A question for the experts
Sep 2, 2011 at 12:29 PM Thread Starter Post #1 of 7

Avro_Arrow

MOT: Soundwerx Designs
Joined
Apr 8, 2010
Posts
2,211
Likes
56
I'm building a desktop amp.
Power will be a center tap transformer.
The power supply is a dual rail 317/337 in the standard arrangement
with filter caps that go from each rail to the center tap of the transformer.
The ground return from the headphone jack also goes directly to the
center tap of the transformer.
Now the question is, since the amp board has only a ground reference for the
input and feedback resistors, should the local filter caps go from rail to rail or
rail to ground? Which would be more effective?
 
Here is a picture of the power supply for clarity.
 

 
Thanks
 
 
Sep 2, 2011 at 12:56 PM Post #2 of 7
for TRS common gnd I think the "star" point should be the TRS gnd - signal gnd can be a gnd plane but I would put power, ps gnd return close to the output devices, return their local supply caps to the pwr gnd as near the TRS gnd as possible
 
the local HF bypass should be close to the chip's ps pins to the plane since it is low inductance
 
drawing shows amp output star - ignore the active splitter since you have a dual supply
 

 
 
 
 
another opion is Bridged output with no gnd return from the "balanced" wired headphone
 
Sep 2, 2011 at 1:02 PM Post #4 of 7
Thanks for the reply JCX, but that is the topology I am using and that wasn't really the question...
 
The "Star" point will be at the center tap of the transformer, where impedance is lowest.
 
Sep 2, 2011 at 1:17 PM Post #5 of 7
Here is the transformer.
 

 
The power supply ground, amplifier board ground and TRS jack ground
will all return to the center tap.
 
Sep 2, 2011 at 1:52 PM Post #7 of 7


 
Quote:
The "Star" point will be at the center tap of the transformer, where impedance is lowest.

 
 
the "lowest impedance" heuristic isn't helpfull in this instance - you want a "hierarchical" gnd - the run from the TRS gnd "Star" to the xfmr CT is a "dirty gnd" - the xmfr secondary happily floats the amount of the Vdrop in the wire from the CT to amp TRS gnd
 
with two separate amp boards and a panel mount TRS their is some unavoidable compromise with one good solution being differential amps in a "gnd breaker" configuration sensing/subtracing out the TRS gnd "gnd contamination"
 
of course at most headphone's current levels you can just use heavy wire and never noitice small differences from "best practice"
 

Users who are viewing this thread

Back
Top