Calibrating the Elecraft N-GEN

The Elecraft N-GEN is a low cost noise source which is quite suited to many applications, more so if the Excess Noise Ratio (ENR) is known.

ENR is a commonly used property to describe the noise power density of a source, it is calculated as ENR=10*log(Tne-T0)/T0 dB where Tne is the effective noise temperature and T0 is 290K.

This article describes a calibration procedure. Note that the calibration is specific to the device and cannot be applied to another N-GEN.

Above is a screenshot of the Spectrum Analyser scan. A text file of the frequency,power pairs is saved for input to a spreadsheet to calculate ENR vs frequency. Continue reading Calibrating the Elecraft N-GEN

Calculation of received noise power given ENB and ENR

At Effective noise bandwidth – IC-7300 SSB Rx Filter2 (2400Hz) the ENB of the receiver was measured at 2088Hz. This article goes on to calculate the power received from a Elecraft N-Gen noise source which has been measured to have Excess Noise Ratio (ENR) at 10.1MHz of 48.2dB.

Lets input the data to Field strength / receive power converter and find the received power.

The measurement is made is preamp off (so that the S meter is more realistic), and the supplied NoiseFigure is a guess… but the noise source is so strong (being some 30+dB above the receiver internal noise) that the result is barely sensitive to that assumption.

The calculator returns many results, we are interested in just the receive power in dBm. The results follow. Continue reading Calculation of received noise power given ENB and ENR

Effective noise bandwidth – IC-7300 SSB Rx Filter2 (2400Hz)

For a lot of experiments, knowledge of the Effective Noise Bandwidth (ENB) of a receiver is necessary. The ENB is the bandwidth of an ideal rectangular filter with the same gain as some reference frequency, 1kHz is usually specified for SSB telephony receiver sensitivity measurement.

Though filters are often specified in terms of bandwidth at x dB down, that metric is of relatively little value, the x is often 6dB but not always, the filters depart significantly from ideal or even common response.

In brief, a white noise source is connected to the receiver input, Filter2 (nominal 2400Hz bandwidth soft response) selected and set to standard PBT, and the audio output captured on a PC based audio spectrum analyser, Spectrogram 16 in this case.

Spectrogram is set to integrate over 30s to average the variations due to the noise excitation. The resulting graph and text spectrum log are saved.

The method is explained in detail at Measure IF Bandwidth.

Above is the spectrum plots, as receivers go this is relatively flat, lacking the usual tapering off above 1kHz (a technique to cheat on sensitivity specs).
Continue reading Effective noise bandwidth – IC-7300 SSB Rx Filter2 (2400Hz)

IC-7300 S-meter calibration accuracy

This article documents measurement of the calibration of an IC-7300 S-meter in SSB mode using a continuous sine wave at 1kHz tone frequency.


There has been a long standing convention that S-meters are calibrated for 50μV in 50Ω to be S9, and S-points laid out at 6dB per S-point. IARU Region 1 formalised this with Technical Recommendation R.1 which defines S9 for the HF bands to be a receiver input power of -73 dBm (equivalent to 50μV in 50Ω).

IC-7300 measurement

A test was conducted where a Standard Signal Generator was connected to the receiver and slowly increased from -125dBm in steps of 1dB and the point at which the S-meter display segments lit was noted.

Above is a chart of the error between the S meter indication and the value per IARU Region 1 Technical Recommendation R.1.  Continue reading IC-7300 S-meter calibration accuracy

Kenwood R5000 thermals

I have a Kenwood R5000 that is now 30+years old and warrants a check of its health.

R5000s are infamous for VCO problems, the early production used ‘yella glue’ to stabilise the VCO components and that decomposed into corrosive components that damage the electronic parts. Repair is not usually economically rational.

This is one of the later model R5000s that used the hard white adhesive which has remained stable.

The R5000 is built on phenolic PCB and operates at relatively high temperature for a simple receiver, reflecting the power consumption of synthesisers of the 1980s.


Above, the case temperature is up to 20° above ambient over the power transformer (upper right of pic).  Continue reading Kenwood R5000 thermals

Polarisation of man made noise – an 80m case

Polarisation of man made noise discussed an explanation for the common observation more ambient noise is captured by a vertically polarised antenna than for a horizontally polarised antenna.

This article documents an analysis of a case on 3.6MHz and is to be read in the context of Polarisation of man made noise.


Remembering that P.368-9 publishes a set of graphs like the one above, and that they show that ground wave attenuation is dependent on distance, soil type and frequency.

Though ground wave attenuation is lower on 80m than 40m, the horizontal antenna used in the example is at a fixed height, so it is electrically lower on 80m which increases horizontal attenuation significantly. Continue reading Polarisation of man made noise – an 80m case

Polarisation of man made noise

Ham lore has it that man made noise on lower HF is radiated predominantly vertically polarised, this is offered and accepted by hams without explanation.

It can be shown by simple observation that the ambient noise level on lower HF is quite different in business or commercial areas, residential areas, and rural areas (ITU-R P.372-12). Not only is there a significant difference, the change happens quite rapidly with distance which suggests there is a dominant component (man made noise) and that the propagation path is a very local one (ground wave).

If you look around a typical residential neighborhood where hams might establish stations, the most obvious conductors that might carry and radiate noise currents from noise generators like appliances, leaky insulators etc are aerial power lines… which are usually closer to horizontal orientation (with horizontal E field) than vertical which seems inconsistent with the common observation that vertically polarised receiving antennas tend to capture more man made noise power than horizontal ones.

This article proposes a mechanism that may explain the apparent inconsistency between noise radiators and noise receivers.

Though this explanation is based on experience, the quantitative analysis here depends on interpretation of Recommendation ITU-R P.368-9 (2/2007) Ground-wave propagation curves for frequencies between 10 kHz and 30 MHz.

screenshot-24_10_16-22_00_39Whilst P.368-9 publishes a set of graphs like the one above for a limited set of grounds, ITU-R also publishes the program (GRWAVE.EXE) which can be used to calculate values for the user’s choice of ground and that is what was used for this article. The graph above is for a vertical monopole over ground with 1000W radiated, the antenna has directivity of 3, and the dashed line (inverse distance curve) is the field strength for a lossless ground (PEC). This can be verified with a spot calculation at 1km. Continue reading Polarisation of man made noise

Expected ambient noise – in practice

This posts shows a measurement of ambient noise and comparison with the data given at Expected ambient noise and its more detailed references.

The test scenario is my 40m station, a G5RV inverted V dipole with tuned feeders, a balun and ATR-30 ATU. Antenna system losses are less than 1dB.

Clip 200

The chart above gives a range for expected ambient noise at 40m.


Above is a screen shot from a spectrum analyser measuring power in 1kHz bandwidth from 7.0 to 7.1MHz. The band is mostly unoccupied, and the mean noise power is about -99dBm, it would be 3dB higher in 2KHz bandwidth (ie -96dBm). Continue reading Expected ambient noise – in practice

Expected ambient noise

One of the casualties of the cessation of was an article on expected ambient noise.

Clip 200

The original work was based on ITU-R P.372-8 which has been updated to -10 and now -12, but the updates do not alter the basis for the original article.

Since the work was a reference cited on my FSM pages, it has been updated and copied to Expected ambient noise level. The graphics and tables in the article and the PDF file all refer to ITU-R P.372-8 but remain correct wrt ITU-R P.372-12 (2015).

Reconciling my #52 choke design tool with G3TXQ’s measurements

A correspondent wrote with concern of the apparent difference between graphs produced by my #52 choke design tool with a graph published by G3TXQ of his measurement of 11t on a pair of stacked FT240-52 cores.

I published a note earlier about my concerns with a similar graph by G3TXQ compared to the Fairrite datasheet, and he reviewed the data, found the error and published a corrected graph.


The corrected graph above might at first glance appear different to my model’s graphs, and the first obvious difference is that G3TXQ uses a log Y scale (which is less common). The effect of the log scale is to compress the variation and give the illusion perhaps that in comparison with other plots, this balun has a broader response.

Screenshot - 09_02_16 , 18_29_42

To compare the two, I have roughly digitised G3TXQ’s graph above and plotted the data over that from my own model (with linear Y scale). Continue reading Reconciling my #52 choke design tool with G3TXQ’s measurements