[CQ-Contest] Summarizing the Skimmer Accuracy Thread

[Pete Smith N4ZR]

Here is most of an e-mail I sent to Bret this morning.  I took it off 
the reflector because I thought at the time that it was too long and 
complicated for this audience, but on second thought figured I needed to 
tell everyone just where we stand on thistopic.

I just ran the numbers for February 16, the first day of the ARRL CW 
test.  The RBN posted 1,957,983 spots that day.

Our analysis used these criteria: same spotter spots the same dx within 
5 seconds and the two frequencies differ by more than 2 khz but less 
than 100 khz.  We recognized that this skips the second-harmonic case, 
but are convinced that it is very uncommon.

We identified 3463 possible wrong-frequency spots, or .17 percent.  
Further analysis showed that 97 percent of the identified spots came 
from just five Skimmers. Three of those used simple I/Q audio receivers 
(Softrock type) while two used QS1Rs.

In many cases, particularly where the simple receivers were used, we 
could identify a consistent center frequency between the two spots, 
signaling an image problem caused by amplitude or phase errors in the 
I/Q output. We worked with all the simple receiver guys to help them 
clean up their situations.

The two QS1Rs (out of over 60 regulars contributing to the RBN) continue 
to baffle us.  They show images like the simple receivers, with a 
defined center frequency, but there's nothing traceable to distinguish 
them from the rest of the QS1Rs.  It may be a truly subtle firmware glitch.

As Bret pointed out, not all wrong-frequency spots result from images.  
As W3LPL and K3LR found out early on, trying to listen with SDRs while 
transmitting can lead to a variety of busts, not only of callsigns but 
also of frequencies, as transients from keying can be copied on many 
different ones across several bands.  I think it's essential to 
effectively disable your receiver while transmitting.

Based on an eyeball review of the data, we also believe that some 
stations were using dueling CQs, occupying two run frequencies on the 
same band.  Other stations had truly egregious key clicks, which may 
have resulted in two spots relatively close together in frequency and 
time.  CW Skimmer assumes that you've QSYed and will re-spot you if you 
move more than about 300 Hz.

Busted callsign spots are another subject entirely. For 98 percent of 
competitors, the ability to filter at the cluster node by spot origin, 
as well as the use of filters like the "uniques >x" filter in AR Cluster 
V6, will essentially eliminate bad spots. For example, in 10 hours and 
1000 mostly S&P QSOs in the ARRL CW, I saw one busted spot, and I ran 
every band dry several times, so if they were getting through I would 
have seen them.

The other 2 percent of users (including the big multis), with big 
antennas, think they will miss things with server side filtering, which 
they cannot control. They prefer to get all the raw spots from 
everywhere and do their own filtering.

The RBN team is actively reviewing and testing options for 
server-side-filtering, while considering the economics of also 
continuing to provide a separate, unfiltered spot stream for users who 
want that. The discussion here and elsewhere has been very useful, and 
we welcome all inputs.  Posting here may not be appropriate at some 
level of detail, but either Skimmertalk or RBN-OPS will make sure they 
get through to us.

73, Pete N4ZR
Check out the Reverse Beacon Network at
http://reversebeacon.net,
blog at reversebeacon.blogspot.com.
For spots, please go to your favorite
ARC V6 or VE7CC DX cluster node.

On 2/26/2013 9:20 PM, VR2BrettGraham wrote:
> Since it's only certain skimmers sending wrong-freq spots, you can 
> imagine all you want Pete, it won't change the fact that the receivers 
> have spurious responses & if signals appear there, they will be 
> spotted on a frequency or even band that is outright WRONG.  I've seen 
> these spurious responses, having used the receivers a lot myself, 
> learned what's behind the responses & trained myself to recognize when 
> they happen. There's apparently also one mechanism that is a bug, 
> results in the agile version of one of the popular QSD receivers 
> spotting freqs with correct kc but wrong Mc digits.
>
> GL overcoming what to me seems like confirmation bias, ex-VR2BG/p.





73, Pete N4ZR
Check out the Reverse Beacon Network at
http://reversebeacon.net,
blog at reversebeacon.blogspot.com.
For spots, please go to your favorite
ARC V6 or VE7CC DX cluster node.

On 2/27/2013 8:40 AM, Marijan Miletic, S56A wrote:
> VR2BG wrote: Since it's only certain skimmers sending wrong-freq spots, you
> can imagine all you want Pete, it won't change the fact that the receivers
> have spurious responses & if signals appear there, they will be spotted on a
> frequency or even band that is outright WRONG.
>
>   
>
> There is no need for great immagination as all the cases can be technically
> explained.  The simpler case is harmonics, especially odd ones.  Strong 7
> MHz CW signal can be easily decoded on 21 MHz.  I guess we all experienced
> that in local environment on the dead bands.
>
>   
>
> Image freq in QSD can be easily distinguished by much lower signal-to-noise
> ratio.  These are physical facts and they would be more often recorded by
> hard working 24/7 Skimmers showing millions of spots.   Unfortunately even
> our latest analog radios tend to have IF in HF range for narrow roofing
> filters.
>
>   
>
> Just beware that we are poluting radiowaves with kW of HF power.
> Professionals are using less in smart cells, military OTH radars much more.
> Robots are more accurate than hams!
>
>   
>
> LP MMM S56A
>
>   
>
> _______________________________________________
> CQ-Contest mailing list
> CQ-Contest at contesting.com
> http://lists.contesting.com/mailman/listinfo/cq-contest
>