EMI issues with the TDA7492E

[jasong]

Hello -

We are having issues with EMC compliance with the TDA7492E. We have broadband radiated EMI between 100MHz and 200MHz that is above the limits. I've included a screenshot of the testing lab's report with the radiation. I've attached our schematic for the amplifier circuit. It was derived using the example datasheet for the TDA7492E. I've spent a fair amount of time having a look at the circuit in operation - I do not see any large ringing on the output square wave, so I believe the filter layout and tuning is good, the snubbers appear to be doing their job. I am currently testing with a 1KHz sine input at 1/8 power into an 8ohm load resistor. I did notice the input signal has tons of HF noise on it while the TDA7492E is on (screenshot attached). The HF noise is in the 100MHz - 200MHz range.

My current working theory is the HF noise is coupling into the amp input signals due to layout issues. We have one PCB with 6 of these amp chips on them, and consequently the input line level signals have to travel a long ways, they are currently running over both the power and ground planes. I had a few questions: for the input to the amp, I'd like to low pass filter as well as AC couple. Will this present a problem to the input stage of the TDA7492E? I'm a bit worried about the interaction with the high impedance input (60k).

I'm also having a look at putting CLC (pi filter) with an Fc ~ 20MHz after the LPF of the output as a last ditch effort, but I'd rather understand how the noise got there in the first place...

Thanks for any and all advice / experience...

[PZHAN.1]

Hello ,

the EMI is quite related to PCB layout even the schematic is same but the traces and components placement different that would cause radiation indeed. before review layout, is it possible to try below suggestions:

• Please try to put a RC(100nF||39Kohm) network as close as possible to the pin 24 of TDA7492E, refer to below a graph.
• Please also disconnecting the pin 25(SYNCLK). When placing the RC network.
• Because the EMI problem is very related to the real PCBA, some comments:
  • Put the VCC decoupling capacitors such as C10, C21, C32 and C33 close to the IC pins.
  • The output PWM route need to be as short as possible, due to high frequency radiation.
  • The EMI need to consider system level, for example the radiation source points, is it from Output Traces or VCC traces or even speaker wires?, once it's been confirmed the workaround countermeasure would be determined accordingly.

[jasong]

Hi PZHAN,

Thank you very much for the detailed reply. Can we send you the layout files for review?

Jason

[PZHAN.1]

Hi Jason,

Yes, you can send me the layout for review.

By the way, have you tried the suggestions I provided you? if yes, please let me know the difference between changes, I can then make more reasonable suggestions based on the clue you shared.

-. as schematic of TDA7492E in your system, you have connected the ROSC&SYNCLK pin to shared with other section, can you let me check whole schematic, means ROSC&SYNCLK related. did you try to use an independent ROSC to check the EMI again? it's very important message from your clew after my suggestions implemented, since the determination mostly based on that. 

thanks!

[LCE]

6 type D amps with that power - that is really hard for the PCB layout.

Alone the distributed power and clocks - the analog signal should be near that anywhere.

[jasong]

Hi PZHAN.1,

Yes, we have placed all bypassing caps as close to the IC pins as possible. The first amp currently has the ROSC connected, it then sends the SYNCCLK out to the remaining 5 amps. We can modify the 6 amps to all use independent ROSCs, but this would probably require a PCB respin.

Thanks again for the reply. Let me get you the complete sch and layout. I agree this is a layout issue, I have some ideas on modifying the layout that I would like feedback on. Please let me know how I can reach you outside of this form to continue troubleshooting as this is a commercial design.

Jason

[PZHAN.1]

Hi Jason,

Ok, you can send me email via peter.zhang@st.com

For the independent ROSCs which means to adjudge and test where the EMI comes from, does not mean let all amplifiers independently. you can verify the radiation one by one the device turn-off or on, then it's possible to find the source of radiation thus to focus on the improper radiation of PCB layout.

Regards,

Peter