ADC128S102: Inter modulation Distortion

muhammed esa

Part Number: ADC128S102

Tool/software:

Hi Team,

Could you please suggest ADC128S102's analog front end circuit to do IMD test at 5V or test circuit for the same,and things to take care in software if any?

The expected 2nd order IMD is below -78dB ,but the below spectrum shows -70dB for Fa-Fb,please help to fix this issue.

Regards,

Muhammed Esa A

10 months ago

0 Joel Meraz 10 months ago

TI__Mastermind 20880 points

Hi Muhammed,

Can you describe the conditions that you used to test for this? Are you able to share a log of the data you collected during this test? I can try to rerun the calculations or review the conditions.

Regards,
Joel

0 Joel Meraz 10 months ago in reply to Joel Meraz

TI__Mastermind 20880 points

As a preliminary test, I would ensure that the input signal is very close to the full-scale input range of the ADC (0-5V). I would target -0.2dB below full-scale (4.886Vpp) or higher.

Regards,
Joel

0 muhammed esa 10 months ago in reply to Joel Meraz

Prodigy 30 points

Hi Joel

Thanks for the input.

Input IMD frequencies:

Fa = 19230.7692307692Hz
Fb = 20833.3333333333Hz
Vmax = 4.27V

Dut side:

Fs = 1MSPS
Data captured = 62400 samples

Regards,

Muhammed Esa A

0 Joel Meraz 10 months ago in reply to muhammed esa

TI__Mastermind 20880 points

HI Muhammed,

How are you generating these frequencies?

I also recommend taking upwards of 2^19 samples. This gives better resolution in the number of frequency bins, and helps reduce noise and harmonics.

Regards,
Joel

0 muhammed esa 10 months ago in reply to Joel Meraz

Prodigy 30 points

Hi Joel,

I am using a 16bit awg instrument to generate the mentioned frequencies.

Could you please help me to understand come it is calculated as 2^19 samples and the reason behind it.

Regards,

Muhammed Esa A

0 Joel Meraz 10 months ago in reply to muhammed esa

TI__Mastermind 20880 points

Hi Muhammed,

Simply, it reduces the spectral leakage in the frequency domain. This is better explained in the videos below. It will help improve SNR and THD without applying a windowing function. Since IMD is a form of distortion, I would expect it to help as well, though I'm not explicitly sure on that point.

https://www.ti.com/video/5834902778001

https://www.ti.com/video/5466154855001

Are you able to collect more samples, or are you limited in some way by memory available? In any case, are you able to share the data itself? I can try to plug it in on my side to see if I get similar results.

I might also be able to try a similar setup with an ADC128S102 EVM board.

Regards,
Joel

0 muhammed esa 10 months ago in reply to Joel Meraz

Prodigy 30 points

Hi Joel,

Sure, thanks for the inputs.

more data can be captured from the DUT since the instrument has enough memory.

i will share the raw data soon.

Regards,

Muhammed Esa A

0 muhammed esa 10 months ago in reply to muhammed esa

Prodigy 30 points

Hi Joel,

The issue has been fixed ,as the signal generator produces those spikes and the same is reflected at the DUT output.

Regards,

Muhammed Esa A

0 Joel Meraz 10 months ago in reply to muhammed esa

TI__Mastermind 20880 points

Interesting. Thanks for sharing. What would be the solution to this? Filtering the output?

Regards,
Joel

Data converters

Data converters forum

ADC128S102: Inter modulation Distortion