AD4005 and ADA4254 with low clock frequency

Hello everybody,

I would like to use an AD4005 with an ADA4254. The signals of interest are below 10kHz on a single input, so there is no multiplexing required. The frontend will presumably be rather far away from the PIC, so I would like to keep the clock frequencies as low as possible.

As I am completely new to the topic of ADCs, I came so far that it would be good to have at least 20kSPS. The chopping clock output of the ADA4254 has 125kHz. I could divide this by two in order to get a synchronized aquisition out of the AD4005 at 62.5kSPS. How jittery is the clock of the ADA4254? Will this lead to problems?

Would it be better to use a 1MHz clock output from the PIC for the ADA4254 and divide it accordingly for the AD4005?

Are there any other options I miss? Is it even worth the trouble with aquisition well below half the chopping frequency?

Thanks in advance for your replies and best regards,


  • 0
    •  Analog Employees 
    •  Super User 
    on Mar 29, 2020 6:50 PM


    First of all you have correctly stated that you will need a conversion rate of at least 20KSPS to represent the bandwidth of interest.  Depending on what you are attempting to do you may want to make the ADC conversion rate at least 100KSPS (10x) or more. 

    My recommendation, with the limited information I have from your post, would be to use the PIC to drive both the ADC CNV and AD4254 GPIO4 pins at 1MHz to ensure that they are fully synchronized.  You will need to use 4-Wire Mode with Turbo (Minimum SCLK = 25 MHZ) to achieve a full throughput rate of 1MSPS on the data converter.  Assuming you're using something akin to the PIC32MX, I would then setup your SPI transaction to make use of the SPI buffers to capture multiple samples and then use na moving average to reduce the effective output data rate (ODR) to something more useful.   By doing so you will get an improvement in the system SNR by 10*LOG(OSR) where OSR is 1MHZ/(ODR) and you'll also achieve natural rejection of the chopping ripple.  

    You may also want to add a simple low pass filter network at the input(s) to the ADC to further reject the output ripple of the ADA4254, just be careful with the selection of the R and C values to avoid introducing settling artifacts/distortion into the measurement.    You can always simulate transient behavior of your amplifier (checking on model availability), ADC and filter component selections using LTSPICE or in some instances ADC Driver tool at the link below.  The ADA4254 does not appear to be an option for the AD4005 at the moment but I will check on availability and get back to you.