Impedance balanced outputs

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

gnd

Well-known member
Joined
Jan 24, 2006
Messages
285
Hi.

I'm trying to figure out impedance balancing in this Forsell summing bus.
http://84.255.203.119/SummingBuss2.pdf

I read about impedance balanced outputs, and it is said to be important that receiver stage (mix bus amp in this case) sees SAME impedances on hot and cold wires regarding to ground.

I cannot figure out why are, in Forsell-Sum-bus, resistors on cold lines 1k in size. Should't they be the same size as summing resistors? And if sum bus is connected directly to pan resistors (no post-pan opamp buffer), wouldn't that pan resistance count too, and even increase cold wire resistors? And how about pre-pan op-amp output impedance, would that count too?

How to determine proper size of these resistors? Are they usually calculated, or are they just set as 1k no matter imbalance?

thnx
gnd
 
I read about impedance balanced outputs, and it is said to be important that receiver stage (mix bus amp in this case) sees the same impedances on hot and cold wires regarding to ground.
It's the ratio that matters not the absolute value. For line driver applications it is convention to use the same impedance on both lines as this will guarantee compatibility with different systems and fully balanced outputs. For summing amplifiers you can do differently and reduce the ground leg impedance for decreased noise.

Samuel
 
[quote author="Samuel Groner"]
I read about impedance balanced outputs, and it is said to be important that receiver stage (mix bus amp in this case) sees the same impedances on hot and cold wires regarding to ground.
It's the ratio that matters not the absolute value. For line driver applications it is convention to use the same impedance on both lines as this will guarantee compatibility with different systems and fully balanced outputs. For summing amplifiers you can do differently and reduce the ground leg impedance for decreased noise.

Samuel[/quote]

Ratio... Let me check if I get it right. Say in example of Forsell mix amp, is ratio determined by 15k sum resistors and 15k feedback resistor on amp? That ratio is 1. So same ratio must be between cold wire termination resistors and resistor from pin1 to ground - both are 1k, so ratio is also 1. Is this the ratio you are mentioning?

And another thing, if mix amp needs to see certain coldwire termination, why is there terminator on each channel? Say 16 channels then reduces total termination to 625ohm, instead of 1k. Why not just put one resistor of required value on the end of cold bus wire? And if there would be just one resistor on the end of wire, should it be 1k, or 625 ohm, to follow ratio?

Is impedance of cold wire terminator dependant on number of channels? Like resistors of 1k in parallel from pin1 on opamp to same ground, are actually like single 625ohm resistor for 16 channels, and like 41ohm resistor for 24channels. ....

I think I may be missing something.... Can you maybe explain a bit more, please?

thnx
gnd
 
What Sam said...

This is not a "balanced" circuit but a "differential" circuit. The important distinction is that this circuit is primarily interested in forward referencing the signals from the local ground at the channel to the master bus ground, "not" to equally pick up and then cancel common mode noise in the course of transmitting that signal through the bus. If indeed the bus was a mile long and exposed to interference there might be merit to using balanced impedance, but that is not the case inside a console frame, and due to the high gain seen by the combined noise at both + and - inputs, it's worth the effort to keep impedances low.

Note: for noise contribution analysis the feedback R is considered in parallel with the - input Rs, but the + input Rs are added in series to the - input Rs.

JR
 
[quote author="gnd"]

Ratio... Let me check if I get it right. Say in example of Forsell mix amp, is ratio determined by 15k sum resistors and 15k feedback resistor on amp? That ratio is 1. So same ratio must be between cold wire termination resistors and resistor from pin1 to ground - both are 1k, so ratio is also 1. Is this the ratio you are mentioning?

And another thing, if mix amp needs to see certain coldwire termination, why is there terminator on each channel? Say 16 channels then reduces total termination to 625ohm, instead of 1k. Why not just put one resistor of required value on the end of cold bus wire? And if there would be just one resistor on the end of wire, should it be 1k, or 625 ohm, to follow ratio?

Is impedance of cold wire terminator dependant on number of channels? Like resistors of 1k in parallel from pin1 on opamp to same ground, are actually like single 625ohm resistor for 16 channels, and like 41ohm resistor for 24channels. ....

I think I may be missing something.... Can you maybe explain a bit more, please?

thnx
gnd[/quote]

You want the ratio for + and - input ports to be same.

From - bus 15k/15k = -1x
from + bus a little more complicated.

If I assume you are mixing 16 channels the signal at the +bus is initially divided by the other 16 1k resistors to 1/17th. This sees an effective gain of 1+(15k/938) or 1/17 x 17 or 1x.

To be picky the differential is imbalanced by the 50k coming from the servo. Technically there should be a similar value applied to - bus, but error will be small.

JR
 
Simplify:

Bal-Sum.png


This is a perfectly good differential amplifier.

It is conventional to make all four resistors the same value, so we can take advantage of R-packs or just the "peas in a pod" matching of resistors from the same batch. This seems to give similar impedance to the two inputs. (And there are reasons to offset the values, depending what kind of balance really matters.)

But the two inputs do NOT have to use the same, even similar, values, to get a differential action. If the source were truly balanced, it may be nice.

In this case, the source is highly UNbalanced. One input is the signal and must be lightly loaded, the other is "ground" and nominally a very low impedance, but a low impedance to left or right end of a 7 foot hunk of mix-desk; IAC not the same "ground" as the mix-amp is screwed to.

This simple 1-input "mixer(?)" handles ground error far better than trusting a long run of "ground" wire, chassis, bus-bar, etc.

It works the same for multiple inputs. The "-" side gets the sum of all signals and all ground-crap, the "+" side gets the sum of all ground-crap, subtract, get sum of all signals with no ground-crap.

It isn't perfect. There's two paths for ground-crap: ground mix resistors, and your physical grounding. Pan-pot impedance could be accounted. As John says, stray bits like the servo ought to be accounted for. But for medium systems, well laid out, in benign rooms, it won't matter. It is still a heap better than hard-grounding. Bigger systems where it matters, you may be "forced" to double-up all your signal paths and run true balanced mixing. Leave that to the guys with unlimited budgets.

The "1K" is pulled from thin air. Mackie uses a lot of 100 ohm, IIRC. It should be much higher than your ground impedance, so you don't get circulating currents, but that should be <1 ohm. It could be 15K, like the other side, like conventional diff-amps, but as John says that would add hiss.

The exact procedure for picking the ground-mix value is: spill all your X-X-Brown and X-X-Red resistors on the floor, close your eyes, pick one at random. (Or see what value you somehow have too many of. You needed one box of 10K, your idiot helper bought ten boxes of 1K.....)

Except: do NOT use, say, 10K and 1K. They look too much alike. In a monotonous build like this, you are bound to get one wrong. You won't know of the problem until days later when you do detailed tone-tests and one (or more) channels don't work right. By then you forgot the phone rang during Ch13's build. And you can stare at that 10K all day and not notice it isn't 1K like it should be. I don't even like the suggested Brn-Grn-Org against Brn-Blk-Red (I know my inner idiot). If I had a heap of 680 or 910, I'd use those: all three color-stripes different from the mix resistors.
 
Thnx for great posts, Samuel, John and PRR.
And PRR, thnx for great practical tips.

I'm not clear about one thing still. Why are ground-mix resistors there on every channel? I'm thinking, that since audio ground for all channels is the same, same ground bus, why use resistors on all channels? Why not use just one resistor, maybe somewhere in the middle of console? And if there would be just one resistor, should it be same size like at op-amp size (1k from pin1 on schematics), or should it be 16times smaller for 16 channel console?

thnx
gnd
 
Why are ground-mix resistors there on every channel? I'm thinking, that since audio ground for all channels is the same.
Audio ground is never the same, unfortunately. Every mm of PCB trace or wire has resistance and inductivity. As soon as there is some current flow through ground we have voltage drop as well.

Samuel
 
[quote author="Samuel Groner"]
Why are ground-mix resistors there on every channel? I'm thinking, that since audio ground for all channels is the same.
Audio ground is never the same, unfortunately. Every mm of PCB trace or wire has resistance and inductivity. As soon as there is some current flow through ground we have voltage drop as well.

Samuel[/quote]

Isn't it, if I'm summing grounds from all channels, I get some average ground, which is different from ground of single channels anyway. So, why not just take one channel, and take that one as refference for ground resistors? Or maybe put two resistors, to have some averaging, on first and last channel?

The point is, that I'm redoing mix buses, and if it is not necessary to put resistors on every channel, I would not put them (lazy...). It is clear that in mixer production resistor is added on each channel for practical reasons, so that all channel boards are the same. I'm just wondering if there is any other reason, technical one, to have ground-mix resistors on all channels.

thnx
gnd
 
This becomes readily apparent when you start working with larger such beasts. "Ground" is a concept, not a voltage.

In the case of a simple 16 channel mixer, the signals being sent to the master bus from those 16 channels is riding on top of 16 slightly different local ground voltages. By also sending those 16 ground voltages to be combined and applied at the opposite input from the signal, mathematically subtracts those individual ground errors.

Perhaps my earlier response was too brief to easily grok, but each channel signal is combined with a gain of +1, each channel ground is combined with a gain of -1 (actually since it's an inverting stage the polarities are reversed but the mathematical effect is the same). So you get signal + ground noise, - ground noise= pure signal.

For this to be accurate and effective the representative ground error signal must be pulled from the same point the signal is referenced to "in each channel".

Note: in smaller, less expensive mixers, this can be brute forced by sending a low impedance bus ground reference back to the individual channels, that each channel can locally reference to, but for this to work the ground reference can't be corrupted by signal currents and needs to be compact with good shielding from magnetic fields.

JR
 
> since audio ground for all channels is the same

Never true.

> lazy...

In simple systems, it is often true enough. Lazy often works OK.

But I've gotten into trouble on a simple passive switch-box. "Ground" here was not the same as "ground" on the other side of a 17" box. I re-thought the grounding ("ground folows signal") and got buzz down below -80dB, "good enough" for the tape I used at the time. Maybe lower; I have not seen buzz in CD masters made since then. But when you get into more complicated systems, or have multiple sources rattling around inside (especially if they don't all go to the same mix, as in a network master control console), good layout is not enough to keep the disco from leaking into the garden channel show.

Back in the 1980s, we did many 6-in and 8-in one-FX-bus mixers just running a #18 ground wire to every channel board and the master board. It works, Crosstalk never seemed to be an issue in our simple mixes.

That's why I was mildly surprised to find ground-difference mixing in a modest Mackie. But it does have a lot more sub-routes than we ever needed, a rather less elegant layout (only way to afford all those frills), and maybe our standards have improved since the old days. (Wasn't so very long ago I did live shows on a passive mike mixer: six faders, one mike-amp.)
 
Back
Top