NYD passive mixer questions

[dmp]

I built a passive mixer this weekend from NYDs design and I want to modify it slightly, but I'm having trouble figuring out the important design principles.

For the input level control I tried using a single 10k pot between the + and - inputs, with the wiper controlling the + output to the mix network (I'm not using the 'assign' feature). Obviously this isn't symmetric on the balanced input, but it seems to work fine in a simple test. I tried parallel compression mixing my sound card output with 1176.  I built the mix network without an input level control for comparison and could not tell a difference in the quality of the sound with and without the input pot. 
- What are the drawbacks from doing it this way? Why the 5ka dual pot balanced design?
- The mix network is stated to have in input impedance of 20k. Adding a 10k fader to the input changes this to 10k || 20k = 6k, yes?  Why not use a higher resistance level pot to keep from lowering the input impedance?

And finally, where does the Rx=((20k/N)*200) / ((20k/N) - 200) equation come from? Doesn't Rx alone set the output impedance? Is it correct to think that a low output impedance is better for bridging with the make up gain amp that follows this, but the lower Rx, the more dB the mixer loses? Is it only a volume level / snr issue or can it degrade the signal in a frequency filtering aspect?

Thanks for the help.
Dan

 

[PRR]

> Doesn't Rx alone set the output impedance?

No, Rx |in parallel with| all the mix network legs.

For a 2-input mixer, "no" difference. 200||20K||20K is == to 200 for practical purpose.

For a 100-input mixer, Rx would be infinity (omitted). One hundred 20K in parallel is 200 ohms already.

Typical mixers will be >2 but <100 inputs, so you ought to do math.

> a low output impedance is better for bridging with the make up

The design is intended for mike preamps as make-up gain.

Transformer-input mike/makeup amps have optimum source impedance. They may not be awful for sources 50 ohms to 600 ohms; but may be more-optimum with 100-300 ohms.

Mike preamps often have minimum gain near 100 (40db). Since mixer sourcs are line-level, and we want line-level at makeup output, we need 100:1 loss in the mix network.

> Why not use a higher resistance level pot to keep from lowering the input impedance?

Do you have sources which won't drive 6K?

 

[dmp]

No, Rx |in parallel with| all the mix network legs.

Thanks, the equation NYD had makes sense now - I worked through the equation for resistances in parallel.

Do you have sources which won't drive 6K?

No, both a 1176 (with 600 ohm output) and my audio interface worked fine. But, I'm wondering what the design criteria that led to a dual 5k input pot was. It seems a higher value pot will not lower the input impedance to the mixer as much (the input impedance is (5k+5k) || (20k + Rx), I believe) and will reduce the effect of the mix network legs on the output impedance. A dual 10k audio pot is easier to get than a dual 5k audio pot. And finally, a regular 10k audio pot is what I tried and seemed to work fine, so I'm not sure why the input level pot needs to be a dual pot.