One vs. Multiple Buffers

[Gobotak]

I'm wondering if there are instances where multiple buffers would be necessary over one.  I'm working on something now that splits the signal at the input to create parallel paths.  Would there be any advantages to using multiple buffers at the beginning of each parallel path?  Or would one buffer at the right at the beginning of the circuit serve the same purpose? 

[GGBB]

If that is the actual circuit - each split directly into a pot that's grounded - then if you don't have individual buffers any pot will adjust the level all of the paths. If one is "off" they will all be off, etc.

[PRR]

The backward pot connection is just wrong here. It shows a high impedance to a load and can show a dead-short to the amp if any pot must be turned to "zero". This also sickens the other outputs by straining the amplifier (most opamps will distort horribly before they get to zero output).

In many cases the pot should be on the destination. The mix-board outputs a +4dBu level on a dozen jacks, and the many monitors and recorders each have their own level control (wired normal, not reverse).

Using one amplifier is a well-proven technique. Each output has an isolation resistor so a shorted line won't put the station (or show) off the air. For a reasonable number of outlets (a musician's rig) each resistor can be like the minimum load the amplifier can drive cleanly (470r works for non-Pro levels).

[ElectricDruid]

As Paul and Gord have said, the screaming mistake in those schematics is the pots having the top and wiper connections the wrong way around.

Aside from that, they look ok, and there's really no reason to use multiple buffers over one. The "one buffer" approach has to drive the load of four pots. That's 4 x 100K in parallel, which is 25K, not a particularly heavy load for an op-amp, so there's really no problem.

On the other hand, the "four buffers" approach puts four 1M resistors in parallel as the input impedance, which then drops to 250K. That's is still not bad, but since there's no need for it, it could be better! The one buffer version wins on this score, as well as simplicity.

[Gobotak]

Whoops.  I thought I had the pots connected the other way but I clearly did not.  What if I replaced the pots with VCAs such as a SSI2164?  What would be the correct way to determine the load and what would be considered a heavy load for the op amp to drive?

[ElectricDruid]

The SSI2164 uses 15K input resistors to a virtual ground node, so has effectively a 15K input impedance. Four of those in parallel brings that down to 3.75K. That's certainly low, but again, I'd expect an op-amp to be able to drive it without too much effort.

"A heavy load" for me in this sense is probably <1K, but it depends on the op-amp. You sometimes see the 5532 used where the loads are very heavy (like headphones, for example) because it can cope better.