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.

BQ25886: FCC certification issues

Part Number: BQ25886

Hello:

We have a product using the BQ25886  that underwent FCC pre-cert and failed on a frequency band near 120 MHz. The noise profile is similar to noise that a boost generator generates and we have isolated the problem to our on-board BQ25886 in the testlab as well as in our lab. The noise completely disappears when the USB charging source is removed. We have used the USB charging source in other devices that have passed FCC, so we eliminated the charger as the issue. I suspect the noise is originating from the SW node - we have nothing else on the board that switches quickly.

I would like to ask TI to review our waveforms and schematics with us, but I cannot post these on a public forum. Is it possible to get a private channel to talk through for this?

Also, I checked the SW pin for ringing, but I really cannot see any significant ringing on the pin. However, every time I touch the pin it seems to make the problem worse. Here is a picture of the scope of pin SW (sorry, USB port is acting up!). I have tried to calculate the RC for a snubber following SLVA790a, but it does not seem to affect the overshoot (tried from 1pF to 47pF and no noticeable difference on frequency, which is the first step of that document).

Pictures of SW pin in mode that fails FCC.

Thank you!

  • Hi Eduardo,

    First step would be to add a small 0.01uF - 0.1uF to the SYS and BAT outputs, as close between the IC SYS and GND pins as possible to damp the output ripple.  If that doesn't help, then you can try a small 1-5 ohm resistor in series with the BTST capacitor.  This should reduce SW node over/undershoot and slow the SW rise time down slightly without too much reduction in efficiency.  Then, if that doesn't work, I would try the snubber.  The snubber capacitor that you used is smaller than any I have tried on other devices. I am not familiar with that appnote, having written my own years ago, but you don't have traditional ringing to help size the components (or more likely, your 100MHz scope is too low of frequency and is filtering out the ringing).   So, you may have to use trial and error. Your goal with the snubber would be to minimize the rising overshoot/ringing and falling undershoot/ringing to get closer to a square wave without too much increase in rise time (<<2x reduction). Maybe start with 100pF and 5ohm. Too much reduction in rise time reduces efficiency and in extreme cases can cause shoot through current in the FETs.    Last option, as the app note says is a ferrite bead tuned as explained in the paper.  I would expect it to work best on the SYS output and BAT output.

    Regards,

    Jeff

  • Hi Jeff:

    Thank you very much for your detailed response! I will try your suggestions and come back with pictures - I will also use our nicer scope to capture the SW pin.

    I was wondering, is it possible for a layout review? I tried to emulate as much as possible from the suggestions of the datasheet and the dev board, but I am still unsure if something was missed.

    I will report back later. Thank you!

    -Eduardo Garcia.

  • Eduardo,

    Yes, I can do a layout review.  Please provide your layout files.  PDF printout is okay.

    Regards,

    Jeff

  • Hi Jeff:

    I found the following ring with the nicer scope. Would you still recommend to do the caps on VSYS, VBAT, and resistor on the BST pin first? Or would you recommend to do the snubber first?

    I am currently trying out the capacitors.

    Thanks!

    -Eduardo

  • Eduardo,

    Other components first.  Snubber will definitely reduce efficiency.

    Regards,

    Jeff

  • Hi Jeff:

    I have made the changes you recommended and have taken RF measurements, and they have helped some in the frequencies of concern (76 MHz and 120 MHz). They have reduced the frequency specially at 300 MHZ, which was the frequency of the ringing on the SW line.

    Here are graphs of the RF signature at CF: 122 MHZ, 30MHz bandwidth, RBW: 3KHz. (Vertical units: dB). Blue lines are the RF level with the USB on, and the orange one is the device working with the USB off (battery always on, with the device always doing the same). The battery is charged at ~0.7A (7.4V Lipo). All changes are cumulative.

    However, after a recommendation of a colleague, it seems that a ferrite bead at the input of the device (voltage from USB) has done the most effect in bringing the frequencies down! I can't quite explain it myself - I wanted to ask you if you have had any similar experience into why this might be the case. Can noise be radiated back into the input to the system?

    Thank you for your time!

    -Eduardo

  • Eduardo,

    Yes.  Ferrite beads are another option.  Glad you found a solution!

    Regards,

    Jeff

  • Hi Jeff, thank you for all your help! The snubber did remove high frequency noise so I will add that too.

    I do have a question though. The ferrite beads removed a component failing FCC at ~122 MHz, but there is a component, 76MHz, that no fix could bring down or improve. In fact, it kind of got worse! It seems to be active also when the USB is connected, and with an RF probe I can see it coming most probably from the charger IC. Do you have any ideas into what I could look for to find a fix? I am stumped into what else I could try.

    Thank you!

    -Eduardo

  • Eduardo,

    Yikes.  You are using all the tricks that I know.  You can try additional 0.01-1uF at input before or after the ferrite bead.  On our EVM that we tested, we added additional 0.1uF at the input and output connectors to the PCB as well as close to PMID and SYS pins of IC in order for it to pass  This implies that the board traces were radiating.

    Regards,

    Jeff

  • Hi Jeff:

    Thanks for the info - I will try the PMID trace caps as that is the only trace I have not tried yet. One question - on the EVM I see some diodes on PMID and VBST traces. Could they help?

    Would it be crazy to add a ferrite bead on the PMID and/or VBST pins?

    Thank you!

    -Eduardo

  • Hi Eduardo,

    The PMID decoupling cap is actually the most important for minimizing EMI.  

    Ferrite beads only help with current flowing through them so no help on PMID, unless you are running in OTG mode.  Although I have never tried, I suspect adding a bead in series with BTST would be bad because it adds inductance to a trace with high di/dt current, which could cause voltage spikes.

    Regards,

    Jeff

  • Hi Jeff:

    We will go to do another FCC precert soon and will come back with results. I think the issue was I used a ferrite bead that had too small of a current capacity for what was needed and was not acting as a ferrite bead anymore. A change in ferrite beads improved the system. However all of the recommendations helped a lot. Thanks!

  • Just to close the loop, I implemented the fixes here and they did the trick. Did not have time to find the minimum amount of changes that did the trick though. But passing is the main concern and parts are very inexpensive. Thanks for the feedback!