[hpsdr] Further Observations on the PowerSDR Noise Blanking Process

[Mike F]

Considering the fact that the frequency we have the greatest interest in 
quieting is our receive frequency, would it make sense to center a 
'noise receiver' at the receive frequency.

Mike, K7SR

> [hpsdr] Further Observations on the PowerSDR Noise Blanking Process
> Phil Harman phil at pharman.org
> Tue Apr 19 20:42:41 PDT 2011
>
>     * Previous message: [hpsdr] Further Observations on the PowerSDR Noise Blanking Process
>     * Next message: [hpsdr] Pandora Enclosure with Antenna Switch
>     * Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
>
> There are a number of ways we could improve the Noise Blanker.
>
> Many years ago Collins (I think) offered a noise blanker that picked up
> wide band local noise in a quiet part of the low VHF spectrum.
>
> I did some experiments with this idea in the 70's using a 1MHz wide
> receiver on about 40MHz. As long as the noise was local and wide band then
> it worked really well.
>
> With HPSDR we could implement a 'noise receiver' in the Mercury FPGA and
> tune that to a frequency band that has the local noise but no strong
> signals.
>
> We could also use a second software receiver in the 192kHz bandwidth that
> could be set to a section of the band where there were no strong signals
> and the bandwidth could be adjustable. You could perhaps show this 'noise
> receiver on the bandscope' to make it easy to move/adjust.
>
> Food for thought.
>
> 73 Phil...VK6APH
>
>
>> ***** High Performance Software Defined Radio Discussion List *****
>>
>> Mike:
>>
>> The way that most noise blankers work is that they turn off or gate the
>> RF signal
>> path during the period that an interfering noise impulse is present.
>>
>> The narrower filters associated with the demodulated signal will "heal"
>> or smooth out the short hole in the RF signal, improving the ability
>> to receive dramatically, in the presence of impulse noise.
>>
>> The bad news is that another way to describe a circuit that turns an RF
>> signal path on and off in time with a control signal is to call it a
>> Mixer.
>>
>> So the gating process that protects the downstream parts of the receiver
>> from the loud impulses, also creates sidebands or mixing products at the
>> noise repetition rate around all of the signals in the band. If the noise
>> repetition rate is somewhat random, then the mixing sidebands are also
>> random noise, and you start to see the noise floor rise across the band,
>> worse in the immediate vicinity of the stronger signals.
>>
>> It is hard to fool Mother Nature, and get something for nothing.
>>
>> --- Graham / KE9H
>>
>> ==
>>
>> On 4/19/2011 10:38 AM, David McQuate wrote:
>>> ***** High Performance Software Defined Radio Discussion List *****
>>>
>>> Mike,
>>>   Your surmise is correct.  Noise blanking is performed before the
>>> bandwidth filters are applied.  This is the same order as is usually
>>> done in analog radios.  In reducing the bandwidth, the filters slow
>>> the rise & fall times of the signal, "smearing out" noise impulses,
>>> making them harder to remove.  And yes, noise processing is usually
>>> done in the time domain.
>>> 73,
>>> Dave
>>> wa8ywq
>>>
>>>
>>> On 4/19/2011 8:04 AM, Mike F wrote:
>>>> ***** High Performance Software Defined Radio Discussion List *****
>>>>
>>>> I made a posting couple of weeks ago on a noise floor phenomena where
>>>> the noise floor across the displayed band varied up and down with the
>>>> presence or absence of strong in-band signals.  Phil Harman explained
>>>> that it had to do with the NB and NB2 functions in PowerSDR.  At that
>>>> time I had it in mind that these functions might be making the noise
>>>> worse.
>>>>
>>>> Recently the noise floor has been quite high at my QTH (S5) and I
>>>> have determined the following.
>>>>
>>>> As I turn the NB and NB2 functions on, the noise floor drops
>>>> significantly.  When a strong signal within the band increases
>>>> amplitude the noise floor noticeably increases, dropping when the
>>>> signal decreases or goes away.  I don't believe that the noise floor
>>>> ever gets as high as it is without the NB-NB2 functions turned on,
>>>> but it does vary.
>>>>
>>>> Based on these observations I would surmise that the noise blanking
>>>> is applied to the whole 192kHz (in my case) bandwidth (as a time
>>>> domain function???) and if there is a significantly large signal
>>>> within the band, the blanking threshold/algorithm becomes less
>>>> sensitive to the lower amplitudes.
>>>>
>>>> Can anyone out there confirm or deny my suppositions?
>>>>
>>>> Thanks.
>>>>
>>>> Mike Fager, K7SR
>>>>


 1303327235.0