[jcx]

I see some more problems; I hadn’t realized –in was the lower pin, in your layout its trace runs under your +in resistor body – a real no-no, you’ve increased capacitance and loop area of the most critical connection in the circuit; check out the evm layout, the feedback resistor should be right up against the –in pin to minimize cap at the –in node

Likewise the output trace picks up a few(?) pF unwanted parasitic C by running under the bypass cap body before the 10 Ohm isolating resistor that lets the amp handle Cload



You like symmetry, try applying it to pwr! – V+ and V- trace to the tpa have the same “average position” adding a extra 1/r to the mag field attn for balanced current flow

Putting the bypass caps under the tpa on the backside (preferably with the gnd ends near the midline of the tpa) get them out of the way and reduce inductance – 0805 really, really is best here without going to totally unsolderable parts

the 0.1 uF value is really too large for a 100 MHz bandwidth circuit –inductance renders bigger C less useful above the self resonance; 0.1 uF probably have <30 MHz srf, I wouldn’t use anything bigger than 10nF at the op amp pwr pins, maybe 0.1uf near the pwr entry could discourage some AM/CB radio input to the board

my idea:

Small green sqs are 0805 caps on backside – big via in pad is bad for automated/reflow soldering but fine for hand soldering – grey is superimposed 0805 outline for reference

I also show 2 pwr input gnds, for any pwr wire length you should consider twisting ea supply w/gnd and twisting the pairs together – or braid

[edit: should short R.L V+ at tpa too, also need to flip/rotate, I started upside down but this shows the ideas]

 

[tyre]

Quote:

I see some more problems; I hadn’t realized -in was the lower pin, in your layout its trace runs under your +in resistor body - a real no-no, you’ve increased capacitance and loop area of the most critical connection in the circuit; check out the evm layout, the feedback resistor should be right up against the in pin to minimize cap at the -in node

I imitated peranders' QRV07 layout which has the + power bypass cap against the side and the feedback loop going in between the caps pads. I didn't think it would be a problem though.

qrv07r0_TPA6120.jpg

Quote:

You like symmetry, try applying it to pwr! - V+ and V- trace to the tpa have the same "average position" adding a extra 1/r to the mag field attn for balanced current flow Putting the bypass caps under the tpa on the backside (preferably with the gnd ends near the midline of the tpa) get them out of the way and reduce inductance - 0805 really, really is best here without going to totally unsolderable parts

Okay, here's another layout.

pic1, pic2

It's a little awkward having the outputs over to the side though. Does it really provide that much better isolation from the tpa6120's power traces over than something like this (pic3) with the output directly above?

 

[jcx]

I could go with the pic1/2 layout now

I'm assuming you're directly soldering wire leads to the output pads and they could end up anywhere in relation to the pwr traces, esp. after stuffing in a box, with the output pads together on the edge its much easier to keep them away from pwr (at least until they've been twisted with the gnd return to greatly reduce external magnetic field sensitivity) - with this layout I know where the output pcb trace crosses pwr and can control the intersection angle to the optimum 90 degrees for min coupling

my remaining quibbles would be that your 1206 pads are too crowded side-to-side, even an extra 20 mils would make soldering without bridging easier and the op amp pwr trace lower horz path is rignt next to an input R, the pwr can move up board ~100 mils and be under gnded pins that aren't even internally connected in the tpa

 

[tyre]

.