06 April 2014

HowTo Calculate Power Using WSPR

Thanks to the hints of Bert PA1B I had to do a revision of my power estimations.

In the last days I was kind of shaken. As told before I wanted to compare my actual data, which I received from my WSPR transmissions using 100 mW with propagation prediction programs. In these programs like VOACAP (more on that later) you have to specify the location, both antennas and the power level of the transmitter. But which figure should I use?

It took me a while, for I didn't know exactly what I was searching for, but finally I found the information I needed on the page of KF6HI SNR . You will find additional information on the page of Bert Power with WSPR, Roger G3XBM WSPR versus CW and in an article from Joe Taylor himself:
K1JT Quest for Optimum Coding and Modulation Schemes for EME  Page 8.

WSPR signals could be decoded down to -32 dB under the noise level when used with a bandwidth of 2500 Hz. If you want to read a CW signal and you have only a filter of 2500 Hz you can read a DX CW signal when the SNR is -15 dB. For a casual CW-QSO you need a SNR of about  -5 dB. And when you want to listen to the voice of the other OM the SNR ratio must be +6 dB. 
But who would listen to CW signals with a SSB filter. I use a 100 Hz filter. Now the relations change. (For the mathematics, please have a look at the pages of KF6HI. (Well written and comprehensible.))
With that 100 Hz filter
  • WSPR could be decoded down to -18 dB
  • CW (DX) up to -1 dB
  • Normal CW +9 dB
  • SSB - you are kidding - with a 100 Hz filter
But the discussion about what good CW operators could decode by brain are depending on the the estimation of the individual capabilities. Joe Taylor says that the threshold S/N is -18 dB compared to WSPR -29 dB, Roger (G3XBM) estimates for a most operators it is -15 dB and Bert (PA1B) comes to the conlusion that a save bet is -13 dB. And I will follow his advice!

So if I use the Ultimate 3, a 3 dB attenuator and set my signal to WSPR I can measure 100 mW at the entrance of my antenna, which has no significant gain. 
  • I have 20 dBm from my TX and 
  • add the difference to the  CW DX-signal decoded by the CW-operators brain = 13 dB
  • So I get 33 dBm = 2 Watt.
  • If I add the difference on a normal +9 dB CW-signal  = 27 dB.
  • I get 47 dBm = 50 Watt.
Now I can choose which power I will fill into the prediction programs. But there is another thought. In CW I use one tone or better: signal on and signal off. In WSPR we send a message with 28 bits for the callsign, 15 bits for locator and 7 for the power level. This message is packed into a datastructure and using a continuous phase 4-FSK. So it is more like a SSB signal. I will set for a compromise and use 10 Watt when trying to predict my signals chances to arrive at another location. These programs calculate with signals over the noise level.

But it still amazes me. If you use 1 Watt on WSPR = 30 dBm and add 13 dB (27 dB) difference, I would need to use  43 dBm = 20 Watt (57 dBm = 500 Watt)  in CW to reach the other station.

1 Watt (WSPR) 30 dBm + 13 dB (CW DX) =  43 dBm =  20 Watt (CW)
1 Watt (WSPR) 30 dBm + 27 dB (CW DX) =  57 dBm =~ 500 Watt (CW) 

 And some people think that 1 Watt is not sufficient.  At QRP-Levels we use 5 Watt.

5 Watt (WSPR) 37 dBm + 13 dB (CW DX) =  50 dBm = 100 Watt (CW)
5 Watt (WSPR) 37 dBm + 27 dB (CW DX) =  64 dBm =~ 2000 Watt (CW) 

I didn't know that I am a big gun for sometimes I used that power level too, but in future I will stay under 1 Watt. That's more than enough ... at last for me.

I have to ask Hans Summers if he is willing to implement JT9-30 in the Ultimate 4. My power level would be even better:

100 mW (JT9-30) 20 dBm + 33 dB (CW) =  53 dBm = 200 Watt (CW equivalent)

I am still dreaming.

Stay Tuned!