By W5JGV - WD2XSH / 7
First Posted on December 14, 2008
An Hybrid Amplifier Designed for the ARRL 600 Meter Research Project
This amplifier is capable of 500 watts of RF output in linear operation on the 600 meter band.
The RF drive requirement is less than 50 milliwatts.
With 3200 Volts on the anode, the amplifier draws 268 MA key down, CW mode. Screen voltage is 600, and the anode idle current is 80 MA. Idle dissipation is 256 watts. DC Input power is 915 watts, and the RF output power to the dummy load is 540 watts, for a calculated efficiency of 59% with good linearity.
In order to get more than a few miles on 600 Meters, it is necessary to use a fairly serious RF power amplifier to feed enough power to the typical antenna used on that band. Antenna efficiency normally will be in the range of 4 to 18 percent, unless the operator makes a heroic and expensive effort to install a really serious antenna. As a member of the ARRL 600 Meter Experimental Project, (WD2XSH/7), my proposed antenna will be in the low to middle point in that range, hence the need for some serious RF from the amplifier. Because the predominant digital mode to be used from this station will be PSK-31 it was necessary to use a linear amplifier in order to obtain a clean signal on the air.
Although I was able to successfully modify the solid-state linear amplifier that I had previously used on 166.5 KHz for experimental station WC2XSR/13 to enable it to operate on 600 Meters, I found the relative fragility of the transistors to be a problem. After vaporizing several of the MOSFET devices while attempting to operate the amplifier into a slightly mismatched load, I decided that perhaps I might be better off with a vacuum tube amplifier for at least the initial phases of testing. Since I had previously had some experience with designing and operating high power vacuum tube amplifiers, I reached into my Junk Box and started to build a new amplifier.
I already had the beginnings of an amplifier that I had planned to use for another medical research project. It was designed to operate in the 27 MHz ISM band, but due to changing requirements, I had never completed the amplifier. All that existed was the power supply and the amplifier chassis which contained a 4-250 tube, an RF choke for 27 MHz, a filament transformer, several meters, and the cooling system for the tube. I figured that this would make the start of a nice linear amplifier for the 600 Meter band.
Since I am retired and I did not want to spend a lot of money on the amplifier, I "made do" with a lot of odd parts and I used whatever I could find around the shack to build the amplifier. I left the 4-250 in place during the construction and initial tune-up tests, since the only real difference between the 4-250 and the 4-400A is the plate dissipation rating. All the interelectrode capacitances, grid voltages and drive requirements are the same for both tubes when operated at the same DC input power level. After construction was complete and all circuit constants had been determined, I replaced the 4-250 with the 4-400A and began full power tests.
Special thanks should be given here to N6LF, Rudy Severns, who graciously provided me with the 4-400A tube - and a spare - plus the sockets and the Eimac chimneys for the tubes. Also, my great thanks go to W5THT, Pat Hamel, who let me bounce lots of ideas off of him during construction.
To help you to quickly follow the circuit discussion in this article, please download the PDF files listed here.
You may wish to print the files for easier reference while reading this web page,
Click to download the Block Diagram of the amplifier.
Click to download the schematic diagram of the RF Driver.
Click to download the schematic diagram of the 4-400A PA.
Click to download the schematic diagram of the Bias Voltage Regulator.
Click to download the schematic diagram of the Power Supply.
Click to download the data sheet for the 4-400A.
Click to download the data sheet for the IRF730.
Click to download the data sheet for the 2N5089.
I grew up (in a technical sense) just as transistors started to appear on the electronics scene. As a result, I feel comfortable working with either vacuum tubes or transistors for whatever project I am working on. After blowing up a few MOSFET's in my solid-state amplifier, I decided that the design of that amplifier was pushing a two-transistor design right up to the limit, and any load or operating errors were going to result in destroyed devices pretty much no matter what I did. With that in mind, I started on the preliminary design of a new solid state amplifier that would use up to twenty MOSFET's arranged to share the load evenly. It "should" be bulletproof, and it would be capable of generating 1500 watts of RF.
After purchasing quite a few of the components for the new design, I came to my senses and realized that the actual implementation of the amplifier would likely be rather more difficult than the calculations indicated (personal experience speaking there!!) so I rethought the idea again. By then, I had recalculated the power budget for the antenna system and decided that about 500 watts of RF would be more than enough to accomplish what I needed to do. It was then that I thought about using the as yet unfinished 27 MHz amplifier.
Having executed a few designs with both solid-state and vacuum tubes, I thought that it might simplify the design of the amplifier if I were to use transistors for the low power portion of the amplifier and a vacuum tube for the output stage. I also knew that I would have to have considerable gain between the input to the amplifier and the output, because the RF drive available from the Starpoint channel modem I was using is quite low. I calculated the gain budget, and found that I could do the job with only two transistors. A single high power vacuum tube as the output amplifier would complete the design.
I knew that using a push-pull amplifier would result is a better output waveform with reduced even harmonics, and also make the amplifier more efficient. The problem was, that I had only one 4-250 available at that time, and the air cooling system was only able to provide enough air for one tube. I also felt that if I added a second tube, there might not be enough room on the rather small chassis (remember, this was for a 27 MHz amplifier) to fit the plate tank circuit. In addition, using a single tube would allow me to use a smaller tank coil and a single-section tuning capacitor for the tank circuit.
Enter the 4-400A. It was a drop-in fit for the 4-250, but would easily handle a Kilowatt DC plate input with a single tube. It would be necessary to operate the tube in the Class AB2 region to both obtain sufficient output power and reasonable linearity at the same time. I also knew that the tank circuit design would be trickier if good linearity was to be obtained.
GENERAL CIRCUIT CONFIGURATION
Please refer to the PDF of the Block Diagram of the amplifier.
RF drive is supplied to the amplifier by my Starpoint RF Channel Modem at a drive level of roughly 31 milliwatts @ 50 Ohms. A signal level of less than 50 milliwatts across a 50 Ohm resistor will be enough to drive the amplifier to full output. This is slightly more than 3.5 volts peak-to-peak. Any suitable source of 600 meter RF may be used instead of the Starpoint modem.
The first RF Driver stage uses a 2N5089 transistor to boost the input RF signal to a maximum level of about 10 volts. This amplified signal is sent to the gate of the second RF Driver stage, which uses an IRF730 MOSFET. This stage boosts the RF drive to the 5-watt level.
The 4-400A power amplifier stage is driven by the secondary of the RF driver output transformer that is driven by the IRF730. Both the primary and secondary of the transformer are tuned to resonance at the operating frequency to produce a clean drive waveform for the 4-400A
AC Mains power is supplied to the Main Power Supply, which is remotely controlled from the amplifier chassis by several front panel switches. Mains AC from the main power supply passes through to the amplifier chassis, where it is fed to the filament transformer for the 4-400A, the blower that cools the tube, and the +41 volt power supply for the RF driver board.
The Main Power Supply produces the required voltages for the 4-400A. They are: +3200 volts for the anode; regulated +600 volts for the screen grid; and -150 volts for the grid voltage regulator for the grid 1 bias.
RF DRIVER BOARD
Please refer to the PDF of the RF Driver.
The input to the first stage in the RF Driver amplifier is shown terminated with a 3300 Ohm resistor. This is because of the low level RF signal available from the Starpoint channel modem I am using. This resistor may be changed to any convenient value between 50 to 4700 Ohms, depending of the drive signal available. It is recommended that the terminating resistor not be omitted as this may cause the first driver amplifier to go into oscillation under low drive level conditions.
The 2N5089 is operated in Class A. Negative feedback is provided by the 680 Ohm resistor in the emitter lead. The 2.5 Kohm RF Gain control acts as a high frequency bypass across the 680 Ohm resistor, and increases the gain of the 2N5089 as the resistance of the potentiometer is reduced. This adjustment will normally be set to about 50% of rotation.
The amplified drive signal is taken from the collector of the first stage through the 0.0022 uF DC blocking capacitor and fed to the gate of the IRF730 MOSFET. DC Bias for the IRF730 is provided by the 120 Kohm resistor and the 20 Kohm potentiometer. The 33 Kohm resistor between the wiper of the Bias pot and the gate of the IRF730 isolates the RF drive from the DC bias line. NOTE: If the +41 volt power supply does not current limit at about 2 amperes, it is possible to destroy the IRF730 if you set the Bias pot too high.
Output Transformer T1 serves a dual purpose, It resonates at the 600 Meter frequency of operation, and it steps up the 41 volt drain voltage waveform to approximately 300 volts maximum peak voltage to drive the grid of the 4-400A. The primary of T1 is brought close to resonance by the use of a 6800 pF poly film or mica capacitor. The secondary is tuned to resonance by using a combination of variable and fixed capacitors.
To adjust the driver amplifier for best performance, Insert the 4-400A (or a 4-250) in the socket. Apply filament voltage, grid 1 bias, and cooling air to the tube. Do NOT apply screen voltage or plate voltage!
Prepare to feed the input of the driver amplifier with an adjustable level two-tone RF drive signal that measures about 3 Volts peak-to-peak at maximum level as measured on an oscilloscope. For now, set the drive level to about 0.5 volts.
Turn the RF Gain pot to MAXIMUM resistance. (Lowest RF Gain.)
Set the IRF730 Bias pot to minimum bias - that's with the wiper set to the ground end of the pot.
Apply power to the RF driver amplifier board. No smoke? Good!
Using the oscilloscope, observe the waveform at the collector of the 2N5089. It should be a sine wave with very little distortion. If it is clipping, reduce the RF drive level.
Adjust the RF drive level to the point where the signal the the collector of the 2N5089 just starts to clip. Then, reduce the signal to 60 - 75% of that value,
Connect one channel of the oscilloscope to the RF input of the driver amplifier, and another oscilloscope channel to the grid of the 4-400A.
Increase the RF drive level while observing the oscilloscope waveforms. Stop increasing the RF drive below the point where the grid waveform clips or distorts badly.
Adjust the tuning capacitor across the secondary of the driver transformer for the best waveform. What you want to do is to adjust everything to get the two-tone grid waveform envelope to match the exact shape of the RF Driver input two-tone waveform envelope.
You will see lots of distortion at the crossover points of the two-tone signal. Increase the bias on the IRF730 to eliminate the crossover distortion. As you increase the bias, the gain will increase, so you will need to reduce the RF drive level to prevent peak distortion.
"Play with" the input RF drive level, the adjustment of the IRF730 Bias, the RF Gain and the T1 secondary tuning capacitor to get the envelopes of the waveforms to match. It is not necessary to have the grid current meter read more than 25 MA during these adjustments, as that will be plenty of drive for full power. The first time I attempted to adjust the driver amplifier, it took me about an hour to "get the hang of it." After that, I was able to readjust the amplifier within a few minutes after making component changes during testing.
You may need to adjust the value of the 6800 pF capacitor that is placed across the primary of transformer T1 in order to improve the waveform. Although this value if not critical, it is important. Correct selection of this capacitor will keep the drain current minimized and allow the driver amplifier to produce more power output.
After completing the adjustment as described here, note the accuracy of the waveform envelopes and how close they match. Take a digital photograph is you can. Now, reverse the leads from the secondary winding of transformer T1 . Repeat the adjustments and compare the waveform accuracy with the results you achieved before you swapped the leads from T1. Leave the leads from T1 connected which ever way gives the best and most accurate waveform envelope.
GRID BIAS REGULATOR
Please refer to the PDF of the Bias Regulator.
The function of the grid bias regulator is to hold the grid bias constant when grid current flows and to allow the bias to be adjusted for the correct voltage to set the no-signal idle plate current through the 4-400A and to reduce distortion and improve amplifier linearity through the 4-400A. In this design, a resting idle current of about 80 MA was sufficient to provide good linearity, as seen my the waveform photos later in this article.
The main power supply generates a regulated -150 volts from a VR-150 regulator tube. This voltage passes through a 100 Kohm resistor located in the power supply and then on through the connecting cable to the bias regulator located in the RF amplifier chassis. The bias regulator is a single stage shunt regulator. The IRF730 MOSFET acts as a variable resistor between the bias voltage line and ground, and allows the operator to adjust the negative grid bias voltage to any desired value between -40 to -130 volts. When grid current flows, the electrons collected by the control grid of the 4-400A would cause the negative bias voltage to increase, but the IRF730 "wants" to hold the voltage constant. It shunts the extra electrons to ground, thereby holding the grid bias constant.
It is important to understand that the power supply feeding the bias regulator must have a limited amount of current available. The simplest way to limit the available current is to place a resistor in series with the output of the power supply feeding the Bias Regulator. Attempting to connect the bias regulator to a "stiff" power supply will cause the IRF730 to draw excessive current. This will damage the power supply or destroy the IRF730, or both.
In this circuit, the 100 Kohm resistor limits the available current to approximately 400 V / 100000 Ohms = 0.004 A or 4 Milliamperes. This is just enough to "fire" the VR tube so it will regulate at -150 volts. At that point, the IRF730 need only shunt less than 4 MA to ground in order to set the bias voltage to the proper value. The only real function of the VR-150 is to place an upper limit on the bias voltage when the IRF730 is biased off by the use of the Cut Off Bias Relay contacts. When grid current flows during normal grid drive conditions, the current the IRF730 shunts to ground will increase, but in any case it will be no more than the operating grid current plus the current through the 100 Kohm resistor. This will be something less than a total of about 40 MA.
The bias voltage is set by adjusting the 10 Kohm potentiometer. Note that this is an "upside down" circuit, that is, the positive side of the power supply is grounded. That's handy for this circuit, since it allows us to simply bolt the IRF730 directly to the grounded chassis with no insulators being required. In any case, the transistor does not get very warm, since it only has to dissipate a few watts at most.
The Cut Off Bias Relay switch shown in the diagram is operated by a switch at the operators position. When the switch is opened, the IRF730 is turned off, and that allows the grid bias to increase to the full -150 volts supplied by the main power supply. This cuts off all plate current flow through the 4-400A, preventing the tube from generating any shot noise that would desensitize the receiver.
Because this is a single transistor regulator, it does not have much gain. It holds the grid bias to within about 2% of the set point, which is adequate. Because of the low gain, there is some "bounce" on speech waveforms. The addition of the 1 uF capacitor from the bias line to ground eliminates most of this bounce. This capacitor may be increased in value if additional bias smoothing is required.
4-400A POWER AMPLIFIER
Please refer to the PDF of the 4-400A PA.
RF drive for the 4-400A comes from the RF Driver output transformer, shown here as transformer T1. One end of the secondary winding of T1 is connected to the output of the bias regulator through the 25 MA grid current meter. RF bypassing to ground is provided by a pair of 0.47 uF and 0.01 uF capacitors.
Screen voltage comes directly from the screen voltage regulator in the main power supply. No metering for the screen current is provided since the current is limited to 80 MA by the series dropping resistors in the main power supply. If the screen attempts to draw more than 80 MA, the screen voltage will drop proportionally, thereby limiting the power dissipation of the screen to a safe value.
This arrangement is not the best use of the VR tubes because during periods when the plate current is cut off, such as during receive, the full current available through the screen dropping resistors must pass through the VR tubes. These tubes are rated for operation at 30 MA, and they will be subjected to an overload condition because 80 MA will pass through them. However, since plate voltage will normally be turned off during receive periods, the current will rapidly decrease as the filter capacitors discharge. A better arrangement would be to use either a solid-state regulator that can handle the extra current or to use a relay to place a resistor across the VR tubes to shunt the extra current away from the VR tubes during receive.
The filament of the 4-400A is fed by a step-down transformer. RF bypassing of the cathode to ground is done with multiple ceramic disk capacitors. The center tap of the filament transformer connects to a 500 MA meter that reads cathode current and screen currents. Although this gives slightly high readings for the plate current, placing the meter in the cathode circuit greatly reduced the shock hazard that would be present if the meter were to be placed in the plate high voltage line.
The cathode current meter is bypassed by a 0.51 Ohm resistor and a 1N4001 diode. Should there be an arc inside the 4-400A, the sudden increase in cathode current will cause a large voltage drop across the meter movement. This could be high enough to burn it out, or at least wrap the needle around the pin. If more than one ampere flows through the circuit, this will cause the voltage across the meter movement and the resistor to exceed the conduction potential of the 1N4001. The diode will then conduct and act as a voltage limiter, shunting excess current around the meter. Should an arc event happen, the overcurrent relay in the main power supply will interrupt the mains power to the HV supply promptly to prevent serious damage.
Plate voltage to the 4-400A is supplied through the 1.65 MHy RF Choke. A DC blocking capacitor of 1000 pF allows the RF from the anode of the tube to go to the plate tank circuit. Because a single variable capacitor of the size needed to tune the tank circuit to resonance would be quite large, a combination of variable and fixed capacitors were used in the amplifier.
The RF Output coupling link coil is fixed in place for simplicity. Loading adjustment is provided by the use of a separate tap on every turn. If it is needed, an additional fine loading adjustment may be accomplished by the use of a large variable capacitor in series with the output of the link coil. See most early editions of the Radio Amateurs' Handbook for more information.
While building the amplifier, I quickly found out that a few things must be considered to get good performance out of the amplifier when running in a linear mode. First, the control grid bias supply must be stable and noise free. Second, the screen voltage supply has to be really well regulated or the linearity will suffer badly. I used VR tubes, because they were available and generally work well enough. I did find that the 30 MA that they can supply in regulated mode was not enough for the amplifier. I had to adjust the resistor value feeding the VR tubes to allow almost 80 MA of current at full load. This is excessive for the VR tubes, but the screen normally pulls enough current so that the VR tubes are not overloaded. During each RF cycle in the tank circuit, if the plate voltage drops below the screen voltage, the screen will draw enough current to cause the screen voltage to go out of regulation. This will cause severe distortion in the RF output of the amplifier.
Another thing I found is that the plate tank circuit must have much more capacity that calculations would indicate. In fact, at least twice the calculated value. If the "C" is too low, the amplifier will not be able to generate very much power with good linearity. When loading the amplifier, it should be loaded quite heavily for best linearity. In fact, at correct load, there is almost no plate current dip visible on the cathode current meter at resonance. At light load, of course, there is a pronounced plate current dip at resonance.
MAIN POWER SUPPLY
Please refer to the PDF of the Power Supply.
The main power supply converts the mains AC voltage to the voltages needed for the 4-400A. Since the power supply was originally designed for use as a test system for a research project, it was built with flexibility in mind. It has been modified slightly to use with this amplifier, so you will see some differences between the description , diagrams and the photos in this article. There is nothing special about this power supply, and any suitable supply providing these voltages may be used with this amplifier.
The high voltage plate transformer was salvaged from a very early microwave oven. This particular transformer weighs almost 20 pounds, and has a lot of iron and copper in it when compared to most present day microwave oven transformers. Because of the extra iron in the core of the transformer, I was able to boost the AC input voltage to the transformer to increase the secondary voltage by about twenty percent. If you try that trick on most of these transformers, they will draw way too much primary excitation current and overheat badly. This transformer was not bothered by the voltage increase because it is a "real" transformer, and not a "toy" transformer that uses the minimum amount of copper and iron.
The output of the HV transformer is full wave bridge rectified and sent to a bank of 16 series connected capacitors, each of which is rated at 2900 uF at 200 VDC. This is the equivalent of 181 uF @ 3200 VDC. Do NOT get your fingers across this thing when it's powered up! Because the supply was designed to have an output of 2500 to 2700 volts under load, the capacitors are operated slightly above their ratings in this application. When in use with the boosted AC mains voltage, the capacitors are charged to between 2700 to 3700 volts depending on the load. There have been no failures - so far!
Each capacitor has a voltage equalizing resistor shunted across it. The negative lead from the capacitor bank is connected to circuit ground through a 5 Ohm resistor. The voltage drop across the 5 Ohm resistor is monitored by a small DC relay that pulls in if the supply current exceeds one Ampere. The power supply is then shut down and locked out until all power is removed and restored.
Screen voltage for the amplifier is obtained by a series of dropping resistors that are connected to 4 series-connected VR-150 voltage regulator tubes. The resistors limit the maximum short circuit current to a low enough level so that the screen grid cannot be damaged during normal operation or circuit malfunctions. Since the screen voltage is derived from the plate voltage supply, it is impossible to apply screen voltage without plate voltage at the same time. The resistors limit the maximum current flow to about 80 MA.
This arrangement is not the best use of the VR tubes because during periods when the plate current is cut off, such as during receive, the full current available through the screen dropping resistors must pass through the VR tubes. These tubes are rated for operation at 30 MA, and they will be subjected to an overload condition because 80 MA will pass through them. However, since plate voltage will normally be turned off during receive periods, the current will rapidly decrease as the filter capacitors discharge. A better arrangement would be to use either a solid-state regulator that can handle the extra current or to use a relay to place a resistor across the VR tubes to shunt the extra current away from the VR tubes during receive.
Grid 1 bias for the 4-400A is provided by a full wave rectifier that is fed by a separate transformer which is energized as soon as the filament voltage is turned on. The transformer is rated at 600 volts center tapped. This voltage, after rectification and filtering, results in a raw bias voltage of about -450 volts. This is much more than is needed, but the original design of the power supply allowed for producing bias voltages of up to -350 volts. The regulator circuit was changed to produce the lower bias voltage for this amplifier, but due to the difficulty of changing the bias supply transformer, the original transformer was left in place.
A step-start circuit is used in the primary circuit of the HV transformer to prevent start up inrush current surges. Due to the large excitation current normally drawn by the microwave oven transformer (which was increased even further by the boosted mains voltage) a power factor compensation capacitor was added to the mains circuit. This would be unnecessary if a real plate transformer were used.
The operational sequence of the power supply is controlled by a series of 12 volt DC relays. A small transformer and bridge rectifier supply the necessary DC control voltage whenever the mains voltage is connected to the power supply. Fuses are used for throughout the power supply for circuit and operator protection, and the entire supply is cooled by the use of a fan salvaged from an old microwave oven.
The operational logic of the control relays will be left as an exercise for the reader.
Now I'll take you through a look at the insides of the amplifier and point out a few of the construction details.
Under chassis view of the amplifier deck.
The front panel of the amplifier is to the left in this photo. The back of the three panel meters are seen at the left of the chassis. To the upper right is the +41 volt power supply for the RF Driver circuit board, which is seen mounted horizontally just below the power supply board. The power supply was salvaged from a discarded Epson inkjet printer. The supply puts out a regulated +41 volts and is current limited at 2 Amperes.
To avoid distortion and ripple in the RF signal, this power supply must be well regulated and filtered. Hum and noise on the +41 volt line measures less than 200 millivolts peak-to-peak. It is very important that the power supply for the driver stage is well regulated and filtered. Any hum or noise present in the power will AM modulate the RF signal and be present in the RF output from the amplifier.
The RF Driver output transformer may be seen just below the RF Driver board, and just above the filament transformer. The black square to the right of the filament transformer is the opening from the centrifugal blower that provides cooling air to the 4-400A tube. The white capacitor to the left of the filament transformer is the phasing capacitor for the split phase, capacitor-run blower motor. The tuning capacitor for the secondary of the RF driver output transformer may be seen mounted on the front panel at the bottom left of the picture.
The +41 Volt RF Driver power supply board.
The RF Driver board.
Note the terrible construction practices and "blobby" soldering. It was built as a quick prototype, but it worked so well that I just stuffed it into the amplifier and called it finished!
The heat sink for the IRF730 was salvaged from an old computer power supply. It receives additional cooling from the air provided by the blower for the 4-400A. If the driver amplifier were to be mounted outside of the sir flow, a better heat sink would be required to prevent the transistor from overheating.
I had to add an extra low-ESR electrolytic capacitor directly across the amplifier power buss to reduce power supply ripple to the required low value. All of the capacitors you see in the picture were salvaged from old computer power supplies.
The square brown capacitor seen above the IRF730 is the tuning capacitor for the primary of the output transformer.
The RF Driver Output Transformer
The transformer was wound with 18 turns consisting of seven parallel twisted together strands of #24 AWG enameled wire for the primary. The wire for the primary was salvaged from a scrapped computer power supply switching transformer. The secondary is wound with a total of 50 turns fashioned from two lengths (connected in series) of #26 AWG wire from some scrap CAT-5 network cable. A single length of wire would have worked just as well, I simply did not have a long enough length of wire handy, so I connected two shorter lengths together for the secondary winding. The transformer core is a T-200-26.
Notice the use of a length of loudspeaker "Zip Cord" wire as the transmission line between the secondary of the driver output transformer and the grid of the 4-400A. It works very well, and no feedback or oscillations were noted in the amplifier under any operating conditions.
Grid Drive and Bias area.
The variable capacitor is the tuning capacitor for the secondary of the RF Driver output transformer. Because this capacitor is not adjusted very often, it has a screwdriver slot adjustment instead of a knob.
The two small brown disk capacitors at the right side of the variable capacitor are the fixed capacitors of 470 and 330 pF that are placed in parallel with the variable capacitor. Since the adjustment of the variable capacitor is not changed after initial adjustment, if desired, it may be replaced with a fixed capacitor after the final value is determined. Because the RF grid voltage and current is fairly low, I was able to use standard 600 volt disk bypass capacitors here.
The large white block-looking capacitor (which was salvaged from a computer power supply) seen above the variable capacitor, is the 1 uF grid voltage bypass capacitor. The actual grid bias regulator circuit is connected to the brown terminal strip mounted on the side wall of the chassis just below the tuning capacitor in this picture. The back of the grid bias adjustment potentiometer is seen in the left side bottom of the picture, next to the tuning capacitor.
RF Amplifier Plate Tank Coil
This end view shows the general construction of the coil. The gray PVC support pipes have a series of slots about 1/4" deep that were cut into them by using a table saw. The wire for the coil is bare #10 AWG copper wire. After winding the coil, I used some clear spray paint to coat the coil to keep the copper wire shiny looking. I forgot to stretch the wire before I wound it, and is sure shows! Brass hardware was used for all the RF connections. All other hardware was kept as small as possible to reduce RF Losses.
This coil was originally intended to be part of a loading coil variometer, and had a rotating link coil installed inside of it. However, it did not work out as planned, so I removed the link coil and used the fixed coil for the amplifiers' plate tank coil. As it finally worked out, the inductance was excessive, and so the completed amplifier only uses about half of the turns you see in the pictures.
As you can see, the link coil was simply wound over the "cold" end of the tank coil. Taps were provided on evert turn for coupling adjusted. This may not look too neat, but it works very well and is stable in operation. The link coil was wound with #12 AWG THHN insulated copper wire.
RF Sample Link
I finally settled on a single turn RF sample link to obtain some RF for monitoring the waveform of the amplifier. I simply wound it over the outside of the coil form. There is nothing special about this configuration, I just happened to have some left over #12 THHN insulated copper wire that was coiled into a circle that size. It fit perfectly, and worked well, so I left it there.
This view of the RF sample link shows it connected to the small "Zip Cord" transmission line that goes to the front panel BNC connector for the RF output sample.
RF Output Link Coil Taps
Oh my! How messy! But, it works.
The RF voltage is fairly low on the link taps, and making a simple twist in the wire while I was winding the link coil provides for a fast and easy way to adjust the coupling. You can see that I thought I had the correct coupling tap, and I had soldered the output wire to the fourth tap from the left. However, I determined that the tap position was incorrect. You can see that the output wire is now just hanging inside the connection on the last turn to the right of the link coil.
If you look carefully, you can see that the yellow RF sample link had not yet been added when this photo was taken. Also, note that one side of the "Zip Cord" transmission line from the RF sample connector si connected to the tap on the first turn of the link coil. This proved to be a big "No-No" because the RF voltage on the first turn of the link coil was high enough to overheat and melt the plastic insulation on the "Zip-Cord."
Here you can see the copper grounding straps for the cold end of the link coil and the cold end of the plate tank coil. They are connected directly to the plate tuning capacitor frame, which is in turn grounded directly to the chassis. Because the tuning capacitor was too long to fit on the chassis, an adapter plate was added to the rear of the tuning capacitor to enable the back end of the tuning capacitor to sit slightly to the rear of the chassis.
Rear View of the Amplifier
You can see the yellow wire that connects the RF blocking capacitor from the 4-400A anode connection to the "hot" end of the tank coil. All of the turns in the coil from the point where the yellow wire connects to the coil to the right side of the coil are not used. They are simply left open circuited. Because there were no unwanted resonances in the open section of the coil, there were no high voltages built up across the open turns, and they could be safely left open. I also tried shorting out the extras turns, but this increased the tank circuit losses by about 25 watts, so I left the extra coil turns disconnected.
The black cylindrical capacitor to the right of the tank tuning capacitor is the 10 KV 830 pF Mica capacitor that is in shunt with the tuning capacitor. The cooling blower is to the right of the right-hand PVC pipe support for the tank coil.
The RF output connector was originally planned to be fed through the rear of the chassis, but it turned out to be better to place it directly on the tuning capacitor. This avoids high RF current flow from the capacitor frame to the chassis and lessens the potential for RF feedback.
The plastic terminal strip has connections to allow the plate voltage supply to be switched on and off from the operating position, and to change the bias on the 4-400A tube from operating to beyond cut-off for receiving.
The black plastic connector has all the interconnect lines between the power supply chassis and the amplifier, including the +3200 volt plate voltage line.
A close-up of the tap on the plate tank coil.
The coil tap is made by taking a small strip of of copper sheet measuring about 1/4" wide by about 1" long and bending it into a "U" shape around a scrap of copper wire of the same gauge as used on the coil. Then the tab is slipped in place over the appropriate turn on the coil. Next, place a pair of slip joint pliers over the tab ends set the end of the pliers flush against the coil turn. Squeeze the tab ends flat against each other. This should form the tab into shape around the wire. Solder the tab ends together, and the tab loop onto the coil turn, then solder the tap wire onto the tab.
This is the small 50 pF fixed vacuum capacitor that is placed across the main tuning capacitor. If the amplifier is operated at plate voltages of 3200-3500 volts, this capacitor may need to be changed to 100 pF to get the tank circuit to hit resonance.
This is my quick mounting method for the vacuum capacitor. The RF current through the capacitor is low, so the twisted wire connections work OK. Also note the use of aluminum tape to hold down the "Zip Cord" transmission line for the RF sample tap.
Not having a nice shiny knob for the main tuning capacitor, I remembered that I had several glass doorknobs in the Junk Box. I quickly drilled out the shaft hole to fit the shaft of the tuning capacitor and installed it on the shaft with a setscrew. It works fine, and since it does not have to be adjusted often, no pointer is necessary. And it does attract attention!
The 4-400A tube installed in the amplifier.
Note the double glass of the cooling chimney. This was accidental. Originally, I had planned to use a 4-250 in the amplifier and not having a standard chimney, I pressed a Coleman lantern globe into service. To save the time and trouble of designing metal clips to hold the glass, I used a bead of RTV adhesive to attach the glass to the chassis. Because the straight sided chimney did not force the cooling air against the plate cap seal, I constructed a rather massive heat sinking plate cap connector (seen here from the reverse side) that has cooling fins that are exposed to the cooling air that flows up and past the side of the tube. This was to keep the anode connection of the 4-250 cool. When I obtained the 4-400A, I also obtained the Eimac chimney. It turned out to be a loose drop-in fit inside the Coleman globe. The turned-in top edge of the Eimac chimney provides the needed air against the anode connection of the tube, but I still used the heat sink plate cap since I had it already. It also adds a bit more cooling for the anode connection.
The plate choke is wound on a ceramic stand-off insulator (See PDF for details) and stays cool in operation. There is no parasitic suppressor in the amplifier. As constructed, the amplifier appears stable under all test conditions, with and without load.
Close up of the tube and RF choke
You can see the RTV adhesive bead under the Coleman glass. The RF choke is held against the chassis by a screw that passes up through the chassis. Cushioning is provided between the chassis and the ceramic by a cardboard washer. When I wound the RF choke, I held the end of the wire in place with some electrical tape. After finishing the winding, I applied a drop of Super Glue to the end turns to prevent the wire from unwinding.
The heat sink anode connector was constructed from a length of 1/2" thick 2" x 2" aluminum angle stock. Two salvaged heat sinks from an old graphics computer system were attached to the side of the assembly. They were designed to be exposed to the air flow as it passed over the tube.
Looking at the connector from the other side shown the two heat sinks.
This is the bottom plastic insulating bracket that holds the choke lead that connects to the +3200 volt supply.
I made a small copper "L" bracket to hold the plate blocking capacitor on to the top of the plate choke. The top of the choke winding is soldered to the copper bracket. I placed another cardboard washer between the copper bracket and the ceramic insulator to prevent cracking the ceramic. The black smudge across the lead-out wire is not from arcing, it is residue from the black electrical tape that got stuck to the Super Glue.
A view of the front panel of the completed amplifier. The glass knob adjusts the PA Tank circuit tuning capacitor. The BNC connector under the chrome handle is the Output RF Sample tap connector. The only thing missing is a window to allow visitors to the shack to see the 4-400A in operation. The original design of the amplifier did not require a window, and so unfortunately I forgot about it until I had completed construction of the RF deck. At that point, I was somewhat reluctant to take a chance on damaging the meters or anything else due to the heavy vibration caused by cutting a large opening through the thick front panel, so no window, at least for now.
The small black knob is the adjustment potentiometer for the grid bias voltage. The screwdriver adjustment is for the tuning capacitor for the RF Driver output transformer T1 secondary. It is adjusted once and left alone, so there is no need for a knob.
RF Input and Output Waveforms Superimposed
Believe it or not, this is an actual photo of both the RF drive signal to the amplifier and a sample of the RF output from the amplifier taken at the dummy load. It is virtually impossible to discern the two waveforms from each other, indicating that the amplifier is quite linear over the operating curve. This photo and the following two waveform photos were taken at the 500 watt peak power level.
RF Input and Output Waveforms Separated
The top waveform is the RF drive, and the bottom waveform is the RF output.
RF Input and Output Waveforms Offset
The top waveform is the RF drive, and the bottom waveform is the RF output. The slight AC-looking ripple riding on the top of the waveforms is the result of artifacts in the RF drive signal from the Starpoint channel modem. It is caused by a leaking transistor in one of the balanced bridge circuits in the modulator.
The amplifier in the rack on the right side of the photo is the 600 meter amplifier discussed in this article. The amplifier on top of the cabinet in the left of the photo is a partially assembled 1-KW grounded-grid amplifier for 160 - 10 meters.
Front view of the amplifier rack
The RF deck sits in the top of the rack so the hot air from the amplifier can exit the vents on the top of the cabinet. (All the panels have been removed for this photo and during testing.) The power supply is in the bottom of the rack. The center shelf holds the Starpoint channel modem and a CD player used for providing the necessary audio signals for testing. The oscilloscope is used for monitoring the output waveform from the amplifier. Sharp eyes will note the missing bias pot knob and the RF driver tuning capacitor. Also missing is the yellow single turn RF sample loop around the right hand side of the plate tank coil. This photo was taken before those items were installed.
My Bird 43 in-line wattmeter is seen hanging from the 1200 Watt Bird dummy load that is mounted on the wall behind the amplifier racks. It's out of the way, and gets plenty of clean cooling air. I just connect it to whatever unit I am testing and then roll up the cable when I am done testing.
A top view of the main power supply with the protective top cover removed.
From top to bottom at the left - the cooling fan, the screen dropping resistors, the microwave oven high voltage transformer.
From top to bottom in the center - the AC power cord and RFI line filter, the remote control cable and connector, some of the control relays, blue bias voltage filter capacitor, single VR-150 bias regulator tube, another control relay, the small silver colored control voltage transformer and gray filter capacitor, open frame overcurrent relay, 4 VR-150 screen voltage regulator tubes, partially obscured grid bias power transformer.
Right side - 16 high voltage filter capacitors with their associated voltage equalizing resistors.
Construction note - I did not have the space to mount all the metal clamps to hold down the filter capacitors, so I took a one inch thick board and used a hole saw to cut a series of 16 holes in the board that just fit the capacitors. I placed that board on top of a second board of the same size and screwed the boards together. Then I inserted each capacitor into one of the holes. I used a small spot of adhesive to hold each capacitor tightly in place. All the capacitors were matched for capacitance and leakage current at full rated voltage before installation and burned in for 48 hours at full voltage, and then retested. Only those capacitors that matched within 2% were used in the power supply.
This is the modified microwave oven power transformer. The high voltage bridge rectifier is between the front panel and the transformer and is not visible in this photo. The screen dropping resistors are seen above the transformer. They are arranged in a zigzag vertical arrangement and supported by two plastic bars, one above and one below the resistors. Behind the resistors and partially out of the photo is the cooling fan with the blue plastic blades that provides air to the power supply.
The 4 VR-150 Screen Voltage Regulator tubes are shown in operation. The metal cover plate that is normally in place over the top of the power supply was removed for this photo. It is left in place during operation both for safety for the operator and visitors to the shack and to ensure that the cooling air from the fan circulates through the power supply in the proper manner.
Here the 4-400A is seen running PSK-31 at 400 watts output. It is quite happy at this power level.
The anode of the 4-400A lights up the Hamshack with a pleasant glow.
[ Home ]
This page Copyright © 2008 by Ralph M. Hartwell - All Rights Reserved.