Hi Leif,

I want to look at a specific range in the waterfall, say from 4000Hz to
21000Hz.

After a long frustrating period of clicking the -> <- arrows on the left
and right upper corners of the display, I am unable to get the display
in this range.  The up/down/left/right arrows do nothing.  I do not have
a wheel mouse.

Is it possible to change one of the par* files to get this?

Or, add a box to allow input?

Tom

Hi Thomas,

I have no idea what you really want to do so my answer might
become a bit too complicated. More below.

The arrows affect the baseband processing and can be used
when you do not have a wheel mouse. Nothing in the main waterfall
should be affected by the arrow keys.

The main waterfall is governed by these parameters in par_xxx_wg
xpoints per pixel [12]
pixels per xpoint [0]
first xpoint [0]
xpoints [8184]

In case you want to use the wide waterfall as the primary information
about what happens in a certain frequency range you should set:
xpoints per pixel [N]
pixels per xpoint [1]

If you do not enable the second FFT use N=1. If you do use the
second FFT, make N the ratio between the transform sizes.

For the second FFT you have a parameter:
Second FFT bandwidth factor in powers of 2 [M]
The relation between M and N should be:
  M    N
  0    1
  1    2
  2    4
  3    8
.
.
.

This way you will get one pixel for each FFT bin in the waterfall display.
Now, if you want 4000 to 21000 Hz on a screen that is e.g. 1100 pixels wide
you would want a bin separation of (21000-4000)/1100 = 15.4545 Hz/pixel.
Linrad can only compute FFT in sizes that are powers of two so you
would have to use an FFT size of 2048 or more. You may choose larger if you
like. The sampling rate that you would have to set would be:

  FFT size     Sampling rate

   2048            31.65 kHz
   4096            63.30 kHz
   8192           126.60 kHz

I would suggest a sampling speed of about 126 kHz which would mean that
the decimation should be 526. Not possible, but 525 will be ok:
525=3*5*5*7
You might use cic2=7, cic5=25 rcf=3

The sampling rate would become 126.984 kHz with 15.500992 Hz per pixel.
Set the first fft bandwidth to M*15 Hz. (M as above)

To make the first pixel = 4kHz in the waterfall, set first xpoint [258]
(you would get 3.999255 kHz for the first bin)

Set xpoints [1100] to make the waterfall use 1100 pixels. (The screen
has to be a bit larger than that, the window is a little wider than
the waterfall.)

The last fft bin would be at 17.051 kHz.

I see your requirement as rather odd. My guess is that there are
better ways of doing what you look for. If you describe what
you are trying to do I might be able to give some hints.

The primary reason for having the main spectrum and waterfall is to
give the user the information required for proper operation of the
noise blanker. Typically the waterfall would have a large number
of bins for each pixel in which case the colour would correspond to
the average RMS power of strongest bin attributed to each pixel.
That procedure enhances narrowband signals (in case that is what
you are interested in the above procedure would not apply.)

It could be a good thing to allow the user to type in a start
frequency and a stop frequency. Since various things have to be
ratios between integers, the result would not always be what
the user would expect though....

Leif / SM5BSZ

Hi Leif,

Thanks for the reply.  I am beginning to understand some of this.

I am looking in VLF, up to 30KHz.  I want to see SID, tweeks, etc.

The VLF stations are in the range of 10 to 25KHz.  So I would use 9KHz
as the lower frequency and 26KHz as the upper.

The other mode is 100Hz to 15KHz.  Depending on activity, I would choose
a subset of this, e.g. 7KHz as the lower frequency and 14KHz as the
upper.

I will play with this.

I have not been able to reliably replicate the failure mode.  I can get
it to happen at random times.  I ran xlinrad several hours while doing
other things, editing, switching windows, moving windows, starting
applications, using gdb, etc.  I had some ft245 entries in
var/log/messages the first time I started xlinrad, but, nothing after.
And, no overrun errors.  I stopped and started xlinrad several times in
this period with no additional entries in var/log/messages.  I will poke
at it now and then.  I keep a log of par* files, so I can capture the
setup if I can reproduce the errors.

Tom

Hi Thomas,

Hmmm, SID and tweaks require quite different displays and etc might
be yet another different thing;-)

For SID?

I would suggest you monitor 9 to 180 kHz or perhaps a little more.
Enable the network and run half a dozen of instances of Linrad.
Select half a dozen of stations at suitable distances over the
frequency range, place the S-meter graph in the upper left part of
the windoe and reduce the window sizes so you see only the S-meter
graphs for all the different stations.

You might have the signal level of the carriers of half a dozen stations
for the latest 12 hours on screen and a SID should show a sudden change
on all of them.

Another thing could be to monitor the frequency at extreme resolution.
I would guess a SID would cause a doppler shift.

For tweeks I think what I wrote previously would apply. If you
want better time resoultion you might want a higher power of
sin for the fft1 window. (Transfoems would overlap a little more.)

Previously I suggested that you should set the fft1 bandwidth
to 15 Hz. That was incorrect. Set 30 Hz if you use a sine squared
window.

Yes. There are several parameters to play with. Record raw data so you
can look at the same sequence with different processinmg parameters
to learn whar effects different parameter settings will have.

Very good. If you find something that seems reproducible I might be able
to reproduce it here. Then I might be able to prevent the problem or
perhaps to make Linrad exit with an error message that would help to
eliminate the problem. ("Make sure that chron is not running" or similar...)

Leif / SM5BSZ