welljim
Newbie level 4
I'm doing some statistical research and need to scan as a wide frequency range as possible (i.e., 300 kHz - 3GHz) and record signal strength (i.e., dBm) for each frequency. I am not interested in demodulating or decrypting the content of the scanned RF signals. The approach I'm considering is to take the I/Q output of a Tuner IC (or a set of Tuner ICs that will cover my range requirements), sample it via ADCs, and FFT the sample to determine signal strength at the tuned frequency. Other ideas welcome.
With respect to the actual implementation, portability and power draw is a major concern. I would really appreciate some help from ppl with experience in DSP / FFT, to give me an idea of how to spec out my parts. I need enough resolution and speed to do the job, but keep the power draw as low as possible at the same time. Circuitry real estate is also critical. I was thinking to use something from the ARM Cortex family since it has integrated ADCs, FFT software libraries, a TFT driver for data output, and dense 32bit instruction set. The M4 also has some hardware DSP extensions, but not sure to what extend they would help in this application. Also, not sure if the M4 would be too power hungry. I don't have in depth experience with any MCU in particular, I will be starting from scratch. Any thoughts?
How do I calculate the sampling / resolution requirements, and what would be the runtime and current draw of each of the available processors that could potentially do the job (ARM Cortex or otherwise)? I'm looking at the spec sheets but it is not straight forward to figure out power consumption as different families need a varying level of CPU cycles to complete relevant instructions. Anybody with experience in this? Optimizations could be done on the software side so I might not have to scan and measure the entire frequency range all the time, so I guess I'm looking to develop a loose metric for my decision making, for example: runtime and power draw to determine signal strength ** per tuned frequency **. Then I can just multiply by the number of frequencies that I need to tune in and measure.
Thank you!!
With respect to the actual implementation, portability and power draw is a major concern. I would really appreciate some help from ppl with experience in DSP / FFT, to give me an idea of how to spec out my parts. I need enough resolution and speed to do the job, but keep the power draw as low as possible at the same time. Circuitry real estate is also critical. I was thinking to use something from the ARM Cortex family since it has integrated ADCs, FFT software libraries, a TFT driver for data output, and dense 32bit instruction set. The M4 also has some hardware DSP extensions, but not sure to what extend they would help in this application. Also, not sure if the M4 would be too power hungry. I don't have in depth experience with any MCU in particular, I will be starting from scratch. Any thoughts?
How do I calculate the sampling / resolution requirements, and what would be the runtime and current draw of each of the available processors that could potentially do the job (ARM Cortex or otherwise)? I'm looking at the spec sheets but it is not straight forward to figure out power consumption as different families need a varying level of CPU cycles to complete relevant instructions. Anybody with experience in this? Optimizations could be done on the software side so I might not have to scan and measure the entire frequency range all the time, so I guess I'm looking to develop a loose metric for my decision making, for example: runtime and power draw to determine signal strength ** per tuned frequency **. Then I can just multiply by the number of frequencies that I need to tune in and measure.
Thank you!!