• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 4 answers
  • Subscribers 46 subscribers
  • Views 594 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.

ADC32RF80EVM

user5856712
Prodigy 220 points
Other Parts Discussed in Thread: ADC32RF80EVM, LMX2582, ADC32RF80, LMK04828

Hello，

TIers！

I have a few questions about ADCRF80EVM. We use ADCRF80EVM to sample a square wave signal with duty cycle of 1% and repetition rate of 1MHz. The result is different from that displayed by spectrum analyzer. I would like to ask you about it.

1. As shown in the figure, where I use the red coil, the spectrum changes periodically, similar to the amplitude modulation of frequency. Why does this happen? What causes this? We use spectrum analyzer to measure the spectrum of the pulse signal, and there is no such phenomenon.

2. The green line will show the spectrum of comb shape. What is the reason for this? We use spectrum analyzer to measure the spectrum of the pulse signal, and there is no such phenomenon.

3. Window functions in HSDC are divided into rectangular window, Hanning window, Hamming window and Blackman window. What are their specific representation functions? How to get the spectrum of original data without using window function?

Best regards,

David

  • 0
    TI__Genius 10865 points

    Hi David,

    For understanding of windowing and coherent sampling, please go through videos 2 and 3 in the training series at below link:

    https://training.ti.com/node/1139105

    You get FFT like this due to windowing artifacts. With coherent sampling and rectangular windowing, you should get FFT as expected theoretically. 

    For coherent sampling, you have to select input frequency such that it falls exactly on an FFT bin and also connect 10MHz reference out from your signal generator to LMK input of ADC32RF80EVM.

    Regards,

    Vijay

  • 0 in reply to Vijayendra Varma Siddamsetty
    Prodigy 220 points

    Hi Vijay，

    Why use a 10MHz reference signal? Do we need to keep it consistent with the external clock input frequency port? Now we set the signal according to the reference circuit in the ADCRF80EVM data manual, which is 2949.12MHz. Can't this set meet the requirements? In order to sample 0-200mhz pulse signal, how should these two clocks be set?

    Best regards,

    David

  • 0 in reply to user5856712
    TI__Genius 10865 points

    Hi David,

    I assumed you were using on-board clocking (from LMX2582). I see that you are using external sampling clock. In this case, you have to frequency lock two signal sources (generating clock and ADC input) by connecting 10 MHz reference out from one signal source to reference input of second source. 

    As I mentioned earlier, you also have to use a input frequency that is falls exactly on an FFT bin.

    As you mentioned that you need to sample from DC, please go through ADC32RF80 datasheet section "9.1.5 Using DC Coupling in the ADC32RF8x" to understand how to use the ADC with DC coupled inputs. 

    Regards,

    Vijay

  • 0 in reply to Vijayendra Varma Siddamsetty
    Prodigy 220 points

    Hi,

    I watched the video link you sent me, and I still have some problems.

    The external clock input and LMK reference clock input shown in the data manual are two signals of the same frequency and represent the maximum of the sampling frequency.As you mentioned, 10 MHz input signal is used, but 10 MHz cannot sample the signal with 200 MHz bandwidth.And if it is caused by the leakage of the spectrum, as shown in the red line in the figure above, for a rectangular pulse, its spectrum is a 0-200 MHz broadband signal, there should be leakage at every frequency, and the result will appear periodically like this?

    Best regards,

    David

  • 0 in reply to user5856712
    TI__Genius 10865 points

    Hi David,

    When on-board clock is used, 2949.12 MHz for sampling is generated by clocking chips (LMK04828; LMX2582). 10 MHz is used as reference frequency to the LMK. So when 10 MHz from input signal source is connected to LMK, 2949.12 MHz generated by it is frequency locked to input signal source. 

    In case of external clock, which is what you are using, you need to have external clock and LMK reference clock both split from one signal source or connected to two signal sources that are reference locked. In addition to this, for coherent sampling, you also need to reference lock input signal source to external clock and LMK reference clock source. 

    I'm not sure how square wave spectrum looks without coherent sampling. I suggest you check with a sine wave at that frequency first.Once you make sure that you can get single tone with no spectral leakage, you can try square wave. 

    Regards,

    Vijay

  • 0 in reply to Vijayendra Varma Siddamsetty
    Prodigy 220 points

    Hi Vijay,

    I use adcrf80 to sample the voltage value of 32310. How can I convert it to the real voltage value? It is a 14 bit ADC. Is the maximum number 2 ^ 14 = 16384?

    Best regards,

    David

  • 0 in reply to user5856712
    TI__Genius 10865 points

    Hi David,

    Are you using DDC with complex output? In DDC mode, output is in 16-bit. (ADC real output is 14 bit. But 16-bit NCO is used)

    ADC full scale is 1.35V p-p. 

    16-bit output codes are mapped from -32768 to 32767. As you get close to 32768 (full scale), voltage at pins should be nearly 1.35 V p-p differential.

    Regards,

    Vijay