Re: RSP Spectrum analyser software available shortly
Posted: Wed May 09, 2018 8:00 pm
Hi John
For sweep widths greater than 10MHz I'm pretty well doing the same thing as your Signal Hound SA. Adding two or more segments together is the only way wide sweeps can be achieved. For a 10MHZ chunk, the usable bandwidth is around 8MHz, with quit heavy attenuation at the band edges. In order to avoid this attenuation, I'm tuning in steps of 5MHZ. I could tune in steps of around 8Mhz, which is the maximum usable bandwidth, but at 5MHZ, the maths works out quiet nicely and gives a nice flat response that does not include a series of dips where a signal of interest would be quiet heavily attenuated.
The problem I have is when the RSP crosses band boundaries at 60, 120, 250, 420 and 1000MHz. There are quiet large jumps in amplitude as the radio is tuned across the boundaries, and it looks like the gain reduction needs to be adjusted for each band. I have tried doing this but it doesn't appear to work as I expected. I seem to be getting some instability at this point, with each band section jumping up and down quiet a bit. Also, signals randomly appear in the wrong segment, indicating a tuning problem. Where the problem seems to be is that while a change of frequency within a band segment can take anything from 0.5 to around 15ms, a frequency step across the band boundary can take up to 150ms. Currently there seems to be no way of knowing when the newly tuned frequency has settled. SDRplay have been most supportive and we are going to be discussing this, and other issues some time in the next couple of days.
I'm getting a similar instability problem with the DC spike, so performing a null scan to establish a baseline will not cure the problem. Having said that, I think the spike problem is due to me doing something wrong somewhere as the DC correction system seems to work very well with SRDuno. I'm going to change the system so it uses a low IF instead of a zero IF at narrower sweep widths, that will eliminate the spike problem completely. I'm not sure if I'll get that done in time for the first aplha release though.
When playing around with the tracking generator side of the equation, using both broadband noise, or a DDS, I was using a similar system in order to establish a flat response across a given sweep width. Having said that, the noise generator seems to give a fairly flat output, making a baseline calibration fairly trouble-free.
I think Phil's idea regarding the RSP generating it's own tracking signal is interesting, but as I'm not familiar with the radio's architecture I can't say much about it. It would be nice if it could be done though.
All in all, I think the problems I'm experiencing are largely of my own making and I'm sure they can be sorted out.
73's - Steve
For sweep widths greater than 10MHz I'm pretty well doing the same thing as your Signal Hound SA. Adding two or more segments together is the only way wide sweeps can be achieved. For a 10MHZ chunk, the usable bandwidth is around 8MHz, with quit heavy attenuation at the band edges. In order to avoid this attenuation, I'm tuning in steps of 5MHZ. I could tune in steps of around 8Mhz, which is the maximum usable bandwidth, but at 5MHZ, the maths works out quiet nicely and gives a nice flat response that does not include a series of dips where a signal of interest would be quiet heavily attenuated.
The problem I have is when the RSP crosses band boundaries at 60, 120, 250, 420 and 1000MHz. There are quiet large jumps in amplitude as the radio is tuned across the boundaries, and it looks like the gain reduction needs to be adjusted for each band. I have tried doing this but it doesn't appear to work as I expected. I seem to be getting some instability at this point, with each band section jumping up and down quiet a bit. Also, signals randomly appear in the wrong segment, indicating a tuning problem. Where the problem seems to be is that while a change of frequency within a band segment can take anything from 0.5 to around 15ms, a frequency step across the band boundary can take up to 150ms. Currently there seems to be no way of knowing when the newly tuned frequency has settled. SDRplay have been most supportive and we are going to be discussing this, and other issues some time in the next couple of days.
I'm getting a similar instability problem with the DC spike, so performing a null scan to establish a baseline will not cure the problem. Having said that, I think the spike problem is due to me doing something wrong somewhere as the DC correction system seems to work very well with SRDuno. I'm going to change the system so it uses a low IF instead of a zero IF at narrower sweep widths, that will eliminate the spike problem completely. I'm not sure if I'll get that done in time for the first aplha release though.
When playing around with the tracking generator side of the equation, using both broadband noise, or a DDS, I was using a similar system in order to establish a flat response across a given sweep width. Having said that, the noise generator seems to give a fairly flat output, making a baseline calibration fairly trouble-free.
I think Phil's idea regarding the RSP generating it's own tracking signal is interesting, but as I'm not familiar with the radio's architecture I can't say much about it. It would be nice if it could be done though.
All in all, I think the problems I'm experiencing are largely of my own making and I'm sure they can be sorted out.
73's - Steve