Newbie question about shunt(?) resistors

[Piedmont Audio]

Hello!

Please forgive the simplistic question as I'm completely self-taught- I've been studying schematics of lots of historical hardware as well as things that interest me for potential DIY projects. (I'm in the middle of a pair of Lola pre's, but I'd like to try something from scratch soonish.) I came across the article on passive summing from DIYRE and am curious about the resistors between pins 2 and 3 of the balanced outputs:

I believe Peterson is referring to these as "shunt" resistors (please correct me if I'm wrong!) and my best guess is that they are there to fine-tune the output impedance. My confusion stems from them connecting the two balanced signal carriers- why are they connected? Is this to ensure that the impedance is perfectly matched on both lines (as opposed to potential small differences between resistors on each line)? And is some portion of the signal on each cancelled out by this connection?

Thanks!

 

[arthurdb]

The shunt resistors are part of a voltage divider which will indeed attenuate (or cancel out if you will) the input signal.

If you plug a symmetrical input (with the exact same voltage on both tip and ring, just inverted), you could split the shunt resistor in half and connect the middle point to ground and you will have exactly the same result. The math does change a little if the balanced input does not have symmetrical voltage but in essence it is the same thing : using a single resistor between hot and cold, or two resistors shunting each side of the balanced output to ground is more or less equivalent. However you have one less part to your design, and don't have to worry about mismatch between resistors. So in other words, superior performance with less parts.

The shunt resistor is not a requirement for passive mixing. You really only need the series input resistors to safely mix signals together. However, the shunt resistor allows the behavior of the mixer to not change too much the more inputs you plug in.

Indeed, if you only plug one input, the series input resistors of the other inputs will be out of circuit. If you don't have a shunt resistor, the output signal will be equal to the input signal (if we ignore whatever load might be connected to the output) and the output impedance, equal to the input resistors (ignoring any source impedance).

But if you plug in another input, the series resistors of both inputs will create a voltage divider and signal will drop. Output impedance will drop too, as the output will now present the input resistors of both channels in parallel. The more inputs you plug in, the more attenuation you will have for each given input, and the lower the output impedance will be.

Now if you have a shunt resistor of much lower value than the input resistors, this will lessen the impact of plugging in additional inputs : indeed, for a given channel, the bottom leg of the resistor divider will be all of the other input resistors and the shunt resistor in parallel. You will always have strong attenuation even if you have only one or two input plugged in. The output impedance would be equal to the parallel value of all of the input resistors as well as the shunt resistor.

The shunt resistor also lessens the impact of the load, and, a single shunt resistor between hot and cold will also attenuate impedance mismatch of the input resistors.

Since a picture is worth a thousand words see the attached schematic : a two channel unbalanced passive mixer, with the schematic drawn to make the voltage divider obvious. Pick some arbitrary resistor values and try to calculate output level and impedance for each configuration, then find what the results would be if the shunt resistor wasn't there.

 

[merlin]

To add to Arthur's excellent description, I thought I would add that (where possible) it is desirable for all the inputs to be back-shorted (tip to ring) when nothing is plugged in. This is standard on a console mix bus for example, to ensure almost no change in mix level, no matter which inputs are being used. Back shorting is easy with relays or switched jack sockets, but not XLR connectors of course.

 

[ruffrecords]

Passive mixers are relatively easy to understand once you understand a couple of basics.

The first thing to note is that passive mixing works on the assumption that the inputs are driven from a source impedance that is low compared to the bus resistors in series with each hot and cold input. The bus resistors are typically around 5K each giving an input impedance of about 10K. The output impedance of the input source signal is usually less than 600 ohms (and typically 200 ohms or less these days).

A consequence of this is that each pair of bus resistors is effectively connected in series across the bus (via the source impedances). If you have ten inputs each with a pair of 5K reistors, you effectively have ten 10K resistors connected in parallel across the bus so the bus impedance will be 10K divided by ten which is 1K. In general, if you have N inputs, the bus impedance will be the two bus resistors in series divided by N. output impedance

Lastly, you want the output impedance to be 600 ohms or less, these days typically 200 ohms. If you have say 24 inputs with 5K bus resistors, you bus impedance is already 10K/24 which is 416 ohms. If you want to tweak this to 200 ohms you can add a shunt resistor directly across the bus. In this case a 390 ohm resistor will get you close to 200 ohms. If you have fewer inputs then you will need a lower shunt resistor value to achieve 200 ohms.

Be aware that the assumption that all inputs are fed from a low source impedance applies also to unconnected inputs. This means that your TRS connecttors whould be arranged to short hot and cold together when nothing is plugged in.

Cheers

Ian

 

[Piedmont Audio]

To add to Arthur's excellent description, I thought I would add that (where possible) it is desirable for all the inputs to be back-shorted (tip to ring) when nothing is plugged in. This is standard on a console mix bus for example, to ensure almost no change in mix level, no matter which inputs are being used. Back shorting is easy with relays or switched jack sockets, but not XLR connectors of course.

And definitely not DB25

Thank you Arthur, Merlin, and Ian! I've got a lot to go on now!