G'day
Greg Zl3IX raised the following great point about noise (see below), which
Phil VK6APH has commented on and should be of interest to those wondering
about how such a system could improve noise reduction on 160m. The
previous post from Phil ('More on 160m receivers and true DSP') is also
useful to look at in this context.
This will be it for now (Phil and I have run out of time...), but I hope
people have found the 'discussions' interesting.
One final point I would like to make is that the idea of my posts was to
bring to the attention of the reflectees the possibilities that direct
conversion receivers, coupled with true DSP offers, and to encourage those
with technical expertise to look at these exciting possibilities with a
view to developing suitable systems that the less technical 160m operator
can use.
My skills are as a scientific journalist - explaining complex ideas in
plain language, which is how I earn my living and I use these skills for
fun on this reflector. I might have been a topbander for over thirty
years, but am not a professional engineer. I see my role is as a
facilitator, which is what I have tried to do in providing a pro DSP/DC
view point.
That being said, the possibilities are very exciting, which is why (with
luck) you will hear me using a direct conversion transceiver, with a KK7P
DSP heart in the next few months.
Vy 73
Steve, VK6VZ
----------------------------
> VK6APH: In terms of system performance, the best you can do is with a
>> matched filter
>> at the receiver - which is trivial to achieve with a DSP, but virtually
>> impossible with conventional
>> analogue filters. Please listen to a properly matched DSP filter -
>ringing is
>> not a problem and if you have subtracted the majority of the noise pulses
>> in the pre-filtered bandwidth, then you are in for a very pleasant new
>> experience.
>>
ZL3IX said:
>Phil, how does this subtraction process work? How can the noise pulses be
>subtracted without subtracting the wanted signal as well?
VK6APH replied:
What we basically need to know is the effect a noise pulse will have on the
received signal. There are a number of ways we can do this. The simplest is
to assume the effect will be that of applying the impulse response of the
DSP filter - which we know. A more complex technique is to 'train' the system,
based on the actual impulse response of the entire receiver.
We then look for a pulse train and estimate its prf, amplitude and relative
phase. Once this is known, we make a negative copy of the pulse and subtract
it from the incoming signal - just like a noise cancelling headset. The nice
part of this process is that it can handle multiple noise sources, each one
is analysed in term and subtracted.
This approach is used by Leif SM5BSZ in his Linrad noise blanker
(http://www.antennspecialisten.com/sm5bsz/linuxdsp/blanker/narrow.htm)
except that he does not handle multiple noise sources. There is much to be
improved in his approach, but it is a great proof of concept and a convincing
demonstration of the potential of this approach.
_______________________________________________
Topband mailing list
Topband@contesting.com
http://lists.contesting.com/mailman/listinfo/topband
|