Post Go back to editing

Spur in AD9375 TX output spectrum

Category: Software
Product Number: AD9375, LTM4644

Hi,

We are using AD9375 powered by LTM4644 switching regulator. Schematic of transceiver and power regulators is attached.

In the TX spectrum of AD9375 we are observing spur at 25KHz offset from the any fundamental frequency we generate.

We have probed the power supply noise level at 25KHz and its harmonics, it is around -70dBm. There was a thread that discuss about the noise levels in the LTM4644, link is here.

We have also observed the clock phase noise, reference clock to the AD9375 does not have any spur at 25KHz and its harmonics. Image is attached below.

It is suspected that 25KHz and its harmonic noise form the switching regulator is causing the spur.

Please provide us insight on reducing or removing the 25KHz spur that we are observing.

  • Can you try by giving external power supply? In this way, you will be able to isolate the cause of the spurs. If its caused because of power supply, then you can add bulk capacitors on the power rails to filter out high frequency noise from power supply.

  • Hi,

    We provided external power supply and the spurs are gone. We found out that 1.3V (VCCA_1V3_RX, VCCA_1V3_TX) is very sensitive to noise peaks.

    We have not prioritized the increasing bulk capacitor approach, will get the results once we do it.

    Meanwhile we have raised another query and this is the link, where the approach is change the design and use a different 4 output regulator,please help us in this regard.

  • We provided external power supply and the spurs are gone. We found out that 1.3V (VCCA_1V3_RX, VCCA_1V3_TX) is very sensitive to noise peaks.

    Yes, the noise considerations for the 1.3 V analog supply are vital.  The synthesizers are more susceptible to low frequency noise than other supplies because they have programmable loop filters. The loop filter bandwidth defaults are 50 kHz for the Rx, Tx, and ORx synthesizers, and 350 kHz for the clock synthesizer. The loop filter bandwidth directly affects the supply noise rejection on the synthesizers. For example, if the loop filter bandwidth is 50 kHz, any noise on the supply less than 50 kHz is not filtered. The roll-off of the loop filter provides a noise rejection of more than 50 kHz.  Hence, the 25KHz noise is not getting filtered out.

    For each of the following power domains listed, a ferrite bead with high isolation at the frequency of operation is recommended to help isolate the pin from the supply source for best performance. It is recommended that the following power domains be powered using a separate trace with extra isolation using a high rejection ferrite bead such as the Taiyo Yuden BK1005LL470-T or a similar device:
    • VDDA_CLKSYNTH
    • VDDA_RXSYNTH
    • VDDA_SNRXSYNTH
    • VDDA_TXSYNTH

    Meanwhile we have raised another query and this is the link, where the approach is change the design and use a different 4 output regulator,please help us in this regard.

    You can refer to the datasheet of the power supply for output noise vs frequency specs form the below link. More details about that part will be provided in power management forum:

    https://www.analog.com/media/en/technical-documentation/data-sheets/ltm8060.pdf