This homebrew transmatch is similar to the design of the famous Murch Ultimate Transmatch, which is the basis for nearly all commercial transmatches in production today. The Ultimate is a HIGH PASS filter, with an input capacitor, a roller inductor to ground, and an output capacitor. This is a T network as explained in the handbooks.
ARRL tests show that an L network, such as the Ten Tec 238, is one of the best for losses. The Johnson transmatches also do well. A T network tuner often will show losses matching into very low impedance loads, especially on 160 meters. Results vary from manufacturer due to use of copper strap or thin wire, roller inductance wire guage, and so on. In any event, losses show up as heating, due to the laws of physics. I did not notice any heating at the power levels during test; that is as scientific as I got. Arcing can occur with high impedance loads. I did not notice any tendency to arc. But I use resonant low impedance coax fed antennas. I also use a MFJ-212 to pre-tune the transmatch in receive mode.
NEVER TUNE A ROLLER INDUCTOR IN A TRANSMATCH OF ANY KIND AT FULL POWER. NEVER TURN AN INDUCTOR SWITCH AT FULL POWER WITH A FIXED INDUCTOR WITH TAPS. These are rookie mistakes that will damage ANY tuner.
My design is an electrical mirror of the Murch design. This configuration was featured in an ARRL Radio Amateur's Handbook and a QST article October 1967 page 22. If you are an ARRL member, you can get that article from their website. It uses commercial inductor stock, but you can make your own inductors. The design is a LOW PASS filter, with an input roller inductor, a capacitor to ground, and an output roller inductor. This also is a T network. The handbooks explain by swapping coils for capacitors and capacitors for coils, the network is changed from a high pass network into a low pass network. Auxiliary coils and a capacitor are switched in to work on 160 meters. This configuration has long been preferred by professional broadcast engineers, since it provides the same matching range, with the enhancement of harmonic suppression.
Adjustment is smooth and precise due to the turns counter dials and roller inductors. The capacitor tuning does not appear to be touchy on frequencies up to 20 meters.
This transmatch is ideal for use with older AM rigs which do not have high circuit Q in the final tank circuit on 160 meters. See the Apache article for 160 meter tank circuit design. Examples: DX100, Viking 2, Valiant. This prevents harmonics from 160 meters occurring on 80 and 40 meters.
It also brings any of the older rigs operating on the 80 thru 10 meter bands without a PI-L output circuit up to modern FCC standards without modifying the internals of the existing radio. Be sure to tune the rig into a dummy load, adjust the transmatch with a MFJ212 Matchmaker, and then touch up the plate tuning. This dramatically reduces tune up time, eliminates any risky on the air adjustment of the transmatch and reduces QRM.
I have easily run 700 watts SSB or CW, or 150 watts output carrier AM with this transmatch with no signs of heating or malfunction.
The chassis is parts of a junked DX60 (top and bottom cover, painted black). The inductor dials are from a junked Tektronix scope, and are commonly used on 10 turn precision pots. The SWR sampling line is from a junked SWR bridge. The coil forms and fixed condensers are from junked WW2 surplus. The switch is from a junked Command Set also. I made the panel labels when I worked at a place with a name badge engraver.
Here are the pictures of the interior: