144MHz 1.5KW GU-84b Linear Amplifier
Matti Vilppula, OH3AWW, 2003

During the summer 1997 I completed my new GU-84b tetrode power amplifier. I used my previous 4CX250R PA power supply for this (Picture 1). Because there have been a few requests for construction information, I decided to write this article. Please do note that this is not made for first PA project builder and I will not take any responsibility about errors of this document.

Picture 1 (at right): GU-84b amplifier and power supply

Click on bordered pictures for zoom

 

GU-84b Tetrode

GU-84b is the military version of the Svetlana 4CX1600b. GU-84b is made for bigger anode dissipation and current. It also needs more powerful blower than 4CX1600b (picture 2, below).


Picture 2: GU74b(x2), GS23b, GU84b and 4CX250R
 

GU-84b socket has air chimney and cooling air is then feed through grid cabinet to anode. Svetlana socket SK3A can also be used but in that case cooling air is feed first to anode cabinet. Socket cooling is important and should not be forgotten. The both sockets have a built-in Screen capacitor and therefore it is possible to connect the cathode to the ground. GU-84b socket has also 0.5W resistor for anode current measurement purpose.

NOTE:
NOS, Old Stock or long time unused tube shouldnt be connected to plate and screen supply without first carefully wakening it up. I would recommend at least 24 hours of heater time before applying other voltages.
 

Picture 3: 4CX1600b & SK3A

PA Working Conditions & Cooling

The power output with my power supply is about 1500 W in linear service. The efficiency is about 65 %. Plate voltage is 1900 V @ 1.2 A. The tube needs also 350 V on screen and 84 V (-120 V or more on rx) on grid (G1). Filament uses 27 V @ 4A. GU-84b maximum anode dissipation power is 2.5 kW. With this power the tube needs really big blower that can give 75 mm H2O pressure and 2.5 m/min airflow. With 1.6 kW dissipation tube needs 30 mm pressure and 1.6 m/min airflow. With my own experience at 1 kW output level and HSCW (high speed CW) in meteor scatter mode the output air temperature (20C room temp) will change from 30C50C within one 2.5 min TX period. The blower is in separate box and cooling air is feed in the 90 mm aluminium tube to PA. On TX the pressure is 32 mm and on RX 20 mm in the input cabinet. Smaller fan could also be used as can be seen from air temperature.

Power Supply

The filament PSU is 27 V @ 4 A DC with current limit circuit. This feature is good for a soft start. I use a L200 regulator and a pair of 2N3055 transistor for this circuit. The filament resistance is 2.5W when it is cold and it takes 3 seconds until current is below 4A. The tube manufacturer recommends that cooling should remain up to 5 minutes after filament is turned off.

The idle anode current is less that 200 mA during TX when G1 is -84 and G2 350V. On RX G1 voltage should be at least -120V, otherwise noise will decrease RX sensitivity. Please be very careful that anode voltage is not disconnected before G2. More valuable information for PA and PSU is found from http://www.svetlana.com.

PA construction

The plate is a l coaxial resonator and grid an LC circuit. Before building I recommend visiting SM5BSZ homepage to see what Leif says. The input is directly from there and the Q-value is decreased with the resistor. But still the gain is about 16 dB. There is no need of neutralizing the PA with this construction.

The output circuit is based on the W9QXP (now W7QX) 2 m and 70 cm PA. Even though Jerry says that l is too big for 2m PA. I didnt agree with Jerry and decided to give a try with this mechanically larger but simplifier construction (Picture 4). Aluminium is not so expensive, I thought.

   
Picture 4: Plate Line Construction

At first I had difficulties to find the right length of the anode resonator. With the help of Jukka, OH1FF I found the correct resonator length. Because of this the cabinet is about 200 mm longer than needed and there are extra holes for tune and load capacitors. However now it is possible to change the tube to 4CX1600b. It needs 75 mm longer resonator.


Picture 5: Plate tune & load capacitors

Output load and tune are both capacitive because more simpler mechanics than inductive version (Picture 5). The efficiency is now a little bit worse with this construction.

Amplifier cabinet is 1 mm aluminium. The 100 mm diameter anode resonator tube is made from this same Al-plate and connected with Al-rivets. Of course copper or aluminium tube would have been much better choice. I just didnt find any. The 1 mm thickness is not too strong so I have to use aluminium profile to make enclosure more rigid. I used very thin household Teflon to lead the cooling air out. This 0.15 mm PTFE sheet is normally used for baking underlay.


Picture 6: Input circuit (note the
non-inductive resistors in the
center of the picture)

Picture 7: Input circuit
view #2

Tube parameters

GU-84b and 4CX1600b maximum values are presented in the following table.

 

GU-84b
(GY-84b)
4CX1600b

Anode voltage

2.2 KV

3.3 KV

Anode current

2.0 A

1.4 A

Anode dissipation

2.5 KW

1.6 KW

Frequency

250 MHz

250 MHz

Screen grid G2

400 V

350 V

Grid G1

-150 V

-150 V

G2 dissipation

30 W

20 W

G1 dissipation

1 W

0.1 W

Filament voltage

27 V

12.6 V

Filament current

3.7 A (3.4-4.0)

4.4 A

Filament/cathode warm-up time

3 min

2.5 min

G1 Cin

102.5 pF (90-115)

86 pF

Anode Cout

20.5 pF (18-23)

12 pF

Cgp

0.2 pF

0.15 pF

Transconductance

71 mA/V (50-92)

50 mA/V

Anode diameter

98 mm

86 mm

Anode temperature

200 C

250 C

Air flow 1600W (flow/pressure H2O)

1.6 m/min, 30 mm

1.0 m/min, 10 mm

Length

111 mm

109 mm

Weight

1.4 kg

0.75 kg



The complete technical data for the Svetlana tubes can be found from http://www.svetlana.com (locate the file 4CX1600B.pdf). There is a lot of other interesting articles e.g. by W9QXP Russian model GU-84b data can be found from DF6NA homepage at http://df6na.mayn.de/~df6na/tubes.htm.

Circuit diagram

C1

5-60pF, Input load

C2

5-60pF, Input tune

C3, C5, C6

1nF, feed-through

C4

1nF, DC block

C7

40nF, G2 built in the socket

C8

1nF, HV capacitor, inside the anode cabinet Cu/PTFE 105 x 200 mm

C9

Output tune, 68 mm 3 mm Al disk (dimensions not critical, larger OK)

C10

Output load, 55 mm 3 mm Al disk (dimensions not critical)

C11

1nF feed-through 400V (or homebrew 0.8 mm PCB 100x200 mm)

L1

Input resonator, 32 x 70 mm, 0.5 mm Cu

L2

Output resonator, length 600 mm (+/- 10 mm) , 100 mm Al tube

L3

Bias choke, 10 turns 5 mm

L4

HV choke, 11 turns 7 mm

L5

VK200 choke

L6

Round ferrite choke, 3 turns (bifillar winding)

R1

50W 4W

R2

2kW, low inductance 10W

R3

25W 40W low inductance, 2.5KV (optional; protects tube during flash over)

R4

50W 10W low inductance, 2.5KV (optional).

Amplifier mechanical dimensions

Picture 8: PA cabinet dimensions

Tuning

The tuning of output and input circuit is not always easy task. Here are some hints and advices I got from Leif SM5BSZ. Please reserve enough time and patience for the tuning job.

Output Circuit Tuning

You will need a DIP-meter, SWR-meter, 2m FM RIG, 50W terminator and 2KW non-inductive resistor. 2000W resistor is used for simulating the tube plate impedance with normal operating conditions: (2KV @ 1A).

 

Connect the tube to its socket and anode resonator to the tube. Do not connect any voltages.

 

Adjust the tune and load capacitors to 70% from ground (30 % from resonator).

 

Connect the 50W terminator to the output RF-connector.

 

Connect 2KW resistor between anode and ground near the tube socket.

 

Mount the front panel temporary but leave small gab near HV choke (current maximum point). You need two panel and few glue press to do this easily.

 

Measure the resonance frequency with DIP meter via the front panel gab and cut the length of anode resonator if necessary. DIP meter is used until you reach the FM transmitter frequency area.

 

Connect the SWR-meter to PA out connector and adjust the SWR to minimum with load and tune.

 

There is possibility that anode resonator length is not correct or that tune/load capacitor disk is too small.

 

The real dimensions are checked only with all tube voltages on. So I recommend leave the anode a little bit long before correct length is known for sure.

 

If on normal operation conditions G2 goes negative, then output load is too small. Move load closer to the anode line and adjust tune to max output power (farther of anode). I tune G2 current to 20-30mA. If you cant do this then anode line is too long.

Input Circuit Tuning

The input circuit was a little bit complicated to tune than output. Jerry W9QXP uses 50W resistor input that needs no tuning at all. I did try this but with poor result. I coulnt get more than 600W output with 100W drive. I decided to use a traditional LC-circuit. Here are some hints how to pre-tune it.

 

Install the tube and connect blower and filament on. Note: disconnect all other wires. Connect 8KW resistor between G1 and ground. This resistor simulates normal operation conditions: 90V @ 11mA.

 

Use a multimeter to measure DC voltage between G1 and ground over the 8 kW resistor.

 

After 3 minutes heating time connect drive for 144 MHz and change tune and load so that there is maximum voltage in G1. The tube works as a diode and rectifies all RF. Do not touch to anode or G2 because there are very high voltages.

 

Input SWR should also be near 1 at the G1 voltage peak settings. The maximum needed 80V is reached with drive level 30 - 40W.

 

Finally the input adjustment is made with normal working conditions with SWR meter. After this there is usually no need to change input tune or load capacitor values.

 
Finally
 

This amplifier has been in use since June 1997. There has been no problem with it. In my mind the construction is simple, even though it uses a lot of aluminium. I can recommend this project without any reservations.