The road to high-frequency tubes
The road to high-frequency tubes
- Foreword
The meaning of term ‘high-frequency’ has been evolving in the years and until WWII the research for vacuum tubes capable of operating at very high frequencies was essentially pushed by communication markets. Military carefully watched at new high-frequency tubes usable for radio localization sets, still experimental in several countries and later known as radars.
Around 1939, with few exceptions, the most advanced television transmitters used frequencies just above the limit of short waves. In 1939 B.B.C. started its experimental high-quality transmission at 45MHz, while RCA operated its vision antenna on the Empire State Building at some 54MHz. Articles on propagation of UHF signals, so defined radio waves just above 30MHz, were worthy of publication in the advanced press. In January 1939 Proceedings of I.R.E. published ‘A study of ultra-high frequency wide-band propagation characteristics’ by R.W. George from RCA., while in September 1939 Proceedings of Radio Club of America gave the article ‘Ultra-high-frequency propagation’ by M. Katzin, a study on propagation of radio waves from 50 to 150MHz.
RF power in the order of few hundreds watts was hardly obtainable at these frequencies. At even higher frequencies few watts were the maximum one could expect to radiate. A 462MHz transmitter, designed by RCA for two-way telephony between Rocky Point and Riverhead, used two UX-852s operating as push-pull Barkhausen oscillator to generate a mere 6W RF. Eight UX-852s were used one year later for a 15W transmitter operating at 432MHz and another set giving 115W at 411MHz required two water-cooled 846s (1200W just for heaters).
Pushed by military, research and development of new high-frequency tubes were suddenly accelerated beyond any imagination at the outbreak of WWII. As result, at the end of war the scenario was radically changed. Countless tubes for microwave frequencies up 24GHz and over were currently in volume production and powers up to several kilowatts CW and to some megawatts pulsed were commonly generated.
The development of new high-frequency tubes involved top industrial groups and even the best scientist of the time, as the research groups at the Birmingham University and at the Clarendon Laboratory in England, or at M.I.T., at Harvard and at Stanford in U.S., or even at R.E.L. in Canada. No information on classified tubes were given during the war. The progress was so fast that many of the tubes developed in the early ‘940s had been already discontinued at the end of the war, so that few traces of their design and characteristics data have survived.
It is interesting to note the primary role played by a well directed research activity in Great Britain, that often set the guidelines for innovative high-frequency components. Many microwave tube designs were transferred to U.S. or Canada through a close collaboration among British scientists and American industries under the coordination of M.I.T. or of R.E.L.
I am trying to offer here incomplete and maybe distorted glimpses of some little known yet excellent tube designs. Further contributions, suggestions or corrections are welcome.
- Micropup VHF and UHF transmitting triodes
In the late 1930s the GEC Transmitting Valve Group was involved in the activities for ’Ultra Short Wave Communications Valve Development’, known as CVD. Virtually all the ways to operate at very high frequencies were investigated, including the experiments at Stanford University on klystrons and those of Gutton at SFR Paris on magnetrons. One of the successful tubes designed in 1939 by W.H. Aldous and J. Bell from GEC was the E1046.
In this tube anode is a heavy copper tube forming the central part of the body, usually with a finned radiator or a copper block all around. Two glass domes are soldered at both sides of anode, using a featheredge or Housekeeper sealing process. The first dome holds a rod supporting a squirrel cage grid, formed by molybdenum or tantalum parallel wires. The opposite dome holds two smaller tungsten rods supporting the thoriated tungsten filament. The assembly process was quite complex, requiring jigs to precisely hold the parts during the sealing operations, in order to ensure the accurate positioning of electrodes.
Fig. 1 – The three basic types of micropup power triodes. 1A shows a 701A, one of the many derivatives of the prototype E1046. Canadian REL7 in 1B is equivalent to the British E1232 or NT99, introduced in 1941 and useable up to about 600MHz. In 1C a conduction cooled CV55.
Details of the electrode structure and of the squirrel cage grid can be better appreciated in the figure below.
Fig. 2 – 2A shows the internal view of NT99 electrode structure. The squirrel-cage grid is firmly held on the top by a copper cup sealed to the upper glass spacer. The nickel cathode cylinder surrounding the heater is supported by the larger tungsten rod from the bottom sealing glass dome. Pictures 2B and 2C show a 3C37, a variant of British CV55 with coaxial heater connections, and its grid with the finned heatsink.
- VT90 and its derivatives
E1046 was the evolution of a previous design using pure tungsten filament, the E1011. It was intended to operate at wavelengths of 50 to 150 cm, depending upon the application. The prototype with thoriated tungsten filamentary cathode went in production as VT90. E1046 gave origin to several variants, covering different CW and pulsed applications, and even to copies in US and in Canada. The early structure, with filamentary cathode, was capable of operation up to about 200MHz. To improve emission and raise the operating frequency to about 600MHz, the tube was soon redesigned. The filament was surrounded by a cylindrical oxide-covered cathode and the inside diameter of the anode block was enlarged, originating higher power and higher frequency variants.
Here are the known variants:
- E1046 early prototype. The tube was standardized as VT90.
- CV46 British service type tested for CW operation at 100MHz, 2KV maximum.
- CV62 British service type tested for pulse operation at 300MHz. 9KV at 7A peak.
- CV1090 British service type tested for pulse operation at 300MHz. 9KV at 5A peak.
- D-160810 Western Electric developmental code for VT90 equivalent, later 710A.
- REL1 Canadian REL version of VT90, production run by Northern Electric.
- WL-538 Westinghouse version of VT90.
- VT90 British Ministry of Service standardized code for E1046. Depending upon the application, continuous or pulsed, VT90 originates the service specs CV46, CV62 and CV1090.
- VT-240 US military code including both 710A and 8011, maybe also WL-538.
- 10E/97 Royal Air Force store reference for CV1090.
- 710A Western Electric version of VT90, also made by National Union.
- 8011 RCA version, also made by Amperex.
After the war, some improved variants of the original design were proposed in US by RCA and other manufacturers for pulse or CW applications and also for industrial RF heating.
- 6C24 RF heating or transmission. 600W plate dissipation, 160MHz max frequency.
- 5786 Similar to 6C24 with center-tapped filament.
- 8014A Pulse rated version; 400W plate dissipation.
Fig. 3 – 3A shows the type 8011, while a 710A is in 3B. 3C and 3D show two high-power derivatives, the 8014A and the 6C24.
- NT99 and its derivatives
In 1941 appeared a new improved design, the E1232. Tubes known as NT99 were sampled in April 1941 and a small production, about 8 units per week, started in July. In Great Britain volume production was launched at MOV in January 1942, but similar tubes went in production also in US and in Canada.
The plate finned block was characterized by an increased diameter with respect to VT90, resulting in a subsequent increase of the cathode diameter and of its emitting surface. Also a different mounting of the grid cage, now hold by a thick copper cup into a short glass spacer, made possible a closer spacing of electrodes. Average plate power dissipation of about 150W and frequency as high as 750MHz were possible with this design.
The new design was quite exceptional when compared with other high-frequency power tubes designed just a couple of years before. A pulsed push-pull oscillator using NT99s could generate over than 150KW at 200MHz, roughly the same power obtainable with the high-efficiency thoriated tungsten VT98s, just sampled in August 1939. But the NT99 was just 90mm high and required 39W heating power while VT98 was 350mm high and wasted 287W of heating power. More, the NT99 was useable up to 600MHz, full power, while the VT98 just reached 250MHz at reduced ratings. In the figure below a compact NT99, on the right, can be compared with a big and heavy VT98.
Fig. 4 –VT98, August 1939, compared against NT99, introduced in April 1941
This design gave origin to high-performance and compact micropup triodes, widely used in the ‘940s and still listed in the early ‘950s, only surpassed by oil-can shaped triodes, 2C39 and its derivatives. Here are some versions and/or variants:
- 4C27 US RMA code for CV92/8026. Exactly interchangeable with them and with Canadian REL7.
- 4C28 RCA and Penta Labs. variant of 4C27 for pulsed operation in SHORAN transmitters. Also sold by Central Sales & Mfg. Corp. around the mid fifties.
- 8026 RMA-EIA code for US made CV92.
- CV92 NT99 derated for 8KV peak voltage.
- CV199 Low emission selection of CV92, with peak cathode current from 20 to 40A.
- CV1256 British service code for NT99. Selected to withstand 12KV peak. 40A minimum cathode emission, 600MHz full ratings frequency.
- E1232 original design.
- NT99 Early service code for production lots of E1232.
- REL7 Canadian version of NT99. Some production run by Rogers.
Fig. 5 – From left, the NT99, a Central Sales & Mfg. 4C28 and a Canadian REL7.
- E1190 and its derivatives
The E1190 was a scaled down conduction-cooled variant of the E1232. The E1190 was followed by the improved design E1458. Higher frequency limits were possible due to low inter-electrode capacitance and to close spacing of electrodes, even if the average anode power dissipation did not exceed 50W. Here we look at a departure of US design from the British one. British tubes retained the flying heater wires of other micropups, while US versions were fitted with a coaxial connector. Some US variants also came with a factory installed grid radiator.
Here are the known types derived from E1190 and E1458 designs:
- CV55 Derived from E1190 for CW applications up to 600MHz. Delivers about 20W per pair
- CV155 Similar to CV55, but intended for pulsed applications up to 1200MHz. 40KW peak power per pair
- CV178 High-mu version derived from E1458. Usable for CW or for pulsed operations up to 1200MHz. 50W steady power dissipation, up to 150W for short periods.
- CV977 Matched-pair CV55
- E1190 Original design.
- E1458 Improved design, higher gain.
- 3C27 US equivalent for CV178. Has a coaxial heater connector.
- 3C27B Same as 3C27, but with grid finned radiator for higher duty cycle pulses.
- 3C37 Similar to 3C27 with increased pulse power capability. Grid radiator.
Fig. 6 – From left, the British CV155 and two U.S. versions, without and with grid radiator, the 3C27 and the 3C27B,
- Other micropup designs
The early design also originated few different tube types as:
Fig. 7 – A CV8 TR vacuum diode.
To thank the Author because you find the post helpful or well done.