Post Go back to editing

LTC2323-16 (Performance not matched with Data sheet )

Category: Hardware
Product Number: LTC2323-16

I bought the demo board DC1996A-A (LTC2323-16) and the PScope board (DC890B) to evaluate the ADC part LTC2323-16.

 

According to the datasheet, I expect the following performance measurement from the ADC.

 

 

 

But I have seen different performances. See the following Pscope Pics.

 

 Picture One:

Input Signal

1-Input Signal= 0 to 4Volt Peak to Peak

2- Input Signal Frequency=1Mhz

 

Measurement

SNR=59.86 dB

SIND= 50.68 dB

THD=-51.24 dB

ENOB=8.13 bits

Flor= -102.74 dBFS

 

 

Picture Two:

Input Signal

1-Input Signal= 0 to 4Volt Peak to Peak

2- Input Signal Frequency=2Mhz

 

Measurement

SNR=59.78 dB

SIND= 48.29 dB

THD=--48.61 dB

ENOB=7.73 bits

Flor= -102.34 dBFS

 

 

Conclusion:

 

Comparing the datasheet performance graph to the Two PScope photo shows a discrepancy in the measurement, clearly showing worse results.

 

If compare two Pscope pics,  You will observe increasing the analog input frequency ( 1Mhz to 2Mhz), and measurement results are getting worse.

 

Please help me with how I can get the same performance mentioned in the datasheet and why the discrepancy in the measurement.

 

Test Setup

 

  • Hi,

    The first suggestion is to use the BLMN-Harris 92dB window instead of the Hamming window. This will reduce frequency spreading and provide a lower noise floor.

    Next, you need a better analog input source or a better filter. The Agilent 8646A only is capable of THD in the -60dB range. I am not aware of any generators that can produce low distortion in the 1-2MHz range but TTE makes low distortion BP Eliptic  filters that can reduce distortion to the -90dB range.

    The clock source you are using is a good choice for this application.

  • Hi Ghoover

    thanks for the reply, I implemented your suggestion which help to improve the THD but still not matched data sheet performance.

    It improves the THD performance but SNR and SAIND we were not to seen too much improvement.

    I am sharing with you one of the analog technical articles that show Pscope result for the same part. (LTC2323-16) that result i am looking for.

    Driving the LTC2323-16 with ±10V Input Signals | Analog Devices

    Test Wave Form

    1-Test setup-1: Picture One

     

    Clock =100Mhz______________________________________________(Agilent N5181A)

    Input Signal Amplitude = 2Volt Peak to peak_______________________(Agilent 8648C)

    Input Signal Frequency = 1Mhz__________________________________(Agilent 8648C)

    Filter= Cutoff Frequency=1.2Mhz________________________________(Model 3945)

    Above picture show Filter reduce the Harmonic less than -90dB (but still we got ENOB=8.70 bits)

    Following Test setup 2 we changed the input frequency 1 Mhz to 500 Khz and filter frequency keep as above setup (1.2Mhz) it shows the Harmonic worse again.

    Test setup-2: Picture Two

     

    Clock =100Mhz_______________________________________________(Agilent N5181A)

    Input Signal Amplitude = 2Volt Peak to peak_______________________(Agilent 8648C)

    Input Signal Frequency = 500Khz__________________________________(Agilent 8648C)

    Filter= Cutoff Frequency=1.2Mhz_________________________________(Model 3945)

    So now we adjust Filter cut off frequency 600Khz, it reduces the Harmonic and improve the THD (ENOB = 9.20 bits)

    but still, we are very far away from data sheet spec

  • It looks like you are still using a low pass filter. For best SNR results you need to use a bandpass filter. That will bring down the noise floor to the left of the fundamental.

  • Hi Ghoover

    thanks for the reply,

    Bandpass filter that is improve the Result but still I am not able get that result what I am hoping for.

    Test setup:

     

    Clock =100Mhz_______________________________________________(Agilent N5181A)

    Input Signal Amplitude = 1.6Volt Peak to peak_______________________(Agilent 8648C)

    Input Signal Frequency = 900Khz________________________________  (Agilent 8648C)

    Band Pass Filter= LPF=850Khz  HPF=950Khz________________________ (Model 3945)

    https://www.krohn-hite.com/3945Data.pdf

    Single ended to differential (J3 Top BNC connector)

    Signal Mid level reference=2V

     

     

     

     

    Test setup:

     Clock =100Mhz_______________________________________________(Agilent N5181A)

    Input Signal Amplitude = 3.91Volt Peak to peak_______________________(Agilent 8648C)

    Input Signal Frequency = 900Khz________________________________  (Agilent 8648C)

    Band Pass Filter= LPF=850Khz  HPF=950Khz_________________________________(Model 3945)

    https://www.krohn-hite.com/3945Data.pdf

    Single ended to differential (J3 Top BNC connector)

    Signal Mid level reference=2V

     

     

    Test setup:

     Clock =100Mhz_______________________________________________(Agilent N5181A)

    Input Signal Amplitude = 2Volt Peak to peak_______________________(Agilent 8648C)

    Input Signal Frequency = 500Khz________________________________  (Agilent 8648C)

    Band Pass Filter= LPF=450Khz  HPF=550Khz________________________(Model 3945)

    Single ended to differential (J3 Top BNC connector)

    Signal Mid level reference=2V

    If I used LTC2325-16 ADC with my sensor, it's mean i need to design high order bandpass filter circuit then I am able to achieve ADC data sheet spec performance Please clarify to me on this?

     

  • I think you need a better filter. If you look at the specs for your filter you will see that distortion is only 60dB and that it is only 4-poles. I recommend getting a low distortion TTE filter. They have 6pole LC filters with 90dB distortion.