You are on page 1of 4

A Single-Stage 1500 W

Amplifier for
160 / 80 / 40 Meters
Do you want lots o’ watts per buck? Here’s your project!

By Frank A. Carcia, WA1GFZ

T
his article demonstrates the 60 W of drive power is required on 80 ation in low-impedance RF circuitry.
ease of designing and building meters. This layout has been scaled up from a
a solid-state, high-efficiency The input and output transformers 160-meter transmitter that I built in
power amplifier using inexpensive are designed to cover 160 through 1996 using 14 IRF840 FETs in push-
switching-power-supply FETs. (Use 40 meters and handle at least 120 V pull parallel running 1500 W peak
static-protective methods while work- peak input voltage during modulation power on AM. The efficiency of that
ing with FETs.—Ed) Although there peaks. The amplifier has been tested final is 90%. It has been in constant
is a wide range of possible operating with three different output-trans- use since 1997, using PWM amplitude
voltages the user may select, the final former turns ratios: modulation with a two-phase current-
goal here is to generate 1500 W out- • A 1:2 ratio produces about 500 W of mode feedback converter.
put from about 50 V dc input. The output power, The amplifier is built on a 10×11-
gates are presently operated at zero • A 1:3 ratio produced about 800 W, inch heat sink 3 inches tall with a
bias, but there is provision for a posi- • A 1:4 ratio produces about 1200 W. total surface area of about 1200
tive bias input to operate the ampli- All of these tests were performed square inches to minimize tempera-
fier in class AB modes. This may be in the class E mode. Drain voltage ture rise. A slot machined in the cen-
tried later, but the present application was limited to 40 V during the test- ter allows the output transformer
is for AM and CW modes only. Fifty to ing. The shunt-loading control in the secondary winding to pass between
tank circuit allows a wide range of the two halves of the output trans-
output power for each output trans- former primary. This heat sink was
former ratio. sized to handle the extra power dis-
181 Columbia Rd The circuit layout has evolved over sipation if the FETs are to be posi-
Enfield, CT 06082 15 years to minimize parasitic reac- tive biased to operate in the linear-
francis.carcia@hs.utc.com tance. This is an important consider- amplifier mode. A quick analysis of

24 Nov/Dec 2005

carcia.pmd 24 10/6/2005, 11:15 AM

This arrangement gate drive with conventional hook-up greatly reduces shunt inductance. a small two-turn inductor nects to the spreaders. tunes the gates to peak the drive sig- The output capacitance of the FETs nal. parts was under $100. Fairchild FQA11N90 rated for 900 a couple minutes of constant carrier ished heat sink with slot and tapped Vds (drain-source voltage) and cost output at 1000 W without a fan on holes for mounting the PC board and about $3 each in small quantities. Although I have not components. The FET the PC board. A sheet of 0. This area is connected to ground capacitors and the heat sink is the but could be used as a point where other. vices are used in the amplifier. Although this inductor can be plus the distributed capacitance of the eliminated. A small area was left for the and release energy over the cycle. The trically insulate the mounting screw end of the transmission line opposite from the heat spreader.4 A. Fig 2 shows the primary of 1. the case of each FET) to the 1/2-inch meters. The heat the drive transformer is loaded with spreaders also serve as low imped. Although this amplifier Fig 2—Primary of the output transformer drive power requirements. Operating class E on 80 meter amplifiers using these FETs. Fig 3—The PC board. two 50 Ω 10 W non-inductive resistors ance connections to the drains and connected in parallel. Their power dissipation is 300 W at tried this amplifier on 40 meters.” This means there would be necessary to limit the peak the final. 50 Ω drive to the amplifier input. ing the “off ” part of the operating gates is about 3000 pF each (result- former primary. addi.003-inch Kapton positive bias is injected. it is impractical to connect the primary tubes.pmd 25 10/6/2005. temperature rise yields “130°C / W / tional external shunt capacitance cycle is about 175 V at 40 V dc on square inch. A PC board was built with two thereby increasing efficiency. Two 3×4×1/8-inch-thick copper heat 25 °C.5:1 across the 80-meter band. the peak drain voltage dur. The input capacitance of the FET copper tubing of the output-trans. The cases of a considerable safety factor even at The FETs used for this project are these devices are barely warm after 1500 W output. wire. Twenty-two of these de- would be about 22°C of temperature voltage on the FET drains. Fig 1 shows the fin. and the total cost of these of this amplifier. Fig 3 shows serves as the dielectric. so they share the load ing source. The match could be used on 160 meters. tionally. Because of the high efficiency thicker if this final was tuned to 40 factor. Nov/Dec 2005 25 carcia. the heat sink. This reduces the output-transformer primaries and reactance of the input circuit and pre- help keep the temperature of each sents a more constant load to the driv- FET constant. The driver-transformer secondary center heat spreaders form one plate of these tap. would have been better in this configu- Fig 1—Heat sink with slot and tapped holes for mounting the PC board and components. mounting screws also attach the The broadband input transformer spreaders to the heat sink and Nylon has a 4:1 turns ratio to match the inserts are used at each FET to elec. 2:21 PM . transmission lines on one side (one for Class E operation requires a shunt each phase) and a ground plane on the capacitor on each FET drain to store other. with a corresponding increase in 80 meters. Further. rise if the final were dissipating the Kapton insulator would need to be which provides a considerable safety 200 W. The heat spreaders ing in an extremely low-impedance are silver-soldered to the transformer circuit). Addi- the output transformer where it con. The input SWR is about properly. the load resistance would heat spreaders provides enough shunt need to be lower to match the drive to capacitance to operate class E on 50 Ω. and their current rating is other amateurs have constructed 40- spreaders connect the FET drains (via 11. where it connects to the heat spreaders. this heat sink has meters.

which provides sufficient. The heat sink tempera- the FET operates in its linear region. inductor value decreases while the its body. We had no way of accu- FET was measured at about 2 ns. termine if it was worth the time to The result of this test proved that The gate drive needs to arrive at package the amplifier into a chassis. old of stability for the tank component thermore. in- sulation. It is also very important to remove all burrs and dirt to prevent damaging the Kapton as the FETs are tightened down. Al. High voltage is also test setup. which appears to be acceptable. present on the secondaries of the step- up output transformers. 43 meter. tor during assembly. My goal was to see how much FET mounting screw dug a hole in the power could be extracted from the insulator while I was trying to find the amplifier with 50 V on the drains and hole in the heat sink.000 V/mil voltage rating. Both sides are tinned with 3% The next logical test step was to try capacitor values must increase to silver solder to minimize the resis. Fig 5 shows the test set former ratio. the amplifier would easily produce each FET simultaneously for the de. brought his peak-reading Bird attached with 22 #4-40 cap screws. plane to the heat sink. The class E drain waveform on the heat sink. as men. heat sink with #6-32 cap screws and This amplifier was tested in the pres- nylon step washers (on the top) to in. Fig 5—The amplifier fier mistuned. the series tun. ration if there were a little more provide better voltage insulation. The threaded the peak power under modulation. much leakage inductance and was sources on the bottom layer (ground). work well and was worth the invest. ence of several hams after the 2003 sulate the gate traces. A small ditional shunt caps for 80-meter op- amount of thermal grease was used eration. Fig 4 shows the doorknob capacitors were added be- assembled amplifier. This test indicated that a bit of sharing the load. could see what this amplifier would bly. fur. in turn.pmd 26 10/6/2005. tank circuit. In this case. 11:16 AM . lowed lower-impedance operation. if not optimum. because my maintain the proper class E wave- tance. vacuum caps for both the tuning and component values. essary. The test verified that the design would 1000 W of carrier. Steve. so the abling his overload interlocks so we Kapton is not damaged during assem. work on the output network was nec- The FET leads are soldered to the ing capacitor failed during these tests. This Fig 4—The assembled amplifier. The output transformer’s secondary lead is #10 (AWG) Teflon insulated wire. to worry about. the nected to the heat spreader through the loading capacitors. tance of each phase by 2000 pF. and the insulator in the 160-meter rig has never failed running at about 80 V peak dc input. tween the drain busses and ground. up. The drain leads are not used. A quick test was performed to de. No test is com. The 1:4 turns ratio of the out- PC board with the gate leads on the so a vacuum-variable cap was in. This was the thresh- vices to share power properly and. is an indication that they are evenly plete without a failure of some sort. My first insulator failed after a dish out. the amplifier worked ture was warm but not hot during The delay between the first and last very well under these sub-optimum these tests. Several CW contacts and rately measuring the dc input current. conditions. The final configuration will have a sleeve of Kapton on the inside of the transformer primary tubes to 26 Nov/Dec 2005 carcia. WA1QIX. increased shunt-capacitor value al- inductance between the drive trans. The PC board is attached to the test supply was severely overloaded. The good reports provided the motivation the heat sink didn’t get warm enough equal temperature rises of the FETs to continue the project. We four #4-40 threaded rods in the cor. former secondary and the PC board. and true to tradition. the FET drain is con. The peak positive volt- age has been measured at about 400 V during testing with the ampli. put transformer introduced too top layer (transmission line) and the stalled to complete the tests. Massachusetts hamfest. Great care is required to protect the so as to increase the shunt capaci- Kapton dielectric of the shunt capaci. culating current is quite high in the later reduced to 1:2 which. spaced off the heat sink with 24 K1JCL. contributed a large dummy 3 /16-inch spacers connecting the source load and RF power meter and Tom. a larger dc power source. Gratifyingly. The board is Marlboro. We found the amplifier would ter the assembly is aligned and the make 1000 W at 50 V but required ad- other cap screws are installed. should have a relatively ment of material and time to complete values because of the output-trans- quick rise time to minimize the time the project. heat spreaders and was cleaned up when four 1000 pF FETs to fill any voids. rods are replaced with cap screws af. provided The assembly is aligned by installing the test shack and modulator. The cir. so the final design uses requires a change in the output-tank tioned earlier. Kapton has a 7. modified Steve’s modulator by dis- ner holes of the heat sink. The FETs are K1JJ.

concepts. ing a series modulator. no duced by 50% with a corresponding should be used in the output trans. to quite low but the layout is very criti. Later are vacuum-variable capacitors. WA1QIX. They as the driver turns on and off.125×1. My testing agrees with this are older switching-power-supply also possible with this amplifier us. and the gate drive until the user is bank of FETs is about 0. set a rial has similar performance with a not be used as a CW linear amplifier rule of thumb at 50 W carrier per little higher permeability. fied to operate on 160 meters with period. margin on the voltage rating of the out a load by mistake without any able core (1. The parts count is depends largely on obtaining a clean value is about 8000 pF per phase. The new total shunt-capacitor final amplifier. form. capacitors. An oscilloscope should be allow a push-pull dc operating im. The A637 because key clicks will be generated device as a good choice for high effi- cores used were flea market loot. tuned to peak the drive signal T2—16 Chrometrics A637 cores (type-43 is heat sink. Nov/Dec 2005 27 carcia. which corresponds to 50 to This amplifier was scaled to use more power at higher voltages. AM is ciency. Figs 7 and 8 show the ampli. use 1/4 W. A capacitor is formed L1—2 turns #14 AWG bare copper wire amplifier transformer. However. C2 and C3 are typical values for or 1:4. better choice. distributed capacitance value. 1500 W Transorbs for FET T1—Broad bandwidth input transformer FET drains and primary of output gate-transient protection. This This amplifier will produce gested A material should be operated amplifier would also make an efficient 1500 W output at 50 V on the drains. and FETs. but the efficiency will be lower than losses low.communication- 0. The value of each capacitor is on 80 meters. larger heat spreader to increase the Conclusion An applications engineer sug. included in the dc power source.5-inch OD. the height of 40 meters without additional shunt 1200 W output at 50 V input. fier module mounted into a chassis for comfortable operating a class E fi- ciency will decrease if the operating additional testing. waveform. 10 W non-inductive resistors voltage. Steve.pmd 27 10/6/2005. 11:16 AM .125 inches. effi. plifier is somewhat different than for eliminate the doorknob caps added Fig 6 shows the schematic of the other modes.) by the two heat-spreader plates and the 1 /2-inch ID. This is in parallel with L2. bypass. www. which seems like a very controlled series switch were to be expected. The amplifier can also be modi. The amplifier will produce a lot 24 VP-P. The RDS ON of each FETs. choice. This least a 2:1 voltage safety factor in be a much better design choice. The gate drive should be at least impedance is decreased further. under 150 gauss at 4 MHz to keep CW transmitter final if a slope. The next drains to provide a comfortable safety This amplifier was operated with- larger size of commercially-avail. capacitors across C1A and C1B. Type-43 mate. with 4:1 impedance. (C2 and C3 wire secondary. Sixty or 70 V would be a conservative number. Inc. There is at having a lot more core area) would output transformer insulation. 1:3 about 3000 pF. will use a larger heat spreader to 1. Turns ratio is 1:2. A few more FETs would al- cores that work well up to 80 meters. than I had in my junk box. 1 turn 1/2-inch OD copper tubing about 3000 pF of output capacitance 80 meters. the tall stack of cores never changed additional shunt capacitors and/or a case temperature during testing. Us. amplifier can be scaled to operate on the drain waveform while running ing these larger cores.09 Ω. This and halve them for 40 meters. with center tapped transformer common point. (Source: Communication is isolated from the heat sink with a capacitors providing a low-impedance dc Concepts. It can. Double them for 160 meters primary and #10 AWG Teflon insulated from the 11 parallel connected FETs. Fig 6—A partial schematic of 80-meter class-E amplifier (FQA11N90).com or an old solid-state C1B—Same as C1A. Unless otherwise specified. shorter. nal. but half as many cores the initial tank capacitor failure. This plate C4—Parallel combination of film and disc secondary. This is sized to work on is the class-E shunt capacitance. other components have failed after reduction in leakage inductance former. but the 60 W of RF input power on 80 larger output transformer cores goal was to operate at 50 V on the meters.003-inch thick Kapton sheet.) 160 meters with 120 V dc (peak) input 4000 pF was added to the shunt value by R1—(2) 50 Ω. connecting two 1000 pF doorknob connected in parallel. cal to insure power sharing of the used to monitor the drain waveform pedance under 2 Ω. C1A—3× ×4-inch copper plate 1/8-inch thick. One of 11 parallel connected circuits shown. CR2—15-V. CR1. extensive testing over a one-year with a better aspect ratio. My next design Type-43 comes in the same size: The adjustment of a Class E am. low 1500 W at 50 V. 5%-tolerance carbon composition or film resistors. After the output transformer could be re. and efficient operation to this configuration. a good substitute). shunt capacitor insulation and protection or failure.