• State Resolved
  • Locked Locked
  • Replies 1 reply
  • Subscribers 33 subscribers
  • Views 109 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TLIN2029A-Q1: TLIN2029A-Q1: Excessive Conducted Emissions When LIN Bus is Active

Karl Mulnik
Prodigy 10 points
Part Number: TLIN2029A-Q1

Tool/software:

Problem:
I am using 5 µH LISNs to test a PCB populated with a TLIN2029A-Q1 LIN Transceiver to CISPR 25 for class 5. However, the PCB fails in the frequency range LW (0.15 MHz to 0.30 MHz).
Request:
Please send any thoughts and suggestions that might reduce the conducted emissions. I have tried countless other filtering components and configurations to eliminate noise, as well as making the LIN Trace as short as possible on the board. The LIN trace is now ~400 mils from Pin 6 of the TXCVR to the output connector.
Setup:
The TLIN2029A-Q1 is configured per Figure 8-4 in the datasheet with the exception of Pin2 and Pin4. Instead of additional filtering or pull-up components, Pin3 and Pin4 of the TI LIN TXCVR go straight to a micro controller pin.
https://www.ti.com/lit/ds/symlink/tlin2029a-q1.pdf?ts=1749413550811
The TLIN2029A-Q1 is setup as the Responder (slave) ONLY. So it cannot toggle RX and TX to see if they are the issue. Is there another way to isolate these pins from LIN to know if they are the root cause?
I have a bench supply providing 12 V or 24 V (Vsup Pin7). This power line is where the conducted emissions measurements are made. There is also an on-board switcher providing 5 V (RXD Pin1 with pullup).
I also have a separate isolated 12V bench supply powering a LIN to USB adaptor that is acting as the Master. The grounds of the two isolated supplies are tied together for a common reference point.
Debugging:
Assuming resonance equation for Series RL, Parallel C: need >10 mH in series to filter LIN @ 150 kHz

This is because LIN limited is limited to 220 pF or less (not incl. tol.)

So not realistic to filter LIN noise with passive filtering as only small surface mount components can fit. Maybe the part by itself fails CISPR 25? TI could you provide EMC data for conducted emissions testing?

So far, I have only looked at the LIN trace since that is the higher voltage and thus most likely since I suspect capacitive coupling.

I know it is related to the TLIN2029A-Q1 because the noise goes away as soon as the LIN bus is inactive.
I know it is NOT the on board switcher because the noise is gone, even if the on board switcher has a load mimicking the LIN TXCR while operating.

Coupling Paths…….

1) Conductive: Unlikely since I attempted powering TLIN2029A-Q1 separately from the power lines measured for conducted emissions.

2) Electromagnetic: Unlikely since the ~400 mil LIN Trace << wavelength of largest noise peak @ 150 kHz as well as << 20 kbs(max data transmission rate for LIN) wavelength

Maybe wavelength assumptions are wildly off for a PCB?

3) Magnetic: Unlikely since lifted Pin6 on the transceiver (breaking loop) and noise remained.

4) Electric: Possible but can’t find significant capacitance between LIN trace and Vsup trace. Noticed that noise rises and falls when Vsup voltage changes from 12 V to 24 V and from 12 V to 9 V respectively. This further hints at electric field coupling from the LIN bus to the power lines.

  • 0
    TI__Guru 51616 points

    Hi Karl,

    I suspect the issue is most likely related to coupling from the LIN trace to the Vsup trace or plane, potentially worsened by sharp edges on the LIN pin and insufficient return path optimization. Please see below for recommendations and reach out to your TI contact in the region to access all available EMC reports utilizing EMC and other compliance reports.

    • Radiated noise can couple into the power line and get measured as conducted and would suggest to wrap the LIN wire (from PCB to connector) in copper tape or foil and ground it to chassis/GND near the connector. Then rerun the test. If emissions drop, it confirms coupling from the LIN to Vsup line. Can also try twisting Vsup/GND or shielding the Vsup line to suppress possible common-mode pickup.
    • Although you’re using the device as a responder, noise on the LIN line can still reflect and bounce due to lack of bus termination and would recommend temporarily adding a 1 kΩ pull-up resistor to LIN, near the connector (at the board edge), even if you don’t use the transceiver in commander mode. This will verify if reflections are dampened and the radiated noise from the LIN pin suppressed.
    • Emissions may generate from sharp transitions on the LIN line. Even at LIN speeds (20 kbps), edge rates of <100 ns can cause harmonics into the 150–300 kHz range and would recommend verifying 10 Ω to 22 Ω in series with the LIN trace further dampens reflections, if not already there. This limits edge rate and dampens high-frequency energy.
    • While you mentioned the LIN trace is ~400 mils, its routing relative to Vsup and return path quality typically matters more for emissions and would recommend ensuring the GND pour under LIN trace is solid and connects to the return path of the Vsup line (EMC best practice).
    • If LIN runs near or parallel to the Vsup trace or in a “gap” in the GND plane, emissions can be impacted and would recommend ensuring adding a grounded shield trace alongside LIN (if re-routing is possible) or even an adjacent grounded via fence.
    • Confirm GND impedance path between TXCVR and Vsup return. Conducted emissions depend not just on differential-mode current into Vsup, but also how return currents flow through GND. Probe the voltage between local GND at the LIN TXCVR and the GND at the LISN, during LIN activity. Add ceramic capacitors from LIN IC GND to connector shield or power GND close to the Vsup input to suppress high-frequency noise currents flowing in GND loops.

    If still stuck, test using a minimal evaluation board (with just the IC, short LIN trace, and power lines well-separated) with TLIN2029EVM for example and repeat the CISPR25 testing to confirm if IC alone is noisy or if your PCB layout is the main contributor, thanks.

    Best Regards,

    Michael.