• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 46 subscribers
  • Views 753 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.

Original question:

AMC3302: 600mV output noise/sine wave

AMC3336: ~400kHz noise in EN 55032 (CISPR 22) Conducted Emissions Test

RCM
Prodigy 10 points
Part Number: AMC3336
Other Parts Discussed in Thread: AMC3301

Hello, all.

I have a prototype application where the AMC3336DWER is closely coupled to some industrial connector leads with 100BASE-TX Ethernet signals.
In Fig. 01 you can see the leads close to the AMC3336 package top face (< 1mm), but they don't touch each other.

 Fig. 01

While testing the Ethernet port through the industrial connector according to EN 55032 (CISPR 22) conducted emissions specs (100kHz to 30MHz BW), there is a ~400kHz noise when a 10MHz, 3.3V, 50% duty cycle clock signal is applied to AMC3336 CLKIN input by an MCU.

Fig. 02 shows the results with no clock applied to CLKIN input. The CLKIN voltage level is kept low (0.02V) by the MCU, despite the CLKIN has a ~1.5MΩ pull-down internal resistor (MCU output is low, not high impedance). No outstanding noise is seen.

Fig. 03 shows the results with the 10MHz, 3.3V, 50% duty cycle clock applied to CLKIN input. You can see 411kHz peak followed by its 2nd, 3rd etc. harmonics.

I tested two prototype samples, and the second one showed a 424kHz, with similar behavior.

 Fig. 02

 Fig. 03

My theory is the AMC3336 has a chopper circuit, or a similar circuit, running at ~400kHz in it, but not documented in the datasheet, and the noise created by the chopper was coupled to the industrial connector leads though radiated emissions, due to the very close proximity, then this coupled noise created the conducted emissions seen on the test.

This related post [ 1059315 ] made me think there is a chopper circuit in the AMC3336 too.

The available literature related to EMC for the AMC3301 family only covers CISPR 11 and CISPR 25 specs (30MHz to 1GHz BW) radiated emissions, outside my target spectrum (100kHz to 30MHz BW, conducted emissions).

My AMC3336 design is according to the [ AMC3336 ] datasheet and the [ AMC3336EVM ] schematics. However, the PCB layout isn't strictly following the layout recommendations, possibly increasing the radiated emissions.

I would like to confirm if there is anything inside the AMC3336 that could create the noise I have described, before I redesign my PCB.

I appreciate your help here.

  • 0
    TI__Guru**** 180931 points

    Hi RCM,

    Welcome to our e2e forum!  You are correct in that there is a chopper inside the AMC3336 device.  I'll see what data we have at the lower frequencies.

    From an isolation standpoint, I'll assume the low side of your circuit is on the connector side in the picture you sent.  I'm curious to know what sort of spacing you have from the left side of the AMC3336 to the Ethernet pins.  What sort of voltages are you dealing with?

  • 0 in reply to Tom Hendrick
    Prodigy 10 points

    Hello, Tom.

    Thanks for the reply!
    I will wait for you to share the lower frequencies information.

    The prototype is a "PCB sandwich". See Fig. 04 below.
    The Ethernet circuitry and the industrial connector belong to the Upper PCB (green PCB).
    The AMC3336 circuit belongs to the Lower PCB (red PCB). The industrial connector's plastic body only touches the Lower PCB (red PCB) due to mechanical constraints.
    The AMC3336 is on a different ground domain than the Ethernet signals on the industrial connectors. They are galvanically isolated from each other.
    The small green vertical rectangle you see in the picture represents the spacing between the Ethernet pins and the AMC3336, it is ~0.5mm.

     Fig. 04

    Regarding your question related to voltage, did you mean the voltage created by the ~400kHz noise on the Ethernet pins?
    If that is the case, I don't know.
    I didn't measure the Ethernet signals after the tests, and I don't have access to the prototype at the moment.

    Did I answer your questions?

  • 0 in reply to RCM
    TI__Guru**** 180931 points

    Oh no!  Jelly side down! 

    No, not quite, the AMC3336 is rated for 1700Vdc (1200Vrms) across its isolation barrier based on the 8mm clearance across the device from the left to right side of the package.  The left side (assuming its the Pin 1 side) would be the high voltage side.  Your clearance from those leads to the ethernet connector pins is looking to be maybe 3mm.  Is that going to be an isolation issue in your system? 

  • 0 in reply to Tom Hendrick
    Prodigy 10 points

    Oh, now I got it!

    No, it isn't a problem for me. The insulation requirement is very low, 100V.

    Besides that, the layout is going to be changed, moving the AMC3336 far away from the connector, assuming it is the cause of the noise. We need to confirm the frequency range.

    Thanks for the hint!

  • +1 in reply to RCM
    TI__Guru**** 180931 points

    So,

    Not exactly the same thing that you are seeing, but the chopper runs at 625kHz as can be seen in this radiated emissions plot.  I don't have a conducted emissions plot to share unfortunately.  There is nothing that I'm aware of running at 400k internal to the AMC3336.