Sunday, July 19, 2015

Closer Look at the XM240 LCA

For my recently-purchased Cushcraft XM240 40 meter yagi I speculated on modifying it for improved performance. The primary options are to convert it into a W6NL Moxon or to replace the low-Q coils with high-Q coils. I built an approximate model of the antenna in EZNEC to evaluate the second option.

The coil is part of the LCA (loading coil assembly). The coil is close-wound on a fibreglass form, an attached to short aluminum tubes at each end. The LCA with the protective coating removed can be see on VE6WZ's web site. You may want to keep that page open while you read the rest of this article since I frequently refer to it.

Calculating the Q of a coil is difficult due to the many factors at play. It is typically easier and more accurate to measure it. Most hams, including me, do not own suitable test equipment. VE6WZ used software to estimate the coil Q and ESR (equivalent series resistance). As I said in my previous article that, if correct, the loss is as much as -3 db and coil heating is excessive. As I said there, I doubted these figures. But how to proceed? It is important to know the loss since it will be a key factor in whether or how I modify the antenna before its expected use in 2016.

The inside story

A close-up of one end of the LCA is shown at right. The ¾" fibreglass form fits snugly inside the ⅞" aluminum tubes, attached by what appear to be rivets. A screw electrically bonds the coil wire to the tube.

Some internet searching told me that the loss tangent of fibreglass is heavily dependent on the formulation. At the high end of the range I calculated an ESR of 8 Ω for the coil at 7.1 MHz, which is what VE6WZ calculated. Presumably that is the loss tangent for fibreglass used in K6STI's software calculator. The calculated coil Q is very low. However at the other end of the loss tangent range the ESR would be a very good 1.5 Ω. Calculation alone is clearly inadequate to gain the required insight.

Puzzled by this difficulty I did continued searching. A passing remark on a ham forum gave me the hint I needed. That person called the form a fibreglass tube. A tube is hollow, not solid. K6STI's calculator appear to assume that the form is a solid rod. I picked up an  LCA, pointed it at an open window and held the other end close to my eye. I saw daylight. Now the trail was hot.

It was difficult to be certain how the interior was structured because the light coming from the other end cast dark shadows from several dark protuberances from the tube walls. But the small diameter tube does not easily allow illumination from the same end I'm looking into with my eye. Fortunately my smart phone camera has a small lens and an adjacent LED "flashbulb" . With some care I was able to get both positioned within the tube opening. I pointed the far end  at daylight and took the following unusual photograph.

The flash was so bright within the LCA's confined space that the metal reflections stopped down the automatic exposure to reduce the far-end daylight to black! Despite this the interior structure is clear.

The rivets are obviously metal (presumably aluminum), as evidenced by their reflectivity. The fibreglass form is definitely a tube and (not clear in the picture) has a narrow wall thickness. The self-tapping screw that bonds the coil to the aluminum is very evident. As an aside, this reminded me that I need to replace the screw with a bolt that goes through the other side of tube, which is a well-known design flaw Cushcraft has persistently failed to fix. Due to the perspective of the camera and hardware alignment the rivets and screw at the far end of the tube are hidden behind those in front.

The upshot of this exercise is that the coil does indeed have an air core. The note by VE6WZ that some have measured the coil Q to be 200 is now entirely credible. Since fibreglass fills only a small fraction of the coil's interior volume its contribution to inductance and Q is small. This is good news.

Inductance, Q and loss

Since those unnamed sources and VE6WZ's calculations for an air-core coil of the LCA's dimensions agree on a Q of 200, let's proceed on that assumption. We can now compare an ideal, zero-loss coil (Q=∞), the stock LCA (Q=200) and VE6WZ's high-Q coil (Q=767).

First we need to agree on the inductance value. VE6WZ's measurement is ~15 μH. An air-core coil of the LCA's dimensions give an inductance of 11 μH. This is a difference that must be explained. I think it is fair to conclude that, per his pictured test apparatus, there is ample stray reactance due the long test leads and the attached tubes to account for the difference. There is no need here to speculate on the accuracy of the pictured test device.

I am also swayed by my EZNEC model which only works well when the load is set to about 11 μH. While I cannot use the Leeson correction, I did use an average element diameter that is roughly consistent with a stepped-diameter correction for an XM240 element. Whereas an inductance of 15 μH in the model is wide of the mark, requiring unrealistic element truncation.

For a Q of 200 and X of 500 (11 μH at 7.1 MHz) the ESR is 2.5 Ω (R=X/Q). I then chose a test frequency where the radiation resistance is a little above its minimum near the frequency of maximum gain. Recall that in a 2-element yagi with a reflector parasite the maximum gain ought to be placed at the bottom of the target band for optimum performance across the band. Since I am targetting the CW and DX SSB segments I chose 7.05 MHz.

The resulting gain figures versus ESR for the several LCA options are as follows:
  • 0 Ω (ideal, zero loss): 5.71 dbi
  • 0.9 Ω (VE6WZ high-Q coils): 5.43 dbi
  • 2.5 Ω (estimated stock LCA): 4.94 dbi
  • 8 Ω (now-invalidated estimate of stock LCA): 3.43 dbi
F/B differences are negligible. Feed point impedance and SWR curve are only modestly different across the first 3 selections. Relative to 7.05 MHz the gain differences (coild loss) will tend higher toward the lower band edge, and gradually decline with increasing frequency. This is due to the radiation resistance change with frequency.

The high-Q coils do very well, being about -0.3 db from the ideal. Compared to these coils the stock LCA is down a further -0.5 db. That's also very good, and better than I expected.


I am now inclined to stick with the stock LCA. In my judgment an additional 0.5 db is not worth the effort involved nor the damage risk due to the exposed and more fragile high-Q coils. The better alternative is to forgo the coils and do the W6NL Moxon conversion which does away with the coils and coil loss while also improving SWR and F/B performance.

However I still need to replace those self-tapping screws on the LCA with through-tube stainless steel bolts. I don't want a coil failure to occur in the midst of a contest during a frigid northern winter.

No comments:

Post a Comment

All comments are moderated, and should appear within one day of submission.