enough that they don't load down the sources and increase distortion.
2) Resistor values low enough that they don't deteriorate S/N from their own internal Johnson noise, or crosstalk, or noise from interaction with the following active gain stage.

So this is pretty wide and fuzzy set of constraints. Since it is the nature of engineers to want to optimize something, lets look at what kind of source impedance the following gain stage wants to see. A 16x1 passive sum will cause some 24dB or so of insertion loss, so you will need roughly that much make up gain. 24 dB gain is little low for mic preamp, and little high for a line input, but most people use a mic preamp for this task.

Your 5K suggestion is not a severe load on most professional gear, while marginal low for some consumer gear, however the source impedance in the 300+ ohm range is a little high to mimic a microphone, but close enough for only 20 odd dB of gain.

For 16 inputs, I suspect you would could optimize the make up gain stage better than using a general purpose mic preamp running at only 24 dB, but that is pretty much what consoles do.

So long answer short.. 5k should work OK. It could be a little lower to interface with a mic preamp optimally, but I would be worried about loading consumer or semi pro gear down much below 5k.

JR

[oFarrellAudio]

I know I'm kicking up an old thread here, but I just finished a passive 16x2 box and I've run into a bit of a snag. I went with 5k resistors (two per channel) making the inputs 10k. My summer has 6 stereo inputs and 4 mono inputs. No switches. The mono inputs have two 2.5k resistors a side. The cold side going to both -L and -R and the hot side going to both +L and +R. All grounds are strapped together.

The problem is that I seem to be getting a lot of crosstalk. If I run a signal into a stereo pair on the summer and pan it hard to one side and get -10db on the panned side it will show up roughly -30 on the other side. I assume this can't be normal, but I can't figure out why I'm getting so much crosstalk. My first though is that it must have something to do with the mono inputs. Any ideas? Would it have anything to do with the grounds all being strapped together? That seems to be the way everyone does it. I appreciate any insight you all can give.

[johnnyc]

Did you put a resistor on the output?

[oFarrellAudio]

Yes. There are shunt resistors across -L +L and -R +R. The output impedance is roughly 300 ohms.

[johnnyc]

Ok just to clarify you are using a total of 4 2.5k resistors for the mono channels, separate resistors feeding L- R- L+ R+ correct? And not tying L- and R- together for example?

Assuming everything was wired correct, two ways to improve crosstalk would be:
1) Increase the size of the mono resistors. Why did you use 2.5k? They should be bigger than 5k, like 7.5k to 10k depending on what pan law you prefer.
2) Lower the output impedance to around 150-200 ohms.

[oFarrellAudio]

Yes, that's correct. Separate 2.5k resistors feeding each bus wire. So it adds up to 10k for the mono input which is the same as each L or R input on my stereo pairs. Each L or R input has a 5k on the cold and a 5k on the hot. I thought they should all be equal. I'm planning to bus out of Pro Tools 9 into this box. What would you suggest for resistor values on the mono inputs if I am trying to keep the pan close to, if not identical to, Pro Tools? Also, how does lowering the output impedance improve crosstalk? Thanks.

[johnnyc]

The mono inputs don't add, they are in parallel. You effectively have 5k on the L and 5k on the R in parallel so the net is actually 2.5k. You would need to use 10k on the mono inputs to get 10k input impedance. So you could use 10k but that would give a roughly 6db pan law, sorry not sure what pro tools is, is it 2.5 or 3db? Think it's adjustable too. 7k would give you around 3db, the exact number will also depend on what output resistor you use, however 7k resistors would be a safe choice. Note your mono channels will have a 7k input impedance while the stereo will have 10k.

Regarding crosstalk think about what happens when you put a resistor on the output, you are effectively attenuating the signal based on the ratio of the output resistor and input resistor. Now the output signal needs to travel back to the input and into the opposite side output for there to be crosstalk. So this attenuation is effectively magnified on the opposite side, so the more attenuation, the less crosstalk. So anything that increases attenuation reduces crosstalk, including lowering the output resistor and/or raising the input resistor. The whole thing is a balancing act between input impedance, crosstalk, and noise.

There was a good discussion here about all this
Optimal resistor values for passive summing

Also you might wanna google New York Dave summing mixer, seemed like everyone had good success with that, it was effectively like yours but with 10k resistors on the stereo and 14k on the mono, forget what was used on the output. Up to you how much re-wiring you want to do.

[oFarrellAudio]

I'm not really into the idea of rewiring the whole box. I guess I'll be going with 7k resistors for the mono inputs. Pro Tools 9 defaults to a 3db pan law. I'll have to do some math to figure out the right output resistor values. Thanks for your help and insight.

One other thing that just crossed my mind though... If Pro Tools is already taking care of the pan law attenuation why would I want the summing box to do it as well? Wouldn't I want all of the inputs on the summing box to have an equal amount of attenuation? Otherwise it seems like the pan law would be magnified by the summing box and throw the whole mix out of whack. My plan is to get my mix roughed in inside the computer and then bus out to the summing box when I'm doing my final mixes. Maybe I'm not properly understanding the concept. Can you clarify this? Thanks.

[johnnyc]

Originally Posted by oFarrellAudio

My plan is to get my mix roughed in inside the computer and then bus out to the summing box when I'm doing my final mixes.

This would be the very reason you would want the pan laws to match. Suppose you used 5k for both mono and stereo channels. Now when you go from ITB to OTB your mono signals will suddenly be too loud. By matching the pan laws the ITB and OTB will be a lot closer to start with.

So around 7.1k (choose the closet practical value) will get you really close to the -3dB of pro tools, thus less drastic tweaking will be required when you switch to OTB.

[Matt Syson]

Hi
The reason it crosstalks is that any MONO inputs MUST be terminated with either a short circuit or the output of an amplifier (typically around 50 to 100 Ohms).
All other 'messing' with resistor values is only going to get you a 'convenient' input to mix level and NOT solve any crosstalk issue.
In ye olden days feeding from say 22 Ohm sources through 15K resistors onto a passive bus set to around 600 Ohms could obtain decent crosstalk figures.
Matt S

[johnnyc]

Originally Posted by Matt Syson

Hi
The reason it crosstalks is that any MONO inputs MUST be terminated with either a short circuit or the output of an amplifier (typically around 50 to 100 Ohms).
All other 'messing' with resistor values is only going to get you a 'convenient' input to mix level and NOT solve any crosstalk issue.
In ye olden days feeding from say 22 Ohm sources through 15K resistors onto a passive bus set to around 600 Ohms could obtain decent crosstalk figures.
Matt S

I assumed the OP was already doing this? (feeding all 16 inputs from 16 d/a outs). Perhaps it was a wrong assumption. So yes I agree fix this first. But the other tweaks will certainly help improve crosstalk given non-zero source impedances.

[oFarrellAudio]

Actually now that you mention it, no. I have the 16 inputs of my summing box wired to my patchbay, so they are always plugged in. However, I hadn't patched all 16 outs of the D/A to the 16 ins of the summing box while I was testing it. I suppose that was my crosstalk problem then? There was a TRS in every input on the summing box but they were not actually all connected to the D/A, just the patchbay. The outputs of the D/A aren't normal to the summing box. I have to patch them manually. Would you say this was the problem with my crosstalk? I'm out of town for a couple of weeks so I can't test it right now. It does appear that I need some different resistor values on my mono channels as well anyway though.

[johnnyc]

Originally Posted by oFarrellAudio

Actually now that you mention it, no. I have the 16 inputs of my summing box wired to my patchbay, so they are always plugged in. However, I hadn't patched all 16 outs of the D/A to the 16 ins of the summing box while I was testing it. I suppose that was my crosstalk problem then? There was a TRS in every input on the summing box but they were not actually all connected to the D/A, just the patchbay. The outputs of the D/A aren't normal to the summing box. I have to patch them manually. Would you say this was the problem with my crosstalk? I'm out of town for a couple of weeks so I can't test it right now. It does appear that I need some different resistor values on my mono channels as well anyway though.

Most likely yes that was the main problem. Matt was right on that one, my bad for not mentioning it. You should still change the mono resistors though as that will further improve things.

[Matt Syson]

Hi
If you short out the switch contacts on your patchbay and remember to REMOVE any cords you are not connecting to gear that is SWITCHED ON the crosstalk will get much better.
Plugged into gear that is switched off is no good as the outputs tend to go 'high impedance'.
Matt S

[matskull]

I just learned about how cheap it would be to build a diy passive summing box.

I’d like to use one of these to free some inputs on my mixer.
My plan would be to plug all of my effect returns into the summing box so they would only use 2 channels on my mixer.

The only thing though is that I’d really like to plug the summing box’s outputs into the aux return channels of my console that are line level (10Kohm).

Would it be possible with some resistors combo to use the summing box into a line in?