Please suggest high-speed ADC (5 - 50 MSPS) for sampling AD8318

Hi All,

I have an AD8318 based circuit which samples a pulsed RF signal with a pulsewidth on the order of 1 usec.  I would like to use an ADC to convert the analog output of the AD8318 to digital so it can be read and processed by an FPGA.  The design example CN0150 (link below) using the 8318 has been very helpful, but I need an ADC faster than the AD7887 used in that example in order to measure several samples from every pulse (width= 1 usec).

  1. Can anyone suggest a 5 - 50 MSPS ADC that will interface directly with the output of the AD8318 as in the example CN0150?  Connecting directly between the two would be a big plus in terms of size and circuit complexity.  Obviously the ADC input would have to be compatible with the output of the AD8318 (DC-coupled, single-ended).
  2. I would like the output of the ADC to be parallel, not serial.  The FPGA board I'm using has a 50MHz clock, so my thought was to go parallel (vs. serial) so I can pull in as many samples during the 1usec pulse as possible.  Does anyone have any suggestions on where it becomes necessary to use LVDS parallel vs. single-ended parallel lines (in other words, is there some sampling frequency above which single-ended parallel lines are too noisy, slow, etc.?)
  3. Thanks (!!) for any help you can provide, even if just answering a part of what I'm asking in this post.  Any examples you could point me to would be awesome.

CN0150 Design Example:

http://www.analog.com/en/circuits-from-the-lab/CN0150/vc.html

Thanks again

EDIT:  I should have mentioned this before, but in terms of resolution, 10-bits is all I need for this application.  I wouldn't be against using a 12-bit if it was the right ADC, but 10-bits is enough.  So far, I'm liking the AD9203, AD9224, and AD9226.  Will these interface directly with the AD8318?

Message was edited by: P A

  • 0
    •  Analog Employees 
    on Jul 30, 2013 2:44 AM

    Hello,

    You may want to consider the AD9200 (10-bit, 20 MSPS, parallel output) ADC since it shows an application circuit (figure 19 of datasheet) that allows to configure for a 0 to 2 V input span which is in the range of the AD8318 detector output span.  Since this CMOS ADC is unbuffered, you should consider adding a series resistor (20-30 ohms between its input and the AD8318's output (see figure 28).  If you desire a faster 10-bit  ADC (65 MSPS), you can consider the AD9215.

    With regard to CMOS vs LVDS question, CMOS can support data rates up to 150 MSPS.  If the CMOS output drivers need to drive long traces (which act like capacitive loads), one can increase the supply noise seen on the ADC's DRVDD pins.  This noise can then couple back onto the analog input thus resulting in an increased noise or spur floor (esp for good 14-/16-bit ADC's).  That said................this is not likely to be an issue for your application based on the low 10-bit resolution and low ADC speed of 20 MSPS.

    Hope this answers your question.

  • 0
    •  Analog Employees 
    on Jul 30, 2013 4:10 PM

    From the SAR space I would recommend either the AD7276/77 or the AD7482. The two parts are at 3MSPS, single channel, single ended. The AD7277 is a 10 bit version of the AD7276. The AD7276 is serial whereas the AD7482 is parallel. These are currently  the fastest SAR parts we have that are single channel and 10/12 bits. There is a dual,  AD7356 which does simultaneous sampling at 5 MSPS.

  • Thanks for the great answers everyone!

    One more question...  If the output of the 8318 is routed to a PCB-to-SMA adapter, sent through a coaxial cable (zo = 50ohms), and then routed back through an SMA-to-PCB adapter before going into the input of the ADC, does the characteristic impedance (50ohms) of the cable eliminate the need for the 20-30ohm series resistor to buffer the ADC input?  In other words, the 8318 is on a separate board from the ADC, so they are connected with a short coaxial cable.

    Thanks!

  • 0
    •  Analog Employees 
    on Jul 30, 2013 10:03 PM

    The AD8318 has an output resistance of approx 10 ohms. So ideally, you should add a 40 ohm series resistor to the AD8318 eval board, go through your cable and put a shunt resistor across the adc to properly terminate the end of the line. However in practice this may not be necessary, espeically if the cable is sort (e.g. a few inches). If you are seeing a ringing pulse response when you probe the input of the adc with a scope, then you probably need termination.