OwenDuffy.net 


Making inferences from QRSS grabber observations

Given observation of propagation on a path using QRSS and a grabber, is it possible to infer the power required for a contact over that path in some other mode, for example CW?

This article describes a method using the commonly used QRSS grabber tool Spectrum Lab.

Firstly, we need to understand the criteria for minimum S/N ratio for a successful contact in the mode of interest. Let us assume that the receiver has a narrow filter, and that the Equivalent Noise Bandwidth (ENB) is 800Hz. In this case, Morse code copied by ear can be read by a competent operator under good conditions down to a S/N ratio of about -6dB. Let's assume S/N=-6dB is the minimum necessary S/N.

So, in summary we need S/N>-6dB in 800Hz ENB.

Path 1: VK2DVK-VK1OD

Fig 1:

Fig 1 shows a capture of signals in the 30m band QRSS segment using Spectrum Lab.

 Let's look closely at the signal at 10140040Hz. The signal is present at the instant of capture, so the spectrum plot on the right hand shows the relative output level of the signal as -60dB. The noise in adjacent spectrum buckets is -80dB, so S/N ratio is -60--80=20dB.

To make much use of this S/N ratio, we need to know the Effective Noise Bandwidth (ENB) of the buckets. The grabber page reports Spectrum Lab's ENB as 252Hz. (This ENB is specific to the grabber setup, and should be reported by the grabber... but usually isn't!)

We also need to know the transmitter power level used for the QRSS capture. In this case, the transmitter output power is taken to be 100mW.

So, the QRSS test tells us that 100mW of transmitter output power achieved a S/N ratio of 20dB in 252Hz ENB.

Table 1: Knowns
Bits S/N (dB) ENB (Hz) ENB (dBHz) Tx power (W) Tx power (dBW)
Requirement -4 800 29.0 ??? ???
QRSS observation 20 252 24.0 0.1 -10

Table 1 contains the known data, and calculated ENB in dB wrt 1Hz and Tx power in dB wrt 1W.

The CW receiver admits more noise than the QRSS receiver, and requires a different S/N ratio to that observed on the QRSS test, so the challenge is to calculate the Tx power that delivers the minimum S/N requirement.

The required Tx power is given by the formula TxPowerR=TxPowerQ+(SNRR-SNRQ)+(ENBR-ENBQ) where R and Q subscripts denote the requirement and QRSS, SNR is the S/N ratio, and the quantities are expressed in dB.

So TxPowerR=-10+(-4-20)+(29-24)=-29dBW. This corresponds to 1.26mW.

The observations above were made over a quite short path, and whilst the power might seem very low, communications is quite feasible at such low power under those conditions.

Path 2: VK2DVK-W4HBK

Fig 2:
 

Fig 2 shows a capture of the same signal a little later on W4HBK's logger over a distance of about 15,000km.

In this case, we take the transmitter power to be 1W. Received relative signal level is -71dB.

In this case, it is not possible to read the S/N ratio with any accuracy due in part to the setup of the grabber, and in part due to the very strong signal at 10140070Hz. However, let's take a guess at relative noise level being about -85dB, and therefore S/N=14dB. Whilst this grabber does not publish its ENB, it was some weeks before this observation 252Hz, so lets assume that remains the case.

Repeating the steps above for the new scenario...

Table 2: Knowns
Bits S/N (dB) ENB (Hz) ENB (dBHz) Tx power (W) Tx power (dBW)
Requirement -4 800 29.0 ??? ???
QRSS observation 14 252 24.0 1 0

Table 2 contains the known data, and calculated ENB in dB wrt 1Hz and Tx power in dB wrt 1W.

The required Tx power is given by the formula TxPowerR=0+(-4-14)+(29-24)=-13dBW. This corresponds to 50.1mW.

Path 3: K1DNR-W4HBK

The very strong signal at 10140070Hz is K1DNR in CT, USA.

In this case, we take the transmitter power to be 100mW. Received relative signal level is sometimes peaking at -50dB.

In this case, it is not possible to read the S/N ratio with any accuracy due in part to the setup of the grabber, and in part due to the very strong signal at 10140070Hz. However, let's take a guess at relative noise level being about -85dB, and therefore S/N=35. Whilst this grabber does not publish its ENB, it was some weeks before this observation 252Hz, so lets assume that remains the case.

Repeating the steps above for the new scenario...

Table 3: Knowns
Bits S/N (dB) ENB (Hz) ENB (dBHz) Tx power (W) Tx power (dBW)
Requirement -4 800 29.0 ??? ???
QRSS observation 35 252 24.0 0.1 -10

Table 3 contains the known data, and calculated ENB in dB wrt 1Hz and Tx power in dB wrt 1W.

The required Tx power is given by the formula TxPowerR=-10+(-4-35)+(29-24)=-44dBW. This corresponds to 39µW.

This might seem astonishingly low, and it is, but it is at the peak of the fade cycles which appear to be 10 to 15dB, so a 15dB increase takes the required Tx power to -29dBW or 1.26mW.

Recommendations for grabbers and transmitters

The QRSS world seems mostly about people doing their own thing, about creating as many mode variants as possible, more graphically oriented than serious propagation measurement, but it can be used for serious propagation measurements and is probably better suited than some competitive technologies.

The spectral dispersion seen in the above captures of long distance paths suggests that there is little benefit in seeking to reduce receiver bandwidth further, and that QRSS6 with ENB around 0.25Hz is a good 'standard' combination.

For the most benefit in making inferences, grabbers should:

For the most benefit in making inferences, transmitters should:

Receive level setup

The AGC in most SSB receivers with around 2.5kHz ENB will limit the audio output power rising more than about 25-30dB above that of the receiver noise alone with a 50Ω load on its input.

When an input signal exceeds the noise floor by more than that 25-30dB, the audio output power due to the signal will not increase further with increases in signal (that is the purpose of the AGC), but the noise will be 'pushed down'.

It is a very strong signal that should be used to calibrate the receive level, ie the level should be set so that a very strong signal is almost at the top of scale.

On quite strong signals, the total noise (ie internal and external) could easily be more than 40dB below the signal. The scale graduations should be set to cover a range at least 40dB below the strong signal to ensure that with moderately strong signals, the noise is still on scale and measurements of S/N can still be made of the signals being received.

The Spectrum Lab grabber at VK1OD is usually setup to that a very strong signal is 10dB below top of scale, and the scale covers 80dB. The 'soundcard' is a Signalink USB, and one of the failings is that it has a rather touchy Rx level control, so setting is approximate.

Links / References

Changes

Version Date Description
1.01 12/09/2012 Initial.
1.02    
1.03    

© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.