[hpsdr] spikes

[Erik Anderson]

Thank you for your responses.  I haven't ran PowerSDR on the machine I'm
currently using for my HPSDR stuff, mostly as the screen goes below the
bottom of my display and I don't really have access to some of the
likely-required buttons down there.  I can't run KK unmodified without some
packet loss without first breaking the main processing loop into two
threads.  I am keeping a fairly close eye on CPU usage-by-thread and I
haven't had missing Metis sequence numbers for a while now so I think I'm
good in terms of performance right now.  The last time I tried to recompile
SharpDSP I ran into some rather strange-looking waterfalls though.

I do recognize that I may be pushing the line here in terms of the amount
of computing power necessary to run HPSDR (which I did underestimate when I
started this project).  I'm not using the desktop machine at the current
time for a number of different reasons, some of which are (*) the antenna
coax won't reach that side of the house and (*) I suspect the majority of
potential in-house QRM would be coming from that general direction.

I still consider myself a relatively new amateur radio operator and I'm
guessing the kind of questions I'm asking kind of emphasis this point, as
many of them seem to be non-HPSDR-specific.  I do recognize that as an
amateur I still need to build up the tools and techniques used to diagnose,
isolate, and attenuate these kinds of issues.  I'm a bit disappointed that
most of the local hams around here seem to have glazed eyes when I ask for
clarification on these kinds of issues, I'm sorry if I'm spamming the list
too much with these kinds of general questions and I do appreciate the
nudges that I get from others.

On Wed, Mar 14, 2012 at 12:38 PM, John Westmoreland <
john at westmorelandengineering.com> wrote:

> Erik,
>
> A few things I have noticed too may tie in with what you're seeing.
>
> While playing with the PowerSDR source, and adding more debug info;
> performance naturally degrades.  Maybe a better term may be program
> response and performance.  This causes the display to appear to have more
> amplitude issues; noise floor issues; etc.  In other words the screen it a
> bit jumpy and it'll eventually get to the point where audio isn't decoded
> if I add too much debug.  So, in a nutshell;
> a high performance computer is important on which to run PowerSDR.   Of
> course it runs a performance test at installation time to determine that;
> so in most cases people should be OK.  I haven't looked to see how well the
> multiple RX version comprehends this issue.  It may be worth your time just
> to try PowerSDR on another machine just to double-check.  I have seen
> phantom spikes too; and I have Alex installed.  I have not yet installed
> the optional Alex ground points but I will soon.
> Adding the debug to the program exacerbates the spikes.  When I tune close
> to them they disappear - sometimes not on 10m - but that is why I need to
> get those optional grounds connected.
>
> Also, I have had smart meters installed recently (didn't ask for them) as
> well as AT&T UVerse (XYL wanted it) - and my shack is currently in close
> proximity to those; and I have definitely seen ingress due to that -
> especially on 160m.  And apparently some on 80m.
>
> I have even done a capture (160m) and sent to a colleague of mine here
> (fellow Ham as well) that is a signals expert and his determination was it
> is definitely man-made.  In other words, ingress from one of those
> sources.  I have not yet determined which one yet.  So, just wanted to
> mention that in case something similar has happend close to your shack.
>
> Anyway, hope this helps in some way.
>
> 73's,
> John
> AJ6BC
>  On Mar 14, 2012 11:40 AM, "Dick Faust" <k9ivb at cox.net> wrote:
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20120314/f2b1b749/attachment-0003.htm>