An AM tube linear in Class 'A', with an efficiency of 20%, would require a plate dissipation of 7500 watts (1500/20%). A pair of 3CX3000F7 triodes was selected for the job. Initial testing of this amp was in class AB2. (Schematic diagrams starting at visual #31).
1
2
3
4
5
1: Oversized SK1406 chimneys, 3CX3000F7 tubes and 1/8" thick, 1" wide, half-circle
aluminum tube clamps.
2: Created a 1/4 inch plywood "drill" template using the tube chimneys and a draftsman's
compass.
3: Three inch tube holes and 5/8 inch cooling holes drilled into 1/8" thick 22" x 17" x 6"
aluminum chassis.
4: The tube grid rigs are larger than 3 inch chassis holes; therefore, the tubes sit on
top of
grid rings and fastened down with 6-32 hardware.
5: The space difference between the oversized chimneys and the tubes was filled by
inserting a 2" x 13 9/16" strip of .060 inch thick teflon and "glueing" them in with GE
Silicon Kitchen and Bath caulk.
6
7
8
9
10
6: Method of mounting the .015 ufd, 2500 V transmitting capacitors to the 8 1/2
inch long
Bakelite terminal strip.
7: Overview of RF input circuit with dual 180 uHy, 76 amp filament chokes and .015
coupling capacitors.
8: Shown is the relay that grounds the tube cathodes and the 12 VDC power supply.
9: Homebrew cathode ammeter shunt (red coil) and front panel test switch.
10: Size of chassis: 22" x 17" x 6". Overview of cathode wiring showing
the two Peter Dahl
180 uHy 75 amp filament chokes.
Choke Construction: 15 inches long, 2 inches in
diameter, 38 bifilar turns
(by count 76 wire turns divided by 2) of #8
wire wound around
5 ferrite rods. (The rods were actually two sets of 7
1/2 inch long rods, 10 in all).
The
choke weighed over 6 pounds.
11 12
13
14
15
11: Front panel turns counter. (From the Palstar Company in Piqua, Ohio).
12: Turns counter to vacuum variable coupling.
13: Mounting bracket for vacuum variables (custom made by Bayshore Metal Products Inc,
Keyport, NJ).
14: 15 KV vacuum variable C1, Jennings UCSL-2750-5N558, measured at 12-540 pf.
15: 5 KV vacuum variable C2, Jennings CSVF-500-0315, Measured at 50-3300 pf.
16
17
18
19
20
16: Tube connection via standard 4.5 inch dryer clamp and Scotch #25 Electrical
Grounding Braid.
17: Five inch tall standoff insulator with 1/4 inch threads, central support for the plate
connection, the 2020 pf, 40KV, 15 amp plate coupling capacitor and the 10 ohm, 10
watt glitch resistors.
18: Pi-network connections. Gates inductor 3 9/16" dia, 7 1/4" long, 28 turns.
19: Inductor clip, 15 amp (Viking Electronics Technologies, Ltd, Lindenhurst, NY).
20: Power out coax connector, AC line connector, RCA keying jack and 1 mHy safety
choke.
21
22
23
24
25
21: Millen high voltage connector and ground lug.
22: B+ bypass caps, each 2720 pF, 30KV, 15 amps.
23: Plate choke #1, Peter Dahl 90 uhy, 3 amps.
CONSTRUCTION: 6.7 inches long, 5/8
inch diameter grooved Delrin Rod, 123
turns of #24 magnet wire.
24: Plate choke #2, 90 uhy, 3 amps.
25: Glitch resistors, (4) 10 ohm, 10 watt units. I have a minority opinion on glitch resistors.
Most designers plan on having the resistor absorb the hit and discharge the capacitors.
When there is a glitch, I want to know about it immediately. The resistors will explode
and then I vacuum out the remains. When 150 MFD capacitors are charged to
5000
VDC there is serious short circuit current. For fusing, I also have a 4 inch piece of
magnet wire in the B+ line.
26
27
28
29
30
26: Label on 1/2 horsepower, 3450 RPM Dayton split phase motor.
27: Label on 9 inch diameter Dayton blower.
28: 4 inch hole drilled into the 3/4 inch plywood shelf which holds the RF DECK.
29: Dryer hose installed.
30: Back of RF DECK. More than enough air pressure comes up through the bottom of the
chassis to cool the tubes.
31
32
33
34
31: Schematic of the plate circuit.
32: Schematic of the cathode circuit.
33. Schematic of the high voltage power supply.
Click
HERE to see that power supply.
34: Schematic of the keying relay power supply.
35
36
37
38
35: Turns counter dial reading vs picofarads for input capacitor C1.
36: Turns counter dial reading vs picofarads for output capacitor C2.
37: Test setup for ballpark pi-network values. With a capacitance checker, I found the
values of the unconnected input and output vacuum variables and set them to the
values of the formulas for a Q of 12. With aligator clips I attached a resistor from plate
to ground equal to the calculated plate resistance. I adjusted the coil taps until I
obtained a low SWR on the MJF-259. This is not perfect test, but it gets the
pi-network into the ballpark. (All power is off for this test).
38: Block diagram of test setup.
1. The initial tests were made in class AB2. In AB2, this amplifier will put out twice the
output of the single 3CX3000F7 and requires twice the drive.
Click
HERE to see that amplifier.
2. Bias diodes are unnecessay with the 3CX3000F7 in a cathode driven grounded grid
configuration. With the grids grounded directly to the chassis, the center tap of the
filament transformer can be switched directly to the chassis to activate the amplifier.
3. Under full load, there is no significant difference in output between the amp running at
4900 volts on the plate or 4300 volts. With the higher plate voltage, the idling current
is higher and slightly less drive is required for the same output power.
4. Tuning for maximum power the amp produces 20 times the input power with fairly good
efficiency; however when adjusted with a signal generator and a scope, the output
power is only 14 times the input at aproximately 30% efficiency. It is clear that an
oscilloscope is an absolute essential when using linear amplifiers.
STATUS: While working perfectly in class AB2, the design and testing with
Class 'A' bias is still incomplete.
