Interesting Grounding on TI EVAL Board

[ssashton]

I got hold of the project files for the TPA3244 TI EVAL board and I thought this was interesting enough to be worth a thread.

On this board there are 4 audio line level inputs which are buffered by a couple of NE5532 op-amps run single-rail with a V-MID supply. They feed a class-D amp IC found in the middle of the PCB.

What I find really interesting is they have chosen to route a separate trace for the audio input ground and connect the V-MID supply filter caps directly to it. This is then joined to the main ground planes next to the big amp IC.

I wanted to know what you guys think of this?

Is it some legacy thinking that audio should be 'star grounded' or does the designer know their stuff? (They work at TI they must be good, right?)

It interrupts the main ground plane, but there are also ground fills on the other layers that tie across the routed trace. Kind of wondering if I should do similar on my own board.

[dobsonr741]

It’s an input ground lift. Allows the other end of the input cable to define your GND and not to force ground current to go through your cable’s shield feeding the amp. Somewhere in between the differential and single ended input. Far worse than differential, might hear a bit better noise/hum performance.

[ssashton]

I'm not sure I follow, can you explain in more detail please? The ground trace here is tied to the ground plane, so it's not lifted in terms of being floating or transformer isolated.

[dobsonr741]

You are right, it's not ground lift. Looks like star ground, if you found the two are directly connected on the gerber. I assumed there is a resistor between the input ground and the system ground. As in TI SLOA143, 5ohm Rgbk on page 2.

[ssashton]

Oh that's a nice document!

Yes I think it is a star ground, but recently I had been reading a lot that star ground is an old design method and single ground plane should be used. So I thought it was unusual to see both here.