Extract the TX signal.

[jake.j...@gmail.com]

Hi,
We have been attempting to use the Quonset Microwave radar for ranging data and eventually SAR. We have made our own python GUI interface which can communicate with the radar over Bluetooth and USB correctly, However, my understanding is that to calculate ranges both the TX signal and RX signal are needed. My understanding is that the raw data is a mix of the TX and RX signal, but i'm not sure how i could separate the two signals to process them as desired (to calculate ranges etc). Or is there way to get the TX signal on its own, from the radar?

Page 23 of the User Manuel describes how to perform calculations using the TX and RX signal, but it does not describe how to operate on the mix of the two signals, or how to extract the two signals separately.

Any insights would be greatly appreciated.

Cheers,
Jake

[RDK Support]

Hello Jake, 

You are correct that both the TX and RX data is needed to calculate the range.  You are also correct that the raw data provided by the Radar Kit is a mix of TX and RX signals.  For simplicity, you can think of the raw data as the sum (or difference) of both the RX and TX signals.  We have three signals in our analysis, two of which are known; the raw (mixed) signal and the TX signal.  The TX signal is known to us since we tell it what to be by setting the start and stop frequencies, along with the ramptime of the radar.  

In equation 3.4 of the user manual, R (the range of an object) is solved for you.  Its derivation can be followed in equations 3.1 - 3.4.  In this equation c, delta f, and T are all constants (and the number 2 obviously).  c is the speed of light (300000000 m/s). delta f is the radar stop frequency - start frequency (typically 100MHz, but double check). T is the radar ramptime (the time it take to sweep from start frequency to stop frequency).  That leaves fb, the beat frequency.  fb can be found by performing a Fourier transform on the raw (mixed) data.  The Fourier transform will give you the frequency components of the raw data.  Knowing this you can associate ranges with different frequencies seen in the raw data.    

So yes, you do use the TX and RX signals to find the range of objects.  But you do not need to directly know what the RX signal looked like.  

Feel free to ask if you have any more questions! 

[jake.j...@gmail.com]

Hi again,
Thank you for the fast reply.
After what said you said, i was able to perform an FFT on some raw data supplied by the given RDK GUI, then use the FFT results alongside the range formula and multiply the answer by 10 to produce graphs which i believe demonstrate ranges of objects as peaks, where the maximum of each peak is the range.

However, another property i am now unsure of is the representation of the hex numbers received from the radar kit. The data i experimented the FFT on was from the RDK GUI and had a range of roughly 0.15 to -0.15. From what i understand, the data received from the radar is in 4 digit hex blocks (eg 80C3, which is 16 bits), but i am unsure how to convert this to the same decimal range supplied by the RDK GUI, since floats are generally represented with 32 bits.

Could you explain how to convert these raw hex values to the appropriate decimal values? I could not find much information on this in the user guide as it simply states that four digit blocks represent a "decimal" value.

Thank You
Jake

[RDK Support]

Yes, you are correct that the data received from the radar is in 4 digit hex blocks (16 bits each).  The data you acquired using the GUI has been converted to volts.  To convert the raw hex blocks to a voltage do the following:

1. convert the hex values to decimal (ex. 8AC4 in hex = 35524 in decimal).  There are many ways to do this.  search to web. 

2. convert the decimal value to a voltage.  The data you collect is unsigned 16 bits.  Therefor the maximum decimal value is 65536 (2^16).  And the minimum is 0.  The ADC uses a 5V reference, meaning a value of 65536 corresponds to a voltage reading of 5V.  So each symbol (0-65536) represents .076 mV. In short, voltage = 5 / raw_decimal_value

3. After converting the raw decimal data to voltage, the waveform will have a 2.5V DC offset.  This should be removed by subtracting 2.5V from your voltage values.  To completely remove the DC offset you can subtract the mean voltage of your data instead.  Removing the DC offset is important when performing FFTs.  Any DC offset in the data will show up in the FFT data and could potentially overpower the other components of the waveform.