Using an attenuator for NoiseLo/NoiseHi in NFM

A common method of making Noise Figure measurements of a receiver is to use a noise generator of known noise power. The output power of the DUT is measured with the generator off (NoiseLo) and on (NoiseHi), a Y factor calculated, and from that Noise Figure is calculated.

Continue reading Using an attenuator for NoiseLo/NoiseHi in NFM

ARRL Test Procedures Manual (Rev L) – Noise Figure calculation

(Allison et al 2011) detail the method used by the ARRL in their test reports on equipment.

Effectively they calculate NF=-174+27-MDS where MDS is measured in the CW mode using the 500 Hz, or closest available IF filter (or audio filters where IF filters are not available).!–more–>

One flaw in this method is that the factor 27dB in the NF formula implies that the Equivalent Rectangular Bandwidth (ERB) of the receiver when measuring MDS is exactly 500Hz. More correctly, the formula should be NF=-174+10*log(ERB)-MDS. The error could be significant, especially with the closest available…rovision in the test requirement.

Clip 159

For example, above is the measured IF response of a TS-2000 set to 500Hz bandwidth. As filter responses go, it is very good, having quite a flat region, better than most crystal filters used for CW, but the ERB is actually 430Hz, that is 0.66dB less noise than a 500Hz ERB filter and any NF based on 500Hz is low by 0.66dB.

The red plot is that of an idealised filter of the same ERB.

Clip 157The plot above is for an R-5000 with 500Hz crystal filter. ERB is 446Hz, 0.5dB lower than the nominal 500Hz bandwidth which would lead to an error of 0.50dB using the ARRL’s method for estimating NF.

A more fundamental flaw is  uncertainty in measured MDS leads to a relatively large uncertainty in NF when NF is small. The test reports do not specify the uncertainty of MDS, it is unlikely (Duffy 2007), (Duffy 2007b) that with receiver bandwidth of 500Hz (as specified for the MDS test) and the HP339A instrument used, that uncertainty to 95% confidence level is as low as ±0.5dB probably closer to ±1dB. The method used is just not suitable to low noise receivers.

Giving NF rounded to 1dB is not very informative for receivers with NF below 5dB, and fairly useless at 2dB as in (Wilson 2012).

References

  • Allison, B; Tracy, , M; Gruber, M. 2011. Test Procedures Manual Rev L. ARRL Newington.
  • Duffy, O. 2007. Uncertainty of the noise sampling process. https://www.owenduffy.net/files/NoiseMeasurementUncertainty.pdf .
  • ———. 2007b. Noise measurement uncertainty calculator. VK1OD.net (offline).
  • ———. 2014. ARRL Test Procedures Manual (Rev L) – Noise Floor test. https://owenduffy.net/blog/?p=1165 (accessed 15/03/2014).
  • Wilson, M. 2012 ICOM IC-9100
    MF/HF/VHF/UHF Transceiver In QST Apr 2012.

ATU voltage verification

I described a method for designing antenna systems to avoid excessive voltages in baluns and ATUs at (Duffy 2011) .

This article reports post implementation measurements of an antenna system designed using that method and using a G5RV Inverted V with tuned feeder and ATR-30 ATU with integral 1:1 current balun. The tuned feeder is a home-made line section of 2mm diameter copper conductors spaced 50mm, and 9m in length. An additional 0.5m of 135Ω line connects from the antenna entrance panel to the ATU.

Continue reading ATU voltage verification

An inexpensive medium power tuner current balun for HF using Jaycar parts

This is a project to design and build a Guanella 1:1 (current) balun suited for up to 100W on HF with wire antennas and an ATU.

For use with a tuner, the most important design criteria are:

  • high voltage withstand;
  • high common mode impedance;
  • power handling.

Continue reading An inexpensive medium power tuner current balun for HF using Jaycar parts

Design / build project: Guanella 1:1 ‘tuner balun for HF’ – #3

Third part in the series..

 

Common mode current measurement

Direct measurement of common mode current in an antenna system is the best indicator or whether there is a common mode current problem.

In Common mode current and coaxial feed lines, I mentioned that common mode current is easily measured.

80mIcm

Continue reading Design / build project: Guanella 1:1 ‘tuner balun for HF’ – #3

Windowed ladder line loss – G3TXQ #2

I tried to glean some useful information from G3TXQ’s measurements of windowed ladder line loss at Windowed ladder line loss – G3TXQ.

In reviewing his article today (05/02/14), there is new information on a further series of measurements of the same line.

450_mll

The shape and position of the two lines does not reconcile with the formulas stated, so I digitised the data points and analysed the data set to try to find the most appropriate model for the reported measurements. Note that although the chart above is in imperial units, my work is usually in ISO metric units, and usually basic units.

The digitised data points were converted to loss in dB/m, and fitted to the model MLL=k0+k1*f^0.5+k2*f using regression techniques. Note that the digitisation process introduces some noise, but it is estimated to be small compared to the noise in the underlying measurement data.

The coefficients k0, k1, k2 were reviewed to test that there was sufficient data to support the hypothesis that they were not zero, and all three passed that test, the standard error of the coefficient was significantly less than the coefficient. Note that k0 is not derived from a DC measurement of resistance as done by some modellers, but from the measurement data over the range of 3.6 to 48MHz in this case, and extrapolation beyond that frequency range increases uncertainty.

Clip 141

The above chart shows G3TXQ’s measurements as digitised from his published graph, and it shows the components of loss indicated from the model I built (the k0 component is allocated as conductor loss).

The “G3TXQ model” line is equivalent to his MLL=0.063+0.063*f^0.5 dB/100′ converted to dB/m, and as you can see it is not a good fit to the measurement data points, nor does MLL=0.063+0,063*f^0.5 dB/100′ reconcile with the blue line on G3TXQ’s chart earlier in this article.

G3TXQ’s measurement points (as digitised) are quite a good fit to the model MLL=0.001456+1.499e-6*f+5.631e-11*f dB/m where f is in Hz, and provide a good predictor of MLL over 3.6 to 48MHz.

 

Foundation manual on power measurement

My article Foundation watts explained triggered some discussion on the thorny issue of compliance with power limits of the LCD.

One correspondent was confident that the Foundation candidates are properly trained, which leads to examining the training materials.

SWRMeter

Continue reading Foundation manual on power measurement

An inexpensive current limiter for flashing and initial testing of ESCs – Mk I

There is a risk of damage when flashing ESCs. It accrues from the fact that ESCs have a three legged H bridge and if a high and low FET are turned on simultaneously, damaging currents may flow. In fact, this can be an issue if the FETs are on together for just microseconds on each PWM cycle. Loading the wrong hex module is a recipe for disaster, it may turn on FETs in an unexpected way.

Continue reading An inexpensive current limiter for flashing and initial testing of ESCs – Mk I