Here is the WA2SRQ data from my personal archive ...

Having access to a HewlettPackard 4193A vector impedance meter at work,
I have made measurements on a number of baluns, coaxial and otherwise.
For my beams I was particularly interested how many turns and on what
diameter are optimum for air core coaxial baluns, and what the effect of
bunching the turns was (formless). Using the remote programming capability
of the HP4193A along with an instrument controller, I measured the magnitude
and phase of each balun's winding impedance at 1 MHz intervals from 1 to 35
MHz. For comparison, I also made measurements on a commercial balun which
consists of a number of ferrite beads slipped over a short length of coax.
I've appended some of these measurements so you can draw your own
conclusions.
PVC pipe was used for coil forms. The 41/4 inch diameter baluns were wound
on thinwalled PVC labeled "4 inch sewer pipe". This material makes an
excellent balun form. It's very light weight and easy to work with, and I
obtained a 10 foot length at the local Home Depot for about 3 dollars. The
65/8 inch diameter forms are 6 inch schedule 40 PVC pipe which is much
thicker, heavier, and more expensive.
Each test choke was closewound on a form as a singlelayer solenoid using
RG213 and taped to hold the turns in place. The lengths of cable were cut
so there was about 2 inches excess at each end. This allowed just enough
wire at the ends for connections to the HP4193A's probe tip. After data was
collected for each singlelayer configuration, the PVC form was removed, the
turns were bunched together and taped formless, and another set of
measurements
was taken. I have only included the "bunched" measurements in the table for
one of the baluns, but the trend was the same in each case. When compared
to
the singlelayer version of the same diameter and number of turns, the
bunched
baluns show a large downward shift in parallel selfresonance frequency and
poor choking reactance at the higher frequencies.
Interpreting the Measurements

All the baluns start out looking inductive at low frequencies, as indicated
by
the positive phase angles. As the frequency is increased, a point is
reached
where the capacitance between the windings forms a parallel resonance with
the
coil's inductance. Above this frequency, the winding reactance is reduced
by
this capacitance. The interwinding capacitance increases with the number of
turns and the diameter of the turns, so "more is not always better".
The effects of a large increase in interwinding capacitance is evident in
the
measurements on the balun with the bunched turns. This is probably a result
of the first and last turns of the coil being much closer together than the
singlelayer coil.
An important requirement of these baluns is that the magnitude of the
winding
reactance be much greater than the load impedance. In the case of a 50 ohm
balanced antenna, the balun's winding impedance is effectively shunted
across
one half the 50 ohm load impedance, or 25 ohms. A reasonable critera for
the
balun's winding impedance for negligible common mode current in the shield
is
that it be at least 20 times this, or 500 ohms. The measurements show, for
example, that 6 turns 41/4 inches in diameter meet this criteria from 14 to
35 MHz.
The measurement data also reveals the power loss these baluns will exhibit.
Each of the measurement points can be transformed from the polar format of
the
table to a parallel equivalent real and reactive shunt impedance. The power
dissipated in the balun is then the square of the voltage across it divided
by
the real parallel equivalent shunt impedance. While this calculation can be
made for each measurement point, an approximate number can be taken directly
from the tables at the parallel resonance points. At 0 degrees phase angle
the magnitude numbers are pure resistive. I didn't record the exact
resonance
points, but it can be seen from the tables that the four singlelayer baluns
are all above 15K ohms, while the ferrite bead balun read about 1.4K. These
baluns see half the load voltage, so at 1500 watts to a 50 ohm load, the
power
dissipated in the coaxial baluns will be less than 1.3 watts, and the
ferrite
bead balun will dissipate about 13.4 watts (neglecting possible core
saturation and other nonlinear effects). These losses are certainly
negligible. At 200 ohms load impedance, the losses are under 5 watts for
the
coaxial baluns and 53.6 watts for the ferrite beads.
Conclusions

 A 1:1 coaxial balun with excellent choking reactance for 10 through 20
meters can be made by winding 6 turns of RG213 on inexpensive 4 inch PVC
sewer pipe.
 For 40 or 30 meters, use 12 turns of RG213 on 4 inch PVC sewer pipe.
 Don't bunch the turns together. Wind them as a single layer on a form.
Bunching the turns kills the choking effect at higher
frequencies.
 Don't use too many turns. For example, the HyGain manuals for my 10 and
15
meter yagis both recommend 12 turns 6 inches in diameter. At the very least
this is about 3 times as much coax as is needed, and these dimensions
actually
give less than the desired choking impedance on 10 and 15 meters.
Measurements

Magnitude in ohms, phase angle in degrees, as a function of frequency in Hz,
for various baluns.
6 Turns 12 Turns 4 Turns 8 Turns 8 Turns
Ferrite
41/4 in 41/4 in 65/8 in 65/8 in 65/8 in
beads
sngl layer sngl layer sngl layer sngl layer bunched
(Aztec)
    

Frequency Mag Phase Mag Phase Mag Phase Mag Phase Mag Phase Mag
Phase
1.00E+06 26 88.1 65 89.2 26 88.3 74 89.2 94 89.3 416
78.1
2.00E+06 51 88.7 131 89.3 52 88.8 150 89.3 202 89.2 795
56.1
3.00E+06 77 88.9 200 89.4 79 89.1 232 89.3 355 88.9 1046
39.8
4.00E+06 103 89.1 273 89.5 106 89.3 324 89.4 620 88.3 1217
26.6
5.00E+06 131 89.1 356 89.4 136 89.2 436 89.3 1300 86.2 1334
14.7
6.00E+06 160 89.3 451 89.5 167 89.3 576 89.1 8530 59.9 1387
3.6
7.00E+06 190 89.4 561 89.5 201 89.4 759 89.1 2120 81.9 1404
5.9
8.00E+06 222 89.4 696 89.6 239 89.4 1033 88.8 1019 85.7 1369
15.4
9.00E+06 258 89.4 869 89.5 283 89.4 1514 87.3 681 86.5 1295
23.7
1.00E+07 298 89.3 1103 89.3 333 89.2 2300 83.1 518 86.9 1210
29.8
1.10E+07 340 89.3 1440 89.1 393 89.2 4700 73.1 418 87.1 1123
35.2
1.20E+07 390 89.3 1983 88.7 467 88.9 15840 5.2 350 87.2 1043
39.9
1.30E+07 447 89.2 3010 87.7 556 88.3 4470 62.6 300 86.9 954
42.7
1.40E+07 514 89.3 5850 85.6 675 88.3 2830 71.6 262 86.9 901
45.2
1.50E+07 594 88.9 42000 44.0 834 87.5 1910 79.9 231 87.0 847
48.1
1.60E+07 694 88.8 7210 81.5 1098 86.9 1375 84.1 203 87.2 778
51.8
1.70E+07 830 88.1 3250 82.0 1651 81.8 991 82.4 180 86.9 684
54.4
1.80E+07 955 86.0 2720 76.1 1796 70.3 986 67.2 164 84.9 623
45.9
1.90E+07 1203 85.4 1860 80.1 3260 44.6 742 71.0 145 85.1 568
51.2
2.00E+07 1419 85.2 1738 83.8 3710 59.0 1123 67.7 138 84.5 654
34.0
2.10E+07 1955 85.7 1368 87.2 12940 31.3 859 84.3 122 86.1 696
49.9
2.20E+07 3010 83.9 1133 87.8 3620 77.5 708 86.1 107 85.9 631
54.8
2.30E+07 6380 76.8 955 88.0 2050 83.0 613 86.9 94 85.5 584
57.4
2.40E+07 15980 29.6 807 86.3 1440 84.6 535 86.3 82 85.0 536
58.8
2.50E+07 5230 56.7 754 82.2 1099 84.1 466 84.1 70 84.3 485
59.2
2.60E+07 3210 78.9 682 86.4 967 83.4 467 81.6 60 82.7 481
56.2
2.70E+07 2000 84.4 578 87.3 809 86.5 419 85.5 49 81.7 463
60.5
2.80E+07 1426 85.6 483 86.5 685 87.1 364 86.2 38 79.6 425
62.5
2.90E+07 1074 85.1 383 84.1 590 87.3 308 85.6 28 75.2 387
63.8
3.00E+07 840 83.2 287 75.0 508 87.0 244 82.1 18 66.3 346
64.4
3.10E+07 661 81.7 188 52.3 442 85.7 174 69.9 9 34.3 305
64.3
3.20E+07 484 78.2 258 20.4 385 83.6 155 18.0 11 37.2 263
63.2
3.30E+07 335 41.4 1162 13.5 326 78.2 569 0.3 21 63.6 212
58.0
3.40E+07 607 32.2 839 45.9 316 63.4 716 57.6 32 71.4 183
40.5
3.50E+07 705 58.2 564 56.3 379 69.5 513 72.5 46 76.0 235
29.6
