This page contains details about the three transmit noise component measurements. The illustrations show how each transceiver varies in terms of the noise roll-off patterns in each radio as well as how individual transceivers vary from one another. The tools below include:
- a sortable table of each noise measurement plus the Index
- a visual distribution of the rank order and metric difference between transceivers on the Index (so the viewer doesn’t get ‘out ranked’)
- scatter plot matrix of the three measurements plus the Index
- a 3D scatter plot of each individual noise measurement to facilitate the viewer’s detailed interrogation of the bench test data
We do not (yet) know how humans discern small variations in total composite transmission noise. As a statistician, I recommend using the “standard deviation” metric to identify bench test differences that are most likely to be worth noting. But viewers themselves vary in terms of what differences are worthy of spending their money on in the marketplace!
The table below is sortable on each of the columns. They are the transmit composite components, the composite index, and related statistics. These include the average noise figure across the three offset bandwidths, the standard deviation for each rig across the three measurements, and the coefficient of variation of the noise-by-offset measurements. The CV illustrates the variation irrespective of the average in the dBc metric of the measurement scale. To illustrate, the Flex 6600 has the highest CV or variation in transmit noise while the Yaesu FTDx3000 has the lowest.
The (non-interactive) ordered dot plot below ranks each transceiver on the basis of the Transmit Composite Noise Index in the above table. A blue vertical line shows the median value on the Index. Note that the lower the index, the better since lower noise is preferred. The viewer should trace each transceiver from the left-hand side along the gray line to the red dot for the Index value for that radio.
To further illustrate the individual components of the Composite Noise Index, the scatter plot matrix below shows that the transmit noise at each offset tends to be imperfectly aligned along each offset frequency. That is, they correlate across transceivers and with the Index highly but not perfectly. This indicates that the weighted index (using the first principal component) represents the full range of transmit noise in each transceiver well.
A parallel coordinates plot below gives an illustration of just how these transmit noise measures vary WITHIN each transceiver. The leading radio from each manufacturer has been highlight in blue lines. Read this chart from left-to-right. Follow the Manufacturer to Rig to each transmit noise value at 10, 20 and 100 kHz. The final column is the Index score which has a median of 100 and a standard deviation of 15. For example, the difference between the Anan Apache 7000 DLE at 72.6 is just over one standard deviation lower than the Elecraft K4D at 88.2.
In short, the three measurements of transmit composite noise at 10, 20 and 100 kHz offsets show variation both between radios and within individual ones. If one prefers less “close in” noise, the Apache 7000 DLE has the lowest score, just edging out the Flex 6700 by just a few points. In contrast, the Icom 7851 has higher noise at each offset than the remaining set of flagship radios. This parallel coordinates chart illustrates these patterns, unlike simpler data visualizations.
Below is a 3D scatter plot allowing interactive examination of each noise measurement. Hovering over the data point shows the transceiver. On the upper right portion, there is a set of clickable steering tools for the viewer to maneuver this chart to suit.