Hi Chris,

> I first read about polyphase FFT on Alex VE3NEA website.
> 
> http://www.dxatlas.com/Rocky/Advanced.asp It is most interesting. 
> 
> However, I may be confusing Polyphase FFT technique with the Rocky 
> waterfall display which is attractively different to other SDR programs.
I was wondering what "Polyphase FFT" might be. Here I find it is just ordinary
power spectra, but averaged in a way that enhances signals of modest duration.
(I thought it might be something that used the phase difference from transform 
to transform.)

It is a non-linear averaging process and my understanding of information
theory tells me that this way a threshold is created. Signals below the 
threshold
become weaker (disappear) while signals above the threshold would get improved
S/N. Now, that is what I expect from signals of long duration. Signals of
short duration should become greatly enhanced.

It would be very interesting to have a recording that contains a signal that
is visible with "Polyphase FFT" but not visible with an "ordinary FFT" or 
waterfall.
I would like to analyze it in Linrad which I believe is at the theoretical
sensitivity limit for weak signals of long duration. (Linrad is flexible
and the user has the freedom to set parameters that optimizes sensitivity in
each particular case.)

In case there are signals of short duration that are of interest to a 
reasonably 
large group it would be easy to modify the way Linrad computes averages, but
before considering it further I would want to know what kind of signals to
look for.

> Second, regarding frequency resolution. No, I am not referring to the 
> excellent lower waterfall window of WinRad - rather to the ability to 
> read/adjust the LO and TUNE frequency readouts with increased resolution 
> - down to 0.1Hz. Winrad HD is already a step in the right direction with 
> 1Hz resolution - as is Sigmira. Sigmira is also worth looking at as it 
> allows modes AM, LSB, USB etc all drive the phase display, allowing the 
> frequency readout to be closely synchronised with WWV or any other signal. 
> So in a few seconds you can see the tue carrier frequency displayed to 
> within 1 Hz. But even 1Hz resolution is not enough to bring the Sigmra 
> phase display to a complete stop on my SDR-IQ. For that 0.1Hz control 
> would be better. Plus the ability - like Sigmira - to make a semi-permanent 
> correction to the inherent NCO error in the SDR-IQ - around 185Hz at 15MHZ 
> in my unit. Maybe I don't understand how to compensate clock frequency 
> in SDR14/IQ control panel? 

It is not clear to me what you are looking for when asking for higher 
frequency resolution, but I think you would find all the tools needed
(and perhaps too many more) in Linrad. Have a look at figure 2 on this
page:
http://www.sm5bsz.com/lir/am640/am640.htm
at a resolution of 0.3 Hz or so there are typically many carriers
on distant MW stations. At higher frequencies different paths via
the ionosphere would give a more complicated picture even if there is
only one station on the frequency.

Linrad can be set to much narrower bandwidth. Look at the spectrum
of the Loran C carrier. Figure 1 here:
http://www.sm5bsz.com/linuxdsp/usage/millihz.htm
The resolution is 0.01 Hz in the baseband window and the Loran C carrier 
shows no spectral broadening although one can see a slow shift upwards in 
frequency in the waterfall.
I do not understand the purpose of a phase indicator. If one would feed it
with a bandwidth of a couple of Hz, one would see the instantaneous phase
of the sum of all the different carriers. What would that be good for?

> One other thing I omitted to mention in my original post is to ask if 
> a future SDR program might be based on seperate server/client architecture? 
> This would allow location of SDR at a remote site and control via internet? 
People are doing that with Linrad. It can be done in many different ways
and I have not done it myself so I can not give any details.

> I think SpectraView 3 can do this but it is not as useful for ham radio 
> applications as WinRad. A question - what broadband data rate would be 
> needed to reliably operate a remote SDR-IQ for instance?
That depends on what you want to do. You might want to send all the raw
data back home so you could start wide band recordings at home now and
then. The data rate then depends on what bandwidth you want to monitor.
If you select i.e.125 kHz you would have to transfer 125000 * 4 bytes
per second or 4 Mbit/s.

In case you run Linrad under X11 in Linux you could run it from
a console in another computer and then you will only need the 
data rate to transfer the audio and the screen updates. With a reasonably
slow waterfall and long averaging times on a reasonably small screen 
the data rates would be very low. Presumably one could run Linrad
from another computer under Windows as well - and Winrad also. Windows
and X11 are servers and the Linrad/Winrad clients should be possible
to run over the network. (But do not ask me for details on how to do it....)

73

Leif / SM5BSZ

Reply via email to