The LEDs are about the same power consumption as the CFLs they replace, the hope was that they had a longer life (you have seen the claims of 100,000 hours).

Two years after cutover, it is time to review their performance.

Of some 25 11W LEDs installed, most would not be used for an hour a month, but 11 are used every day for an average of around 4 hours per day.

The pic above shows the failures of two years operation, 5 of 11 have failed. The average life of the lamps that failed is less than 3000 hours. probably in the region of 2000 hours, certainly a long way short of the claims of 50,000 to 100,000 hours.

Annual power consumption of the 11 lamps is 11*11*4*365/1000=177kWh which at current rates costs about $40. Total running costs then are $40 for electricity and about $35 for 5 replacement lamps ignoring costs of procurement, testing and installation. Repairs and maintenance has doubled the cost of operation if labour costs are ignored.

One has to question the energy efficiency of the things in the wider sense, the rubbish in the picture is mostly aluminium, one of the most carbon intensive metals to manufacture.

The embodied energy is conveniently ignored in almost all cost benefit analyses.

Are the failed lamps recyclable with expensive dismantling? Probably not.

Beyond total failure, most of the remaining lamps have changed colour as the yellow phosphor is exhausted progressively, and with just a few thousand hours of service original warm white lamps are distinctly blue alongside their new replacements.

The colour change will probably result in replacement of otherwise working lamps because their colour has become unacceptably blue, and that is likely to happen way short of 100,000 hours.

Another green hoax where the total cost of ownership and real world savings are not part of the justification!

]]>There are many possible explanations for why an antenna captures less noise power than another, this article discusses the distribution of electric and magnetic fields (E and H) very near to a radiator, and the power captured by antennas that respond more to E or H fields.

Electromagnetic radiation consists of both and E field and a H field, and they are in the ratio of *η*0=µ0*c0Ω, the so-called impedance of free space, often approximated to 120πΩ or 377Ω. Close to a radiator there are components of E and H additional to the radiation components, the ratio of E/H is not simply 377Ω.

Fig 1 shows the magnitude of the ratio E/H near a quarter wave vertical over average ground at 3.6MHz. |E/H| depends on location near the antenna, and with increasing distance it converges on 377Ω.

Whilst the chart is of an intentional radiator, the effect occurs around any radiator, be it house wiring or antenna. Near to high current parts of a radiator, H field is relatively high, and near charge parts of a radiator, E field is relatively high.

Figure 2 shows the error in dB calculating power flux density as S=E^2/*η*0. The error in calculating flux density as S=H^2**η*0 would be the negative of the value in the chart.

An alternative way of interpreting the chart is that an antenna responding principally to H field (eg a small loop) at ground level 5m from the radiator would capture 6dB more power than you would expect given the actual power density at that point.

Fig 3 is a map of power flux density near the antenna for 1W input to the antenna. Taking a check point, at 46.5m out and 10m height, S=-40dBW/m^2.

Fig 4 shows the expected power density ignoring ground attenuation, it is about 0.4dB higher than the NEC model (which does capture ground attenuation), so the model reconciles.

Now in the real world, antennas aren’t tiny and E or H dominated, nor are radiators (intentional or unintentional) tiny, they typically sustain standing waves and in regions of current antinodes (maxima) H field will be higher, and in charge antinodes (maxima) E field will be higher.

Whilst it might be observed that some antenna in some particular location seems less sensitive to noise from local sources at some frequency, it is unlikely that the effect can be exploited generally (different locations, radiators, frequencies).

Fig 3 shows that power flux density decays very quickly near the radiator, a small increase in distance can result in a large reduction in power flux density and captured power from that source.

The simple facts are:

- some antennas may be dominated by E or H fields, more so for small antennas;
- the ratio E/H is fixed in the far field (more than say a half wavelength from a radiator);
- near field of radiators may contain a higher or lower E/H ratio at some locations on some frequencies;
- separating receive antennas from noise radiators will reduce local noise pickup;
- effective measures to eliminate feed line common mode current is a measure towards separating receive antennas from noise radiators.

This article is a revision to take account of recently updated information published by LNR filling in some of the gaps in their original page. It is encouraging to see better product descriptions and measurement data.

The antenna is described at (LNR Precision 2016).

The loop itself appears to be 3/8 Heliax or similar (nominally 9.5mm outer conductor diameter) in a rough circle of 45″ (1.143m) diameter.

Little information is given of the internals, but the promotional material gives a VSWR curve for a matched antenna at 7.065MHz. To their credit, they give the height above ground and ground type for their tests.

The VSWR=3 bandwidth scaled from the graph is 18kHz.

If we assume for a moment that the VSWR measurement was captured at a substantial height above ground, its behaviour approaches that of the antenna in free space. Taking the assumption that the published curve is similar to the antenna in free space, we can estimate the gain and efficiency based on earlier assumptions.

The assumed values and published VSWR curve indicate an antenna system half power bandwidth of 5.6kHz and Q of 453 which implies efficiency of 2.8%.

The actual value for radiation resistance is likely to be with -50-+100% of the free space value used, and that rolls up as an uncertainty of +/-3dB in the calculated efficiency and gain.

For example, Duffy (2014b) analyses the Chameleon M loop. Based on the larger diameter of the conductor (25.4mm) and smaller loop perimeter (2.873m), we might expect it to have a loop inductance of around 61% that of the LNR Precision loop. Lower inductance is one of the factors of the equation for Q, and for half power bandwidth. The smaller loop of larger diameter conductor will tend to have lower Q, wider bandwidth, all else equal. You cannot simply infer relative efficiencies of these antennas based on half power bandwidth alone… any competent antenna designer know that.

Taking (Milazzo 2014) measurements for 7.1MHz, MinVSWR=1.345, Z at minVSWR=57-j14, BW at VSWR=3 35kHz (from the graph), we can explore the likely efficiency if we assume that radiation resistance was close to that of free space,

Above is the result of a more complete estimate of efficiency that includes not only half power bandwidth, but loop perimeter and conductor diameter. The estimated efficiency at 0.88% is significantly lower than a similar estimate for the LNR Precision loop, some 5dB lower but measurement of field strength from both loops in a relevant setting is the best way to compare performance.

The actual value for radiation resistance is likely to be with -50-+100% of the free space value used, and that rolls up as an uncertainty of +/-3dB in the calculated efficiency and gain.

One could be forgiven for thinking that the market for these type of antennas are hams who aren’t much concerned with efficiency, but there remains a small number of hams interested in the technology of radiocommunications.

Based on measurements published by both suppliers, it is likely that the LNR Precision loop is significantly better than one of its competitors, the Chameleon M-Loop, but the difference between them cannot simply be inferred from the bandwidth measurements of each.

The calculator Calculate small transmitting loop gain from bandwidth measurement which captures most of the relevant factors, properly applied may provide the basis of a useful comparison between different implementations.

- Chameleon. Oct 2014a. CHA M-LOOP Analysis. http://chameleonantenna.com/resources/CHA_M-LOOP_SWR_ANALYSIS.pdf(accessed 30/11/2014).
- Chameleon. 2014b. CHA M-LOOP. http://chameleonantenna.com/PORTABLE%20ANTENNA/CHA%20M-LOOP/Chameleon%20M-LOOP.html (accessed 30/11/2014).
- Duffy, O. May 2014. Small transmitting loop calculators – a comparison. http://owenduffy.net/blog/?p=1693.
- Duffy, O. Nov 2014b. Chameleon CHA M-LOOP. http://owenduffy.net/blog/?p=2999.
- Efficiency and gain of Small Transmitting Loops (STL)
- LNR Precision. 2016, W4OP Small Transmit/Receive Loop. http://www.lnrprecision.com/loop-antennas/
- Milazzo, C. May 2015. Chameleon CHA F-Loop Antenna Parameters: 5-30 MHz. http://www.qsl.net/kp4md/chafloop.htm.
- Small transmitting loops
- Yates, S. April 2009. Small Magnetic Loop Antenna Calculator ver. 1.22a.

Above is a model hypothetical 1m diameter loop of 10mm conductor on 40m with 1% radiation efficiency.

Lets say it is rated for input power being the lesser of 10W continuous, or 30W PEP SSB.

The continuous power rating implies the tolerance of components to current that heats them.

Main loop current at 10W input is given by I=(P/R)^0.5=(10/0.603)^0.5=4.07Arms. This current flows through the main loop conductor and its tuning capacitor. In poorer designs, this current often flows through spring contacts / bushings / bearings in the capacitor contributing to low efficiency.

Note that common mode chokes that might be part of the configuration (on the main RF feed or on tuning motor feed), whilst not directly heated by the main loop current, also exhibit a slow response to average power.

Voltage breakdown is usually a near instantaneous mechanism, and the SSB PEP rating is limited by this failure.

The loop current at 25W input is given by I=(P/R)^0.5=(25/0.603)^0.5=6.44Arms.

From the calculator, the main loop reactance is 131Ω, so it will tune with a capacitor of reactance -131Ω and the voltage impressed on the capacitor will be I*Xc*2^0.5=1193Vpk. The capacitor must withstand this voltage, and its dielectric loss is proportional to its voltage. Poorer designs using plastic film insulation between plates tend to exhibit high capacitor loss contributing to low overall efficiency.

Main loop current and capacitor voltage are constraints on the maximum continuous and peak power respectively. The expected values for a particular loop can easily be calculated from values returned Calculate small transmitting loop gain from bandwidth measurement.

Whilst the current and voltage calculated here seem moderate and easy to accommodate, the same loop with higher efficiency (eg reduced conductor and capacitor loss) would operate at higher current and voltage.

]]>The antenna is described at http://chameleonantenna.com/CHA%20P-LOOP%202.0/CHA%20P-LOOP%202.0.html.

This analysis does not consider the proprietary Power Compensator option for lack of sufficient information.

The loop itself appears to be LMR400 coax or similar (nominally 8.0mm outer conductor diameter) in a rough circle of 34″ (0.863m) diameter.

Little information is given of the internals, but the promotional material gives a VSWR curve for a matched antenna at 7.15MHz. To their credit, they give the height above ground and ground type for their tests, though elevation above ground was between 1/2 diameter to a full diameter of the P-LOOP 2.0

is a little vague.

The VSWR=3 bandwidth scaled from the graph is 27.0kHz. The shape of the curve near minimum suggests that were the scan points sufficiently close, the minimum VSWR would be very close to 1.0 and it is taken as 1.0.

If we assume for a moment that the VSWR measurement was captured at a substantial height above ground, its behaviour approaches that of the antenna in free space. Taking the assumption that the published curve is similar to the antenna in free space, we can estimate the gain and efficiency based on earlier assumptions.

The assumed values and published VSWR curve indicate an antenna system half power bandwidth of 22.0kHz and Q of 324 which implies efficiency of 0.95% (-20.2dB).

The actual value for radiation resistance is likely to be with -50-+100% of the free space value used, and that rolls up as an uncertainty of +/-3dB in the calculated efficiency and gain.

The estimated efficiency is considerably lower than Chameleon’s specified 5.67%, but they offer no justification for that value.

The VSWR=3 bandwidth scaled from the graph is 25.3kHz. The shape of the curve near minimum suggests that were the scan points sufficiently close, the minimum VSWR would be very close to 1.0 and it is taken as 1.0.

If we assume for a moment that the VSWR measurement was captured at a substantial height above ground, its behaviour approaches that of the antenna in free space. Taking the assumption that the published curve is similar to the antenna in free space, we can estimate the gain and efficiency based on earlier assumptions.

The assumed values and published VSWR curve indicate an antenna system half power bandwidth of 21.9kHz and Q of 326 which implies efficiency of 2.55% (-15.9dB).

The estimated efficiency is considerably lower than Chameleon’s specified 14.4%, but they offer no justification for that value.

The booster kit increases diameter by 48/34=1.41 times, and for a STL, that implies an increase in Rr of dia^4=4 times.

If all other losses were constant, for such a loop an increase in Rr by a factor or 4 would increase efficiency by close to the same amount. But, some losses will increase significantly for the larger loop, and it seems unlikely that efficiency would increase by quite 4 times.

The calculator suggests the likely efficiency improvement was around 2.7 times which is credible.

The comparison shows how sensitive efficiency is to diameter when the loop perimeter is less than λ/10.

At some point though there is a trade-off between size and performance, and indeed the 48″ loop on 7MHz only just qualifies as a STL where the conventional formula for Rr is sufficiently accurate for most purposes (perimeter is less than λ/10).

Chameleon gives sufficient information on their ‘basic’ CHA-P-Loop 2.0 to allow a reasonable estimate of its efficiency on 40m to be 0.95% (-20.2dB) +/-3dB.

Chameleon gives sufficient information on their CHA-P-Loop 2.0 with 48″ booster kit to allow a reasonable estimate of its efficiency on 40m to be 2.55% (-15.9dB) +/-3dB.

- Chameleon. nd. CHA P-LOOP 2,0. http://chameleonantenna.com/resources/CHA_M-LOOP_SWR_ANALYSIS.pdf (accessed 16/04/2017).
- Duffy, O. May 2014. Small transmitting loop calculators – a comparison. http://owenduffy.net/blog/?p=1693.
- Duffy, O. Nov 2014b. Chameleon CHA M-LOOP. http://owenduffy.net/blog/?p=2999.
- Efficiency and gain of Small Transmitting Loops (STL)
- Small transmitting loops
- Yates, S. April 2009. Small Magnetic Loop Antenna Calculator ver. 1.22a.

Steel MIG wire is often coated with copper and is claimed by some online experts to “work real good”, particularly as a stealth antenna.

But is it the makings of a reasonably efficient antenna?

This article applies the model developed at A model of current distribution in copper clad steel conductors at RF to estimate the effective RF resistance of the wire at 3.5MHz.

In fact copper is an undesirable and restricted contaminant of steel welding wire, high grade MIG wire is not copper coated.

Copper content is held to less than 0.05% in the core, and less than 0.05% in the coating… which on my calcs says the coating of common 0.91mm MIG wire is less than 0.125µm…. basically it is a small diameter wire with low conductivity and high permeability.

The permeability contributes to higher inductance per metre of wire, and very small skin depth which contributes to higher resistance per metre of wire.

Analysis of the current distribution in the wire shows that almost all the current flows in the core, and that the RF resistance with and without the coating is almost the same. The copper plating can be ignored as significant, in fact the copper plating will usually fall off with a few weeks of exposure to weather as rust develops under the copper.

Lets model a 40m long centre fed dipole in free space, this would be a half wave dipole on 80m if using a copper conductor but we will use the MIG wire equivalent, a steel wire (σ=2.1e6, µ=800). We will use NEC-4.2 (NEC2 cannot natively model ferromagnetic wires.)

The current distribution is nothing like a half wave dipole, it is more like a three half waves dipole due to the magnetic effect of the wire.

Conductor loss is a staggering 91% whereas you might expect more like 3% for a copper dipole.

Ferromagnetic wire changes the tuning of the antenna, radically.

If the objective is a stealth antenna, this is doubly stealthy, not only can no one see it, no one can hear it!

]]>A small roll of galvanised tie wire can be purchased from Bunnings hardware for about $10 for 95m… so at $0.10/m it looks like an economical solution.

But is it the makings of a reasonably efficient antenna?

This article applies the model developed at A model of current distribution in copper clad steel conductors at RF to estimate the effective RF resistance of the wire at 3.5MHz.

A sample of new unweathered wire was measured to determine the approximate zinc coating depth, it was 15µm. Note that zinc is a sacrificial coating and it will erode through life, so this study is an optimistic one of wire when new.

Above is a plot of the current density distribution in the surface region of the wire. Essentially the current flows in the zinc coating and negligible current flows in the steel core.

At 15µm, the cladding depth is just a quarter of the skin depth (60µm), so we might expect high effective RF resistance, it is not the zinc so much as that it is very thin.

The current distribution allows calculation of the effective RF resistance, it is 805mΩ/m, as against a 1.5mm copper conductor’s 0.106Ω.

Whilst you might anticipate the conductor loss in a 80m half wave dipole made from 1.5mm copper conductor to be around 3% (0.13dB), the galvanised wire above is likely to be more like 20% (1dB).

As mentioned earlier, as the zinc erodes with weathering, resistance goes up. When say half the zinc thickness has eroded, effective RF resistance will roughly double.

It is cheap, little risk of the wire being stolen, but it is low efficiency. As a half wave dipole, it would be more efficient on higher bands where the wire length is shorter.

We can reasonably assume that a multistrand wire that uses strands thinner than this wire will be worse.

]]>Two late posts as I write this were:

There really is no real issue with skin effect on HF bands with copper clad materials.

and…

At 1.8 MHz, the skin depth in copper is 0.654 micro-meters (.0000654 mm), so the copper cladding on the center conductor of most RG-11 type coaxial cables is more than sufficient for any of our current bands.

The specific advice above looks interesting, convincing even… but thankfully, the skin depth in copper is nowhere near either of the figures he gave.

If it was 0.654 micro-meters as he first states, the RF resistance of a 1.6mm solid copper centre conductor would be 2.5Ω/m.

If it was .0000654 mm as he secondly states, the RF resistance of a 1.6mm solid copper centre conductor would be 25.6Ω/m.

In fact the skin depth in copper at 1.8MHz is 48.6µm, and the RF resistance of a 1.6mm solid copper centre conductor would be 0.0349Ω/m.

You do have to be discerning in what online advice you swallow.

To the original poster’s question, if the cladding is more than say 100µm in thickness, it will have almost copper like performance at 1.8MHz.

I measured a piece of quad shield RG-6 yesterday to have 10µm cladding, copper like performance above about 150MHz, certainly sub-optimal for HF, much less MF.

Copper thickness is relevant on CCS conductors at HF and MF, and the problem is that lost of cables lack either specifications for matched line loss or cladding thickness data down to low HF and MF. If the datasheet stops at 30MHz, it is a warning that it may be significantly poorer than copper at lower frequencies.

]]>(Stewart 1999) published a set of measurements of the popular Wireman windowed ladder line products. His measurements were in the range 50-150MHz. They form the basis for most calculators on quantitative analyses at HF, despite the fact that it is a dangerous extrapolation for CCS construction.

Nevertheless, the directly stated measurements at 50MHz are a useful calibration point for reconciliation.

Above is Table 1 from Stewart, it sets out measurements of four Wireman m.products and a plain copper line.

The table below compares Stewart’s measurements with the CCS model and with TLDetails results (where available).

The third case is a smooth round conductor of equivalent diameter to Wireman 553’s 19 strand bundle with cladding of the thickness expected on the individual strands. The CCS model is for conductor loss only of a single core conductor, measurement of the real line might reveal some additional loss due to current passing between strand contact points, and dielectric loss (though that should be very small). Importantly, approximate agreement validates the assumption that current in the 19 strand conductor will flow mainly in the circumferential copper of the bundle and current flow in internal copper will be insignificant.

Line | N7WS measurement (dB/m (dB/100′)) | Duffy CCS model conductor loss (dB/m) | TLDetails (dB/m) |

1.291mm (#16) Copper @ 19.05mm (0.75″) | 0.00984 (0.30) | 0.009712 | N/A |

1.024mm (#18) CCS 67µm cladding (Wireman 551) | 0.01083 (0.33) | 0.011078 | 0.01403 |

1.024mm (#18) CCS 15.3µm cladding (similar to Wireman 553) | 0.01247 (0.38) | 0.012276 | 0.01983 |

Skin depth in copper at 50MHz is 9.225µm.

The CCS model quite closely agrees with Stewart’s measurements. TLDetails departs from both quite significantly.

The CCS model is in close agreement with Stewart’s measurements of the three line types, and the agreement on the latter case validates the assumption that the 19 strand CVS conductor behaves much like a single core #18 conductor with cladding thickness equivalent to that of the individual strands.

Stewart, W (N7WS). Mar 1999. Balanced Transmission Lines in Current Amateur Practice.

]]>Fig 1 is a plot of the current distribution in a 1mm dia (#18) round copper conductor at 1.8MHz as implied by the model. Note that while the magnitude of current decays exponentially with depth, there is an imaginary component that hints a curl of the E and H fields within the conductor.

The sum of these complex currents over the cross section area is the total current passing a point, I will call it inti.

Power lost in the internal resistance of the conductor in any vanishingly small part of cross section area is simply |I|^2R for the area. The sum of |I|^2R divided by init calculated above gives the effective resistance.

Fig 2 above is a plot of the current density per metre along the radius.

Fig 3 above is a plot of the power density per metre of radius along a radial. It is this that will be summed over the radius to calculate total power lost at current inti.

The effective RF resistance is this case is 112mΩ, 5.5 times the DC resistance as a result of the current flowing mainly near the surface of the conductor and much of the conductor cross section not utilised significantly.

This case uses a copper cladding of 67µm copper over a steel core for an overall diameter of 1.024mm (#18).

Fig 4 above is the current distribution of the CCS conductor. The cladding is not quite enough to have a distribution almost exactly like the copper case, but it is very close and in fact the effective RF resistance is calculated to be slightly better at 111mΩ. So even down to 1.8MHz, the conductor delivers near to copper like resistance.

This case uses a 19 strand conductor with copper cladding of 15.3µm copper over a steel core for an overall diameter of about 1.024mm (#18). At radio frequencies, the current will tend to flow in the outside surface of the outside layer of strands, and can be roughly approximated by a 1.024mm (#18) conductor with 15.3µm copper cladding.

Fig 5 above shows the current of the equivalent CCS conductor. The cladding is too thin to exhibit the copper like distribution, rather we see a truncated part of the distribution which will drive loss up as even less of the conductor cross section is utilised effectively.

RF resistance is calculated to be significantly higher at 343mΩ. A transmission line using the 19 strand #18 conductor is likely to have about three times the effective RF resistance to the #18 single core conductor at 1.8MHz.

Conductor loss in a matched 400Ω transmission line would be around 0.75dB/100m. That might seem to be a small number but these lines are usually used at high to extreme VSWR and that matched line loss may drive rather high loss under mismatch.

Beware of line loss calculators that mostly return unduly optimistic performance figures for this type of line.

You might wonder why the ‘premium’ 19 strand line is so popular when it does not deliver copper like performance at MF/HF.

]]>