abbey road d enfer said:
JohnRoberts said:
leigh said:
Would a load being driven to a virtual earth input count as being "driven to ground", or as "driven differentially"?
It would count as driven differentially since the AC input current is matched by current supplied by the feedback resistor, which comes from the op amp output and ultimately the PS rails, so no current flows into ground for that example.
JR
That's not my analysis. If the PS rails were infinitely stiff, the current would be flowing into the PS 0volt, but they are not, and the decoupling caps are there to provide a stiffer path to whatever "ground" they are connected to, thus diverting the current(s) to this local "ground".
Yes but..... You could make the same exact argument for differential loads, while in theory that PS ground current would net out to 0. .
I see a difference here between a grounded load and VE summer since the current is not flowing directly into an actual ground trace, but directly into the PS rails. Where it goes after that is a slightly different consideration.
As I posted earlier with my AM radio station anecdote signal ground and power supply ground can be too separate for system stability.
In fact, to my knowledge, just about every mixer has either protectyion resistors or local regulators that definitely prevent local currents to be reflected to the PS rails.
I have used per channel regulators and flameproof resistors to provide system robustness. A single channel failure should never take down the whole console.
I advocate always following the current. It is shame that many sensitive PCB layouts are made by mechanical PCB artists and not the circuit designers who have more appreciation for subtle current routing issues, but major layout problems generally reveal themselves during prototype development and testing. In critical circuitry it is not unusual to create multiple schematic grounds (or 0V nodes) that ultimately get connected together. This can annoy PCB artists... ;D
Driving faders. pan pots, and send pots, will generate signal related local 0V (or local ground) currents that need to be accounted for so they do not corrupt other channels causing crosstalk.
I even killed some brain cells back in the day trying to envision a topology where all local channel signal currents netted out to zero. I gave up because it seemed impossible (or at least very impractical), so a little bit of local brute force is indicated to maintain signal integrity. When sending signals across any distance they will often benefit from differential treatment (IMO).
I try to avoid sweeping pronouncement about grounding (even while making them) 8). I don't even like to think of ground as part of signal flow while is clearly is impossible to ignore.
JR