Antenna Modelling... A Follow-up

VA7NF Helps Explain What is Going On Here… 

In the last post, I described my experience with modeling and construction of a quarter wave 80 m inverted L antenna.  I stated that, while the model reliably predicted the real-life behaviour of the antenna, there were also some puzzling results beyond my understanding with the use of various antenna analyzers,  so I asked Stan VA7NF to apply his technical expertise to interpretation.   Other readers’ comments are also invited. 

Part 1 – Measurements Made at the Antenna

Here is a summary of my questions and Stan’s explanations:
Because the feedpoint of the antenna is at ground level, I was able to make one set of measurements without the feedline present then a second set at the station location with the feedline present.  

An AIM 4170 analyzer was connected to the antenna at the feedpoint by a 1 m long RG8X jumper.  I first adjusted the series capacitor at the feedpoint to bring series reactance (Xs) to zero at 3750 KHz.  Initially, each scan and re-scan produced the SWR curves below, showing a large number of random spikes, with each re-scan producing another set of spikes at different frequencies.  The first scan [below] is in red; the re-scan is in orange.

Could the noisy behaviour be the result of strong signals being pickup up by the antenna and interfering with the reflected wave or electronics of the instrument?  To check for the presence of out-of-band signals, I used the AIM 4170 band sweep function from 0.500 MHz (the low end of the broadcast band) to 5 MHz (above the 80 m band) as shown in the graph on the next page. This illustrates two strong, off-scale signals in the AM band (CKST at 1040 and CFTE at 1410 KHz) plus another unidentified signal at 2.0 MHz.  

As I continued with my measurements, the odd behavior shown in the first graph abruptly ceased and I was unable to reproduce it again.  I believe that the improvement came about after all the connections were thoroughly tightened, but it was also coincidentally about the time the weather changed from a prolonged cold, dry spell to very wet.  From that time on, the curves looked “smoother”.

Stan’s comments: 
Your assumptions above are well founded.  Antenna analyzers will send short bursts of RF into the test device (antenna) while monitoring the strength of the return signal.  For short antennae this will normally only be the reflection of its signal; when longer (80M or 160M or our FD antenna 320m) there will be many other strong signals especially the broadcast stations as you noted.

This is where analyzers differ.  Newer ones have Rx filtering that follow the transmit frequency; older ones take a voltage reading of everything on the antenna.  That means the readings are significantly affected by “stray” signals, making them useless in strong fields anywhere in the spectrum.

Your comment about the “noise” changing appears to be directly related to tightening the connectors.  Those two AM stations will mix with themselves, other lesser strength signals, and noise presenting the peaks you saw.  By making a better connection the diode effect of the poor junction has been removed eliminating the mixing products you saw.  Aside from analyzers, any diode (rusty tower bolts, wire fence supports, etc) will mix these strong stations with noise that becomes several S unit Rx background noise.  Doing the math, the 4 large spikes (3620 -3760) appear to be the sum of (1040 + 1410) mixing with 4 other local broadcast stations; also a 3rd peak exists at 2450 being 1040 + 1410.
Incidentally, our 80M antenna at the OTC is producing such images and should also be investigated.

Once the noisy data quietened down, and again scanning with the AIM 4170, the graph [below] showed an SWR close to 1.0, reactance near 0 and R not far off 50 Ω at 3750 KHz.  It doesn’t ever get much better than this, but could I believe it?  I decided to check with another instrument.

Measurements were again made at the antenna, but now using the AA-600 analyzer – results were somewhat different.

As seen in the graphs below, the minimum SWR was now higher at 1.5 but also shifted lower in frequency from 3750 KHz.  Note that at 3750 KHz, the reactance was still showing zero but Rs was around 80 Ω, thereby accounting for the higher SWR.   These results were more in line with my expectations but why did two high-quality instruments show such different results?

Stan’s Comments:
A major part of your low SWR is by design; you have tuned out Xs by adjusting the series capacitor; at 3750 your Xs went from -ve to +ve (tuned resonance) and the Rx value at that frequency as reported by the AIM was 50 ohms.  The RigExpert agrees with AIM that Xs crosses over at 3750 but it says Rs is around 80 ohms and drops as the frequency drops.  The >50 ohm Rx and -ve Xs combine below 3750 for the lower SWR.  It is unknown why there a difference in Rs from 50 to 80 ohms.

Conclusion:  Because they develop different Rs values the SWR curves are different.  By observation there is significant noise still on the wire and that may be the basis of the differences.

Continuing at the antenna, measurements were now made with an MFJ 269 analyzer. 

At 3750 MHz regardless of frequency, the SWR was off-scale (∞) consistent with an unchanging Rs=0 and, in addition, Xs varying erratically, i.e. it was not possible to get a meaningful measurement (note it was confirmed that the analyzer returned appropriate readings when tested with a 50 ohm dummy load and at 20 m on the author’s beam, so the meter seemed to be “working”).   Was this possibly another strong-AM signal interference problem?

Stan’s Comments: 
Agree.  Too much broadband noise for this meter to function.  Not usable on a real antenna with a long length of copper.

Part 2—Measurements Made at Station

Next, a length of new LMR-400 coax plus a short jumper were attached to the feed point along with a second common mode choke close to the station end, but otherwise conditions were unchanged from earlier.  All the following measurements were made at the station end of the transmission line and jumper.

The AIM 4170 produced SWR, Rs and Xs curves across the band. [left]  Zero reactance occurred now at a higher frequency than previously, and instead of rising into positive (inductive reactance) territory at frequencies above 3750 MHz, Xs decreased, showing a net capacitive reactance. How can this be explained?

Stan’s Comments:
Coaxial cable when connected to a matching impedance (50 ohms) will not affect the apparent impedance regardless of length;  however:
1) When connected to a non-matching impedance there will be returning currents with a +/- Xs.
2) As the current flows back it will be a “normal” sine wave changing phase until at 1 wavelength it again matches the phase of the forward wave.
3) RF on a coax cable does not travel the same speed as RF in free space.  Specifications show this as a velocity factor (VF) which is .85 for LMR-400; this means the RF will cycle its phase over less length (1/VF wavelength).  As the test frequency changes this means that the length for a full sine wave will vary being longer at 3.5 MHz and shorter at 4.0 MHz.  Viewing it from a single point (like the feed point) the phase of the reflected signal will vary with frequency.

Putting this together, a reflected signal starting at the antenna connection will be out-of-phase with the forward signal (The +/- Xs as you measured at the antenna).  As that reflected signal moves down the cable its relationship to the forward signal will change, starting at (for example) a +ve phase that will shift to in-phase, then -ve phase, to in-phase, then back to the +ve.

Depending how long your cable is (in wavelengths times VF) the reflected signal will appear + or - phase (Xs) and at places exactly in phase with the signal going the other way.  This combination is what your analyzer will see.  In your image the transformer effect of the length of LMR-400 modifies what is seen.

Another effect of coax transformer effect will be, when the antenna has a high SWR, artificial dips in APPARENT SWR appear; these artificial SWR dips appear when the forward signal is cancelled out by the reverse signal with a resulting Xs=0.  This cancellation effect is frequency dependent so a long coax may show several artificial dips typically 5-10Mhz apart.

Accepted practice is to have feed lines that are ½ wavelength (after applying the VF) but only when attempting to measure at-antenna SWR.  They will NOT appear if the antenna is low SWR as there is no reflected signal to mix with the transmitted one.  This will not eliminate the artificial dips as the length will not be a ½ wavelength across all those frequencies.  This also explains why an antenna that will not tune, may tune if a short length of coax is added between  tuner/transmitter and the antenna.
Continuing with measurements at the station end of the transmission line using the RigExpert AA-600, both the SWR and Rs-Xs graphs now agreed fairly closely with the results for the AIM 4170. [below]

Next using the MFJ 269, the behaviour was once again erratic, with no minimum SWR discernable at any frequency and no meaningful SWR measurable.

I took one further step when I noted that with the AIM 4170 instrument, conditions at the antenna can be measured from the far end of the feedline, if its length, velocity factor and matched loss are known.   As per the Times Microwave specs for LMR-400 giving a velocity factor of 0.84 and matched loss of 0.1 dB/100 ft. (at 1 MHz),  the total length of the cable was known to be 112 ft.  Under these conditions, a “refer to antenna” set of curves measured from the transmitter end of the coax looked like the following.  It was reassuring to note that these curves were virtually identical to those obtained when the measurements were made right at the antenna. 

Stan’s Comments:
The AIM is calculating the coax length effect and applying it to the reflected signal.  Smart!

Conclusions:

On the assumption that strong signals in the AM band (not sufficiently filtered out by the instrument) were the cause of anomalous results, the MFJ 269 readings were rendered useless.

The AIM 4170 analyzer appeared to be influenced by the strong off-band signals, but considerably less so than the MFJ.  

Only the RigExpert AA-600 appeared to be immune to the strong AM signals, and give relatively consistent results at both locations of measurement.

When adjusting for resonance of the antenna/feedline combination, conditions at the station end of the feedline would appear to be the place that counts, as presence of the feedline will influence the resultant impedance seen by the transmitter.  

~ John Brodie VA7XB
  18-02