Balanced link coupled tuner may be the best solution

[k4kyv]

Looks to me like the good ol' link coupled balanced tuner may be the most versatile after all.  My 80m tuner uses a 3-turn link with a series capacitor, fed with 50Ω line. The output coil is series-tuned to the resonant feeders, which are approximately 1/2λ long, making the feed point low-Z somewhere near 70Ω ± the reactance. (Wish I could find a reliable antenna impedance bridge that would directly measure a balanced load, not just unbalanced loads grounded on one side).

Series tuning is accomplished by splitting the main coil exactly at the mid-point and attaching the OWL across the split, and resonating the main coil with a split stator capacitor that has the same plate spacing as the transmitter tank capacitor. When I first built it decades ago, I was able to quickly adjust it to a perfect match at 3750 by adjusting the series capacitor in the link circuit along with the main split stator cap.  I can then QSY anywhere between 3500 and 4000 and get within a 1.1:1 match simply by rotating the main split stator cap without touching the series link capacitor.

I have been trying to come up with something equally versatile to be be fed with a 440Ω balanced untuned line.  First, I tried substituting the 3-turn link with a couple of taps on the main coil near the mid-point, placed equidistant from the split.  I was able to carefully adjust the placement of the taps to get a perfect 440Ω match at 3750.  But the damned thing had such narrow bandwidth, that I couldn't QSY more than ± about 50 kc/s and still maintain anywhere near a  reasonable match, without moving the taps on the coil in addition to rotating the capacitor.

Next, I tried a balanced L-network.  Calculated the L and C values using one of the on-line calculators, and  constructed a prototype tuner.  The tuner was first used to feed a 75Ω dummy load, made from Glo-bar resistors.  I got a near-perfect match the very first try, and could QSY all the way from 3.5 to 4.0 and maintain a perfect match with nothing but a slight adjustment of the C in the  L-network.  But when I tried feeding the real antenna, same problem.  Very narrow bandwidth, and tuning beyond  ± 50 kc/s required varying the inductance (by moving taps on the coil) as well as adjusting the capacitance. Apparently, the simple L-network couldn't handle the variations in the reactance by adjusting only the C. I spent the entire day yesterday working on the L-network, with nothing to show but the knowledge that it won't work as planned.

So, it's back to the drawing board. The next attempt will be to go back to the original link-coupled balanced tuner, but increase the number of turns in the link to make XL equal to 440 ohms, and tune it with a couple of series capacitors, one at each end of the link. The present 3-turn link calculates almost exactly to XL=50Ω. So the 440Ω link in the form of a separate coil should work exactly the same as the 3-turn low-Z link, with equal bandwidth and not affect the tuning of the main coil any differently, unless I am completely overlooking something major that should be very obvious.

I can't believe it could be so easy to make a link coupled tuner to match a low-Z source to the tuned line, adjustable across the entire band by merely rotating a single capacitor, but impossible to come up with some kind of tuner that will match a medium-Z line (440Ω) that doesn't require moving taps around on the coil in addition to rotating the capacitor.

[W4NEQ]

Two suggestions:

1.  Using my GR1606B or 2180VNA to measure balanced loads -  I have had decent luck using a #61 ferrite torroidial balun wound conventionally with two bifilar conductors for a 1:4 step up.  The low impedance (unbalanced) side has enough turns to produce Xl >= 6 times your highest expected impedance / 4.   For instance, if you expect to see, maybe 600 ohms on the balanced side, one winding by itself should present at least 900 ohms Xl at the lowest expected frequency.  THEN:

Connect a known 200 ohm resistance to the balanced side, and balance your bridge at 50 ohms.  Your bridge will indicate 25% of the Z.
My results have been reasonably accurate (within 10-20%) until the loads' reactive component exceeds the resistive portion.

2.  With your link coupled tuner - I recall seeing a pretty high turns ratio, and I thought you mentioned arcing issues.   Might be worth a try using more parallel C (if you have it) tapped down to significantly fewer turns - less circulating current, less voltage, lower Q network, better bandwidth ...

Edited clarification:  When I said "wound conventionally"  I meant Ruthroff style.

[K5UJ]

Don, it appears you are the victim of an extremely efficient antenna    It's a problem but it's also a nice problem to have.  This circuit is the link coupling to get you from the balanced line to the unbalanced load at the tower?  or is this a matching network at the tx?  Sorry for being confused.

[k4kyv]

This is the balanced 80m matching network from the balanced line to feed the balanced dipole on 80m. Nothing to do with the unbalanced vertical.

[aa5wg]

Hi Don:

By cutting your tank coil in half is a good way to go.  This provides TWO opptions for current feed (low impedance).  As you know, you still have traditional parallel feed for high impeance, voltage feeding.

If there is a low impedance point at you tuner and the line is just a tad short then use series feed from the split tank coil.  However, if the low impeadace point at your tuner is a tad long then use series tune with two output tank capacitors for each side of the line.  

With low power, i.e. 5 watts or so, try tuning the coupler with no line attached to coupler and write down your coupler settings, high Z.  Next, tune the coupler with the left and right output sides shorted
together, low Z.  Write down these settings.

Next, attach the line to coupler and pre set the coupler to above settings.  If the coupler is not close to working then shunt capacitance or inductance acrosse the line and try again.   By doing so the coupler will handle the reistive portion of match and the added shunt coil or capacitor will handle the reactance portion of mathing.

Make sure you are using a variable input capacitor with your link coil.    "The reactance of the link coil (and hence the reactance of the capaciotor setting which will resonate the coil) should be about 3 or 4 times the impedance of the transmission line between the antenna coupler and the harmonic filter, so that the link coupling circuits will have an operating Q of 3 or 4." Reference: radio handbook, 20th ed., William t. Orr, W6SAI, page 26.19, second printing, 1978.

Chuck

[k4kyv]

I see a lot of recent interest in balanced link coupled tuners.

Here's the latest episode in the saga of my 80m tuner.  

Yesterday, jury-rigged (or JSed) 7 additional turns onto the link, making the total 10 turns.  With a 360pf variable in series with each end of the link,  connected it to the 450Ω OWL and checked for reflected power with the Micromatch.  Reflected power was enough to make the meter almost hit the pin. Shorted out one of the capacitors, and could get a slight reduction with the remaining cap fully meshed.  So removed both capacitors and attached the OWL directly to the link without any tuning caps.

With caps removed, I could null out the reflected power at 3750 almost to zero by rotating the main cap. At 3500 it would null right to zero, but only with the tuning cap fully meshed.  At 4000 it would null close, but not quite to zero with the cap less than half-meshed.  With this set-up all settings of the main cap at the 3 frequencies are shifted towards more capacitance, but over a wider range of settings than with the tuned 3-turn link fed with 50Ω line.  

Next step is to try removing a turn from the link to see if that improves anything, or add to the link if necessary.  Also, to remove the line that goes to the antenna from the tuner, short out the gap, and use a neon lamp to find the natural resonance setting with the main coil & cap, to see if the antenna shifts the main capacitor setting towards more or less capacitance.

I believe the secret will be to find the exact number of turns on the link that will retain something close to the original settings with the higher Z feedline and no capacitor, since this arrangement seems to offer the best bandwidth with the simplest adjustment.

As a final test, an RF ammeter will be inserted in one leg of the tuned feeder and the TX loaded up to a pre-determined power level, and rf current readings compared between the 50Ω link back to the shack, and the 440Ω one.  Apparently, there is substantial reactance at the tuned transmission line feed point, since the L-network behaved completely differently from with the 75Ω dummy  load substituted for the feed line.

The antenna is almost exactly cut for a half-wave in the middle of the band, and the tuned feeder is a little over 127' long between antenna feed point and tuner.  With the near half-wave long tuned feeder, impedance at the feed line input should be close to resistive at 75Ω, but apparently it varies too much and is too reactive for the balanced L-network to handle without varying the number of turns in the coil along with the setting of the capacitor. The balanced link coupled tuner appears to adequately compensate for impedance variations across the entire band when adjusting only the main tuning capacitor.

[aa5wg]

This is one approach I use with your antenna system setup, current feed, low impedance.

(1)  Power from rig is in series to a single input link variable capacitor.  Xc at 3.5 MHz is 227 pf for this capacitor.  Your 360 pf variable capacitor is perfect for this.  

Use a female input coax connector.  Male coax connector, from rig, plugs into this female coax connector which is bolted down on your antenna tuner.  The output of this female coax connector goes directly to your link input variable capacitor.

(2)  Output of of above link variable capacitor is in series with link coil.  This coil is built to have 200 ohms of XL or 9.1 uH at 3.5 MHz.   This coil should be variable.

(3)  Output of link coil is connected/bolted to a small brass plate (i.e. 1 inch x 4 inches) which the input coax is connected/bolted to.  There is no extra connection to ground here.  

For the tank coil and capacitor try this for a change.

(4)  Physically cut your output tank inductor electrically in half.  Connect your variable output tank capacitor electrically in the middle of tank coil.  You are physically cutting the tank coil electrically in two parts and now have a coil-capacitor-coil series combination. 

The tank coil should have a minimum of 23 uH or 500 ohm of XL (22.7 uH). This is 11.35  uH (12 uH rounding up) on each side of the tank variable capacitor achieving a total of 22.7uH (or 23 - 24 uH rounding up).  

(5)  Leave the ends of the tank coil floating.  Or, loop the end of each coil back onto itself.

(6)  Taps, from transmission line, are attached symmetrically on left and right sides of tank coil.  

(7)  Use maximum values of input link capacitor and link coil to achieve resonance.

Use what ever tank inductance required for resonance.

(9)  Use flashlight lights on both side so line to indicate maximum current output when tuning.

Chuck

[aa5wg]

Don:

I was thinking about your antenna system this morning and remember reading
something about overall antenna system length.  I am looking for the reference.

I believe the reference stated the overall length of the antenna system should be 10% longer than ideal, i.e. your 80 meter 3 quarter wavelength long system would be considered ideal.

It seems the reference said this 10% additional antenna system length facilitates the antenna coupler when tuning antenna system to resonance.  I think what the author is saying is this added 10% length gives the antenna couple more to work with.

I could be all wet but will continue to look for this reference and then post it.

Chuck

[k4kyv]

Mine may be slightly on the short side, since the halfwave feed line is only 119' plus the 8' or so running from the base of the tower to the dawg house.  With the 50-ohm input to the tuner that I have been using, it doesn't seem  to matter; the tuner will make a perfect match from 3.5 to 4.0.  Now, it's just a matter of getting the link circuit properly adjusted for the higher Z feed line.  Reactances transfer through the tuner, so if the tuner is looking into a reactive load in the tuned transmission  line running up to the  dipole, the reactance may be tuned out by introducing an equal and opposite reactance, using a combination of altering the setting of the main capacitor, and introducing additional reactance in the link circuit. Or the transmission line back to the shack could run a little SWR, and the remaining reactance tuned out with the transmitter-to-transmission line coupler in the shack. This is my interpretation of what the much-discussed "conjugate match" is all about.

This explains why a homebrew link-coupled tuner or one like the Johnson Matchbox with capacitor in series with the link can usually be adjusted to a perfect match by a combination of settings of the main tuning capacitor and of the capacitor in series with the link. If the inductance of the link is carefully adjusted, the series capacitor can usually be eliminated from the link circuit altogether for a perfect match at the middle of the band, and the match will still be acceptable at the extreme ends. A variable link will usually do the same thing as the series capacitor, by adjusting the mutual coupling between link and main coil.

If the total overall length is slightly too long or too short, that allows the tuning adjustment to work with only one polarity of reactance, either a varying degree of capacitive or of inductive.  You are not having to tune out inductive reactance at one end of the band and capacitive at the other using the same tuner.

Please do post a link to that reference if you find it.

[W0BTU]

I finally started using my homebrew balanced link coupled tuner last week. I built it over 30 years ago, but it had just been sitting here collecting dust for over 8 years.

I almost forgot how wonderful such a design is. It's good to be able to tune to almost a perfect match anywhere on any band from 80 through 10.

Photos at http://www.w0btu.com/files/antenna/Balanced_antenna_tuner/ . (Sorry about the duplicates, I need to delete a few.)

The two SWR bridges were so I could read forward and reflected simultaneously while I figured out where to connect the feedline, and which coils to use for 60, 30, 17, and 12 meters.

One thing that puzzles me is that on one or two bands (forget which ones), the SWR at the rig (45' away from the tuner at the opposite end of the basement) is a little high, even though those SWR bridges at the tuner show a 1:1 match. The only thing I can think of is that the ~60' run of Wireman "Mil-spec" RG-213/U is not exactly 50 ohms (and yes, those SWR bridges are not exactly lab quality). :-)

[aa5wg]

Here is the quote regarding the 10% longer tip.  This would be about 6 feet and 3 inches for 3.75 Mhz at .95 velocity factor.  I only have this quote and no link.

Quote:

"Usually it is preferable with tuned feeders to have a current loop (voltage minimum) at the transmitter end of the line.  This means that when voltage-feeding an antenna the tuned feeders should be made an odd number of quarter wavelengths long, and when current-feeding an antenna the feeders should be made an even number of quarter wavelengths long.  Actually, the feeders are made about 10 per cent of a quarter wave longer than the calculated value (the same value given in the tables) when they are to be series tuned to resonance by means of a condenser instead of being trimmed and pruned to resonance."  

End Quote.

I believe this information is reference to a half wave doublet.

The reference is from Radio Handbook, 17th edition, issued by the Editors of "Radio", page 405, published an distributed by Editors and Engineers Limited, 1300 Kenwood Road, Santa Barbara, CA, First Printing October, 1940.

Chuck

[aa5wg]

Don:

Last year I was testing a 75 meter doublet and was current feeding it.  It was interesting when I tried current feeding via a split tank coil (one coil in series with each side of transmillsion line) and then with a single tank coil with two output capacitors (one capacitor in series with each side of transmission line).

One series system worked better than the other.  At present, can't say which one was best but there was a significant benifit with one series system over the other series system.

With another set up, and current feeding, the other series system worked best.

I try all feeding circuits and go with best results.

Chuck

[W4NEQ]

We are all familiar with using tuned feeders with a 130 foot dipole and tuning it in the shack with a link-coupled tuner such as a Johnson Matchbox.  That's what I use.

Don's arrangement is a bit unique.  He has essentially that antenna with nearly an electrical half wave line length so that he can expect close to the dipole's true impedance at ground level.

THEN he transport's it another 140 feet horizontally to his shack.  His elaborate tuner at the tower base required only adjusting a vari cap remotely to tune the whole band to 50 ohm coax.  I thought that was pretty slick, but to produce lowest possible line loss, he is wanting to transport the RF horizontally on 10 gauge 440 ohm open-wire feeder, OPERATING IT AT ITS CHARACTERISTIC IMPEDANCE.  Not many folks do that.

You might ask why not just connect the two open wire feeders and tune it in the shack?

As would be expected with long tuned feeders, bandwidth is poor and every QSY requires re-tweaking. 

As slick as he had it working with 50 ohm coax on the horizontal run, I think I would just upgrade to 1/2" heliax.  (about 0.13 dB/100 ft @ 4MHz)

http://www.rfparts.com/pdf_docs/AndrewHeliax/LDF4-50a.pdf

And I'm a big fan of open-wire feedline.  I've never been lucky enough to be able to tune just one network element and get a nice match across the band.  Hope you get it worked out Don.   I know you hate to use the newfangled stuff.

Chris

[k4kyv]

I measured the balanced line to be 98% efficient, while the RG-214 and 213 were both only about 92% efficient. Heliax might be close to the OWL in efficiency, but I have a fetish for symmetry; dipole fed with balanced line, balanced tuner, push-pull balanced final, therefore balanced line from transmitter to tuner at the base of the tower.

Looks like the balanced link-coupled configuration has enough bandwidth to get by with tuning only one network element (the variable capacitor in the main tuned circuit) to QSY across the band with the 440Ω line, just as with the 50Ω coax. This is achieved without a variable capacitor in series with the 440Ω link, by carefully adjusting the number of turns in the link.  The OWL feeds directly into the link.  I tried 9 turns on the link, 9 1/2 and 10.  This seems to be right at the sweet spot.

I observe that the main tuned circuit resonates with much less capacitance as a natural tuned circuit with minimal coupling using a 1-turn link and no load (dead short across the split coil, tuned for maximum brilliance with a neon lamp), compared to working into the real antenna or a 75Ω dummy load using the full size link.

Next test will be to wire an rf ammeter in series with the output of the tuner and compare readings at the same power level, running the balanced OWL from the shack, to the 50Ω coax line, using the appropriate matching networks with each.

80/75m is the most problematic in this respect, since percentage-wise to the operating frequency, this band is wider than 160m or any of the HF bands, including 10m.