VOACAP reliability, the end of the myth ? part2

Following are a few more plots concerning the F to ZL2 path. The above plot shows the signal levels reported by ZL2TLD over a longer period than the earlier published one. You can see that the delta between the lower and higher reported levels, is around 10dB, which can be explained by day to day propagation variation, but also by some punctual intereferences (we are not alone on this part of the band). The long path "average" is around -22dB for 1W TX power in a -1dBi gain antenna (simulated gain of my vertical at 10degrees). The short path average, under the same conditions, is some 2.5dB higher. Now here are the VOACAP simulations for the short and long path. This time I used method 30, CCIR monthly coefficients (and NOAA monthly SSN smoothed indexes). I used "isotropic antennas" (both antennas have negative gain at considered 10degrees angle) and switched the multipath "off". I used -150dBW/Hz noise level at 3Mhz (5dB quieter than the average recommended -145dBW), and pushed the required SNR to 12dB (-22dB observed +34dB for BW relation). The result is about the same... 50dB are missing to make the link possible...

Above is the short path... about 40dB are still missing to match the observed -20dB average level in 2500Hz, to the VOACAP simulation (forgetting that both antennas have negative gain, when compared to the isotropic used in the models.)
It is not all, but of course I kept the best for the end... stay tuned !!!

VOACAP reliability, the end of the myth ?

Sorry, I have been away from the blog for quite some time... Just that I was blogging elsewhere ! Following is a series of articles recently published on the wsprnet.org site, which question the actual reliability of VOACAP ionospheric simulations. Just as a reminder WSPR is a bidirectionnal (TX/RX) digital beacon mode invented by K1JT, which provides a SNR measurement and automatically uploads it to a Worlwide database. The sensitivy is extreme (-30 dB in 2500Hz bandwidth, equivalent to +4dB in 1Hz bandwidth). Here is the first article (slightly modified).

In April 2008, I published an article on my personnal blog showing some offset between the levels reported by WSPR and the VOACAP models. This article was based on reports provided by W1BW on my 30m WSPR signal. I am now working on the data collected during the 40m daily tests with ZL2TLD. The first results are even more surprising...
Without entering now into details, I find some 50dB offset between the WSPR reports and the VOACAP models...

There are a 3 hypothesis:
1- I did a big mistake somewhere... (to get 50dB difference it has to be a very big mistake !)
2- VOACAP is not properly calibrated for >10,000kms pathes, or for vy low SSN's, such as those we are currently facing.
3- We all took for granted that VOACAP was the ultimate ionospheric prediction tool, without seeing that it could be completly "out of range" in a certain number of cases.

To convince you that there is a problem, have a look at the attached. It has been obtained using the popular VE3NEA's HAMCAP (free on his site). With this interface to VOACAP the risks of mistake are minimal (you just setup, the TX power, and the type of antennas at both sides). The solar data is obtained automaticaly through IONOPROBE (or manualy through the usual means).

- First is the setup screen (you may click to enlarge), showing that I configured Hamcap for 100W TX power (for antennas I used a 1/4GP on TX side and a dipole at 55 feet on RX side).
- Following is the chart screen, showing an expected level of -25dB, for the "long path" (had put the mouse in the middle of the yellow rectangle, on the 7Mhz line, to get the 0530z SNR displayed at the bottom)




- Finaly is a plot of the long path WSPR-reports at 1W normalised TX power (the offset between the 2 series of reports, can easily be explained by a higher noise level at my location)

Now just a few maths: - I used 100W as TX power in hamcap, thus 20dB above the 1W normalised power used for the data plot
- VOACAP or HAMCAP are using SNR's in 1Hz bandwidth; WSPR in 2500Hz bandwidth... the bandwidth relation translated in dB's is 34. Which means that you need +4dB in 1 Hz bandwidth to get WSPR to report -30dB in 2500Hz.
Here the expected level is -25dB, thus lacking some 29dB to reach the required +4dB SNR in 1Hz bandwidth.
In short VOACAP tells you that the path is impossible, lacking some 49dB of antenna or power gain... Even 55 if you consider that the average bilateral levels peak at -24dB and even more if you consider that I used a dipole at 55 feet in the model and that Glenn is using a windom at 15 feet !!!



The attached VOACAP plot, is for the Required Power and Antenna Gain, vs time. You can see that according to VOACAP 51.47dB are missing to get the required +4dB in 1Hz BW (for 33% reliablility) - moreless the same value provided by HAMCAP in the above-example.
For this one I used "method 21" (long path forced model), but the classical "Method 30" gives exactly the same value.
While the minimum RPWRG is reached at 0600z, which matches exactly the observed maximum, the "required Power and antenna gain" is completly out of range... According to this graph, I should run some 400kW to have 33% probablity to get a -25 WSPR report from ZL2TLD, while it works every day at 2 W (even at .5W) !
Serious offset isn't it ?

Now; If you believe that I did a big mistake somewhere, I invite you to access the wsprnet.org database, and work on some of the huge data available there.

More to follow...

SV*/F6IRF/M

Je serais en Grèce, à partir du 24 Juillet et jusqu'au 12 Août. Au programme, tourisme, plage et radio dans le Nord: SV6, SV2, SV4, SV1, peut-être SV7 et probablement une des îles de la mer Ionienne ou de la mer Égée (SV8).
QRV mobile essentiellement sur 10m et 6m. Les fréquences de prédilection 28024, 28424 et 50224. Pour tester la propagation 10m SV>F, le mieux est d'écouter la balise SV3AQR sur 28182.6. Pour info le path était ouvert quasiment tous les jours pendant ces 2 dernières semaines.
Au plaisir d'un QSO !

SY3M - A few audio recordings

80m sunday at sunrise. From Greece, end of may, the US windows on 80m is not very wide... But the 6 points QSO's are always welcome, worth making the effort ! (op Pat)

Sunday morning on 40m, just after sunrise. As usual in the WPX , 40m is the "money band" with the highest point/QSO average. Unfortunately, we missed the recording of the first hours, which provided the higher rates on this band... We could not explain the rythmic noise on the K3 recording, maybe a "knocking spirit"... (Op Nico)

Saturday morning on 10m. The 10m band provided good rates, but with a low point average despite a few DX's. (Op Pat)

The bigger 10m surprise has been a one hour opening to JA on sunday morning. We worked 20 JA's and 1 HL A selection of a few of them. Most likely multi-hop E's. BTW, the ghost is back ! (Op Pat)

20m also provided good rates despite our low gain/low height antenna, but difficult to find and keep a clear frequency. (Op Nico)

An amazing recording done on 40m a couple of hours before the contest end. (Op ?)

A vertical half-moxon for the 40m band

The MMANA model: Not obvious here, but there is a 20cm gap between the 2 horizontal parts.
Following is the description of the 40m array we used at SY3M. The antenna was designed "on the spot" and constructed very quickly using 2x 10m fiber masts. It is composed of 2x inverted L's facing each other with 6m spacing between the 7.4m vertical parts and 20cm separation between the 2 ends of the horizontal parts. We used two sets of 10x10m ground mounted radials for each monopole. As we did not know the "velocity factor"of the plastic insulated electric wire, we used the same technic as we used for constructing the wire beams at CN2WW for the ARRL. Suppressing the reflector on the model gives X=0 for 7.372Mhz.
The MMANA simulation, using MININEC "average" ground.
We cut the driven element accordingly, which gave us the velocity factor of the wire (0.96). The R at X=0 was measured at 33 ohms; exactly what was predicted by the model inserting a 1 ohm resistance, simulating the ground losses at the base of the monopole. We then cut the reflector taking into account the 0.96 velocity factor. After a few minor adjustments, consisting in inserting small pieces of wire at the base of the elements, we obtained X=0 for 7.025... this was kind of useless, as the bandwidth is extremely broad, as predicted by the simulation. In fact the only major difference with the model, was that the R value, was found at 50 ohms, and not at 36 ohms as predicted by the simulation using mininec ground, but it did not worry me, as I know that Mininec is not very accurate for the R-value as this calculation is done on a "perfect ground" basis.
The MININEC simulated SWR response. The very broad bandwidth (about 300kHz at 1.5), was confirmed on the real antenna. The impedance was found very close from 50 ohms on the real antenna.
On top of this, the practical antenna was somewhat different from the model, due to the flexion of the fiber masts (we had actually 6.5m between the 2 masts bases).
First "on the air" tests were done on the Russian C-beacon (Moscow) on 7.039 a the end of the afternoon; compared to the 16m vertical, tuned on 7Mhz, the difference was constantly over 2 S-points in favor of the half-moxon (about 6dB on the pro2 S-meter). We found this difference to be quite constant for all stations in the North sector, the difference becoming minimal on the EA stations located West of us.
MININEC simulated Gain and Front to Rear ratio vs frequency.
Our 40m score during the contest and the fact that we often won the "power battle" and got first in the pile-ups, confirmed that the antenna was working fine. The simulated gain using mininec average ground and 1 ohm resistor simulating ground losses, gives 3.3dBi at X=0 and near to 20dB F/B, the gain being -0.1 dBi for the radiator alone. This value might be a bit pessismistic, as the simulated gain for the radiator alone, using NEC2 finite ground and 10x10meters radials is about 1dB higher than the model using mininec ground. Unfortunately NEC2 finite ground does not allow to specify 2 sets of ground mounted radials with different centers. Later (on the ship) we simulated a NEC2 model using a Sommefeld-norton ground and 2 sets of 4 elevated radials, which confirmed an impedance close to 50 ohms, a bandwidth matching the measured one,a gain slightly over 3dBi and a F/B ratio of 25dB.
MININEC patterns from 6.8 to 7.25 Mhz.
Of course we could have gone for a 2 vertical monopoles passive array (driven+reflector), but the horizontal pattern would have been sharper and the array more difficult to match, due to the lower radiation resistance. We liked the cardioid pattern of the half Moxon and his very broadband characteristics, making the construction somewhat uncritical.
Dimensions (for 1.6mm diameter bare copper wire with vel.coeff 1)
and Fo= 7.025
vertical parts height: 7.40m
Driven horizontal part: 2.7m
Reflector horizontal part: 3.20m
Gap between the horizontal wires: 20cm
Separation between the verticals: 6.10m

Above, the NEC2 simulations using a model with 2x4radials, over a "Sommerfeld-Norton" average ground.

SY3M in the WPX, the video

The "rats terminators" in the CQ WPX. More than a "contest video"... hope you enjoy !
This video is also avalaible on dailymotion (better quality)

SY3M, the photo album

You can see the SY3M expedition photo-album by following this link

SY3M 2008 WPX-CW by the "rats terminators" contest group

Operators: Nico SV3SJ, Pat F6IRF
QSO details and claimed score here TRX: K3 + TT Titan, pro2 + AL1500ANT: 2el SteppIR @9m, Ground-mounted vertical Half-Moxon (40m), 16m ground mounted vertical (80/160), K9AY RX-loop and 120m beverage to JA
MSC: N1MM, Microkeyers, laptops, ICE filters, 700m of wire, 250m of coaxial, 200m of rope and a lot of enthusiasm and energy to setup/dismantle all this !!!
Under Max's hass, some of the hardware used for the operation !
My initial personnal plan was to go back to CN with some ambitions in the SOAB category, but for some obscur administrative reason the licence never arrived. On his side, Nico had already planned a trip to his home country... I could not miss such an opportunity to go back to wonderful Greece, so we quickly improvised something to be on the air; not to be competitive, but to take part, which is the most important as says the Olympic games chart! (another famous Greek/French joint venture !)Among the difficulties to setup a all-bands M/S station from scratch in a remote location, the first one was to rehabilitate a house, unnocupied by humans for some 20 years (adopted by rats and mices and invaded by wild vegetation); It took a few days to make the house usable, but we managed it... The cohabitation with the dense rats population was not always easy, but they finaly accepted to free-up some space for us (must say that we had to insist a bit, using some powerful unfair arguments... the main problem for me with the Greek rats, being that that they understand only Greek and they start working at sunset until sunrise...).Waiting for Nico to complete some "non-ham" obligations and beside some 6m and peri-contest activity (about 800 Q's as SV3/F6IRF ), I started by installing my 16m multiband portable HB vertical, tested the week before in britanny.
The 16m vertical used for 80 and 160. Alone it consummed some 300m of radials...
After a few tests on the air, I started to think about some more efficient solution for 40m, which is key band in this contest. We needed something simple, that we could setup quickly with the hardware we had (2 fiber 10m telescopic masts), so after a few simulation sessions we decided to go for a vertical "half-moxon" mainly for his broad horizontal pattern (a description will be published later here). Must say that the results overtook my expectations, the difference with the vertical being obvious right from the first listening tests. It was confirmed during the contest, as we did not get any problem to break the pile-ups and get DX stations to call us.
The vertical base; despite the poor ground quality, it worked fine...
Obviously the weak band has been 20 meters, due to the low height and limited gain of the antenna, but we could not do better for this time...10m provided some nice openings, including to JA (most probably multi-hop E's) and 15m a nice combination of E's and F2 propagation.
The driven-element of the half-moxon which gave us a strong signal on 40m. Another 200m of radials !!!
At the end we are just short of multipliers to be "almost competitive"- we had initialy planned DX-cluster through a VHF packet-link to Kalamata, but the distance and terrain profile between Kalamata and Pilos area did not allow it to work. We obviously also missed a second antenna on the high bands - but we did no expect them to be so productive and to say the truth we ended up short of time and energy (the temperature on those days did not help) ! - Anyway, in may still be a new Greek M/S record...
The SteppIR at 9m. On 20m difficult to compete with fixed stations using large antennas or stacks at a decent height.
I really enjoyed those 2 weeks in this still preserved paradisiac wild area... unfortunately not for long, as in many other places, the "beton" is about to win (a large touristic ressort is already in construction, a Motorway is scheduled and the promoters all around). A special mention to the Greek local people for their "way of life", sometime a bit annoying for a stranger, but so nice at the end !
73's from Sunny Greece - Pat
PS: More to follow here soon including a video for which I have a lot of raw material...

WSPR 80m tests from IN87

The above plot shows the 80m WSPR-data collected from W1XP (FN42fo) and my portable station (IN87nn) during 4 nights, from May3 to May7 (about 250 spots). 4 to 10W were used on both sides but all plotted levels are normalized to 1W.
The general trend, over the 4 nights is a 6 to 7dB offset in favor of W1XP transmitter, with a 10dB peak around 01:30. This may indicate a higher TX antenna performance or a higher noise floor (or both) at W1XP. It is interesting to note that at W1XP sunset (when the band opens for him), he seems to hear me better. As suggested by ON4UN in "low band DXíng", the explanation may be linked to a lower noise level while the D-layer is not yet fully dissipated. This is confirmed by the below VOACAP Noise-power plot for Boston area, where we can see that on 3.5Mhz, the noise is minimal 3 hours before sunset and maximal in the middle of the night (The below graph is for a very quiet site, the variation may be partly or totaly cancelled or changed if the main source of noise is local). Those variations, although the power and antennas gain remain constant, do not mean that the propagation is not bilateral,just than the noise is not distributed equaly at both locations. The 7 dB offset trend clearly means, that my TX antenna does not work as well as W1XP's one (unless some problem on W1XP's Receiver, signal treatment, QRM,QRN, birdies, etc...). The only way to know would be to compair with a third station used as reference.

F6IRF/P IN87NN

One example of last night 80m session. As usual all levels are normalized to 1W. Apparently my 80m portable antenna is working fine...

The portable antenna, just a bit shorter (16.5m) than my home one, but built on the same principle (see the best of this blog). Built with a light telesopic mast, a steppIR fiber element, and a a light fishing rod.

Just a dozen of 10m radials laying on the ground, enough to get the radiation resistance close to 28 ohms on 80m. Matching is just done by a serial coil and 28/50 ohms UNUN.