Waterfall settings

I've been busy trying to figure out some interference problems with one of my KiWi's.

The main issue I have seen is with strong signals in the MW broadcast band, where the 'skirts' of the strongest signals spread out and mask weaker adjacent signals.

I've found that changing the FFT to "Blackman-Harris" in the waterfall 'more' control panel and adding the /?wfi=0 switch to the KiWi url, in my opinion makes a huge difference to the quality of the waterfall and spectrum displays.

However I'm still not entirely sure about the other settings that I've tried, and I'm also not sure how to make my changes to these settings 'stick'



Anyone else got any 'tweaks' they would like to share.




  • jksjks
    edited October 3

    Well, that's pretty interesting. Looks like the Blackman-Harris window function should be the new default. I never considered it because the Wikipedia article on spectral leakage suggests B-H has the property of raising the noise floor slightly in exchange for improving the leakage issue. But a quick test on weak 20m CW signals shows it actually cleans up the waterfall very slightly. So it appears there is really no reason not to make it the default.

    Can't remember if I mentioned this before, but there is a URL parameter for setting the window function: wfw=N where N=0 to 3 corresponding to the four window function menu entries. For me changing the waterfall sampling mode (wfi=) didn't change things too much.

    Below confirms Hanning vs B-H using the Kiwi's build-in signal gen. Note the URL params "gen" and "attn" to use without having to open the sig gen extension (*). This is because you want the waterfall extension open instead to be able to change the window function back and forth. "attn=22" is needed even though the gen is producing -35 dBm (see S-meter for confirmation) because you have to manually add the waterfall manual calibration factor (13 dB in my case). Note the yellow peak hold line showing the side-lobe difference in the spectrum as with your screenshots.

    I can make wfi and wfw "sticky" like other parameters.

    (*) There is a bug enabling the sig gen from the URL. Once set it will remain enabled until attn is set to zero. So if you reload the page on that Kiwi using some other URL you'll still get the sig gen output instead of antenna signals. You could bring up then close the sig gen extension. Or just specify attn=0 in the URL. Also remember that the sig gen only works on the first rx channel (rx0). All the other channels operate normally using antenna signals.

  • Hi John,

    "spectral leakage suggests B-H has the property of raising the noise floor slightly in exchange for improving the leakage issue."

    Maybe if B-H is supposed to have these drawbacks, the existing Hanning & Hamming parameters could be further tweaked ?

    I included the wfi=0 paramter, as I was tracing interference when I noticed the difference in the spectrum.

    Whilst playing around with the extended controls a bit further, I found that I preferred the 'Last' interpolation setting instead of the 'Drop sample' or 'CMA' options, although I must admit I'm not entirely sure I understand the differences between the various settings.

    Can you remind me what the CIC filter option does ? I seem to remember that it was associated with the waterfall display when fully zoomed in, but I can't find a reference to it.

    The effect of the 'Apature auto mode' also seems fairly subtle. I've tried various settings but don't seem to be able to observe any immediately obvious differences between them.

    Thanks again for the updates.



  • jksjks
    edited October 4

    Maybe if B-H is supposed to have these drawbacks, the existing Hanning & Hamming parameters could be further tweaked ?

    Window functions like Hanning & Hamming have fixed definitions with no adjustable parameters. Although there are plenty of window functions that do. Literally dozens: https://en.wikipedia.org/wiki/Window_function But I think switching to B-H as the default is probably pretty safe from the experiments I've run.

    Can you remind me what the CIC filter option does?

    It just removes the little bit of droop at the high end of the waterfall that occurred because we're not using a proper post-CIC FIR filter due to lack of resources (no FPGA space, no Beagle CPU cycles). I added the checkbox in case anyone thought it was giving them trouble.

    Aperture auto mode..

    Aperture auto mode attempts to make the WF min/max adjustment "auto ranging". Essentially an automatic periodic pressing of the "auto scale" button. When it's enabled you'll note the WF min/max sliders change to become floor/ceiling values centered around 0 dB. When the auto ranging makes a measurement the floor & ceiling values are added to the measured result and then the total is applied to the final WF min/max values that are applied.

    Example: Open the waterfall extension. Enable auto aperture mode. There will be an equation in the middle of the extension panel that says something like:

    Min/max: -155/-75 = -145/-80 (computed) + -10/+5 (floor/ceil)

    That line will flash each time the auto ranging measurement occurs. It means the new values of WF min/max is the computed values from the auto ranging measurement (-145/-80) plus the additional floor/ceil values. If you want a lower total WF min result on each auto range then set a lower (more negative) WF floor value. Similarly for WF ceil. Another way to think about it is that it's adding more "headroom" to the auto ranging result.

    The function used to make the auto ranging measurement is the same as the "auto scale" button and similar to the SNR measurement stuff added a while ago. The FFT bins are sorted, and the 50th percentile value is taken as WF min and the 95th as WF max (if I remember correctly).

    Averaging is then applied to several successive measurements using functions identical to those used for the waterfall & spectrum averaging controls on the main control panel, WF tab (IIR, MMA, EMA). The adjustments are at the top of the waterfall extension panel (Aperture auto mode: Averaging, Parameter). If you turn averaging off then auto ranging only occurs when the waterfall is zoomed or panned.

  • OK thanks John,

    Now I understand teh controls a bit better, I'll continue trying different settings.



Sign In or Register to comment.