• State
  • Locked Locked
  • Replies 8 replies
  • Subscribers 46 subscribers
  • Views 373 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.

DATACONVERTERPRO-SW: Can only read data in frequency domain, not time.

Matthew Snowdon
Prodigy 30 points
Part Number: DATACONVERTERPRO-SW
Other Parts Discussed in Thread: ADC3444EVM, ADC3443EVM

Dear Ti,

I am trying to capture signals with a TSW1400EVM and ADC3444EVM. I have and external clock and connected the ADC3444EVM to a function generator to test. I set the function generator to output a sin wave with 100kHz (or any other frequency) and when I read the data I just captured the FFT has one peak at 100kHz like expected, however, when I try to read the time domain there is no signal, only zero volts and a little bit of noise. I am confused because my system seams to work in the frequency domain but not the time domain despite the ADC sampling in the time domain. Does anybody have an idea of what is going on or how to get my system to work?

Your help is greatly appreciated. 

-Matthew

  • 0
    TI__Mastermind 21415 points

    Hi Matthew,

    Can you share images of hsdc pro FFT and also the time domain/codes view? What is the input level you are supplying in terms of full scale(dBFs)? This will better help us see the problem.

    Thanks, Chase

  • 0 in reply to Chase W
    Prodigy 30 points

     

    I hope these screen shots will be able to help you. The system is reading 2Vpp sinewave at 150kHz frequency. The FFT has its peak at -80dBfs and 150kHz, and I was think about it and -80dBs seams quite small and maybe it is only reading some trace harmonics in data which is otherwise just a bit of noise around zero. If this theory is correct do you have any idea on how I can properly read the signal and not just noise?

    Thanks, Matthew

  • 0 in reply to Matthew Snowdon
    TI__Mastermind 21415 points

    Matthew,

    Can you zoom in on the x axis in the time domain plot? I think your signal is there but you’re looking at 524k samples. In the bottom right corner, where it says 524288, can you double click and enter 100 to view the first 100 codes? 

    I’m not sure why a 2Vpp input would only show at -80dBFS. It should be much higher. Can you try at higher input frequency such as 1.5MHz?

    Thanks, Chase

  • 0 in reply to Chase W
    Prodigy 30 points

     

    I am running this test at with a 2Vpp sinewave at 2.5MHz and I was able to remove some noise so now you can notice the signal, when you zoom in, in the time domain. However, it is still only at like -55dBfs. I hope this will hep you. 

    Thanks, Matthew

  • 0 in reply to Matthew Snowdon
    TI__Mastermind 21415 points

    Hi Matthew,

    What is the signal generator supplying this 2Vpp signal? Can you share the model number?

    Have you checked on any of the other inputs? It might be an assembly problem on just that one input. Our EVM are meant to be tested (to avoid issues which may cause something like this) but we have seen a few instances where it seems they were not properly tested. Checking on another channel should help rule out any assembly issues.

    Further testing can be done if you have access to an oscilloscope or spectrum analyzer, you can measure the input signal directly to verify level is correct (I am suspect of this) and if the signal generator seems to be outputting an accurate level, we can look at measuring the signal at the ADC side of the front end network on one of the differential legs.

    Regards, Chase

  • 0 in reply to Chase W
    Prodigy 30 points

    The function generator I am using is the GW Instek afg-2105 and I have made sure to use n oscilloscope to verify that the output of the function generator is the proper 2Vpp. Making sure to check all of the channels to look for assembly issues was a good idea, however, when I ran the test they exhibited the same behavior as the first channel.

  • 0 in reply to Matthew Snowdon
    TI__Mastermind 21415 points

    Hi Matthew,

    Sorry for the delay in response, 

    I've checked and see the output of that signal generator is 50Ω, so no issue there. Are you able to probe the input signal on the EVM at the connector? Since you checked the output of the signal generator is at 2Vpp, I think we can rule out the signal generator and the cabling as well. Can you let me know of the board rev so I can check on my end as well. I haven't noticed any problem like this with this board before but there may still be an assembly issue. If you can probe along the front end and find where the major loss occurs, that would find the issue.

    The reason before the signal increased from -80dBFs to -50dBFs was strictly due to the balun bandwidth being 400kHz to 800MHz. Once you moved the input to within the bandwidth of the ADT1-1WT the signal no longer was attenuated by 30dB from operating out of band.

    Thanks, Chase

  • 0 in reply to Chase W
    TI__Mastermind 21415 points

    Hi Matthew,

    I have checked on RevC ADC3443EVM (same PCB, just different speed ADC) and everything is as expected. If I apply a near 2Vpp signal into the device (measured on a 300MHz BW scope), the FFT and time domain view look as below. I have checked on all 4 channels and all appear the same. This is probing right at the first series resistor for each channel, R4, R16, R32, R46 respectively. I am reaching close to -1dBFS for all cases. My clock supply amplitude is 10dBm (roughly 1.7Vpp) when measured on C43.

     

    Time domain view:

    Input signal on scope, measuring 2Vpp:

    Have you had the chance to check this?

    Regards, Chase