• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 927 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

AFE4403: AFE4403 noise pickup during EMC testing causes low frequency aliasing.

Morgan Smith26
Prodigy 60 points
Part Number: AFE4403
Other Parts Discussed in Thread: AFE4404,

Hello

We are using the AFE4403 in our device and during EMC Radiated Immunity testing we are seeing low-frequency signals being picked up.

The signals are not at the carrier or modulation frequency, they are at a much lower frequency, around 2-3Hz.

The Carrier is in the range of 2.3GHz with a modulation of 80% 1kHz.

When the interference is enabled, a near perfect 2Hz sign wave is seen on the PPG signal.

On closer inspection of the data sheet, buried on page 30 is the following paragraph:

TI highly recommends that a single clock source be used to generate the clock required by the AFE as well as the clock needed by the microcontroller (MCU).
If an independent clock source is used by the MCU, then any energy coupling into the AFE supply or ground or input pins can cause aliased spurious tones close to the heart rate being measured.

This is not the default set up of the AFW4403EVM, which is surprising as it is "highly recommended", we modified our circuit to drive the clock from the MSP430 and sure enough, the noise frequency appears to have been reduced.

However, we are still picking up level shifts in the signal when the noise is enabled and disabled. We are also still picking up burst modulated noise as well.

Is anyone from TI able to explain the noise coupling effect and why modifying the circuit todrive both Micro and AFE chip from the same clock source has this effect.

We would like to understand the effect, so we can attempt to further mitigate the noise pick-up on the signals.

The explanation is the datasheet, doesn't make sense as the AFE handles the A/D conversion and only SPI interface goes between the Micro and the AFE.

If anyone can explain why it makes a difference we would be very grateful.

Thanks.

  • 0
    TI__Genius 12555 points
    Hi Morgan,

    We need few details as given below to able to locate the issue.
    1) Is the circuitry for EMC ( Carrier signal and modulating signal) connected to the AFE4404's circuitry? Is the GND planes are shorted? Or they are electrically isolated.
    2) Is AFE4403 is configured as Crystal Mode or Crystal Disable?

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Prodigy 60 points

    thi Prabin

    1) No, the EMC is radiated immunity, the device is irradiated using a large antenna in the EMC chamber, they are not electrically connected.

    2) We initially had the AFE is crystal mode, as per the AFE4403EVM and we received the low frequency pick up at ~2Hz. The crystal is now out of circuit and the clock is driven from the MSP430 (crystal disabled), like the EVM can be by replacing R15 and removing R16 & R17. Now it only gets level shifted when the circuit is irradiated. Although there is also a, very low frequency, drift on the signal (approx 0.025Hz).

    Thanks

  • 0 in reply to Morgan Smith26
    TI__Genius 12555 points
    Hi Morgan,

    Following is my response for your queries.

    Any high frequency asynchronous energy when sampled by a low sampling system, will fold back into the system as a low frequency signal.
    This can be mitigated by having a synchronous system ( sharing a same clock) such that the external interference signal is seen as a DC signal.
    This can be further removed by having a differential architecture ( as present in AFE4403) instead of single ended ones. However the cancellation will not be perfect and very low portion of interference will still creep into the system. This is the reason that you see low signal (reduced as compared to asynchronous clock)pick up when you have a same clock source in MSP430 and AFE4403.

    From our past experience we have seen that this kind of interference is picked by the AFE when a high frequency communication block such as BLE, WiFi coexist with our AFE. In such scenario we recommend to use an external common mode choke between the photodiode and the AFE.
    In fact, EVM for newer generation devices have this choke be default. The part number of choke is DLW31SN900SQ2L.

    Hope this clears all your queries and help to remove the interference.

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Prodigy 60 points
    Hi Prabin

    We are using the recommended common mode choke on the design. It is effectively the same as the dev kit.
    I understand how signal aliasing causes the noise, but I am unclear on why the MSP430 providing the clock should help to limit the noise.
    The AFE does all the A/D conversion, the MSP is only interfaced via the SPI and the SPI clock is not used to drive the sampling.
    Could there be some other effect at play?

    Thanks
  • 0 in reply to Morgan Smith26
    TI__Genius 12555 points
    Hi Morgan,

    Looks like in your case somehow the EVM is picking some noise and sharing the clock between the MSP430 and AFE makes it synchronous.

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Prodigy 60 points
    It's not just the EVM, it's our board as well. The above answers don't really answer my question, regarding the mechanism.
  • 0 in reply to Morgan Smith26
    TI__Prodigy 215 points

    Any source of energy that is asynchronous to the sampling frequency of a conversion system (like what the AFE4403 is) has the potential to show up as a tone at the output of the converter. This includes a case where you have digital logic in the MCU that runs off an independent clock (that is not same or synchronous to the AFE clock). Such digital logic switching at the MCU clock rate can inject energy into the PCB (for example through the supplies or ground). It could get picked up by the Photodiode or it could appear on the supply/ ground rails of the AFE. Or such energy could cause a periodic jitter in the AFE's sampling clock. Any of these mechanisms can cause such energy to appear as a tone at the AFE output. The differential signal chain of the AFE and the internal high PSRR LDO helps to suppress some of the effects of such asynchronous energy so that only an attenuated portion of this energy shows up as a tone at the output. 

    To completely eliminate the effect, you might need to eliminate the very source of such asynchronous energy - and the way to do that is by sharing a common clock between the AFE and the MCU. When you do that, there is still energy coupled into the PCB by the digital switching logic in the MCU. However, since such energy is synchronous to the AFE clock, the sampling operation by the AFE shifts this energy to DC.