Post Go back to editing

AD9233 oscillation

Hello, we are developing a new acquisition unit with AD9233, we design the  antialea and the FDA chain to drive the ADC. 

We product more or less 20 sample (ten layers board with Zynq ecc) but in only one sample we found a 27MHz  oscillation (500mVpp)  from the converter..

the first time my through  was the FDA amplifier which oscillate for that i revome the inductor to isolate the ADC and FDA, and the oscillation goes to 15MHz about  (quite sinewave, not perfect) is from the ADC. this signal is common to Vin+ and Vin-  and CML. 

Could a bad quality capacitor (0402 ) 100nF connected to REFT and REFB do this problem?

Thanks in advance.


  • Hello Matteo,

    Let me make my understanding clear. Your signal chain with ADC is FDA +ADC(AD9233). And there was a 27MHz oscillation spur on AD9233 digital output (after FFT). This oscillation spur moved to 15MHz when you remove the inductor. You can see this oscillation spur at VIN+/VIN-/CML pins.

    If this is the case,

      - where is the inductor you removed?  Is it in FDA circuit?

      - To isolate ADC at this oscillation spur issue,

        1. please check FDA output (with oscilloscope or spectrum analyzer) while only running FDA (apply the supply voltage to FDA only , not to ADC, disconnect ADC from FDA by removing a component in series) and don't apply any input at FDA input

        2. If there is an oscillation spur(it will not be at fixed frequency, it will move around when you zoom-in), oscillation spur comes by FDA at some reason. And if this spur goes into ADC, of course there is an this spur at ADC output because there is a signal at ADC input.

        3. If there is no spur at FDA output but can see the oscillation spur at ADC output, then we need to go into ADC in detail. Let's talk it after checking above.

    A decoupling capacitor between REFT and REFB is intended to reduce the noise from the internal reference buffer and the value is related to the power on/power down time.   



  • Hi Tony, i don't use FFT because the oscillation is very clear (about 500mVpp) , i separate the FDA and the ADC in removing  two inductors,( only to understand the source). i suspect that is the bias, see the attached picture



  • Hi Tony, problem solved, i hope, it was the 100nF X5R (in my schematic C25C) connected to CML unfortunatly we use a 0201  and the component deseapper when i try to solder it  (0201 it was very thin for my eyes..). know my guess is :

    it was a problem of soldering  or wrong /bad quality capacitor? i lost this small capacitor and in couldn't measure it. 

    thanks for you support.


  • Hi Matteo,

    Good to hear the problem solved. Let us know if any support is necessary.