• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 3 answers
  • Subscribers 46 subscribers
  • Views 586 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.

ADC3563: info about full scale range and noise figure

Matteo Ricci
Prodigy 180 points
Part Number: ADC3563
Other Parts Discussed in Thread: ADC3561, ADC3660, ADC3660EVM

Hi team,

i'm analyzing the datasheet of the ADC3563.

I did not find the following data:

  1. differential input resistance: it's 8 kOhm but what is the value of external termination? In the figure 8.6 the termination is 50 Ohm (25 Ohm on P and N). Must this termination be fixed 50 Ohm or can it assume different values ​​for example 100 Ohm, 200 Ohm ecc..? Obviously, the choice of balun depends on the value of the external termination (for example, impedance ratio 1:1 with termination 50 Ohm, impedance ratio 1:2 with termination 100 Ohm). Is it correct?
  2. full scale range in dBm: in the datasheet the full scale range is in voltage (=3.2 Vpp), but there isn't the full scale range in dBm. What is this value in dBm? I suppose thath this value depends on the value of the external termination: for example, if the external termination is 50 Ohm the full scale range is 14 dBm, if the external termination is 100 Ohm the full scale range is 11 dBm. Is it correct?
  3. Noise Figure in dB: in the datasheet of the ADC3563 there is the value of the NSD of -158 dBFS/Hz (with not input signal). To get the noise figure in dB, i have to convert the NSD in dBFS/Hz to dBm/Hz  and add it to -174 dBm/Hz but i don't know the full scale range in dBm (NSD(dBm/Hz)=FSR(dBm)+NSD(dBFS/Hz). So, can you provide the noise figure value of the ADC?
  4. Noise spectral Density: in the datasheet the NSD for the ADC3563 is -158 dBFS/Hz with no input signal. So if i want the NSD value with input signal (for example at 20 MHz) i can take the SNR value at 20 MHz (=81.6 dBFS) and add it to the Nyquist factor 10log10(Fs/2) with Fs=65 MHz. So the NSD at 20 MHz = -81.6-75.12=-156.7 dBFS/Hz. Is it correct?

Thank you in advance.

Best Regards,

Matteo Ricci

  • 0
    TI__Genius 14251 points

    Hi Matteo,

    1. You are correct - depending on the component before the ADC, the ideal termination can be determined. Per the datasheet, "The termination is located within the glitch filter network. When using a balun on the input, the termination impedance has to be adjusted to account for the turns ratio of the transformer. When using an amplifier, the termination impedance can be adjusted to optimize the amplifier performance."

    2. The full scale range in dBm can be converted 3.2 Vpp = 14.08 dBm. 

    3. Using the 3.2 Vpp to reach full scale -> 

    NSD (dBm / Hz) = 14.08 dBm + (-150 dBFS/Hz) = -135.92 dBm / Hz

    Using F = 1 + 10^((-135.92 + 174)/10) =

    Then NF = 10log(F) = 38.08

    You refer to this app note: https://www.ti.com/lit/an/slyt090/slyt090.pdf 

    4. Use the formula below to calculate this:

      -1 * (SNR (dBFS) + 10*log10(Fs/2))  -> where Fs is the sample rate

    Regards, Amy

  • 0 in reply to Amy
    Prodigy 180 points

    Hi Amy,

    thank you for the answer.

    For the NSD in dBm/Hz =14.08 (dBm) + (-158 dBFS/Hz) = -143.92 dBm/Hz. So the Noise Figure = -143.92 - (-174) = 30.08 dB and not 38.08 dB. Is it correct?

    Another question: are there A/D converter by Texas Instruments with 16 bit, max sampling clock about 100 MHz but with lower noise figure values?

    Thank you in advance .

    Best regards

    Matteo Ricci  

  • 0 in reply to Matteo Ricci
    TI__Genius 14251 points

    Hi Matteo,

    Oh yes, good catch - you are correct. I used the NSD value for the ADC3561 (-150 dBFS/Hz). Yes please substitute the -158 dBFS/Hz for the ADC3563 into the equation.

    I did a search with these parameters on our device page. I had to back calculate the Noise Figure using the equations provided. I attached my excel sheet so you have a quick-reference calculator to use to back-calculate using SNR provided in the majority of the datasheets. 

    If your system allows you to run at 65 MSPS, and want to improve the noise figure, the decimation feature in the ADC35/36xx devices will improve the SNR by a factor of 10*log10(decimation factor). For example, these devices can decimate up to /32, so *theoretically* the SNR can improve by ~15 dB from bypass mode. 

    This is why the Noise Figure for the ADC3660 (serial CMOS) running at 65MSPS @10M in complex decimation /16 is greatly improved over the bypass mode. This SNR improvement factor will also be the case for the ADC35/3663, if your system requires SLVDS. 

    Regards, Amy

    1273.Noise_Fig-Calcs.xlsx

  • 0 in reply to Amy
    Prodigy 180 points

    Hi Amy,

    thank you so much for the detailed answer and for the excel sheet.

    Why for the ADC3660 the Noise Figure is 23.68 dB?

    I agree that decimation increases the SNR but the NSD (dBFS/Hz) and then the Noise Figure shoud remain constant (see the TI application note: How to simplify AFE filtering via high-speed ADCs with internal digital filters, at the paragraph 3 SNR improvement).

    Is this osservation correct?

    Thank you,

    Matteo Ricci

  • 0 in reply to Matteo Ricci
    TI__Genius 14251 points

    Hi Matteo,

    Oh yes, you are correct - my mistake. Thanks for reading up. NSD value should remain constant regardless of the decimation factor. If you divide the fs /16 in the excel calculator, then you will see the expected NSD is ~152 dBFS/Hz. I setup an ADC3660EVM on the bench and confirmed that this is the case for complex decimation /16 sampling at 65 MSPS. However, instead of improving the noise figure, this appears to make it worse (using the calculator I saw a noise figure value of 35.704). Hope this helps clarify. 

    Regards, Amy