No.20
Amateur Radio
4CX1600B

Background

There was a film some time ago titled, "The Russians are Coming." The introduction of a rather complete line of high quality RF amplifier tubes manufactured in St. Petersburg, Russia, which employ the modern external anode technology, makes this a reality. A very large company -- Svetlana Electron Devices, Inc., privatized in 1992 -- now sells its products worldwide. Recent descriptions in Communications Quarterly of two of their tubes, gave me the incentive to try one to revitalize my needy homebrewed Class AB1 amplifier. The application data and results are presented here.

Svetlana 4CX1600B characteristics

The tube, and its custom SK3A socket, are shown in Photos A and B. It's a ceramic-to-metal external anode tetrode whose original application was in a military transmitter, which attests to its ruggedness and quality construction. (This tube was called the 4CX1600A, and had a much smaller cooler.) Thanks to several unique design features, the 4CX1600B exhibits high-performance when operated in class AB1 with relatively low anode voltage.

The anode was recently enlarged and is now essentially identical to the 8877 in size and configuration. Unfortunately, its matching chimney hasn't yet been modified to fit. To overcome this problem, I designed one of my own. I've been told that a compatible chimney will be available in the near future. For the general tube mounting outline, dimensions, and construction details of my homebrewed chimney, please refer to Figure 1.

Figure 2 shows the tube's specifications, along with my actual operating parameters, while running the tube as a grid driven amplifier under single tone conditions. As you can see, it's conservatively rated for 1500 watts of continuous output at 75 MHz -- employing a recommended 50-ohm passive grid configuration that greatly simplifies design and construction while minimizing the risk of parasitics and the need for neutralization. It requires only a net of 25 watts of PEP drive in class AB1, with grounded cathode for full output in passive grid. Because of its internal geometry, the 4CX1600B performs efficiently at a recommended relatively low anode voltage of 2.4 kV for 1.5 kW output.

The tube requires the use of Svetlana's SK3A socket. This is a very modestly priced custom unit with a built-in, removable 0.01 mF annular screen bypass capacitor (mine measures 0.0114 mF). The upper frequency limit of this tube, at full power, is specified at 250 MHz. The bottom views of the socket and tube base are shown in Figure 3A and B.

The intermodulation distortion (IMD) specification for the 4CX1600B is a very respectable -36 dB for the 3rd order products, as measured by the conservative method to one of two equal tones. In amateur applications, which are referenced to the PEP, this would translate to -42 dB which, to my knowledge, exceeds all current transceiver specifications. If operated properly, without "all knobs to the right," you shouldn't experience any significant degradation of the tube's performance. This assumption is based upon the manufacturer's specified use in a grounded cathode configuration. Using my HP141T spectrum analyzer at 1500 watts PEP output at 14 MHz into a dummy load, I was essentially able to verify this, as the two-tone test results showed no change -- with or without the final amplifier. I used my Kenwood TS950SDX as the driver, with 30 watts PEP (see Photo C). The scale was 10 dB/div and the sweep was 1/2 kHz/div.

Invariably, most amplifiers -- that is, those in transceivers when operated at significantly reduced power -- will exhibit a corresponding improvement in IMD performance as evidenced in Photo C. As specified by the manufacturer, but not tested (see Figure 2), with increased grid bias and degenerative cathode bias of 24 ohms, the tube's specified efficiency rises from 61 to 72 percent. Unfortunately the IMD would suffer, increasing to -30 dB on the third order products from -36 dB. On the plus side, the zero signal anode current decreases from 500 to 200 mA; however, the drive requirement increases from 25 to 77 watts.

The chimney fabrication

As Figure 4A and 4B illustrates, the chimney may be constructed inexpensively using commonly available materials. The parts are attached using clear silicone rubber cement (RTV). Scuff mating parts with a coarse, 100-grit sandpaper, apply the adhesive sparingly, and an excellent bond will result. My base mounting plate was fabricated from a sheet of 3/16-inch Lexan®, a polycarbonate. Almost any similar material will do, as there is no mechanical nor thermal stress on the mounting plate. I used a fly-cutter to fabricate the base mounting ring. Of course, this concept is applicable to almost any tube requiring a chimney for forced air cooling.

Some provisos

Svetlana recommends that the regulated screen grid power supply be capable of handling any negative current that may occur. This is relatively common to the tetrode class of tubes because they lack a suppressor grid. As a result, the supply -- no matter what type -- should be capable of maintaining good regulation under all conditions, e.g. positive or negative current. A very simple shunt regulator is therefore recommended, and could consist of a pair of OA2s or VR150s (OD3A) in series. This setup should just do the job, but only for the grounded cathode configuration, as they are specified at 30 and 40 mA, respectively. I'm cautious about zeners because of their inherent positive drift characteristic. Drift occurs with zeners unless they are extremely well heat sunk, oversized, and air cooled. I obtained good stability with eleven 30-volt, 5-watt zeners in series exposed to forced air cooling inside the plenum. A shunt regulator is a preferred means of control for the source of the screen voltage because it offers the fail-safe advantage of fold back. It may also be available from the high-voltage power supply, providing additional safety for purposes of interlocking the two.

It's possible to use an adjustable, electronically, series-regulated power supply (I've used one in the past2), but it must be capable of handling possible negative screen current. To that end, I'd bleed a fixed amount of current consistent with an assumed worst-case negative current of 20 mA. Make sure the output voltage is clamped to prevent over-voltage. I'd use a pair of 1N2845Bs in series (180-volt 50-watt zeners) for this purpose. Also, to protect the screen from excessive dissipation, set the "roof" voltage about 90 volts above the regulated output voltage. This is just at the edge of the required differential for regulation while using a 40-mA transformer. In the event of excessive screen current, that combination causes the voltage to fold back, or "starve," the regulator -- thereby falling out of regulation. The manufacturer recommends that a source impedance of not more than 9,000 ohms be established as a current path from the screen to the cathode. Also, use a minimum of 50 ohms source impedance to the screen grid to further protect the tube in case of an arc. Both of these requirements are easily met with a shunt regulator. To read the screen current properly in either direction, I reset the zero forward manually -- fully beyond the normal range of the adjustment screw -- to 20 mA on a separate zero-to-one meter movement and calibrate for 100 mA full scale. I did this as a precaution in case a negative screen current occurred in addition to my normal, internal multimeter setup.

Svetlana also recommends that a 50-watt fixed resistor of at least 25 ohms be placed in series with the anode supply as a safety measure. This resistor is not used as a fuse, but to provide protection from infrequent internal arcs that are known to strike tubes at this power level. As with all tetrodes, the tube should never be operated without the anode voltage if the screen voltage is applied. It's an essential practice to only activate the two supplies together, or to interlock them accordingly. The heater voltage should never be applied without the proper cooling. The manufacturer has suggested the inclusion of a small inductance, as an L section in series with the 50-ohm input load resistance to ground, to cancel the 86 pF plus strays of the input capacitance of the tube for operation at 10 meters. I would recommend that this be modified into a fixed Pi section with a 91-pF input capacitor with a 0.33-µH inductor (5 turns #16 spaced to occupy 1 inch on a Micrometals T130-6; yellow core), which is effectively followed by my estimated total of 90 pF of grid input capacitance -- allowing for strays. The calculated value for 28.5 MHz alone would be 0.27 µH. This Pi section will have negligible effect on frequencies below 10 meters, and will enhance the match on that band while offering a small compromise for a slightly better match at 14 MHz. I provide this information for the purist, as most modern tuners will accommodate the basic mismatch resulting from the total input capacitance at the grid.

The heater is specified at 12.6 volts plus or minus 5 percent at 4.4 A. It's best to be slightly on the low side, as this enhances tube life along with providing some minor IMD reduction. To test for this, reduce the heater voltage just until the power decreases slightly, then increase it a few tenths of a volt above that point until full output is restored. You can repeat this test periodically during the life of the tube.

The heater voltage can be obtained conveniently through series connection of two properly phased 6.3-volt windings, or by commonly available transformers for 12 volts DC power supplies followed by a precise adjustment with an autotransformer or variable resistance. I used an old UTC S-70 -- a dual 6.3-volt, 5-A transformer with no problem; it's air cooled in the plenum, too. Two and one half minutes of warm up time is required for the heater to reach 90 percent of its safe current demand, although I conservatively elected to use three minutes. The tube may be operated in any position.

To enhance tube life, I recommend a Keystone Carbon Company Inrush Current Protector type CL-60, a negative coefficient thermistor, in series with the heater circuit. This device is specified to offer 10 ohms at 25 degrees C, dropping to 0.18 ohms at 5 A. These specifications closely fit the heater current drain and offer a smooth ramp of 10 seconds to bring the heater up to its rated voltage and temperature - reducing thermal stress to a reasonable minimum. Bear in mind that the thermistor must be placed outside the cooling airstream to function, and away from heat-sensitive objects for safety. In my amplifier, it was necessary to use Keystone's CL-190 in series with the primary because of my air cooling system. This device is rated at 25 ohms at 25 degrees C and ultimately falls to 0.41 ohms. In my case, it drops to 1.79 ohms at less than full load; that is, about 0.7 A. This unit also provides a nice time constant of approximately 10 seconds. The cathode, when operated above ground, is specified at a maximum of 100 volts - regardless of polarity, between it and the heater.

Last, but definitely not least, the cooling requirement for 1600 watts dissipation at sea level, at 25 degrees ambient intake air temperature, is 44 c.f.m. with a back pressure of 0.50 inches. The anode temperature must be limited to 225 degrees C (433 degrees F). It's a well-known fact that tube life is enhanced with low envelope temperatures. I found that the recommended air flow specification is most conservative for ICAS service. In my amplifier, the exit air temperature only increased 46 degrees F at 1500 watts output during heavy duty SSB operation.

The SK3A socket

I like the way the tube's element configuration and related socket contacts (all of which are spring loaded, see Photo B) are designed for low inductances. The heater contacts are brought out coaxially. The cathode is a ring surrounding the control grid pins, brought out through four equally spaced pins connected in parallel inside the base. The screen grid is also a ring at the base of the tube to which contacts are made via eight points around its circumference. Four contacts are made in the tube socket to the cathode ring. These are brought out to the four mounting ears so, with operation such as mine using an electronic bias switch (EBS),3 the cathode contacts must be isolated from ground. Otherwise, these terminals provide a nice, low-inductance means of directly grounding the cathode.

To isolate the four cathode connections that also serve as the socket mounting holes, I chose nylon insulating bushings used with the TO-3 type transistor. These fit the existing socket mounting holes precisely. By cutting off their stems, I could also use them to make insulating washers for the nut end of the mounting screws. These bushings are readily available at parts houses and are very inexpensive. The mounting job to the deck was completed using flat head 4-40 machine screws. The general characteristics for the 4CX1600B are given in Table 1.

Table 1. 4CX1600B general characteristics

Construction

In my homebrew amplifier, the layout and format are quite conventional with a Pi-L output. The amplifier is essentially designed for 20 meters. All capacitors and RF relays are vacuum type for reliability and low inductance. One of the two 4-inch panel meters serves as a center zero 500-µA tuning phase detector4 and the other, a zero-to-one milliameter, serves as a multimeter that reads all the currents and low voltages. Their layouts are shown in Photos D and E. For details of the Pi-L output network, refer to Table 2. By today's conventions, my unit is pretty large and probably overdesigned. However, I intended it to run continuously at 1500 watts without a whimper! I use a low velocity axial exhaust fan located at the top of the cabinet to enhance cooling. This type of construction is a near necessity for making IMD measurements because of the heat generation, and results in higher reliability. The schematic of my amplifier is shown in Figure 5.

Amplifier performance

By generally following Svetlana's typical operating voltage and current guidelines (2.45 kV on the anode with 330 volts on the screen), as given in Figure 2, I was able to generate 1500 watts of continuous carrier at 14 MHz. The tube ran quite cool into the dummy load. In SSB or CW service, using my EBS (see Figure 5), it ran much cooler than before. It saves power consumption, thereby reducing heat, increases tube life, and eliminates all background noise like fans, etc. -- so I wouldn't be without it. Note that this tube's quiescent anode current is specified at 500 mA, under the condition of grounded cathode for Class AB1, which at 2.45 kV results in just over 1200 watts of pure power dissipation. That, alone, justifies the use of the EBS. From the manufacturer's specifications, along with its low inductance mechanical configuration, it appears that this tube would make an ideal 1.5 kW amplifier for VHF.

One caveat: the control grid dissipation is rated at 0.1 watts, so you must ensure that essentially zero grid current flows - and for good reason. First, it precludes one form of nonlinearity that creates distortion and resultant splatter and second, it protects the grid dissipation and enhances the life of the tube. I recommend high-level ALC, as shown in Figure 5. It senses a few microamperes of grid current to generate bias that automatically reduces the drive from the exciter, ensuring the integrity of the grid dissipation and linearity while providing the correct drive power level. Further, due to the scales and/or ballistics of most meter movements, especially under voice conditions, it is very difficult, if not impossible, to detect grid current. This current, if present, will seriously and almost exponentially degrade the linearity of the amplifier. At the same time, it will cause an upswinging of the wattmeter due to the distorted, ("flat-topped") waveform. It's almost impossible to see small increments of this phenomena on an oscilloscope, but the result is very apparent over the air and quite visible on a spectrum analyzer. As a simple rule, if you observe any movement of any grid current meter, you're either underloaded and/or overdriving the amplifier -- creating distortion.

The typical Christmas tree pattern isn't totally reliable for detecting nonlinearity, even if precautions are taken. However, it can be more informative if you follow the procedure below, which is illustrated in Figure 6. First, lower the baseline of the oscilloscope to the bottom of the CRT, doubling the effective size of the resulting half pattern. Second, increase the sweep speed to display no more than two or three tree patterns. Now, load the amplifier under voice conditions, for the highest and sharpest single peak. If you want an alternative sensor for improper or inadequate loading, watch the screen current. Look for times when the current appears high or takes an abrupt, severe upward excursion at the onset of grid current. I know of one ALC system that employs this characteristic as a means of generating its control voltage.

For best combined SSB power out and IMD performance, it's necessary that the amplifier be heavily loaded; not by a single tone, as is often practiced, but by voice, a high speed pulser that simulates the voice, or two tones -- in that order of preference. As an alternative approximation and a last resort, load up fully using a single tone and continue loading approximately 10 percent beyond that point until a slight power decrease is noted. This will only be in effect under single-tone (CW) operation; under SSB conditions it will represent the maximum linearity and power out because of the differing anode impedances between a single tone versus the complex, voice waveform. The power output phenomenon, under voice operation, is easily verified on an oscilloscope, but often overlooked -- causing needless and annoying distortion products.

Initial operation

There are a few things you can do to make initial operation a bit easier and safer. First, grid dip the anode tank circuit for the frequency of operation. Then, check all circuits for continuity, shorts or opens, etc. Next, place a 100-ohm, 100-watt resistor in series with the high voltage, which I would initially reduce to around 1800 volts.

Insert a 5-k, 20-watt resistor in series with the screen voltage, which would also be preliminarily reduced to 250 volts. Allow proper time for warmup, and measure the heater voltage. Now, with the anode and screen voltages applied and cathode grounded, adjust the grid bias and/or screen voltage for approximately 325 mA of quiescent anode current. Then, slowly bring up the grid drive power to a few watts and watch for any peculiar meter readings or other signs of instability while peaking the power out. If all goes well, cautiously bring up the anode and screen voltages, and the drive, and reduce the values of the series current-limiting resistors until all the normal values are obtained. Of course, make sure to use a good dummy load and remember high voltage kills! Always short the high voltage lead to ground after any high voltage has been deactivated, and before commencing any other work regardless if the high voltage meter reads zero. It's also a good safety precaution to work with just one hand when any high voltage is on, or even if it has been deactivated.

Conclusion

The tube performed precisely to my expectations, and proved itself a very cost-effective means for providing a stable legal-limit linear amplifier using a passive grid input. The trade-offs, as Littlefield pointed out,1 of using tetrodes instead of triodes have become more significant in recent years. As a result, tetrodes may now be given much more serious consideration for projects such as mine. Using one in the manner described here is a significant way to provide a high quality signal, while avoiding a number of the common electrical and mechanical problems associated with amplifier construction and performance. I highly recommend its consideration.

Acknowledgements

I wish to gratefully acknowledge and express my sincere appreciation to W6JAZ and W6TC. Their contributions significantly assisted me in my work.

REFERENCES
1. Rick Littlefield, K1BQT, "Quarterly Devices," Communications Quarterly, Summer 1993, page 89. 3
2. Richard Measures, AG6K, "A Regulated Screen Grid Power Supply," Ham Radio, June 1986, page 51.
3. Marv Gonsior, W6FR, "Electronic Bias Switching," Ham Radio, March 1975, page 58.
4. Amateur Single Sideband, 1st edition, Collins Radio Company 1962, page 77.

Figure 1 - Tube mounting with chimney
Figure 2 - 4CX1600B specifications
Figure 3A and B - bottom view of the SK3A socket and 4CX1600B
Figure 4A - chimney mounting flange and design pattern, top view
Figure 4B - chimney design pattern
Figure 5 - PA schematic
Figure 6 - improved loading procedure
Photo A - Svetlana 4CX1600B
Photo B - SK3A socket
Photo C - IMD performance
Photo D - PA layout, interior
Photo E - PA layout, front panel

Table 2. Pi-L values¹ Q=12

Table 2 Notes:

Zin from the formula:

RL = 2500/ 1.5 x 1 Å 1666 *

1. ARRL Handbook, 1989 edition, page 15-3, 15-8.
2. Collins, Amateur Single Sideband, 1962, page 68.
3. Includes output capacitance plus strays (12 + 5 pF).

For Figure 5

C - 0.1/600 volt discs
C1 - 0.01 integral to SK3A
C2 - 0.001/5 kV Centralab 850
C3 - 0.33/100 volt paper
C4 - 0.001/600 volt discs
R1 - Globar Type A, 25 watt, 50 ohm, noninductive
R2 - 25 ohms, 50 watts, wire wound (see text)
R3 - 25 k, 10 watt, wire wound
R4 - 100 k, 2 watts (fail safe for R3)
T1 - Audio transformer; 500 ohms, 1:1
D1 - 1N914 or HP2800
S1 - Test/Disable switch
Q1 - 2N3439 or 2N3440
Q2 - 2N3902
D2 - 1N56644 ÒTransorbÓ for optional spike protection

Keystone Inrush Limiter type CL-60 negative coefficient thermistor (see text).

All resistors are 1/4 watt, except as noted.

The Manufacturer

Svetlana Electron Devices, Inc. manufacturer of the 4CX1600B. The word svetlana means light in Russian, which was appropriate enough when the company initially produced light bulbs. Svetlana is a relatively large, old line company, organized in 1898. The company began producing electron tubes in 1928. Svetlana's vacuum tube engineering and production staff, alone, consists of more than 2,000 persons. Apparently little was known about them until recently, as their production was solely for domestic and military consumption. The company manufactures a wide variety of quality tubes from triodes to pentodes. They range in size from the smallest tubes to extremely large devices, manufactured under rather impressively strict quality control procedures. For instance, Svetlana tubes are processed at higher temperatures than those in the West. An extensive two year warranty is offered and their products, including compatible sockets, are far more inexpensive than those prevailing for similar devices. A number of their tubes are directly interchangeable with existing Western types, and more will be coming out in the future.

For more information on Svetlana Electron, Inc. and their line of tubes, write: Svetlana Electron, Inc., 8200 S. Memorial Parkway, Huntsville, Alabama 35802. You can also phone the company at (800) 239-6900.

**The information provided in this application note is intended for general design guidance only. The user assumes all responsibility for correct and safe usage of this information. Svetlana Electron Devices does not guarantee the usefulness or marketability of products based on this material.

Back to Technical Bulletins Listing