Need to convert Parallel LVDS stream from ADC to USB at sampling rate of 250MHz, 9 bits resolution

Discussion in 'Digital Circuit Design' started by DaMooseMan77, Feb 15, 2016.

  1. DaMooseMan77

    Thread Starter New Member

    Feb 15, 2016
    5
    0
    I need to convert the data stream from this ADC (see link below) to USB.
    http://www.ti.com/product/ADS58B19/description
    The ADC will be configured for 9 bits of resolution at a 250MHz sampling rate and output a parallel LVDS digital data stream. I need to somehow get this data stream to a PC via a USB port. I foresee needing a LVDS serilalizer and receiver, as well as a USB interface IC of some sort. If anyone has suggestions or alternative methods to transport the data from this IC to a PC in real time, it would be greatly appreciated.
     
  2. Papabravo

    Expert

    Feb 24, 2006
    10,138
    1,787
    Which flavor of USB are you considering? Have you procured the USB specification yet?
    That is a sample every 4 nanoseconds if I'm not mistaken, and that works out to 2.25 GB/sec. I'm not sure it can be done easily, and I think you need more than a USB interface chip. That is just less than half the bandwidth of a USB 3.0 connection.
     
    Last edited: Feb 15, 2016
  3. DaMooseMan77

    Thread Starter New Member

    Feb 15, 2016
    5
    0
    I suppose it would have to be USB 3.0 to support the bandwidth I am asking. 250MHz at 9 bits gives 2.250 Gb/s in a serialized stream, which exceeds the 480 Mb/s high end of the high speed USB 2.0. USB 3.0 SuperSpeed supports up to 5.0 Gb/s, therefore I feel that is my only option. Again, I am open to any alternatives - the objective is to get this data from the ADC model provided to a PC for processing.
     
  4. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,529
    1,248
    At those speeds the physics of the universe change. How familiar are you with microwave circuit design? I'm coming off of a 10.1 GHz project, and it's basically "everything you know is wrong" territory.

    ak
     
  5. Papabravo

    Expert

    Feb 24, 2006
    10,138
    1,787
    My best advice is to hire someone who knows what they are doing. This is no place for amateur wannabes.
     
  6. DaMooseMan77

    Thread Starter New Member

    Feb 15, 2016
    5
    0
    Thank you both for the replies. I apologize for amateur nature of my inquiry, I do admit I am in unknown territory. Any suggestions for the creation of a solution? Simply put, if I needed to sample a signal at 250Mhz with a minimum of 8 bits resolution and provide the digitized signal to a personal computer, what would the best method be?
     
  7. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,010
    3,784
    A) Prayer.
    B) Hope.
    C) An introduction to an expert
    D) quit and have a beer instead.

    Maybe some details. What kind of an analog signal are you measuring?
    What will you do with 250Mb of data you collect every second?
    Is the analog signal modulated in some way (AM/FM/PwM, ...).
    What is the frequency of rate of change of the signal you are measuring at a rate of 250,000,000 every second to 9 bits of accuracy?

    Are you sure this is the best method to achieve your goal?

    A very smart man visits this site occasionally. He once said, "let us help you solve your problem, not your solution."

    That means, unless you are doing something completely new and unique, it has likely been done before. There are very few completely new problems in the world so I doubt you have something new. I think you just cam up with a very convoluted solution to a problem.
     
  8. Papabravo

    Expert

    Feb 24, 2006
    10,138
    1,787
    Well, the problem is that this application is just a bit out of the mainstream for most hobbyists. As near as I can tell the best way to do this is to have a data acquisition system with a really large storage device, like a 1 Terabyte storage drive. In data acquisition mode that is all it does. In data delivery mode it looks to the PC like another hard drive. I don't know why you seem to think there is a simple solution to this problem.
     
  9. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,529
    1,248
    Buy something. Seriously.

    At 250 MHz each step of the task - condition the signal, digitize it, FIFO buffer it, serialize it, capture and record it - is a serious piece of hardware or software design. I would start with the USB oscilloscope people. They may not be able to handle long-term continuous stream recording, but they probably have the front end circuits you need.

    ak
     
    JWHassler likes this.
  10. DaMooseMan77

    Thread Starter New Member

    Feb 15, 2016
    5
    0
    Thank you for the suggestions everyone - I did not believe this was a trivial task but I did believe prior to reading the suggestions that it was possible. GopherT, I assume you are correct in saying that this problem is not new - it has clearly been achieved before. Additionally, there may exist a more appropriate option that I am unaware of. Papabravo - what would be involved in a data acquisition scheme of that magnitude? AnalogKid - I will look into a USB oscilloscope, thank you.
     
  11. Papabravo

    Expert

    Feb 24, 2006
    10,138
    1,787
    I guess I would start with a method of buffering the data in memory. Next I would implement a memory to disk drive interface, SATA-3, I think. Last I would do the drive to USB part for transfer to the PC.
     
  12. DaMooseMan77

    Thread Starter New Member

    Feb 15, 2016
    5
    0
    Thank you for the pointers, Papabravo. Do you know of an example circuit or design guideline of some sort to aid in design of such a system?
     
  13. GetDeviceInfo

    Senior Member

    Jun 7, 2009
    1,571
    230
    Personally, I'd be looking at a pcie bus card from NI or similar. Your USB specs put your source in close proximity to the host, why bother.
     
Loading...