One - Do- Tree... Tests with the 2x2x2 40M Yagi stack

[WA1GFZ]

Tom,
The VSWR measured at the TX reflects the fact that I lopped off a foot of wire at each insulator when I was closer to the ground. I'll dial the rest of the system in when the coax lengths when I get the antenna right. I agree with HUZ the antennas will not be perfect 50 ohms in a force fed system. Read the article i sent you. fc

[K1JJ]

What do you figure the swr (input impedance) will be for each dipole feedpoint and what is the expected total length of RG-213 coax for each dipole?

Remember that even 1/2 db of additional loss due to swr = ~12% of your forward gain.  In contrast, the  parasitic array competition has none of this additonal loss...

T

[WA1GFZ]

Good point 10 MHz. RG213 loss is .66 dB/100 feet plus VSWR loss. I guess this is the price I pay for force feed with the option to flip the pattern.
I wonder what loss you would see in a Stub switching system like Steve has? I wonder what loss you see in a 1/12 wave transition or a wilkensen.
A single antenna in a single direction is best for sure.

[WA1GFZ]

Tom the article I sent you said the two dipoles look like 122 -J41 and 34 J39.
I dumped the values into the transmission line calculator to see what 100 feet of RG213 did with the load. Matches loss at 7.15 MHz .53 dB. Total loss of each is .76 dB and .74 dB. So we are talking less than 1/2 dB  additional loss for both lines.
I wonder how that compares to restive loss of two feedlines at QRO.   

[K1JJ]

There was a thread back a few months ago where I axed W2DU what the return loss was on a 1/12 wave balun.  I believe the db loss was down around 1/10th db, but I'd have to check.  Though, my stub switched delta loops don't use a 1/12 balun.

The loop's input impedance is something like 100 ohms, depending on how tighly tuned it is with the other element.  So using 75 ohms coax = about 1.3:1 swr on the stubs?

What does your model say for the two dipole input impedances when 135 degrees out and working well?

T

[K1JJ]

Tom the article I sent you said the two dipoles look like 122 -J41 and 34 J39.
I dumped the values into the transmission line calculator to see what 100 feet of RG213 did with the load. Matches loss at 7.15 MHz .53 dB. Total loss of each is .76 dB and .74 dB. So we are talking less than 1/2 dB  additional loss for both lines.
I wonder how that compares to restive loss of two feedlines at QRO.   

That's a moderate swr, but the resultant losses don't sound too bad.

But it sure hurts to give up 18% of our forward gain in coax losses though...   Heck, I have 420' of underground hardline going to the stack from the shack... plus the phasing harnesses -  don't feel bad, OM.

T

[WA1GFZ]

I bet a PL259 loss is interesting. They are around 35 ohm Z and that is if you solder them correctly.
Tom, you could always move the shack out to the field if it ever gets to you.

[Steve - WB3HUZ]

I calculated the stub loss at a few tenths of a dB. They are quite short and the SWR is low, well less than 2:1.

Good point 10 MHz. RG213 loss is .66 dB/100 feet plus VSWR loss. I guess this is the price I pay for force feed with the option to flip the pattern.
I wonder what loss you would see in a Stub switching system like Steve has? I wonder what loss you see in a 1/12 wave transition or a wilkensen.
A single antenna in a single direction is best for sure.

[K1JJ]

PL-259's?

I've had poor luck with them out in the field. They eventually go intermittent on me from moisture, even though they are well sealed to the WX elements.

All of my antenna phasing lines are hardwired together. I do my best to simulate a 50 ohm situation by keeping the inner conductor covered to the last 1/2".  I mean, after all, how much of an impedance bump is an open frame relay out there compared to a hardwired coax connection?  

When measuring I rarely see 1.1:1 difference using hardwired connections.  I do it all the way to 50 mhz. That old 16 bay 6M stack used the technique and the swr was FB.

Darn, my Milwakee drill just crapped out. I'm dead in the water and need to hit Home Depot.   Gotta have a drill.

T

[WA1GFZ]

Tom it might just need brushes...Check Sears they might be cheaper for the same item.

So sounds like Huz configuration and mine are a wash for loss.

I've been a type N guy forever. I like the tall connectors with the bug rubber gasket. Then cover them with a good layer of tape. HN even nicer. They go well with Bird switches

[Steve - WB3HUZ]

I modeled two dipoles spaced at 20 feet, phased at 135 degrees and at a height of 55 feet above average ground on 7.160 MHz. I first set the length of each dipole by itself to be exactly resonant at 7.160 MHz.

The feedpoint impedances turn out to be as follows.

Leading element (zero phase dipole): 18.2 - j35.6
Lagging element (135 phase dipole):  48.0 + j61.9

To get the proper phasing with RG-213 cable, the leading (zero phase dipole) will be fed with 16.333 feet of cable. This will result in a loss of 0.32 dB. The lagging element (the -135 degree phased dipole) will be fed with 72.85 feet of cable. This will result in a loss of 0.6 dB.

The impedance at the point where the two cables above are joined will be 14.57 + j0.86. That makes for an SWR of 3.44 and loss of 0.97 dB in 100 feet of cable back the shack. All told, there is about 1.5 dB total loss. Placing a matching network where the two phasing cables join and the loss is reduced by about 1 dB (minus whatever loss the matching network may have).

[K1JJ]

Wow. That's quite a message!

Looks like Frank should make up a transformation out on the tower.   14 ohms is VERY low at the junction.

I have the same problem here.  When I join my three 50 ohm 40M Yagis, I'm down to 16.6 ohms. I use a 1/12 wave balun to bring it back up to 50 ohms for the hardline fed at 125' high on the tower. Works great. Maybe Frank already plans on some kind of coaxial transformation there - or is he running both dipole coaxes into the shack? - dunno.

T

[Steve - WB3HUZ]

What if you cut the dipoles a little short?

I modeled two dipoles spaced at 20 feet, phased at 135 degrees and at a height of 55 feet above average ground on 7.160 MHz. I first set the length of each dipole by itself to be exactly resonant at 7.260 MHz.

The feedpoint impedances (at 7.16 MHz) turn out to be as follows.

Leading element (zero phase dipole): 17.235 - j56.138
Lagging element (135 phase dipole):  46.478 + j37.478

To get the proper phasing with RG-213 cable, the leading (zero phase dipole) will be fed with 18.998 feet of cable. This will result in a loss of 0.54 dB. The lagging element (the -135 degree phased dipole) will be fed with 74.858 feet of cable. This will result in a loss of 0.6 dB.

The impedance at the point where the two cables above are joined will be 15.15 + j7.43. That makes for an SWR of 3.4 and loss of 0.94 dB in 100 feet of cable back the shack. Looks like the loss is a little worse here.

[WA1GFZ]

Steve,
I simulate 41.6 -j3.4  and 16 -J93  the the feed point so we are pretty close. My feed lines are 65 and 100 feet to get into the shack. I just added 15 feet to the 100 footer and 50 to the 65 footer. Just to make them even  Signals to the west great FB but not great to the east. I used this ULA 400 cable for the 50 footer that turns out to be foam dielectric with a VF of .85. I guess it is a version of LMR400.  So East cable somewhat longer (electric length) than I thought.  I'll swap it out with something that matches. Also tonight I added the foot of wire on each 4 wires back on. I really need to invest in an antenna analyzer. I was thinking of hooking up a Z bridge. I have a couple kicking around. I could js an analyzer to find a Z null at 50 ohms but wanted to fix the FB problem. I was seeing a solid 20 dB on the HPSDR to the west. Back to the pit

[WA1GFZ]

Well I need to look at that and what happens with an odd quarter wave length of coax which should raise it up. I'm looking at around 5 quarter wavelengths for feed line. I could get by with 3 if I did switching at the base of the tower.

[K1JJ]

This is an off-the-wall solution, but what if you made a compact variable (un to bal) L/C matching network for each dipole that hung out on the end of each boom mast at each feedpoint?  Using a 1/2 wave multiple feedline you could zero in on the proper match from the ground. Knowing the approx modeled impedance you could pre-set the network in the shack and be close.   This way you could run the coaxes to the shack with no matching and a good swr...   

There was a time when I did that with remote controlled tuning on a rotary 75M Yagi. Worked FB.

But all in all, the parasitic stub technique wud be far simpler.

T

[Steve - WB3HUZ]

What if you cut the dipoles a little long?

I modeled two dipoles spaced at 20 feet, phased at 135 degrees and at a height of 55 feet above average ground on 7.160 MHz. I first set the length of each dipole by itself to be exactly resonant at 7.060 MHz.

The feedpoint impedances (at 7.160 MHz) turn out to be as follows.

Leading element (zero phase dipole): 19.147 - j15.496
Lagging element (-135 phase dipole):  49.805 + j86.012

To get the proper phasing with RG-213 cable, the leading (zero phase dipole) will be fed with 12.644 feet of cable. This will result in a loss of 0.18 dB. The lagging element (the -135 degree phased dipole) will be fed with 71.356 feet of cable. This will result in a loss of 0.79 dB.

The impedance at the point where the two cables above are joined will be 12.63 - j4.09. That makes for an SWR of 3.97 and loss of 1.09 dB in 100 feet of cable back the shack.

[WA1GFZ]

Steve PM me your email address and I'll send you an article a found in the morning.
The guy who wrote it also set the Resonant length off frequency. I'm at 34 feet 5 inches each half right now which is about 7.15
I don't have it here at home but interesting reading. I need to run about 100 feet or so of feed line to get inside or do 3 quarter waves to the ground. I was just loooking at the transmission line calculator and the low z is pretty lossy getting it to the ground.  So it boilds down to 3 quarter waves, 4 quarter waves or 5 quarter wavelengths of coax.
Tom the problem with lumped values is the Z changes a lot when you swap the pattern

[K1JJ]

Tom the problem with lumped values is the Z changes a lot when you swap the pattern

Plus the phase changes when the L/C is used, so it becomes even more complex to tune. 

Yes, the two feedlines are different lengths when swapped... I see the problem there.

O'well, bag that idea...

T

[Steve - WB3HUZ]

Using your Z numbers you have two phasing cable options. The first would be to feed the leading element with a 9.592 foot long cable. This will give you 0.37 dB loss in this cable. The lagging element would be fed with a 40.059 foot cable and you would get 0.21 dB loss.

The impedance at the point where the two cables above are joined will be 9.54 - j15.25. That makes for an SWR of 5.66 and loss of 1.47 dB in 100 feet of cable back the shack.

The other option would be to feed the leading element with a 22.182 foot long cable. This will give you 1.12 dB loss in this cable. The lagging element would be fed with a 80.076 foot cable and you would get 0.43 dB loss.

The impedance at the point where the two cables above are joined will be 12.94 - j18.05. That makes for an SWR of 4.34 and loss of 1.19 dB in 100 feet of cable back the shack.

Steve,
I simulate 41.6 -j3.4  and 16 -J93  the the feed point so we are pretty close. My feed lines are 65 and 100 feet to get into the shack. I just added 15 feet to the 100 footer and 50 to the 65 footer. Just to make them even  Signals to the west great FB but not great to the east. I used this ULA 400 cable for the 50 footer that turns out to be foam dielectric with a VF of .85. I guess it is a version of LMR400.  So East cable somewhat longer (electric length) than I thought.  I'll swap it out with something that matches. Also tonight I added the foot of wire on each 4 wires back on. I really need to invest in an antenna analyzer. I was thinking of hooking up a Z bridge. I have a couple kicking around. I could js an analyzer to find a Z null at 50 ohms but wanted to fix the FB problem. I was seeing a solid 20 dB on the HPSDR to the west. Back to the pit

[Steve - WB3HUZ]

Lumped values where? And why would anything change when switching the pattern if everything is symmetrical?

Tom the problem with lumped values is the Z changes a lot when you swap the pattern

Yeah, plus the phase changes when the L/C is used, so it becomes even more complex to tune.  O'well, bag that idea...


BTW, why wud the Z change when the pattern is swapped? Are you talking about real whirl imperfections between the two dipoles hanging out there?

T

[K1JJ]

Frank's referring to an earlier post where I suggested the possibility of using L/C matching at each dipole feedpoint.

Another thing to consider - we're swinging around big swr mismatches here.  When running QRO, this is a consideration for cable damage, caw mawn.


BTW, Frank - I'm drilling and tapping the second 600w SS FET module.  Remember the trouble I had breaking taps?  Someone suggested to use larger than recommended drills.  They recommend #43 for 4-40 screw. I tried a #41 drill today and the taps goes thru much easier - and the screw is still tight.  Shuda thought of this long ago instead of struggling with those tight holes...

T

[Steve - WB3HUZ]

OK. Got it.

BTW, the loss in the stubs on my loops is about 0.14 dB. The entire system, including the cable back to the shack has less than 0.5 dB loss at the lowest SWR frequency and never gets worse than about 0.61 across the entire span of 7.1 to 7.3. These are modeled numbers, so I'm sure it's not quite that good in reality but if you start out low, you are more likely to end up low. Tight? That's another story.

Frank's referring to an earlier post where I suggested the possibility of using L/C matching at each dipole feedpoint.

Another thing to consider - we're swinging around big swr mismatches here.  When running QRO, this is a consideration for cable damage, caw mawn.


BTW, Frank - I'm drilling and tapping the second 600w SS FET module.  Remember the trouble I had breaking taps?  Someone suggested to use larger than recommended drills.  They recommend #43 for 4-40 screw. I tried a #41 drill today and the taps goes thru much easier - and the screw is still tight.  Shuda thought of this long ago instead of struggling with those tight holes...

T

[WA1GFZ]

Well, Just swapped out the foam coax and had a 50 foot hunk of RG 8 then I found a couple short jumpers to extend the delay line out past 90 degrees. SWR looks real good at 7.09. Conditions look crappy tonight so figured I'll quit here.
Well Steve maybe it is worth looking at setting the elements inductive but sure looks like you have a better configuration. I suppose I could hang a box off the center pulley for switching so I could lower if if it needs repair in the middle of the winter.
Maybe I'll simulate all the losses in the article and see if it makes sense. Since you have done this about a million times by now what do you think I will need to do to duplicate your feed method using dipoles as elements with the ends at about 45 feet?
I also want to give folded dipoles another look since they have a higher z or look at a 1:2 transformer up in the air with force fed switching in the air
Tom Cool on the tapping. Whenever I start a new project like that I buy a new name brand tap and sometimes it comes with a drill. I go very slowly and use plenty of oil. I  usually go 1/4 turn in 1/2 turn out and if it feels tight I go back and forth a few times before I go any deeper. Jay found a source of strapping heat sinks if you need any.

[K1JJ]

Frank said:

"Another option is build a 2 element for North East and hang a second driven element from ropes behind it to do South West sharing the reflector."


But then your f-b is only 12-15db at most with a 2el Yagi. You need more. 

Another idea is to try modeling a parasitic pair of delta loops (apex up)  but with extended bases. (and corresponding shorter sides)  Make the bases much longer than the sides and then avg height goes up.  You can get 20-30db f-b easily and I'll bet the TO angle isn't degraded that much from dipoles.  Plus, use the parasitic stub method and you're back to a clean NE or SW switchable pattern. You can pull the loops out with less space too.

I ran a pair of these "extended base" deltas on 75M for years and they worked out really FB.

Model them and see how far you can push the base length. You will be surprised at the pattern results and f-b of a close spaced parasitic pair.  Even 15' spacing is good. Now, with a 30' boom, you COULD do a back to back common reflector array that kicked ass and NO relays... just two coaxes into the shack. Use a 2:1  1/12 wave balun at each loop feedpoint and you're golden.  Feed them at the top or bottom, whatever is best for you.

T