comparing SNR dB reports between JTDX .159 and WSJT

Neil G0JHC
Users

Posts: 23

comparing SNR dB reports between JTDX .159 and WSJT Mar 4, 2023 7:04:10 GMT

Quote

Post by Neil G0JHC on Mar 4, 2023 7:04:10 GMT

Can someone please explain how JTDX calculates the SNR (dB) report compared with WSJT?
Recently I have been comparing the SNR report (dB) in JTDX against WSJT-X.
Putting to one side the fact that JTDX decodes far more signals this post is regarding stronger signals and the reporting.Anything that is S5 and above on the “meter”.
An example on 40m over a 5 min period the meter is stuck at S9+10.
JTDX rarely gives me a dB report (SNR) above 0 to +02
WSJT gives me many in the +10 to +20 range.
SNR (dB) reports of weaker stations seem similar in that -20 to -10 , range , around 5dB difference in the -10 to 0 range. But can be wild in WSJT sometime 10-20db different when the signal is loud.
A final example on 6m a local who is 59+10 on S meter, shows as +4 on JTDX and +18 on WSJT.
I appreciate the reports are SNR based rather than signal “strength” , and at the end of the day it doesn’t matter what the dB column reads so long as you are decoding well (I am).

I am just interested in the way both bits of software interoperate this. These reports are with no noise, S meter sits at zero all bands, Im in the middle of fields.

Arvo ES1JA
Administrator

Posts: 429

comparing SNR dB reports between JTDX .159 and WSJT Mar 4, 2023 7:28:18 GMT Neil G0JHC and Don - KS0USA like this

Quote

Post by Arvo ES1JA on Mar 4, 2023 7:28:18 GMT

JTDX calculates SNR differently, It does not depend on signal strength (S meter value) to much, it is pure signal/noise ratio. So comparing to WSJT-X JTDX reports are much more conservative.

Yoshiharu JP1LRT
Users

Posts: 13

comparing SNR dB reports between JTDX .159 and WSJT Mar 7, 2023 0:10:29 GMT Neil G0JHC likes this

Quote

Post by Yoshiharu JP1LRT on Mar 7, 2023 0:10:29 GMT

Igor -san said as below.

There are several reasons we had withdrawn WSJT-X approach to calculate SNR of FT8 signals:

- WSJT-X algorithm completely breaking SNR calculation if narrow filter is selected in receiver while keeping wide audio bandwidth in SW waterfall.

- WSJT-X algorithm is based on the code we had to delete to improve FT8 decoder sensitivity.

- WSJT-X approach to evaluate noise in wide bandwidth would increase number of CPU cycles and FT8 interval decoding time if implemented in JTDX multithreaded FT8 decoder where we split whole spectrum to several threads.

For this reasons we did develop own algorithm of SNR calculation in FT8 signal bandwidth, 50Hz, that is why in overcrowded band conditions one can often observe SNR different to WSJT-X.

You can find our SNR algorithm in ft8b.f90 source code, it had several changes\patches before and IMD it is in good shape now, while any possible improvement would be appreciated.

Due to the multipath decoding approach with signal subtraction It is difficult to say which algorithm has better accuracy, as any overlapped signal will get different SNR value if decoded before or after subtraction of any QRM-ing signal.

There is the only difference: JTDX evaluates noise in the 50Hz bandwidth of FT8 signal, not whole waterfall bandwidth. Any signal compression(if signal is not fit in the receiver's/RX path's dynamic range) or IMD in receiver will affect SNR in JTDX, while WSJT-X will not show user has issues in the receiver.