A 400 Watt Low Frequency SSB Linear Amplifier
Boost the output of your Motorola Starpoint Modem from QRP to QRO!
by Ralph Hartwell W5JGV
This page was first posted on April 13, 2005
After obtaining a pair of surplus Motorola Starpoint channel modems to use as SSB exciters at WC2XSR/13 on 168 KC, I wanted to build an amplifier to boost the power to about 400 watts, which is the maximum licensed transmitter output power for WC2XSR/13.
The Starpoint generates a LSB signal at a power level of about -20 dBm across a 75 Ohm load. I wanted to raise that to about +55 dBm or 400 watts output. To do this, I planned to build an amplifier using a pair of MOSFET's as the final amplifier stage. I had already used MOSFET's in my 400 watt output QRSS30 / CW transmitter with good results.
I knew the Starpoint would not directly drive the MOSFET's, because it just didn't have the RF power output required to handle the gate charge required by the MOSFETS. I knew that I would have to build driver amplifier to put ahead of the PA stage.
The final design of the IPA driver (intermediate Power Amplifier) stage uses a pair of International Rectifier IRF510 MOSFET's. These have the necessary gate sensitivity so that they can be driven directly from the Starpoint without further amplification. Note that the IR version of the IRF510 has characteristics better suited to this application than do the same type number available from other manufacturers.
The input from the Starpoint exciter is transformer coupled to the gates of the IRF510's. The output of the IPA is transformer coupled by input transformer T1 to the gates of the Power Amplifier MOSFET's.
The PA stage uses a pair of IRFP260N MOSFET's. Other suitable transistors are types IRFP3415 and HUF75645P3.
The IRFP260's are recommended for power levels above 250 watts, the other transistors will work well at power levels below 250 watts. The IRFP3415 has a lower voltage rating than the IRFP260, and the HUF75645P3, although having a higher power rating than the IRFP260, is harder to handle thermally because of the smaller TO-220 case size. During developmental testing, it proved almost impossible to remove the heat from the smaller case at full power. Several of the HUF75645P3 transistors sacrificed their all during the design of this amplifier.
The IPA output transformer T2 is resonated close to the desired carrier frequency of 168 KC to improve the gate drive waveform at the PA transistors.
The PA output is coupled by transformer T3 to the antenna load. The transformer T3 is resonated close to the desired carrier frequency of 168 KC. This greatly improves the output carrier waveform as compared to the conventional broad-banded amplifier design. Nevertheless, the use of a low-pass filter should be considered mandatory to comply with FCC regulations.
An analog meter is provided for transmitter adjustment, but an oscilloscope is required to make sure everything is working correctly, at least for the first power-up and on-the-air-testing.
WARNING ! ! Do not apply DC power to the amplifier without a 50 Ohm load connected to the output connector. If DC power is applied to the PA stage with no load connected to the input of the amplifier, should the amplifier go into oscillation, excessive voltage will build up across the output transformer. This may cause the PA stage MOSFET's to fail shorted. If the power supply can supply enough current, the transistors may explode violently. It is OK to turn on the power to the IPA stage without the 30 Volt DC power applied to the output PA stage.. This will not harm the output PA transistors.
DO NOT OPERATE THE AMPLIFIER WITHOUT A 50 OHM LOAD CONNECTED TO THE OUTPUT CONNECTOR.
You can download these data files in PDF format by clicking on the following links:
Schematic Diagram of Amplifier with component values
Making the Non-Inductive Current Shunts
Every project has to start somewhere.
This is the prototype version of the linear amplifier during initial testing.
At the right of this picture is my "Tower of Power", a stack of 6 series connected computer power supplies which produces 30 volts at 35 amperes.
As you read this article, you will see that I have taken many of the parts for this project from old computer power supplies. They are today's equivalent of the old vacuum tube TV sets of days past. In those days, Amateur operators would gleefully strip old TV sets for the valuable components they contained. Many an Amateur radio operator went on the air with a transmitter built almost entirely from old TV parts.
The large black unit in the right of the picture is my home made KW dummy load, consisting of 6 large 300 ohm non-inductive resistors inside a pair of one gallon cans carefully soldered together and filled with transformer oil for cooling.
The Motorola Starpoint modem is on the bench in the center of the picture. The modem is powered by a small computer-style switching power supply taken from some old networking equipment. The RF output from the modem goes to the prototype amplifier, shown here at the left side of the picture.
At this point, the IPA was still a single ended design with just one IRF510. It was unable to provide a symmetrical drive waveform to the PA stage, so that design was abandoned. This version of the amplifier was eventually greatly modified, and the second revision became the finished unit you will see in the following pictures.
The finished amplifier.
It was built into a discarded commercial battery charger case. This turned out to be a very good choice, because the entire rear of the case is just a large heat sink which turned out to be able to handle the heat load from the final PA transistors.
The heat sink on the front panel is for the two IPA (Intermediate Power Amplifier) transistors. It was placed here because it covered up a large hole in the chassis. In the same fashion, the meter was placed where it is because it covered a hole in the front panel where a large connector used to be installed.
The two LED's are used to indicate the presence of the +15 volt IPA supply voltage and the +30 volt PA supply voltage. These LED's use a pair of existing holes in the panel. There is a small rectangular hole to the left of the heat sink; a switch was there in the original battery charger, now the hole is covered by a filler panel. The rest of the items seen on the front panel were placed where they were most functional.
The heavy red and black wires carry the 30 volts for the PA stage. The black wire is the 120 volt AC line cord for the IPA and bias power supply.
There is an RF sample connector just to the left of the small silver AC power switch. The sample connector is fed by a two turn winding on the final amplifier transformer T3. This sample connector will provide an RF voltage of 28.6 volts peak-to-peak when the amplifier is producing 400 watts output.
The black knob selects the various metering positions. The knobless potentiometer shaft to the right of the meter switch is the adjustment for the IPA bias. By not placing a knob on the shaft, there was less chance of accidental misadjustment of the bias setting.
The meter is a Radio Shack 15 volt DC meter. This meter is actually a 1 milliampere meter with the appropriate scale calibration and meter resistors. The original 0-15 scale was left in place, because the full scale readings are 1.5, 15, or 30, so the 0-15 scale is correct for all metering positions except for the final PA drain voltage reading. On that position, it is necessary to multiply the meter reading X2. Any suitable meter may be used by changing the series and shunt resistors as required to obtain the proper scale readings.
The heat sink extends beyond the rear of the cabinet to allow effective cooling. I added a pair of old computer power supply fans to the heat sink. If the PA transistors get too hot, their idle current rises and they waste a lot of power, so adequate cooling is essential. With heat sinks, bigger IS better!
The battery charger had a handle on the top of the case for ease in transporting it. I decided to leave the handle on the case, since the finished weight is about 12 pounds. Notice that there are a lot of vent holes already in the case. These allow lots of cooling air to flow through the amplifier and keep the interior components nice and cool.
These fans came from a couple of old Dell computer cases. These are the same size as the fans commonly used in computer power supplies. They are connected in series and placed across the main filter capacitor of the IPA power supply. The fans run whenever the IPA power is turned on.
After sticking my finger into the cooling fans a couple of times, I gave up and installed the original fan guards on the fans. Although the guards do reduce the airflow slightly, the airflow is still more than enough to keep the heat sink cool and my fingers stay unbruised.
The two silver-capped knobs between the fans are the PA stage bias adjustment potentiometers. Since the PA bias adjustments are essentially "set and forget" so this was a good place to put them. Anyway, there was a pair of holes just the right size in that location, so I used them.
Behind the front panel of the amplifier.
The IPA stage is to the left in the picture, and the meter is directly below the IPA stage. The IPA input transformer (with the orange and white windings) is seen to the upper left of the meter. The IPA bias potentiometer is in the center, and the metering selector switch is just to the right of the bias pot and right above the RF output connector. The RF input connector is at the far right of the picture. The IPA power and bias supply may be seen in the center and left of the picture. The rectangular cased capacitors in the lower center of the picture are for tuning the PA tank circuit. The large metal cased capacitor (partially seen at the lower right) is the oil filled motor run capacitor used as the RF bypass capacitor for the PA stage.
As you can see, this amplifier was built using the "haywire" method of construction. I started breadboarding the amplifier in this manner, but it worked so well, I did not want to tear it apart and try rewiring it, so In left it the way you see it. Luckily, it looks quite a bit better with the cover on!
The IPA stage is seen in this picture.
Transistors Q1 and Q2 are a pair of IRF510 MOSFET's which are mounted on the back of the heat sink which was salvaged from a Pentium Pro microprocessor taken from an old computer. The heat sink was carefully drilled and threaded with a pair of 4-40 holes for mounting the transistors. A pair of 1/4-20 threaded holes were drilled and tapped into the heat sink so the heat sink could be attached to the front panel. No cooling fan is required; convection cooling is sufficient.
Select a matched pair of transistors for Q1 and Q2 by testing for an identical drain current of about ½ ampere while under test to within 5%. Use the same bias voltage on each transistor during the matching test. Note that the metering position for the IPA stage reads the combined drain current for both of the IRF510 transistors.
The plastic insulators between the transistors and the heat sink came from a computer power supply. The transistors in the IPA do not get very hot, and the heat sink runs at only about 118 degrees F. The heat load from the transistors is about 7.5 watts. In this circuit, the transistors are running in Class A mode therefore they will dump most of the DC power they draw as heat and deliver very little RF power to the PA transistor gates.
Note that the transistors are mounted so that their source leads may be soldered directly together. This helps to prevent parasitic oscillations. These transistors have lots of gain, especially when operating Class A.
The metering shunt N1 is seen to the right of the terminal strip and is covered with semi-transparent tape. It is mounted on the terminal strip seen to the right of the heat sink. The shunt is hand wound to be non-inductive. It is rated for a voltage drop of 0.085 Volts at 1.5 Amperes, which is correct for the Radio Shack meter used here. It is wound using #24 copper wire on a non-metallic form.
Current metering shunts N1, N2, and N3 are hand wound to be non-inductive. They provide the correct voltage drop for the meter and also provide some negative feedback for the amplifier stages.
To make a non-inductive shunt, take the wire you are using for the shunt and fold it double (so that it is in the shape of a "U" and then wind it up into a tight coil. I used a length of wooden pencil for the form for the IPA shunt. The shunts for the PA stage are made from heavier wire so they are self-supporting, and will not need to use a solid form.
For more information about the shunts, click HERE to download a PDF information sheet.
This view shows the two small ferrite beads FB1 and FB2 that are slipped over the gate leads on both of the MOSFET's. (Look closely; they're there!) . The inductance of each bead is 4.0 uH for a single wire pass-through. The exact value is not critical. These ferrite beads came from an old computer power supply.
With the meter selector switch in the IPA DRAIN CURRENT position, adjust the bias for the IRF510 IPA to set the idle drain current to 1/2 ampere as shown on the meter. After the heat sink warms up, the drain current will need to be readjusted slightly. There should be RF no drive signal applied to the input connector during the idle bias adjustment. Note that the metering position for the IPA stage reads the combined drain current for both of the IRF510 transistors.
The blue disc capacitor (salvaged from a computer power supply) is the tuning capacitor for the IPA to PA interstage transformer T2 primary winding.
NOTE: When selecting the tuning capacitors for transformers T2 and T3, the waveform will look better if the capacitor is chosen to tune the tank circuit to the desired carrier frequency (Fc) or a little bit lower than Fc. Tuning the tank circuit higher than Fc will result in a waveform containing "bobbles" near the peak of the waveform.
The brown and white twisted pair wires from the MOSFET's to the interstage transformer may be seen leading off to the right of the photo.
The rectangular case bypass capacitors were also salvaged from old computer power supplies.
The input transformer between the RF input and the IRF510's is seen here with orange and white windings. It is bifilar wound, and the ratio is 1:1:1. It has more turns than are actually required for the frequency used here, but this was desirable in order to ensure that it would resonate at a frequency below the resonant frequency of the PA stage output tank circuit to prevent unwanted oscillations. Note that this transformer is not tuned with a capacitor.
Transformer T1 specifications are as follows: Primary - 20 Turns #26. Inductance is 3.8 mH. Secondary - 20 bifilar turns #26 twisted. Inductance is 3.8 mHy each section. Core #43 material. OD 1", ID 1/2", HT 1/2". All Electronics part # FB-45. The wire was salvaged from Cat-5 cable.
This view of the back of the front panel shows the PA stage B+ power wires going through the panel.
The wires are protected from abrasion by the edges of the panel holes by rubber grommets. The AC line cord is clamped at the panel by a retaining clamp salvaged from a computer power supply. In fact, the power cord also came from a scrapped computer. The power line is fused at 1/2 ampere. The AC power switch is directly above the fuseholder.
Directly above the power cord clamp is the BNC connector which is the RF input to the IPA stage. Note that coax cable is not used, but instead, some of the twisted pair salvaged from CAT-5 cable is used. Because this amplifier operates at low RF frequencies, wiring techniques similar to that used in audio amplifiers may be used. Twisted wire works just as well as coax cable. No problems were encountered by using twisted wires instead of coax.
Note the larger UHF style connector with the larger twisted wires attached to it just to the left of the fuseholder. This is the RF output connector from the PA stage. When the amplifier is operating at 400 watts output, there is a peak-to-peak voltage of 400 volts on the connector. Watch your fingers!
The metering switch and the meter calibration resistors may be seen mounted on the front panel above the RF output connector. The voltage readings are calibrated by using a series resistor between the voltage to be measures and the meter.
A NOTE ABOUT THE METERING CALIBRATION:
Since the meter used here is a 1 mA movement, a resistance of 1000 Ohms will give a current of 1 mA through the meter at an applied voltage of 1 Volt. (This is what is known as a "1000 Ohms-per-Volt" meter, which says that for a full scale reading on the meter, we need to put 1000 Ohms in series with the meter.)
This simplified calculation neglects the internal resistance of the meter movement itself, which, in this case, is specified by the manufacturer as 85 Ohms. If we wish to be exact, the correct value for the external resistor required in the case just mentioned would be 1000 Ohms - 85 Ohms = 915 Ohms. If we are measuring the 15 volt IPA power buss we can use 15 Volts X 1000 Ohms = 15,000 Ohms for the series resistor. If we neglect the 85 Ohm internal meter resistance the error will cause the meter to read 14.92 volts at full scale instead of 15.0 volts. Since this is less than the specified error for the meter, unless we need extreme accuracy, we can safely neglect the meter's internal resistance when choosing the metering resistors.
For the current shunts N1, N2 and N3, the case is a little different. We have to send most of the current we are trying to measure through a very low resistance resistor and allow only a small part of the total current (0.001 Ampere in the case of this particular meter) to flow through the meter.
Commercial ammeter shunts use materials for the resistor that have a very low amount of resistance change with temperature, i.e., they have a low temperature coefficient. Copper wire, which we used for the shunts here, has a high temperature coefficient, but if we use large enough wire and keep the wire at a fairly constant temperature, the resistance change will be small enough so that the accuracy of the meter readings will be adequate for our requirements.
We are trying to read the drain current through the PA transistors. This is a pulsing DC current, but it is pulsing at 166 KC. We have to consider this pulsed DC as an RF signal. If we simply wind up the shunt wire, we will form an inductor which will have enough reactance to cause considerable RF voltage drop across the shunt when the transistor draws current. This will cause a large amount of negative feedback, and greatly reduce the gain of the amplifier at the carrier frequency. It may also cause the amplifier stage to go into high frequency parasitic oscillations. This will make the transistors overheat and possibly destroy them. We will need to use a shunt with the smallest amount of residual inductance possible.
To make a non-inductive shunt, take the wire you are using for the shunt and fold it double (so that it is in the shape of a "U" and then wind it up into a tight coil. You may also use a zigzag arrangement. Just be sure to wind or form the wire so that the inductance is minimized. Consider that you want to wind half of the coil clockwise, then wind the other half of the coil counter-clockwise. That way the magnetic field from each half of the coil will cancel the field from the other half of the coil. The shorter the coil is from end-to-end, the better the cancellation will be.
For more information about the shunts, click HERE to download a PDF information sheet.
As shown on the schematic diagram, one side of the meter is connected to ground through the metering switch, but the metering ground return must be split so that the PA current meter ground connects directly to the same point as do the meter shunt resistors. If this is not done, the meter readings will be inaccurate due to the circulating DC current in the chassis. The ground return to the metering switch for the voltage readings and the IPA Drain Current reading may be connected to any convenient chassis ground point.
The IPA and bias voltage power supply section.
The transformer seen at the top of the photo was salvaged from an old wall-wart power supply from a defunct computer modem. It may be replaced by any transformer capable of supplying 16 to 20 volts at 1 ampere. The bridge rectifier used in the power supply came from a computer power supply. The large gray electrolytic capacitor is the main filter capacitor. About 20 volts DC appears across this capacitor, so I was able to connect both of the heat sink cooling fans in series with each other and connect them across the filter capacitor.
The voltage from this filter capacitor goes to the LM317 voltage regulator which is seen here bolted to the bottom of the cabinet. This provides adequate heat sinking for the regulator. Both the regulator and the plastic insulating sheet between the regulator and the chassis were salvaged from an old computer power supply. The horizontal black electrolytic capacitor seen just to the right of the gray filter capacitor is connected across the output of the LM317 regulator. It came from an old computer power supply, too.
The dropping resistors and Zener diode for the IPA and PA stage bias voltage are seen mounted on the terminal strip which is hanging from the terminals of the transformer. The large potentiometer seen "flying" from the terminal strip is the +15 volt regulator adjustment control.
Note that it is not advisable to take the IPA power from the PA power supply since any voltage variations on the supply line of the IPA stage may be amplified and fed to the gates of the PA transistors, resulting in "motorboating" or other unwanted oscillations.
Interstage transformer T2 between the IPA and the PA stages.
Each winding on T2's core is composed of four bifilar turns of 26 gauge wire salvaged from some CAT-5 network cable. The transformer is fastened to the heat sink by a bolt which gently squeezes the transformer between a pair of Teflon washers - which I happened to have on hand. Cardboard or wooden washers would work as well. Be sure not to form a shorted turn through the transformer core with the mounting hardware.
Transformer T2 specifications are: Primary - 4 bifilar turns #26. Inductance is 162 uH each section. Secondary - 4 bifilar turns #26 twisted. Inductance is 162 uH each section. Core #43 material. OD 1", ID 1/2", HT 1/2". All Electronics part # FB-45. The wire was salvaged from Cat-5 cable.
A view of the PA section of the amplifier
Q3 and Q4 are a pair of matched IRFP260N transistors. Match them at a drain test current of 3 amperes. When under test, the transistors should be mounted on a heat sink and kept at the same temperature while testing.
Adjust the idle bias for each of the PA transistors so that the meter just begins to indicate some drain current with no drive signal applied to the input of the amplifier. This will be about 100-150 mA and will just begin to register on the meter. The heat sink should be at or below a temperature of 85 degrees F during this adjustment. When the transistors are hot, the idle bias will rise to about 1/2 ampere but will drop rapidly as the transistors cool down. Note that setting the bias too high will cause excessive power dissipation in the PA transistors and cause the transistors and the heat sink to run much hotter than necessary.
The final PA transistors MUST be very well heat sinked. Use of thermally conductive rubber, plastic or Mica insulators is not recommended. Suggested insulators are anodized aluminum, solid aluminum oxide, or none at all - the best method is to bolt each transistor directly against a separate electrically insulated heat sink. The transistors dissipate considerable power on voice peaks and may be damaged by single tone testing or by extended periods of two-tone testing unless care is given to heat sinking them.
The final PA power supply must be "stiff" or the signal linearity will suffer badly. Use heavy wire, such as #8, to connect the amplifier to the power supply.
Looking down at the PA stage in the amplifier.
The red and black power leads for the 30 Volt PA supply voltage connect directly to the RF bypass capacitor. The +30 lead connects with a short length of #12 wire to the center tap of the primary winding on T3. The black 30 volt return wire connects to the heat sink with a short length of bare #12 copper wire. The black wire is also connected to the case of the bypass capacitor which is grounded to the case of the amplifier.
You can see the T3 tuning capacitors that are connected to the junction of the MOSFET's and the ends of the primary winding of T3 by using a couple of lengths of #12 copper wire. Use heavy wire for these caps, there's lots of RF current at full power.
The green case current shunt RF bypass capacitors are connected directly across the silver colored current shunts.
Seen here in the center is the PA stage output transformer T3. It is tuned to the carrier frequency by the multi-colored stack of capacitors seen connected to the heavy copper wires going to the MOSFET's. The heavy wire on T3 is the primary winding, and the smaller wire beneath the primary is the secondary winding. The core itself is insulated by a layer of Fiberglass tape.
The specifications for transformer T3 are: Primary - 4 turns each side of center tap, #12 THHN, solid or stranded wire is OK. Inductance is 450 uH. Secondary - 28 Turns #17 solid enameled wire. Inductance is 5.7 mH. Sample Port Winding - 2 turns #26 wire salvaged from CAT-5 cable. Core is Amidon type 193A-J, OD = 1.932", ID = 1.252" HT = 0.75".
The metering shunts N2 and N3 for the PA transistors are hand wound non-inductive coils made from #18 copper wire. They are rated for a voltage drop of 0.085 Volts at a current of 15 Amperes. The shunts must be placed directly between MOSFET source connections and ground.
Note the silver colored bolt in the center of the assembly - the one with the stack of nuts on it. That's the common ground point for the amplifier stage. The shunts, the shunt bypass capacitors and the negative side of the 30 Volt supply all connect to that point. The heat sink serves as the common ground plane for the PA stage.
The negative power lead for the PA stage must be connected as closely as possible to the common ground connection of metering shunts N2 and N3. In this picture you can see the bare copper wire leading from the black negative power wire and going to the heat sink. The common ground connection of metering shunts is the bolt attached to the heat sink. The bolt is also the common connection for the source leads of the transistors.
The 3900-uF 6.3 volt capacitors must be placed directly across shunts N1 and N2 and the common ground connection. The capacitors may need to have their leads extended. If so, the extensions should be made from heavy wire to minimize the lead resistance. These capacitors were salvaged from old computer power supplies. The capacitors are used to bypass the RF voltage drop across the shunts during amplifier operation. Since they must handle close to 15 Amperes of RF, they should be selected for the lowest AC resistance.
Most of the capacitors found in computer power supplies have very low AC and RF resistance, and are eminently suitable for bypass service. The units used here have a 105 degree C temperature rating. Why the 1900 uF value? Because that was the value I had in the Ham Radio Junque Box!
I selected the capacitors from the ones I had on hand by connecting the capacitor and a 50-Ohm resistor in series across the output of my RF signal generator. I then measured the voltage that appeared across the capacitor when the signal generator was turned on. The signal generator was adjusted to provide a 10-volt sine wave signal at 166 KC. I chose the capacitors that showed the lowest RF voltage across the capacitor. A reading of less than 10 mV is suitable. If the capacitor has an excessive AC resistance as indicated by a higher than normal voltage reading during the test, it may get too hot during use and short circuit or even explode.
The back-to back Zener diodes are used as protection to prevent excessive voltage from being applied to the gates of the PA MOSFETS should the IPA stage be seriously overdriven or break into parasitic oscillation. In normal operation, the Zener diodes will not conduct. Any 1 watt Zener diode may be used as long as the total breakdown voltage of the two diodes in series is less than +/- 20 volts. This is sufficient to protect the gates of the MOSFET's.
Note that there are two types of zener diodes. One type has a voltage drop in the reverse (non-Zener) direction of approximately 0.7 Volt, just as an ordinary rectifier diode has when conducting in the forward direction. That type of diode was used in this amplifier. This type of Zener diode allows you to connect two diodes in series, but with the same polarity leads connected together like this: +(diode)- -(diode)+ That way, the diode that is being biased in the reverse direction will only add 0.7 Volts to the Zener voltage of the opposite diode. You will have made an AC Zener diode which will clip the applied voltage to the Zener value, whether the voltage is positive or negative.
The other type of Zener diode, which is not very common any more, has a normal Zener voltage in one direction, but the reverse voltage is very high - sometimes as high as 100 volts. Obviously you cannot use the back-to-back series connection with a pair of these diodes, because the diode which is not working in Zener mode will be non-conducting and will effectively disconnect the other Zener diode. If you use this type of Zener diode, you will have to connect the two diodes in anti-parallel instead of in series.
The PA tank circuit tuning capacitors.
Transformer T3 is tuned to Fc by the use of three parallel pairs series connected capacitors. Each of the capacitors in a pair have the same value, but the pairs themselves have different values as needed to tune T3 correctly. See the schematic diagram for more information. There's nothing critical about this capacitor arrangement, it was just the best way to make use of the parts I had on hand.
The capacitors were salvaged from some old computer power supplies where they are commonly used as RF and noise bypass caps on the AC power line. These are "X" rated caps, so they can handle 250 VAC across each capacitor. ("Y" rated caps can handle 125 VAC.)
Voltage rating is not a problem here - the tank circuit voltages are fairly low - but the RF current can easily approach 35 amperes or more, so the capacitors must have low RF losses. Mica capacitors would be excellent here, but the cost would be high. These free capacitors have low enough losses to work well in this circuit.
The PA tank circuit B+ bypass capacitor must handle RF at 2Fc, or 2 X 168 KC = 336 KC. The RF current is also high, and can approach 25 Amperes at full power. The best capacitor I found for this application was this oil filled motor run capacitor. It exhibits very low loss and provides very good RF bypassing. Not having a factory made clamp available with which to mount the capacitor, I made one from some copper roof flashing. Note that the PA stage B+ power wires go directly to the capacitor.
The blue 6800 pF capacitor seen attached to one of the PA transistors is not used in the finished amplifier.
A PA transistor mounted on the heat sink with a home made anodized aluminum insulator.
Note that the transistor is mounted against the heat sink without the use of a Mica insulator. This was done by using an anodized aluminum plate which is placed between the transistor and the heat sink. This "metal insulator" provides excellent heat transfer from the transistor to the heat sink.
I made the insulator by carefully cutting some fins off of an old Pentium Pro heat sink (the same as the one used in this amplifier for the IPA stage.) I smoothed the cut edges to prevent shorting the bare aluminum of the insulator to the heat sink. Then I drilled a hole in the insulator for the transistor mounting screw. The edges of the hole were then carefully smoothed to prevent shorting the transistor to the heat sink.
Note that during the machine work on the insulator, extreme care must be used to keep the workpiece and work area clean and free of any debris which could puncture the anodized surface of the insulator. Any punctures in the anodizing can cause the transistor drain to short to the heat sink.
After finishing the insulator, the transistor, heat sink, and insulator were carefully cleaned with alcohol and a soft cloth. A small amount of heat sink grease was then applied to both sides of the insulator. The insulator was then carefully placed against the heat sink, and the transistor was attached to the heat sink and the mounting bolt carefully tightened.
Notice the large washer between the bolt head and the transistor. The washer is used in order to spread the mechanical compression load over a larger area on the body of the transistor so as not to crush the transistor case.
This is what the RF carrier wave looks like at 200 watts. Not bad!
It's not quite a perfect sine wave, but the signal generator I was using for these tests does not produce a perfect sine wave. It generates a slightly triangular waveform.
The RF carrier wave at 400 watts.
Still very good.
The RF envelope with a two-tone test (actually DSB) at the 200 watt level.
This is the RF envelope with a two-tone test (actually DSB) at the 400 watt level.
There's some compression on the signal peaks, but after passing through a low pass filter, it's good to go!
This is a stepped linearity test signal at the 200 watt level.
The finished amplifier installed and on the air at WC2XSR / 13
The Motorola modem and its power supply are fastened together with some tape and placed next to the amplifier.
There's a portable CD player laying on top of the amplifier. The CD player plays the audio ID loop that feeds the Motorola modem audio input. I crank up the audio from the CD player until the AGC amplifier in the modem starts to compress the audio.
I then adjust the RF carrier output level control in the Modem to set the amplifier power output to 400 watts. An oscilloscope is used to monitor the RF output of the amplifier. The same low pass filter used for the 400 watt QRSS30 / CW transmitter is used with this amplifier.
The amplifier B+ power leads go to the same power supply that is used to run the QRSS30 transmitter which is seen directly below the power supply. I did have to add a 65000 uF capacitor (seen here) across the power supply to keep the supply voltage from following the audio modulation.
You can download these data files in PDF format by clicking on the following links:
Schematic Diagram of Amplifier with component values
Making the Non-Inductive Current Shunts
[Home]