AnsweredAssumed Answered

AD6649 Evaluation

Question asked by avian on Jul 24, 2013
Latest reply on Jul 25, 2013 by David.Buchanan


I am working on evaluating the AD6649 ADC for use in a wideband Software Defined Radio application. I have the evaluation board and the HSC-ADC-EVALCZ interface board. For initial testing for wideband use everything is great. The problem comes when I attempt to prove that the chip can be used for narrowband applications. My original plan was to make a recording of the spectrum using a basic antenna and decode some AM and FM stations from the data. The VisualAnalog software only allows recording of a limited number of consecutive samples, which at 250 MHz is an extremely short period of time. Even lowering the clock to the minimum of 40 MHz there is still the same problem. That short of a time period doesn't allow proving anything with an actual audio sample. I also attempted to use the FIFOInterface code available on your FTP server to make some longer recordings. I made a modification to extend the number of samples to provide a meaningful period of recording time and made some recordings that way. The input variable was a Long data type, so that didn't seem to be a problem. The recording I made did contain the requested number of samples, but there were only 64k (2^16) good samples followed by all zeros for the remaining samples. I assume at some point that requested number of samples is converted to a 16 bit Unsigned Short and that is what creates the limitation.


So that leads to the question, how can I prove that this chip will work properly for narrowband applications? Just about the only thing I can show with that small number of samples is basic AM/FM demodulation from a signal generator. I don't believe just a sine wave is enough to prove that it will work for our purposes. Spending an extensive period of time modifying the FPGA code or rewriting the driver library (not sure which part is the problem) to allow more samples is not feasible just to show that the chip will work for our purposes. Without being able to prove that it can work, there is no way that management can be convinced to spend a large amount of money developing a product using the chip.

Any help would be appreciated.


Andrew Vian