Who decodes more WSPR, WSPRDaemon or KiwiSDR?"

I have read that WSPRDaemon decodes more WSPR spots than KiwiSDR, but I couldn't find quantitative data proving it, so I set up a small test.

I configured three WSPR decoding channels (40m, 20m, and 10m) on the KiwiSDR2 (rx_sign = EA5JTTSWL, version = 1.4A Kiwi) and another three on WSPRDaemon (rx_sign = EA5JTT, version = WD_3.4.5-2).

WSPRDaemon is running on a Raspberry Pi 3B  (4-core running Debian Linux) with no dedicated resources shared with other applications.

I started both WSPR decoding tasks and let them run for a few hours (ultimately aiming for a 24-hour run to analyze any changes over time relative to signal propagation and the volume of spots emitted). Afterward, I compared the results using the following query on WSPR Rocks

SELECT
    toDate(time) AS fecha,
    toHour(time) AS hora,
    countIf(band = 7 AND rx_sign = 'EA5JTT')     AS WSPRDaemon_40,
    countIf(band = 7 AND rx_sign = 'EA5JTTSWL')  AS KiwiSDR_40,
    countIf(band = 14 AND rx_sign = 'EA5JTT')    AS WSPRDaemon_20,
    countIf(band = 14 AND rx_sign = 'EA5JTTSWL') AS KiwiSDR_20,
    countIf(band = 28 AND rx_sign = 'EA5JTT')    AS WSPRDaemon_10,
    countIf(band = 28 AND rx_sign = 'EA5JTTSWL') AS KiwiSDR_10
FROM wspr.rx
WHERE band IN (7, 14, 28)
  AND rx_sign IN ('EA5JTT', 'EA5JTTSWL')
  AND time >= toDateTime('2026-08-26 03:00:00')
  AND time <  now()
GROUP BY fecha, hora
ORDER BY fecha ASC, hora ASC;


Results

Over a 24-hour period, a total of 11,293 WSPR spots were received (7,010 by WSPRDaemon and 4,283 by KiwiSDR) across the 40m, 20m, and 10m bands.

Decoding the exact same signal, WSPRDaemon decoded 64% more WSPR spots than KiwiSDR overall.

Furthermore, it was observed that this percentage gap grows wider on higher-activity bands: 

  • 20m: WSPRDaemon spots were 61% higher (6,302 vs 3,771)
  • 40m: WSPRDaemon spots were 44% higher (561 vs 390)
  • 10m: WSPRDaemon spots were 20% higher (147 vs 122)
One of the primary reasons for this performance difference is that WSPRDaemon is capable of decoding at a lower SNR. Over the 24-hour test, the minimum observed SNR for KiwiSDR was -30 dB, while for WSPRDaemon it was -33 dBa 3 dB advantage, which translates to decoding signals at half the power level.

It was also noted that even at higher SNR levels, KiwiSDR misses more spots than WSPRDaemon during periods of peak traffic, which is expected given that WSPRDaemon offloads processing to an additional dedicated computer.

In conclusion, WSPRDaemon outperforms KiwiSDR in raw spot reception (delivering 64% more decodes). However, this comes at the expense of requiring an extra device (a Raspberry Pi 3B in this test), leading to higher upfront costs, increased power consumption, added installation/maintenance complexity, and a higher potential point of failure—all factors that should be weighed for individual setups.


Spot decoder comparison for a random 10-minute interval


Spots decoded by both programs from 08:00 to 09:00 CET on 7MHz (4 vs 1)

Since the results are stored and publicly accessible from WSPR Rocks, feel free to conduct your own studies of SNR, spot density, etc.


A picture is worth a thousand words.

Spots received in 1 hour by EA5JTTSWL (KiwiSDR):
51 different stations, the furthest being 5,820 km away

Spots received at the same time by EA5JTT (WSPRDaemon):
69 different stations (35% more), the furthest being 17,075 km away (293% further away)



References

Important Disclaimer

The modifications, configurations, and procedures described on this site may involve technical, legal, or safety risks. The author assumes no responsibility for equipment malfunction, permanent damage, voided warranties, or potential legal violations resulting from the use of this information.

The reader assumes full responsibility for any action taken based on the content of this blog.

© 2026 EA5JTT. All rights reserved. Reproduction in whole or in part without express written permission from the author is prohibited.

No hay comentarios:

Publicar un comentario

Xiegu X6200: R1CBU

Dado que después de 9 meses Xiegu sigue sin resolver los problemas de la versión 1.0.7 que ya arrastraba de versiones anteriores  que he an...