Technical Topics



QST Logo


GS35B Russian Triode Amp

GS35B Photo

This amplifier is designed around the low cost Russian GS35B triode, and features:

  • 1.5KW output using grounded grid circuitry with about 100 Watts drive

  • A commercial (expensive) tube socket is not required

  • Easy to fabricate low cost air system chimney

  • Fast vacuum relay T/R switching

  • Variable cathode bias to accommodate various operating conditions

  • Tube protection circuitry helps prevent damage from mistakes or circuit failures

  • Tube filament "step start" circuit limits cold inrush current; extends tube life

  • Cooling air blower remotely located for noise reduction

  • Modest filament voltage requirement of 12.6 volts at 3.2 amps

  • Modular construction using die cast enclosures

Click on each picture for larger image

GS35B Front View
Front View

Front Panel Removed
Front Removed

Top Panel Removed
Top Panel Removed

Chassis Top
Chassis Top

Chassis Rear
Chassis Rear

QSK Relay Box
QSK Relay Box

Chassis Bottom
Chassis Bottom

Grid Compartment
Grid Compartment

Click download link to see complete five page schematic diagram in Adobe format.

Download complete 5 page schematic PDF format

Build a high power Low Pass Filter for 50 MHz Amplifier.

Go to low pass filter plans page

Go to GS35B liquid cooling page

GS35B Cooling requirements

Some Russian data sheets show the GS35B anode cooling requirement to be 2500 liters/minute. Other GS35B specification sources include information on the cooling requirements of the GS35B cathode and grid. By definition, one cubic foot equals 28.327 liters. Converting to cubic feet per minute (CFM), the anode cooling requirements are about 88.3 CFM.
GS35B Cooling Graphic

Refer to the airflow diagram to the right. In this amplifier, the under chassis area is pressurized. A number of screened airflow holes are placed in the grid compartment and main chassis mounting area. An inspection of the photograph for the grid compartment will show some of the eight one inch diameter screened inlet air paths. Two 2.5 inch diameter holes in the top chassis plate provides an air path towards the bottom of the tube radiator. Providing a large airflow surface as possible will reduce the back pressure requirements of the external air blower fan. Passing all of the airflow directly through the under-chassis shielded grid compartment will easily satisfy the cathode and grid cooling needs. It is very important to not ignore the smaller cathode and grid airflow cooling requirements. 

The table below shows GS35B element cooling requirements in various units of measure:

GS35B Cooling requirements
Volumetric Units Anode Cathode Grid
Cubic feet
per min (CFM)
88.3 1.8 10.6
Cubic meters
per hour
150 3 18
Liters per minute 2500 50 300

Click to hear  this amp from 3C5I in Africa.

Alan, 3C5I kindly sent me an audio file of this amplifier as he copied it on 4 December, 1999 at 1649Z from his African QTH in Equatorial Guinea. Click the Six Meter QSL cards to to hear this unmodified RealAudio recording made by Alan.

Narrative Description of Amplifier Circuit Function

This amplifier provides 1500 watts power output for the Six-meter band. It uses the low cost GS35B Russian power triode in a grounded grid or cathode driven configuration and requires about 100 Watts of drive power. The Russian tube is currently available from USA and international sources1 at attractive prices, especially when compared to the cost of traditional American tubes of this power rating.


The classic cathode driven or "grounded grid" circuit is used. The secondary of the filament transformer is isolated from the vacuum tube with a homemade bifilar choke wound on a 1/2" diameter six-inch long ferrite rod. The center tap of the filament secondary provides the connection to the tubes bias circuitry. RF driving power from the transceiver is applied to the tube's cathode using an input matching circuit.  

Partial RF DECK Schematic

Partial RF Deck Schematic

The vacuum tube output circuitry is the familiar PI network design. Home made high voltage RF chokes wound on 3/4" ceramic forms are used to keep RF out of the high voltage supply. The output Pi-network coil is wound from 1/4" diameter copper tubing. A 5 to 30-pF vacuum tuning capacitor is shown on the schematic. At plate voltages approaching 4kV, this capacitor may be eliminated. The tube circuit can be resonated using the vacuum tube output capacitance in conjunction with the output Pi-network coil. A safe, easy way to adjust the output coil is described in the section "Adjusting the Pi-Network". This is also a "set and forget" adjustment. Considerable expense is saved by elimination of the vacuum variable.  

The plate blocking capacitor value needs to exhibit a reactance of less than about 5% of the plate load impedance. For this amplifier with a plate voltage of 4 kV, this works out to a minimum capacitance value of 200 pF. The current rating of the plate blocking capacitor when used in this amplifier is 10 Amps at 50.1 MHz. If you find a capacitor that has no RF current rating marked on it or has no current rating certified by the manufacturer, don't use it. It is probably intended for power supply and not RF service. The basic specification for the blocking cap required by this amplifier is 10 kV minimum voltage breakdown, 200-pF minimum capacitance, and 10 Amps RF current at 50 MHz. A suitable commercial component would be one Centralab HTY57Y102 capacitor 5.  

The only remaining output circuit components are an inexpensive 200 pF 1kV air variable used as the output loading control, and a tuned, 1/2 wave length shorted coaxial cable "stub". This stub is located on the output of the RF deck enclosure and serves two purposes. The first and most important is to provide a dc path to ground should the plate blocking capacitor fail. If you choose to omit this stub from your amplifier, you MUST add an RF choke from the amplifier antenna output to ground. The size is not important; it must present a high impedance at 50 MHz. A one mH choke is adequate. The dc path this choke provides will blow the fuse in the high voltage power supply if the plate blocking capacitor were to fail. Leaving the shorted stub or RF choke out is asking for BIG trouble. Essentially, the plate voltage will appear on your antenna line when the capacitor shorts, a potentially lethal situation. This voltage on the output coax has caused welded vacuum relays, tube destruction, and high VSWR antenna/feed line problems. 

The second purpose of the shorted stub is to help suppress second harmonic radiation from the amplifier. This stub is made from RG/8 type coax and is tuned for 100.2 Mhz with a SWR analyzer. Physically mine is made from 52 inches of International # 9086 coax, with one end soldered to a standard PL259 connector, and the far end has the center lead and shield braid shorted together. (Hence, shorted stub). When hooked up to a SWR analyzer (like the MFJ-259), and the frequency set to 100.2 MHz (the second harmonic of six meters) the instrument panel resistance meter shows zero Ohms. This means the shorted stub acts like a short at this frequency, and helps stop the second harmonic from radiating. The desired 50.1 MHz energy is not affected.

If you use a different type of coax, just trim the coax length until you get a zero Ohm reading on your SWR analyzer. This type of filter is somewhat broadband, easy to make, and does help in reducing the second harmonic. Again, the primary function of this stub is to blow the fuse should the plate blocking cap fail short, and not for harmonic reduction. This tuned stub does not eliminate the need for an external low pass filter. A suitable design is available on this web site

A small transistor will turn on when supplied with plate high voltage reduced by a large resistance divider. This signal is used by the tube protection circuitry.


The transmit relay is built into a separate die cast aluminum enclosure and is also mounted on the RF deck rear panel. This vacuum relay is a Jennings RB3. The 15 msec operate and release times for the RB3 relay are not particularly fast, but the driving digital circuitry is designed to compensate for this. Of course, a faster relay may be used. This relay has been flawless in this application. The CW keying or PTT line first goes directly to the amplifier. This activates the vacuum relay, and a digitally delayed output is then sent to the transceiver. This insures that no "hot switching" of the vacuum relay occurs. When this amplifier is turned off, the CW key signal simply passes straight through to the transceiver as usual. A small 5-volt fast acting reed relay in the T/R control circuit is activated to again regenerate the PTT or CW keying signal. This approach will key any transceiver you wish to use, regardless of the type of keying circuit it employs. The digital delay circuit is crystal controlled, and uses two series connected shift registers to generate a 15 millisecond delayed keying wave form These commonly available digital circuits are low priced. The schematic page labeled "T/R Control Timing" shows detailed digital timing diagrams. No trouble has been experienced with this circuit in over three years of operation.  

Two front panel mounted panel meters measure plate current (0 to 1 Amp) and grid current (0 to 500 mA). The basic meter movement in both meters is 50 micro Amps. The four 1% resistors are meter shunt resistors. Other meter movements may be used, just scale the shunt resistors accordingly. If you don't have 1% values in stock, simply measure with your ohmmeter and make up the necessary resistor series/parallel string to get within 1% of the target values. This will insure your panel meter readings are correct. 

The 6 amp 1 kV diodes protect the mechanical meter movements in the event of circuit failure.  

The cathode driven circuit of this amplifier requires that a positive voltage be applied to the cathode of the vacuum tube with respect to the grounded grid. This establishes the operating point of the tube, and determines the class of operation. Older circuit designs have used high power discrete Zener diodes. Some designers have used low wattage Zener diodes driving a high power external series pass discrete transistor. Both approaches were tried in earlier versions of this amplifier. A third circuit using a variable voltage source was suggested by G3SEK 2. This circuit uses a precision programmable current source driving an external PNP transistor. This emulates a variable high power Zener diode. The external transistor is mounted on a heat sink. This circuit is superior and provides smooth variable tube bias that is stable over the full power output range of this amplifier. Ian White, G3SEK mentioned in an email message that when plate voltages of 4 kV++ are used, the 36 Volt maximum rating of the TL431 is exceeded. He has an application note available on his website that describes this. The circuit here was suggested by Ian and provides variable bias in the range of 27 to 45 Volts, without exceeding the regulator voltage rating. The TL431 is available in two package styles, costs less than one dollar, and is widely available. The schematic diagram shows the pin connections for the eight-pin mini-dip package. The variable resistor voltage adjust control needs to be available from outside the amplifier, so mount it accordingly. The one amp fast blow fuse in series with the filament transformer center tap protects the tube from excessive current in the event of circuit failure. Likewise, the fuse needs to be replaced without disassembly of the amplifier, so mount it in a good spot on the chassis wall.  

The low voltage regulated power supplies provide regulated +26 volts and + 5 volts for the amplifier control circuitry.  


This lamp driver and circuit protection logic is assembled on a small circuit card and resides under the main chassis. Feed through capacitors of 1000 pF are mounted in the under chassis walls, and decouple the circuitry signals. Signals from the high voltage power supply (warm-up and high voltage OK), RF deck (grid fault), and front panel (operate/standby) are combined to provide a signal to allow application of RF drive to the amplifier. RF drive is prohibited if high voltage is not present, if a grid fault has been detected, or if the front panel function switch switches the amp to standby. Four colored status lamps are used. This amplifier uses 28-volt incandescent lamps and jeweled glass lens covers. These look good, are bright, have a very wide viewing angle, and are highly visible in a busy ham shack. If you prefer to use LED indicators, simply substitute the LED of your choice for the lamp, connect the cathode of the LED to the integrated circuit driver, and connect a (typical) 1.6 kOhm resistor in series with each LED.  

The amplifier front panel has a three-position rotary mode switch. The first position is OFF. The second is STANDBY, while the third position is OPERATE. When the amplifier is first turned on, both the WARM-UP and STANDBY lamps are illuminated. After a two-minute filament warm-up period, high voltage is applied to the amplifier RF deck. If the mode selector is on STANDBY and the high voltage detector logic senses proper high voltage, the WARM-UP lamp goes out, but the STANDBY lamp remains illuminated. If the WARM-UP lamp is off, the mode selector is on OPERATE, but the STANDBY lamp stays on, suspect that the high voltage is not connected to the RF deck. Turn everything off and then check the high voltage connections. This protection circuit will not allow RF drive if proper high voltage is not present on the RF deck. Normally, turning the mode selector to OPERATE removes the STANDBY lamp and illuminates the OPERATE lamp. Should a grid over-current condition be detected, the FAULT and STANDBY lamps only will illuminate, and the protection circuitry will automatically bypass the amplifier preventing damage. Momentarily depressing the front panel RESET pushbutton will restore the amplifier to the normal OPERATE condition.  


The two highest priced single components in RF amplifier construction have usually been the vacuum tube and the high voltage power transformer. The availability of Russian tubes has brought the tube part of this cost down to manageable levels, but the power transformer price can remain a stumbling block 3. I had an existing 2 KV transformer from an earlier project. Rather than use 2 kV in this new GS35B amplifier, a voltage doubler circuit was employed so that the existing transformer could be used. If you are starting from scratch, consider a design voltage of about 3 kV This will keep the plate voltage within published tube specifications. 4KV has presented no problems with this amplifier; especially since the tube plate dissipation rating is not being exceeded. Some amateurs in Europe (PA3CSG, 9H1PA, DL4MEA, G0RUZ) and the USA (K0PW, K7CW) are now using this approximate voltage level on the GS35B with good success 4.

A power on surge delay relay is used to reduce the initial power on current surge required to charge the filter capacitors. Special high voltage connectors are used to connect the high voltage to the RF deck. This amplifier has the power supply remotely located, so high voltage wire is used to make this connection 5. Exercise extreme caution when working with the voltages present in this power supply.  


The tubes tested in this amplifier are surplus items from the Russian military. The tubes may be unused, but have probably been subjected to long-term storage. Filament conditioning is recommended 4, and a high voltage breakdown tester described on this web site will be valuable in identifying tubes with problems before they are placed into the amplifier. 

Some circuits in this amplifier are built as small sub-systems. They can be tested before placing them into the amplifier. Pre-testing your circuitry will greatly help when the time comes to actually insert your vacuum tube for final checkout. In particular, the RF deck plate tank circuit, grid trip circuit, and QSK T/R relay circuitry are easily adjusted and tested ahead of time. The front panel meter Grid Current and Plate Current readings can be verified easily as well. Having a tuned plate tank pi-network ahead of time prevents subjecting your tube to grossly mistuned conditions. The same is true of the grid trip circuit. Adjusting the plate pi-network ahead of time is a practical necessity if you choose to eliminate the vacuum variable tuning capacitor.  

Adjusting the pi-network: 

ABSOLUTELY REMOVE ALL TUBE VOLTAGES. Leave the tube in circuit. The plate load impedance of your amplifier is expressed approximately as: [(plate voltage in volts) divided by (1.8 times the plate current in amps)]. Assuming 4000 plate volts and 750 ma of plate current, this works out to about 2963 ohms. Make up a resistance value close to this number with low inductance resistors, and temporarily place this resistor string from the tube anode connection to ground. This simulates the plate load impedance of the amplifier. The purpose of the pi-network in the amplifier is to change this relatively high plate load impedance value to 50 ohms for your transmission line. Now, hook up a SWR analyzer to the amplifier's RF OUTPUT connector. Adjust the SWR analyzer for a frequency of 50.1 MHz Adjust the copper coil windings (slightly expand or squeeze together) in conjunction with adjusting the output loading capacitor. Adjust for a 1 to 1 SWR reading on the SWR analyzer. When the SWR reading is flat, your adjustment is finished. The top RF deck shielding cover on this amplifier had a small effect on this setting, so a tiny adjustment was necessary when the amplifier was running at full output. It's amazing how close this procedure gets your amp to the final settings. If you don't have a SWR analyzer, consider getting one. Remember to remove the temporary resistor you installed during this procedure.  

Adjusting the grid trip circuit:

ABSOLUTELY REMOVE ALL TUBE VOLTAGES. The GS35B and GS31B tubes are designed to run much higher grid currents than American tubes of the 8877 variety. A typical Russian GS35B will run about 25 to 30 percent of the plate current value for a grid current. So, if the plate current is 800 ma, a grid current of about 240 ma is common. A grid trip setting of about 300 ma or so is an approximate number for this amplifier. To set up the grid trip variable resistor, you will need the low voltage plus five VDC supply activated. You will also need a current limited low voltage external power supply, and a multimeter. Connect the minus lead of the external power supply to the amplifier chassis ground. Set the voltage setting of the external power supply to zero volts. Connect the plus lead of the external power supply to the "B minus" connection of the amplifier. Using the multimeter, adjust the external supply for 300 milliamps. The front panel Grid Current meter should also now read 300 ma Adjust the grid trip variable resistor until the small LED indicator illuminates, indicating the grid trip relay is now latched. Remove the external supply. Pressing the front panel RESET button should extinguish the indicator LED and the relay should now be deactivated. This completes the grid trip adjust. This is a "set and forget" adjustment. The variable resistor and LED indicator can be buried inside the amplifier since no external adjustment is required.  

The Plate Current meter operation can be verified in a similar manner. Using the minus lead of the variable external supply connected to the amplifier "B Minus" connection, and the plus lead connected to the plus terminal of the Plate Current meter, your multimeter current reading should be the same as indicated by the front panel meter.

 The T/R relay circuit can be tested ahead of time. ABSOLUTELY REMOVE ALL TUBE VOLTAGES. The low voltage +5 and +26 volt regulated supplies need to be active for this test. Temporarily remove the "Standby Signal" input connection. Apply a hand key or CW keyer to the "Key In" connection. Hook up a code practice oscillator to the "Key Output" line (or otherwise verify activity at the output line). Keying the input causes the output to key, and the keying relay should activate. If you have an oscilloscope, you can verify that the delayed keying activity is 15 milliseconds delayed. The logic timing diagram shown on the schematic page 4 shows the circuit activity for typical CW keying speeds. The delayed keying times are fixed, and no adjustments are necessary for proper relay activity. Momentary grounding of the "Standby Signal" input should disable the relay activity, but normal output at the "Key Output" will still occur.

The variable bias circuit is adjusted with all normal tube voltages applied. With ZERO power input (turn your transceiver off), ground the "Key input" signal line, and adjust the variable bias resistor for a resting plate current reading of 100 mA. 

The input SWR adjust is best set with the amplifier running normally, and about ten watts or so of RF power applied to the amplifier. A suitable dummy load or antenna must be connected to the amplifier output. Apply RF drive, and adjust the small variable capacitor for a 1 to 1 SWR between the transceiver output and the amplifier input. You should be able to obtain a very low SWR reading. Six different tubes were tested in this amplifier, and each was easily matched with a slight adjustment of the capacitor. This is a "set and forget" alignment, but make sure that you place the capacitor so you can adjust it while the amplifier is running. This amplifier has the adjustment shaft available on the rear RF deck chassis.  

At this point, more input power can be applied to the amplifier and the output loading capacitor adjusted slightly for best power output.  

Typical operating conditions:

  • 4100 volts plate voltage

  • 700 ma plate current

  • 100 ma idle current

  • 230 ma grid current

  • 85 watts drive

  • 1500 watts output

These numbers will change with different tubes. Some tubes tested in this amplifier show outputs greater or less by a couple hundred watts or so. 


Special thanks for help and support go to B-N "Bob" Alper of Svetlana, Jim Tonne W4ENE, Paul Kiesel K7CW, and Paul Goble ND2X/5 (SK).