earthsled said:
That may be your first problem. I guess you found out that your first attenuation step is something like 0.3-0.4dB. If you want to use 1k resistors, the load impedance must be higher than 50kohms.
Using an 11-step switch with a -3dB sinusoidal pan law, I'm getting a range of about 0.1 dB to 16 dB of attenuation at the extremes.
haven't done the calc, but don't forget you want one extreme at minus infinity.
I'm glad you pointed out the 1k value as an issue - I really need to understand this better.
The big dilemma here is that in order to produce fractional dB's of attenuation, the ratio of the series resistance (2x 1k in your example) to the mixing resistors (bus injection resistors) must be quite small. If you really want 0.1dB attenuation the mixing resistors must be at least 85k. Then the bus load would be 7k, which is too high for obtaining a good noise factor from a mic amp. But would be good for a dedicated balanced virtual-earth mixer based on 5534's. Anyway, I think the 0.1dB position should be discarded, being totally inaudible. Then the next position is 0.4dB; this would favor the choice of 20k mixing resistors. The bus load would be 1.6k, which many mic preamp will be content with.
So, excuse me for asking a fundamental question, but how should I calculate the min source and max load impedance of these pan-attenuators?
The input impedance must be at least equal to the min. recommended load impedance of the preceding equipment. 50 ohms is the actual source impedance, but min. rec. Z is probably 600-2000R. The min impedance presented by the attenuator is when the shunt resistor is zero; then the impedance is 1200R, which should be adequate for most equipment.
Also, how do I calculate for 8-channels vs. 24-channels?
No difference. But you'll have to crank up the gain of the make-up amp.
I would like this pan-bus unit to interface well with modern DAW equipment. A Digi 002 has an output impedance of 50 ohms and an input impedance of 10M (on the line inputs).
Wouldn't it be 10
k ? Anyway, the line input impedance is irrelevant, since mixing involves attenuating all signals
a lot , then you neeed a mic input or a dedicated virtual-earth mixing node.
I noticed that this passive summing schematic is using 5k resistors on the inputs:
Would 5k be a better choice for series resistors?
With 5k, the bus load is down to 200R, which will give marginally better noise figure for the make-up amp, but then, the series resistor would be 150R and the load to the sources would be 300R, which most equipment will choke on.