I needed a station unit, not a super compact field unit. It had to have enough heft to support the RG-213 I use to feed the antennas (a GAP-VI and a Butterfly beam). And it had to be soon. That meant finding what I needed in the junkbox, rather than making up a design and exploring hamfests for several years to find the parts.
Here is what I found: 2 10 ÁH LaPointe rotary inductors from a remotely tuned EDZ beam project, a few HF-100 variable capacitors, a 550- 550 pF 2 section capacitor from my father's junkbox (probably from a military rig), a ceramic multi-wafer 7-position rotary switch about 7" long, knobs galore, some 3/4" aluminum L-stock from a portable antenna no longer in use, and about 2.5 square feet of plexiglass the previous owner of my house left behind. I also found a bicentennial quarter and two long- lost screwdrivers among the junk. That was almost everything I needed for an L-C-L Tee tuner.
The plexiglass would make the panels and chassis, supported and bound by the L-stock. That decision alone would save almost $160 (the cost of 2 new turns-counters). Since plexiglass is transparent, I could count the turns myself. Well-wired ATUs do not radiate, consisting of passive components only, so a metal case is not required. All I needed were 4 insulated shaft couplings, which I obtained from Buckeye Electronics for $3.00.
The inductors have spread turns at one end to maintain Q at low inductance values. Therefore, the short from the rotary contact goes to the close-spaced end of the coil.
10 ÁH coils are sufficient for most situations, down to 80 meters. However, I can imagine some load types for which they might not provide an efficient match on the lower bands. If you short out all of the output coil, you have an L-circuit, suitable for end-fed random wires. Alternatively, to get more inductance into the circuit, I added S2 to move the capacitor set to the output terminal and use both coils. On low bands, one coil might be at full inductance and the other varied for the match.
Although the schematic shows 3 switches, all switching can be combined in a single multi-section rotary switch. Ceramic wafers are best.
I have placed two photographs--an open-case front view and a closed-case rear view--on a separate page for reference. Because of the photo file sizes, downloading them may be time-consuming.
Two strips of L-stock hold the chassis plate off the table. The coils and capacitors mount on this plate. A set of "ground" connections run under the plate and link all parts, as well as the rails and a small aluminum plate at the rear on which the input and output coax connectors are mounted. Exact techniques depend on the components one finds in a junk box. A handful of 6-32 machine screws, nuts, and lockwashers (with a few 8-32 pieces to mount the coils) is all the hardware the project requires.
The ceramic wafer switch had enough positions and sections to handle the switching jobs shown in the schematic. I shortened the rods (and rethreaded them) so that with the shaft mounted to the front panel, the rods just projected through the rear panel. This technique keeps the switch from sagging, reduces strain at the front mounting, and holds the front and rear panels apart at the correct distance. I read switch position by a pointer knob that aligns with the visible detents in the switch support plate behind the front panel.
I wired the chassis components first, the switch second, and made interconnections when I mounted the switch. Quarter-inch wooden dowel provides the coil and capacitor shaft extensions through the front panel to hold the tuning knobs.
Aluminum L-stock pieces around the case perimeter link the sides and bottom of the case, with an independent top piece of plexiglass and a rim of L-stock to hold it in place. Vertical L-stock at the corners is electrically connected to the grounded chassis rails. Only the top perimeter pieces float, but have shown no detectable RF, either to my finger tips or by detuning the circuit from semi-assembled settings. There are no panel markings because none are needed.
Despite the use of two inductors, L-C-L tuners are capable of high efficiencies. W. I Everitt, in the 1930s, provided the basic analysis of fundamental networks, including their losses. His work is summarized in Terman's Radio Engineers' Handbook (McGraw-Hill, 1943, pp. 210-215). The key term for determining losses is delta, based on inductor losses in each type of network. Although Terman provides graphs of delta, popular in the days before computers and pocket calculators, Brian Egan, ZL1LE, has derived the delta equations and added them to his very useful program, TUNER.BAS. This versatile program is now included in the collection of programs called HAMCALC, made available by George Murphy. (Contact George Murphy, VE3ERP, 77 McKenzie Street, Orillia, Ontario L3V 6A6, Canada. The software is free and can be distributed. However, to cover the costs of disks and Canadian postage, Murph asks a donation of $5.00. He donates the excess over his disk and mailing costs to the amateur radio program of the Canadian National Institute for the Blind.)
Using an L-C-L (or any other ATU) circuit effectively requires some forethought. Multiple matching settings are possible, and we always want to tune for minimum delta and maximum efficiency at the best match (lowest SWR to the transmitter). The operative rule of thumb is this: Choose the lowest value of L2 (the antenna-side inductor) that permits a match. This setting will ensure the lowest obtainable value for delta, whatever its actual value.
The L-C-L ATU has met all my design specifications and is doing its
work well with the QRP+. The fact that the transparent case reveals the
components and connections and, therefore, mystifies shack visitors is
simply an unintended but welcome bonus.
first printed in 72: The New England QRP Newsletter, April, 1996.