Date 1 - 2 of 2
Receiver comparison charts in Grafana - Topic for Arne
With interest I saw your developments on your grafana page, (wspr.live), where you try to visualize receiver performance comparisons. There is the experimental query page based (non-wsprdaemon data) as well as the wsprdaemon page, which both target to show the performance difference of two receivers.
However at this stage there is a fundamental problem:
- there should be only two spots compared from the same spotted call at the same time stamp. There should be no difference calculated and thus not visualized, if one of the receivers do NOT here, what the other receiver hears at a specific time. In other words, only difference data, which is a valid report number for the same station at the same time should be calculated.
Remember many reported S/N numbers are negative.
As an example here is what happens actually:
RX1 reports station ABC with -20 for S/N while RX2 do NOT report the station at all. The station is too weak at RX2- and better antenna is perhaps necessary!
What is calculated actually is RX1 minus RX2, and it calculated -20 minus zero = -20
In the graph then RX1 is 20db less performant than RX2 ... ???
Unfurtunatly that is why the graphs are not useful at the moment...
Only one typical example - only unique spots at the same time should be compared and calculated...
I hope my feedback can help you a bit with getting it right, as the charts are very useful for the dayly experimention - I am so happy with the work you have done for our community...
Hello,toggle quoted message Show quoted text
thanks for letting me know about those issues. I where not able to find any issues on the "SNR Comparision" page. There where three issues with the "Receiver SNR Comparsion" page I just fixed.
1. The "SNR differences" map had an error in the SNR difference calculation so it came up with the behavior you described.
2. The "Mean receiver SNR differences" graph did not take the selected receiver into account (data from all receivers was used).
3. The "Count of common spots" and "Count of common transmitters" did not work as expected.
The SNR differences are calculated per time bucket (point on the graphs) as follows:
1. All spots of the same transmitting station are counted and their average SNR is calculated. This is done independently for both receivers.
2. Transmitting stations with less than "minspots" spots inside the time bucket are removed.
3. For each time bucket the difference between the SNR values from step 1 are calculated to compare the receiving stations. This is done independently for each transmitting station.
4. The Min/Mean/Max/Quartiles of all SNR differences of all transmitting stations are calculated for each time bucket.
Hope that helps
On 03.02.23 11:31, ON5KQ wrote:
|1 - 2 of 2|