TPA3113D2 Crosstalk issue

Matthew MacDonald

Other Parts Discussed in Thread: TPA3113D2

Good Day,

I've developed an audio amplifier using the TPA3113D2 and am experiencing some issues with crosstalk between the left and right channels. The amplifier is being fed from two single-ended sources (one at each IN+), with each IN- impedance matched, as suggested in the application note. The outputs are composed of 33uF ferrite-core inductors and 1uF capacitors for the LPF, feeding 8-Ohm loads.

My layout appears to be rather sound, but I am experiencing higher-than-expected crosstalk. The crosstalk presents itself as a 2nd-order harmonic at -40db on the opposite channel to which I am feeding the audio. So for example when I feed 1kHz into the Left side, I see distortion with a main component of 2kHz on the right side, attenutated by about 40dB, and vice vera if I feed the audio into the Right channel.

I have poked around the board with a scope, and have verified that the following:

a) No crosstalk/output frequencies are present in the supply rails or at the supply pins for the TPA3113D2

b) No crosstalk frequencies are present at the input pins for the adjacent channel (feeding audio into left IN+, no crosstalk at right IN+)

c) Crosstalk frequencies are present at the amplifier outputs.

From this, it appears as if the crosstalk is localized to within the TPA3113D2, but that seems incorrect, given the info on the datasheet. Nothing is being over-driven, and I believe I have followed all of the recommendations in the design note, so I am at a loss to understand why the crosstalk performance is so low.

Any help will be greatly appreciated.

Regards,

Matthew

over 13 years ago

0 Don Dapkus over 13 years ago

TI__Guru**** 189401 points

Hi, Matthew,

One place where cross-talk can occur is via the output filters. If they are unshielded and placed close together, you can get cross-talk. Can you look into that and see if that's the cause of your cross-talk?

I also highlighted your post to my colleague working on the TPA3113D2 (also named Matthew!). Let's see if he has any further ideas.

-d2

0 Matthew MacDonald over 13 years ago in reply to Don Dapkus

Prodigy 30 points

The output filtering does not appear to be causing a problem. I am using Bourns SRR6038-330Y shielded inductors, and there is about a 6mm spacing between the Left and Right side inductors. Just to rule out mutual indictance between PCB traces, I even tried cutting the traces going to the output filter on the right side of the amp, but the crosstalk still remains.

I am wondering if additional filtering is needed for AVCC, GVDD or PLIMIT. Each are bypassed with 1uF ceramic caps presently.

0 Matthew MacDonald over 13 years ago in reply to Matthew MacDonald

Prodigy 30 points

Any more thoughts on this?

0 Matt Lauer over 13 years ago in reply to Matthew MacDonald

TI__Expert 7925 points

Hi, Matthew. (I'm Matt(hew).) Thank you for testing what you've already tested. You mentioned there was no detectable crosstalk at the input of the TPA3113D2. May I ask what signal feeds into your TPA3113D2? Does it come from anything multiplexed or any digital to analog converter? I'm trying to cross that off as a source of crosstalk that could be amplified in the TPA3113D2 and only detectable at the output.

You haven't tried to make a ground loop break on your board, have you? (See this TI Application Note, section 7, page 5 on the subject.) There is a potential crosstalk issue when there is resistance along any part of the ground plane introduced to reduce ground loops.

Another thing to check related to grounds is that your PGND on pin 24 is not connected to your PGND on pin 19 as much as possible. For instance, the right side switching output never touches traces from or around pin 24, and the left side switching output never goes near pin 19. Shared grounds are a potential source of crosstalk. In our evaluation module, we did our best to separate what we call PGND1 (left) and PGND 2 (right). You may try to look at the two PGND pins and see what signals are picked up on them, or how they differ from each other.

You may also want to observe the signals picked up on the AGND pin, to see if we need more filtering between AGND and AVCC, GVDD, or PLIMIT. We have never needed AVCC, GVDD, or PLIMIT to have additional filtering beyond what you've done. See the TPA3113D2 EVM schematic after this sentence. It uses the same filtering you've suggested. 3225.TPA3113D2 EVM Schematic.pdf

Touching on the topic of mutual inductance, our inductors for right side and left side of our evaluation module are very far apart. (See the layout files below.)

You can review your layout against what we have done with the TPA3113D2 evaluation module using the files after this sentence.
4201.TPA3113D2 EVM Top Composite.pdf
4118.TPA3113D2 EVM Top Xray.pdf
3632.TPA3113D2 EVM Bottom Composite.pdf

I hope this is helpful!

Best Regards,
Matt

0 Matthew MacDonald over 13 years ago in reply to Matt Lauer

Prodigy 30 points

Thanks for the tips, Matt. Poking around on the board further I can see that the noise is coming from the 3113, going into the ground plane and affecting the unamplified audio that way. I am going to make some adjustments to my layout, and add a ground pour on the top layer (presently only had a ground pour on the bottom layer). I will report back with the results!

Cheers,

Matthew

0 Matt Lauer over 13 years ago in reply to Matthew MacDonald

TI__Expert 7925 points

Hi, Matthew. I wanted to follow up and see if my recommendations were useful to you in solving this issue.

Best,
Matt

0 Matt Lauer over 12 years ago in reply to Matthew MacDonald

TI__Expert 7925 points

Hi, Matthew. I'm writing again to see if your adjustments to layout have helped at all in your design. How are things coming along? Any better?

Best Regards,
Matt

Audio

Audio forum

TPA3113D2 Crosstalk issue