Noise blanker reduces noise heard, but increases it on waterfall

[ofadam]

Why would turning on the Kiwi's noise blanker result in less noise heard and a drop in the S-meter, but show an increased noise level on the waterfall?

[jks-prv]

Short answer: The next release will allow the WF blanking to be applied separately from audio blanking.

Longer answer: Try zooming in. You'll note that at some point WF blanking will start behaving more like you expect. This is because there are not a million strong signals in the WF passband. The blanking is applied in the time-domain ahead of an FFT. And it is done as a rectangular window blank -- the same as for the time-domain audio. This is fine for the audio but causes all sorts of grief when an FFT is applied. So a shaped pulse is probably needed. It's the same issue as why you have to apply a window function to discrete samples to an FFT. And you must also remember the fundamental premise of time-to-frequency domain conversion: that a change to a single sample in the time domain essentially effects all the bins in the frequency domain (exactly what your image shows!)