Monday, August 27, 2018

Tuning & Testing the 80 Meter Vertical

It seems everything I do takes longer than expected. This is essentially a progress report on the construction of a 3-element, 4-direction 80 meter vertical yagi. When last I checked in the driven element (a tower with stinger at the top) was completed. Since only a few radials were installed and the antenna not tuned it was premature to talk about performance.

Before the 80 meter vertical yagi array can be built it is first necessary to get the driven element working as a simple vertical. I have slowly made progress to the point that I now have a very effective full-size vertical for 80 meters. This will also be its function after the array is built for the omni-directional mode. The character of the vertical changes markedly as radials are added.

It is important in a project of this complexity that it be broken into a sequence of steps with testing after each step is completed. Surprises can be investigated and dealt with before they can be obscured by further changes. I see this as an advantage rather than a burden since it is a tremendous opportunity to learn about antennas and propagation.


I'll step through the process as I go from the basic vertical with 4 radials and a monopole to a tuned 34 radial vertical for 80 meters, including what I learned along the way. Turning the antenna into a yagi will take more time. Reasons for the slow progress so far include: lack of 80 meter activity during the summer needed for on-air evaluation; connecting the antenna was inconvenient until I repaired the antenna switch; many other concurrent projects; and, non-radio summertime activities.

4 radials

Despite the stinger being fully retracted the resonant frequency with only 4 radials of 20 meter length the fell well below 3.5 MHz. The radial length dominates the short monopole (~18 m). This is expected since not only are the sparse radials resonant they are electrically much longer than 20 meters due to ground proximity lowering the velocity factor.

I didn't bother to precisely measure the resonant frequency since 4 radials was a transitory configuration. On air the vertical performed poorly when compared to the 32 meter high inverted vee. Both signals and noise were noticably attenuated. The match was good because the high ground loss raised the feed point impedance close to 50 Ω. Loss is in series with the radiation resistance.

With a perfect ground a full-size vertical ought to have feed point impedance of 37 Ω, and can be much lower as the monopole diameter increases, as it does when it is a lattice tower. The estimated ground loss with 4 radials was a minimum of -3 db, based on feed point impedance, but likely closer to -6 db. Measuring ground loss accurately isn't easy and I didn't try.

Some insights can be had even though the ground loss is high. By using perceived SNR (purely by listening) an inkling of how the antenna will perform with more radials can be ascertained. SNR on the vertical was better on the longest paths, such as to PY. The inverted vee SNR was always better before sunset when elevation angles on all paths is higher due to absorption at low angles.

8 radials

The addition of 4 more radials made a dramatic difference. Resonance made a big jump to ~3.8 MHz. This shows rapid progression towards non-resonant radials system as radial count increases from a low number. Efficiency improved so that it was more equatable to the inverted vee on reception tests. The inverted vee continued to outperform the vertical within eastern North America. On longer paths the received signals with the vertical were often equal or better than the inverted vee.

The impedance changed little, only dropping several ohms. Obviously there is more affecting the impedance than simply ground loss. I suspect the main reason is that with radials longer than an electrical λ/4 the current distribution on the radials places the current peak away from the feed point. Lower current at the feed point is associated with a higher impedance, thus partially masking the effect of the decreasing serial resistance due to ground loss. This has amply documented by N6LF in his extensive experimentation with verticals.

12 and 20 radials

I did not do on air tests for these radial count for the aforementioned reason of inaccessibility to the antenna from the shack. The only testing was measurement with my antenna analyzer.

The next 4 radials were not symmetrically interlaced with the existing 8. These radials went to the hubs of the 4 parasitic elements, elements which have yet to be installed. Collinear with these short 10.5 meter long radials is a 15 meter radial, bringing the total radial length to 25.5 meters. The parasitic element radials will all be 15 meters and these systems will overlap rather than employ boundary busses. These radials for the parasitic elements are the only ones bonded to the driven element's radial system.

Once these radials were buried the antenna's impedance was measured. The resonant frequency continued to increase though at the expected slower rate due to the monopole beginning to dominate as the radial system becomes increasingly non-resonant. The resonant frequency was ~3.9 MHz with an impedance in the low 40s.

The next 8 radials are symmetric with respect to the original 8. After adding them the resonant frequency increased to ~4.02 MHz.

Measurement precautions

The photo of the analyzer display shows a lower resonant frequency than what I quote above. This is due to the 1 meter length of transmission line between the analyzer and feed point. It adds several ohms of reactance which lowers the apparent resonant frequency by more than 50 kHz. That is ~1.5%, a not inconsiderable amount.


This is easily simulated using TLW to determine the impedance at the feed point. In this example notice the difference between the source and load ends of the transmission line. The amount may seem trivial, but it is enough to impact the performance of a yagi when tuning the parasitic elements, as I will be doing with this antenna. It is far less critical for single element antennas such as ordinary dipoles and verticals. Better analyzers include a feature to compensate for the transmission line so that you can directly read the feed point impedance.

When measured at the true resonant frequency -- with regard to the analyzer reading, which is limited to its inherent accuracy -- the resistance part of the impedance is ~40 Ω. This is more reasonable since the "ideal vertical" has a radiation resistance of ~37 Ω. Ground loss is certainly present. As we'll see below even this value is misleading since the ground loss is significantly greater than 3 Ω (40 - 37).

Another factor to keep in mind is the placement of the analyzer, the coax and your body during the measurement. Although all are small relative to the wavelength I have experienced up to ±30 kHz (±1%) variation with this antenna. Don't let the coax drape over the monopole or the radials, don't lean against the antenna or lie on the ground (on top of the radials). In some cases it can help to prop the analyzer on an insulating platform rather than hold it in your hand.

To bury or not to bury

I stopped burying radials after these first 20. It's a lot of work and I found that if I was careful to put little tension on a radial and walk along it to press it down that it was soon hidden under the vegetation in the hay field. Large dips in the field were levelled to help this along. I can mow over the entire radial field with ease. However I caution visitors to step carefully to avoid tripping and ripping them out. After a year even than warning may become unnecessary.

The other concern with leaving radials on the surface is when my neighbour harvests the hay. That equipment is not like a lawn mower and will tear up surface radials. When he did the harvesting this summer he was exceedingly cautious following my line of stakes and brick markers and would not have hit a radial had the parasitic elements used surface radials.

In light of this experience I have decided to stop burying radials. It's a lot of effort that I am happy to avoid.

The temporary run of RG-213 from the antenna was replaced with ~90' of LDF4-50A. It was laid in the trench along with the control cable and covered over. Heliax is rated for direct burial. From the tree line (see top photo) there is an overhead run of RG-213.

34 radials

This may seem an unusual number of radials until you consider the geometry of the radial system. My objective was to double the 16 radials. With the 4 additional ones that connect to the parasitic element base hubs there were 20. When I added the next set I should have only needed 12 more to reach 32. However two of the hubs were not in the required position for symmetric placement. Therefore I needed two more.

Should I eventually go for 64 radials on the driven element (1,280 meters of wire!) those hubs are aligned as required so I will only need to add 30 radials.


With the analyzer attached it is clear that we are approaching the limit with respect to the antenna's true resonance with so many radials creating a non-resonant ground plane. The additional radials increased the resonant frequency by only about 15 to 20 kHz. This is effectively nil since measurement repeatability at this level of accuracy is unlikely.

Of greater interest is the resistance value. Notice that it is 5 to 6 Ω lower than with 20 radials. That's a big improvement that is evident from on the air receive tests. On all DX paths during full dark the vertical now equals or exceeds the inverted vee with respect to absolute signal level and often SNR as well. This bodes well. I have worked several DX stations with the vertical and, while I did not do A-B comparisons, it does seem to get out well.

Notice that the resistance is well below that of the ideal vertical. That ideal is for a thin vertical (with respect to wavelength) such as a wire. For a fat monopole such as mine the radiation resistance is lower. Unfortunately NEC2 does not handle fat verticals very well and with experimentation I could not push a model lower than 35 Ω over perfect ground. Many hams report fat vertical radiation resistance below 30 Ω. Without a viable model I can only estimate based on the trend line that my vertical's radiation resistance is probably no lower than 25 Ω.

For the moment I will therefore assume that the ground loss is ~6 Ω. This approaches my initial goal of 5 Ω. With the high currents (low impedance) when the antennas becomes a yagi I should see a performance improvement by doubling the radials to 64, since I expect that this would lower ground loss to ~4 Ω and perhaps lower. But not this year.


After reaching this milestone I climbed the tower and raised the stinger to its full height. The resonant frequency moved ~200 kHz lower. This is ~75 kHz higher than the objective I set in the antenna design. I am not concerned since the low radiation resistance compels me to build a matching network for the antenna in its omni-directional mode to meet my SWR objective. Originally I planned a matching network to be switched in when the antenna is in directional (yagi) mode.

With the vertical working -- albeit using the rig's ATU -- the inverted vee will come down shortly for relocation to a different tower. It can have a lower apex and do well for short and medium distance contest QSOs and for select grey line DX conditions.

The ropes hanging from the stinger have been untangled and prepared to act as the upper catenaries for the 4 parasitic wire elements. In the coming weeks these elements will be built and installed and full radial systems laid for them. My objective is 16 radials for each parasite this year, possibly rising to 32 next year. Then I'll complete the matching networks and switching system. Unless I get sidelined by other projects this 3-element vertical yagi for 80 meter will be operational this year.

Monday, August 20, 2018

6 Meter Season Wrap-up: FT8 Conquers All!

It may seem premature to pen a write-up of the summer sporadic E season while the band continues to open periodically, helped along by the ability of FT8 to tease signals out of the noise. Despite this there is little doubt that it is ending. A few days make a big difference on the shoulder of the season, such as a noticable lack of signals when I returned from a family trip to FN30 last week. Now is a good time for reflection, before details slip from my mind.

For me these have been a very good few months on 6 meters. FT8 is obviously the big story due to its universal popularity and, more importantly, it delivered results. Big time. The difference is obvious by comparing this article with the one I wrote after last year's sporadic E season.

In two months using FT8 I made almost 500 contacts and worked 56 DXCC countries. On August 4 alone I put 40 European stations in my log. That was a phenomenal opening that lasted several hours. Many others on both sides of the Atlantic have been similar successful. I haven't seen DX activity like that since my last F-layer openings in 1990, and never via sporadic E.

Now I'll dive into the details. The past 2-½ months have been very educational to me and I've learned a lot about just how far one can push marginal propagation with the help of technology. These paths must have always been there but missed without the assistance of FT8.

Discovery

Some say CW can do as well as FT8 under weak signal conditions using good narrow filters. This may be true, give or take a few decibels. That means little when the opening to any particular station is fleeting because on CW you likely will not find the station since you must spin, spin, spin the dial hunting for them.

When they're spotted it gets easier since you can park on their frequency and wait, perhaps flipping among several spotted stations. It quickly becomes a tedious chore. Yet it can work and has worked for me in the past. FT8 is very different since you are in essence copying the entire band without lifting a finger. When a station is workable on FT8 you'll know immediately.

Discovery is perhaps the greatest advantage of FT8 on 6 meter sporadic E. You need no beacons, no spotting network and no arduous tuning around. When the station is workable you know it. Some DX stations -- CT1HZE in particular come to mind -- CQ continuously when conditions are favourable, essentially becoming beacons themselves, so you may not even need to check for actual beacons at the low end of the band.

Time is of the essence

Of course you must still work the station. This is a challenge since the typical FT8 QSO takes more than 1 minute. If your experience with FT8 is on HF this may seem inconsequential. On 6 meter sporadic E it is often critical since many of the longer DX paths last only this long if not shorter before rapidly fading out.

I have many partial QSOs, including perhaps another 10 DXCC countries, because signals faded, often never to return. I gave examples in my earlier FT8 article.

Despite this duration difficulty FT8 is still superior to CW since with FT8 you have more of a chance to make the contact because of the discovery challenge I described. In particular, you decode the desired station when the path opens so you can take advantage of its full length. Even so you'll need to get used to regular disappointment as you watch a signal gradually fade to nothing in 30 seconds, and there's nothing you can do about it. You can't speed up FT8 transmissions.

To ameliorate this difficulty more callers omit the grid square and jump straight to the signal report on the initial transmission, thereby reducing the time to complete a QSO. They are not being rude by omitting this information. Similarly they'll use RR73 rather than the slower RRR sequence, even if conditions are not truly good enough for RR73 to be a reliable concluding message.

WSJT-X supports these methods for shortening QSOs and they do come in handy when DX openings are breathtakingly short. Give it a try yourself.

Agony of the single decode

Often you never get a chance to attempt a QSO since there is only one decode of the DX signal. In these cases of marginal propagation a signal may peak above the noise for just one 15-second window. This behaviour is normal on the shoulders of an opening, while other times it really is just a single decode. Waiting for the opening to progress may be fruitless.

Examples of single decodes can be seen in the accompanying activity monitor from a few weeks back. There are two of them. Notice that the signal strength is very low. This is typical in the case of single decodes in these types of fleeting openings.

There are other peculiar single decodes that I've experienced. One was a CQ from a YO station with a SNR of -8 db. There were no other decodes and no other signals from Europe, although others in eastern North America were seeing a few marginally copied Europeans. I copied several stations from VE3 down to W4 reply to the YO. No one got through and there were no other decodes.

Unusual propagation must have been present for this to happen. Perhaps a bolide over the North Atlantic briefly provided a bridge between E-layer clouds. Intense meteor trails can support propagation on 6 meters long enough for a 15-second FT8 message. That's just speculation since I have no idea what happened. FT8 is providing us with new insights into propagation.

Signal reciprocity

When I first got on 6 meter FT8 I was surprised at the asymmetry of signal reports I sent and received; the reports I received were mostly worse than the ones I sent. I attributed it to the several decibels of loss in the 40 meters of ancient RG213 coax feeding the yagi. Now I'm not so sure.

As I said in my earlier article FT8 is no more a low power mode than any other. Many 6 meter operators have long used high power and they continue to do so on FT8 since it pays dividends in teasing QSOs out of marginal propagation, which is typical for long DX paths on sporadic E and other transitory propagation modes. I have become accustomed to signal report differences of 10 db or more, and often not being decoded by stations received well here.

There may be other reasons for the difference. For example, noise. Because WSJT-X signal reports are the calculated SNR (signal-to-noise ratio) each station's noise floor is a factor. Although external noise (QRN) is usually lower at VHF than HF there is still the receiver noise to be considered. I live in a low noise environment where I am able to turn up the receiver's pre-amp to improve SNR, a common requirement on VHF and higher where atmospheric and man-made noise is below the receiver's thermal noise level.

When is a QSO complete?

I was pleased to see this issue mentioned in the recently arrived September QST. There has been some confusion on the air, including those I try to work. For transitory 6 meter openings one's view of the matter becomes important.

The issue is exemplified by a QSO that proceeds as follows:
QQ8ABC VE3VN FN24
 
VE3VN QQ8ABC -10

QQ8ABC VE3VN R-08
  VE3VN Q8ABC RR73
Q8ABC VE3VN 73
  VE3VN Q8ABC RR73
Q8ABC VE3VN 73
The other station insists on receiving a "73" message to complete the QSO. It isn't necessary since both stations have exchanged and confirmed call signs and signal reports. When conditions fade -- as they frequently do for marginal DX paths -- the "73" is never received. Should I log the QSO?

My concern is whether the other station logged the QSO, not whether the QSO is valid. It is valid. My choice then is to log the QSO or not. In almost every case I do. Whether he logged it may only be known when it is confirmed in LoTW or via some other system. In one instance the questionable QSO was a new country, which I logged but did not include in my DXCC worked list.

Quirks

In my previous FT8 article I mentioned a few quirks of the software and its users. Here are a few more.

I call CQ, someone answers on my transmit frequency and we complete the QSO. They then proceed to call CQ on frequency. This happens more often than you might imagine. I suspect carelessness rather than malice. Most likely they forgot to QSY back to their previous transmit frequency after we've worked. All I can do is QSY, which is a minor inconvenience on FT8 compared to CW or SSB.


WSJT-X occasionally decodes noise as a valid message. The frequency of these occurrences increases with how aggressive you set the decoding options. This is no surprise to FT8 operators. What does surprise me is that some of these messages are addressed to me! If it occurs while I'm CQing with "Call 1st" selected the software will happily reply to the noise. I've seen numerous instances of others answering what are obviously nonsense call signs magicked into existence by WSJT-X. There is something in WSJT-X that causes it to try to preferentially interpret noise as your call sign.

WSJT-X occasionally loses track of its transmit state when a transmission is manually halted within the first second. You press the appropriate button and it goes on transmitting, while also receiving. With the rig's monitor feature enabled (which is usually the case here) the transmission is received and decoded. It is highlighted since my call sign appears in the message. It takes a couple of clicks of the "Tune" or "Enable TX" button to fix the problem.

In a good opening the full 3,000 Hz is densely occupied. Stations overlap since, due to the spotlight nature of sporadic E propagation, I can hear both stations but they can't hear each other. This is less of a problem that it might be because WSJT-X is astonishingly capable of decoding each of the overlapping signals, even when they differ in strength by 10 db. That's impressive, and useful. It's a skill even the best DXpedition operators would envy.

Contests

After observing FT8 activity on 6 meters during ARRL Field Day and the CQ WW VHF contest I've chosen to not contest with FT8. I like DXing and I'm happy to periodically work US and Canada on FT8 (especially west coast), but for contests it's just too slow for my taste.

There are other difficulties with FT8 use in contests that may be corrected when WSJT-X 2.0 arrives in several months. If you're already using FT8 on VHF the following should be familiar.

First, there is inconsistent use of NA VHF contest mode in WSJT-X. When selected the messages change such that stations decoding your transmissions will see messages that are seemingly incomprehensible. Unless both stations select (or deselect) this option QSOs aren't possible. Endless repeats between incompatible stations were often heard both weekends. Selecting the option midstream or between QSOs does not retroactively translate messages already received. I was puzzled by what I saw during Field Day since I hadn't yet learned about NA VHF contest mode. The manual enlightened me. Others remained confused.

Many NA stations incorrectly selected this option during the CQ WW contest. It is not intended for worldwide contests and none of the European and other DX used it. Once again confusion reigned. I avoided the option entirely while working DX during both contests.

After each contest many forgot to deselect the option. They had little luck working anybody! Eventually normalcy returned to the bands.

Unfortunately when I did try CW and SSB in the CQ WW contest I logged exactly zero contacts. I heard a few stations fading in and out, but failed to work the few that were present. Ten minutes of continuous CQing netted no replies at all. Yet FT8 activity was high. Perhaps I'll avoid VHF contests for a year or two until something changes to rekindle my interest.

50.323 MHz

Used a great deal for intercontinental openings early in the season its use recently has been almost nil. During the exceptional opening to Europe on August 4 the 50.313 MHz segment was packed with signals from 200 Hz to 3,000 Hz. During each of the several times I checked 50.323 MHz there were no signals found. Neither did my band scope show a blip +10 kHz the many times I glanced at it while active on 50.313 MHz.

Is the idea behind 50.323 MHz dead? We'll have to wait and see how this develops. My suspicion is that for the majority the importance of discovery (discussed earlier) is greater than reduced QRM from intracontinental activity. As matter stand I see no reason to monitor 50.323 MHz.

Signal quality

Monitor FT8 on any band and you'll soon discover poor signals. Perhaps because these digital modes are new, the software unfamiliar or the operator carries on with poor practices learned on other modes many seem unable to consistently transmit a clean FT8 signal. This not only hurts them but everyone using FT8.

Based on the WSJT-X spectrum display and other displayed data it is possible in many cases to make a good guess what they're doing wrong. Some that I'm aware of include:
  • Misadjusted clock: WSJT-X will forgive small clock errors but can be intolerant of errors greater than 1 second. It is mystifying that many operators don't know they have a problem since WSJT-X displays the clock error on every message. You'd think they'd notice that every signal has substantial error. Either they don't notice, imagine everyone else is wrong or don't understand the importance. The most common consequence is that they can successfully monitor but their transmissions overlap with others resulting failed decoding when a QSO is attempted.
  • Drift: In this era of rock solid digital oscillators we have grown used to frequency accuracy of 10 Hz or better. Imagine my surprise to see that a small fraction of signals drift. Sometimes the drift is only 1 to 2 Hz every 30 seconds, and in extreme cases I have seen drift of up to 10 Hz. Since WSJT-X highlights messages addressed to you (your call sign) you still see messages that drift outside your receive window, so it is usually not a serious problem.
  • Compression: Speech compressors by their nature -- both AF and RF -- are non-linear devices. When properly adjusted on SSB they achieve a balance between distortion and comprehensibility while improving average talk power. On FT8 the compressor produces mixing products that create QRM and reduce the power of their fundamental audio signal.
  • Transmitter ALC: When the ALC meter moves there are distortion products due to the introduction of non-linearity in the RF amplification chain. This may be tolerable at very slight ALC action but is always dangerous. When it kicks a lot you can be certain that you are generating spurious signals.
  • Amplifier over-driven: In the quest for squeezing out those elusive DX QSOs many run a lot of power on 6 meter FT8, and it is no surprise that those amplifiers are sometimes over-driven. Since negative feedback to the transmitter ALC is frequently not used there may be no immediate indication to the operator that there is a problem. But as we've seen ALC is not the cure for a poor FT8 signal. Carefully adjust the transmitter power, and observe its ALC meter if it has one, to ensure the amplifier remains in its linear range.
Despite the presence of distortion products that pollute the spectrum for everyone else you can still be make QSOs since the fundamental signal is also present. IMD products can actually spread your power among 1 or 2 adjacent signals that are decoded. You may not notice these problems unless a friend (or one of the ubiquitous policemen) tells you. Pay attention. Everyone, including you, will benefit when your transmissions are clean.

WSJT-X decoding is so good at discarding distortion products that most of us can survive the onslaught of QRM. However that is no excuse for transmitting a poor signal.

CW and SSB

Well, what about CW and SSB? Once I made the move to FT8 in early June I have worked very little CW and SSB. Not only has FT8 been far more productive for my primary objective to work DX -- despite the painfully lengthy QSOs and sudden QSB -- when I switch to CW and SSB there is often no one there. Few signals other than beacons are heard and CQs go unanswered. Yet 50.313 MHz is bursting with activity.

I have no answer to this issue, and I can't even claim that it is an issue. There are many who shun FT8 and similar digital modes. To them it can be annoying to have fewer stations to work. I felt the same before June. As a practical matter, when I'm set up for FT8 it is not convenient to tune the CW and SSB segments. I would have to come up with a system whereby a second rig can share the antenna.

While I am not prepared to give up on the "legacy" modes there seem to be fewer reasons to use them as time progresses. In this I refer only to 6 meters, not HF. On HF my interest in FT8 is very low. The only band I may play with it is 160 meters, if only to see what it can do.

My plans

My move to FT8 on 6 meters was one of necessity. It was that or make far fewer QSOs. After one season on FT8 I've become a believer. FT8 delivers the goods: an astounding amount of DX, which is my primary operating objective on 6 meters.

This success motivates me to improve my performance on 6 meters in 2019. First up is to replace the poor coax with low loss Heliax. That will improve my signal by at least 3 db. When I shop for an amplifier I would now like one that includes 6 meters. In time I may add one or two more antennas, for stacking and elevation angle diversity.

Together these measures should improve my DX results on 6 meters. Despite my renewed enthusiasm for 6 meters this work remains lower priority than towers and antennas for HF contesting. We'll just have to see how far I get by next spring. I am also considering playing with other WSJT-X modes to try out other forms of VHF propagation such as meteor scatter.

Despite FT8 being less hands-on that traditional modes it has earned a place in my station. That is perhaps the biggest surprise to me this year. Who knows, 6 meter DXCC in another year or two could be mine thanks to FT8. That this is possible during a solar minimum is extraordinary.

Monday, August 6, 2018

Hamplus AS-82 2×8 Antenna Switch

Last year I purchased a used Hamplus AS82 2×8 antenna switch from a fellow contester. The specs are good and the brand seems to be in good regard. At first glance its construction appears sound and it is quite easy to wire up and use with a custom controller. That is, once I discovered how to drive the unit. Documentation is sparse and schematics do not appear to be published.

Pictures are from the Hamplus web page linked to above

From the start I had problems with it. I thought, perhaps, it was due to abuse, and I was unhappy that the seller did not disclose the problems. Due to the nature of the problems and his particular application it is possible he didn't know, therefore I will give him the benefit of the doubt. Regardless of this circumstance the problems are real and deserve an airing, since as I discovered the problems are inherent in the product.

As the winter season progressed the problems became increasingly manifest. Perhaps the intense cold played a part -- it is mounted outdoors with a precipitation cover. Making trips outdoors at night with a flashlight when the temperature is -20° C and the wind howling to effect repairs in the midst of a major contest is not acceptable! However, spring warmth did not bring relief.

As part of the re-cabling of the station this summer I had to repair the switch or replace it. It is critical to station operation. An antenna switch, even one designed for multi-op and SO2R antenna sharing, is not complex. The challenges are in reliability, electrical safety, fail safe to prevent one antenna being selected by both stations (lockout) and port-to-port isolation. I was not deterred by the lack of a schematic.

While not intended as a review that is in part what this article contains. I'll relate what I discovered about the product as I searched for and repaired the problems I experienced. The process was educational.

Problems

I'll start with a concise list of the problems I experienced with the switch:
  • One antenna port (#8) could not be selected on one side since selecting that control line caused a power supply short. My power supply has electronic short circuit protection so nothing dreadful occurs.
  • One side experienced intermittent disconnection or high resistance on most antenna ports.
  • The other side had one non-functioning antenna port and one that was intermittent.
The lack of a schematic I already mentioned. Fortunately the circuit is simple enough and the parts large enough that signal tracing and testing isn't too taxing an effort. However removing the PCB proved to be quite a puzzle. That is, it could be removed with some difficulty but I could see no way to reassemble the unit. Yet the PCB had to be removed to diagnose and repair the unit.


Modifying the mechanical design

In the picture notice the narrow opening between PCB and enclosure and the blocking of access by the vertical flange. There are two sets of screws and nuts securing each of the 10 SO-239 UHF bulkhead connectors and the connectors are soldered to the PCB. The nuts are inaccessible except for one on each of the 6 outer connectors.

With some difficulty the nuts can be held in place while removing the screws. Holding them in place for reassembly is beyond difficult. I needed a better way, one decided before opening it up. After some thought I tentatively chose a technique that would work. I asked some friends if they had any better ideas but none was forthcoming. That convinced me to proceed on my chosen path.

Since the product is made in Brazil it is not surprising that metric fasteners are used on the SO-239 connectors. Despite the connectors being specified in English units -- the flange holes are ⅛" (~3.2 mm -- the 3 mm stainless machine screws and nuts are a comfortable fit. The screws have bevelled heads to seat within countersunk holes in the enclosure. The screws provide the electrical connection between the connector flanges and the enclosure to ensure a continuous ground plane.

A narrow steel tool was slipped between PCB and enclosure to jam the nuts against the connector bodies. Most screws had to be loosened this way. The quality of the stainless steel alloy is so soft that several screw heads (Phillips) stripped under moderate torque. These were drilled out on a drill press.


I elected to tap the connectors so that nuts are not required to secure the PCB. It is not an ideal solution because the metals used in SO-239 connectors are not designed for this usage it. But it does work. The base metal is typically brass with thin nickel plating. Tapping removes the nickel within the flange holes and so cuts threads into the base metal. Some care is required to avoid stripping the threads when tightening the new screws.  Lubricating the threads can help to avoid excess torque for future removal provided that the lubricant does not prevent a low impedance path from screw to flange. Dielectric grease may be a good choice.

Original 3 mm fastener (left) and new 4 mm screw
The smallest standard screw sizes that will do the job are either #8 or 4 mm. A #6 screw will dig into the nickel a small amount, even without tapping, but it is not enough. Tapping the flange holes for either #8 or 4 mm does not require drilling the holes to a larger diameter since the metal is soft enough for the tap to bite in.

As you can see in the picture I was not concerned with cosmetics since the flanges are not visible when the unit is assembled. Where you must be careful is to keep the tap at a right angle to the flange surface so that the threads are not cut at an angle. The flange is thin enough and the metal soft enough that this is an easy mistake to make.

I chose 4 mm screws for this project because of the local fastener emporium's wide selection of metric stainless steel machine screws. I could choose the perfect length to fully pierce the connector flange without getting near the PCB. Also, my metric taps get very little exercise!


I widened the enclosure holes for the larger screws using a large drill bit -- visually similar angle on the screw bevel -- to broaden the countersunk holes. When all was done it was easy to fit together the PCB and enclosure and use the new screws to hold it all together. Mission accomplished.

Connector attachment to the PCB

Did you notice that there's a connector missing in these pictures? When I removed all the fasteners and lifted the PCB from the enclosure that connector fell off. Both solder joints had failed. That certainly explains the observed intermittent behaviour of one side of the switch!

Each connector has two wires attached. A stranded wire is soldered to the centre pin and pierces the PCB, soldered to pads on both sides of the PCB. The hole is much larger than the wire so there are collars to bridge the gap. You can get an idea of the arrangement from the pictures shown above.

A solid wire is encased in a large blob of solder on the flange surface and pierces the PCB in the same manner as the centre pin but with a narrower hole. This wire is quite fragile and prone to fatigue and breakage as the connector wiggles on its wire supports. The wires are only protected when the connector is screwed to the enclosure. In a couple of case the solder blob did not hold the wire securely and broke off, and another broke during repair. The wires are not mechanically connected to the connector flange; solder alone does not make for a reliable connection.

This is good evidence of how the unit was most likely manufactured. However I cannot guarantee it was done this way, and I can imagine a couple of variations. But here goes:
  1. Fasten the connectors to the enclosure with the screws and nuts.
  2. Solder wires to the connector and position them in alignment with the PCB holes.
  3. Place the PCB on top of the connectors and wires and solder them to them PCB.
Otherwise they must have special tools for slipping in and holding in position the nuts for the connector fasteners. Either way this is not a product intended for servicing by the owner, and perhaps not even for servicing by the dealer. It is therefore highly unlikely the previous owner had disassembled the unit. There is also no evidence of modification or post-manufacture soldering.

I replaced wires that had broken off from the pins and flanges. To ensure the affected connectors would sit flush on the enclosure I used the new screws to secure the connectors to the enclosure, sat the PCB over them (with the unmodified connectors slipping through their respective holes on the enclosure, then soldered the wires to the top of the PCB. With the wire lengths now assured to be the correct length I removed the PCB and completed soldering to the pads on both sides of the PCB.

Resolving the short

Finding the short on the port #8 actuator took a little detective work. First I examined the voltages and resistances on the working ports for comparison. This led me to conclude there was a short across the coil of the front relay in the chain. Since the relays are PCB mounted and the board is two-sided the fault had to be on the PCB traces under the relay body, in the relay itself or elsewhere. Lifting the relay is no simple feat.

I decided to focus on the third option since relays of this style rarely fail in this manner and the PCB layout and soldering is generally clean. Excess solder flux along the board traces were removed to ensure no solder bridges were hiding beneath. With nothing found I was left with a capacitor and a diode to examine. The capacitor shunts RF from the control lead to ground and the diode across the coil protects against back EMF when the control voltage is removed.

The capacitor checked okay so I lifted one end of the protection diode with a soldering iron. That solved the problem. The diode tested short. Look at the (fuzzy) picture for a close-up of the bank of back EMF protection diodes. From the package design and lettering it is evident these are 1N4148 switching diodes. They are not suitable for this application. The diode junction probably failed due to an unusually high voltage spike from the collapse of relay coil's magnetic field the diode is there to protect the circuit.

I replaced the failed diode with a 1N4007 from my ample stock and all was good. I really should replace the other 7 protection diodes at some point but chose not to do so at this time. With the changes documented earlier the PCB can be quickly removed to do it when I am ready.

Testing, reassembly and installation

With the unit reassembled I built a temporary wired connector (DE25) to apply power to the unit and a probe to ground the selector pins. With an ohmmeter I tested each port for continuity and equal low resistance value, then did the same for the other side of the switch. The PCB was then fully attached to the enclosure and again tested in the shack with real antennas and rigs. This test was for expected port isolation, SWR and receive and transmit performance.

Before I began repairs I was concerned that one or more relay contacts had worn due to poor quality or hot switching with high transmitter power. The previous owner, like me, used a custom control unit that may not have had protection against hot switching. A TX enabled signal is available from most current transceivers that control units can use to prevent antenna switching while the transmitter is active.

Luckily all the relays tested okay. Although these sealed DPDT PCB-mount relays have an industry standard form factor (2c) and are inexpensive it would be a lot of work to de-solder the 8 pins on each, and there are 24 of them. Plug compatible replacements that may be more available in North America include the Omron G2RL-24 with 12 VDC coils. I have not confirmed this so do your research before you replace one of the Tianbo relays.

When that test was passed the enclosure was fully boxed and reinstalled at its outdoor location. It now performs flawlessly. Hopefully this time it will continue to work as it should.

Last words

My purchase of the Hamplus AS-82 2×8 antenna switch was done on impulse to solve an immediate need. I did not want to buy new since it was the early days and I did not have a definitive plan for my antenna switching architecture, including physical layout, antenna/band hierarchy (flat or switching per band) and control system for SO2R or multi-op. I still need this switch as an interim solution for at least the coming winter contest season.

The design of the AS-82 seems sound and when it works it works well. The PCB quality is excellent. What is not so excellent is its construction and parts selection. I expect better for the price. I also expect a schematic and detailed interfacing specs. I can only hope their other products are better built. 

Many of the big guns design their own switching systems, eschewing commercial products entirely. Others buy from among several available high-end systems, including those that support more than two operating positions. The small number of dedicated multi-multi stations have simpler requirements since each operating position is only concerned with selecting from one of a small number of dedicated mono-band antennas.

As my station grows it is possible that the Hamplus switch or one very similar could form a core component of my switching system, supplemented with secondary switches for antennas on each band. The switching and control system design is ongoing. That is a subject that is very likely to appear in an article later this winter.