Manufacturing of experimental magnetrons (1953 – 1955)
In April 1948, the Dutch Minister of War established a national Radar Commission. Its task was to review the potential market for radar and to prepare a set of recommendations. The recommendations had to be about how the Dutch market would be best served: own R&D and production of radar, or choose for licensing of foreign equipment. Prof. jhr. ir. J.L.W.C. von Weiler was appointed as Chairman of the committee.
The committee quickly came up with the recommendation for radar R&D and production in the Netherlands, with as core facilities the Physics Laboratory in The Hague, the LEO (later LEOK) in Oegstgeest, and Philips in Eindhoven.
For that reason, the Physics Laboratory TNO and Philips worked closely together in the early 1950s, in particular on the development of magnetrons for radar transmitters. The following internal note (dated 1954) explains the manufacturing steps of magnetrons at the TNO Physics Laboratory, Waalsdorp, The Hague. The slightly edited text provides a good insight into the inventiveness of people at TNO to manufacture magnetrons:
Magnetrons were described already in 1921. Starting in 1922, the split-anode structure had been developed [Radiomuseum]
No detailed description is given of the various actions; these can be found in the Philips report no.55100. Only the soldering process of the three spouts of the magnetron is described more extensively because TNO uses a different method for heating the block.
The material and components, with one exception, are purchased from Philips. The oxygen-free high thermal conductivity (OFHC) copper for the blocks and the endplates is supplied as a staff of 50 mm diameter respectively a strip of 50 x 2 mm2. The blocks and lids are manufactured in our own (TNO) workshop. The cathodes are delivered in a partially assembled state.
The glass-metal connections are delivered finished so that they can be soldered into the blocks immediately after cleaning. The silver solder, gold wire, molybdenum wire and copper wire for the ‘straps’ (flexible connection) are supplied in lengths for creating 10 magnetrons.
The cooling jackets form an exception as they are manufactured by TNO’s workshop.
For the construction of the high-vacuum set, some glass parts are also required from Philips, as well as a folder of drawings and regulations, marked no.55100, for the manufacturing of magnetron no. CV 76.
The block and the covers are, as already mentioned, manufactured in our workshop according to the drawing specifications in the Philips leaflet no. 55100. No special difficulties arise in this regard.
First, the straps are bent in a mould and later made to length, after which they are put in the holes drilled in the block. The straps are fixed in the block with a specially-made punch.
Before proceeding to solder, the block and spouts must be degreased and pickled. The degreasing is done in gasoline and the pickling in respectively:
- in warm hydrochloric acid; then rinse in water;
- in chromic acid; rinsing in water;
- in glass stain; rinsing in water;
- neutralising in ammonia; rinsing in water;
- rinsing in distilled water;
- rinsing in alcohol.
These operations also apply to the 3 spouts, except for the glass stain step. After these operations, the block and spouts may no longer be touched by hand, as this would again contaminate the cleaned metal surfaces. To prevent this from happening, they must be handled only while wearing nylon gloves.
The following operation is the soldering of the spouts. The blocks are heated at the Philips manufacturing facilities using High-Frequency heating. This is a very sensible method of working. We [TNO] can not apply that method because we lack a device of such capacity. Therefore, it was necessary to look for another possibility for heating, while keeping the block with the spouts in a reducing atmosphere during the soldering process. It was thought, among other things, to heat the block using a gas flame. The difficulty was to keep the combustion gases and the reducing atmosphere separate. Finally, a solution was found which later proved to be perfectly satisfactory in practice, not only because the soldering process runs smoothly, but also because of a time saving of +/- 50% in comparison to the Philips HF-heating method.
The anode block is clamped between the two sockets A (see drawing above), one of which is fixed in housing B, while the other is pulled to the anode block using two springs. The endplates of these sockets consist of a chromium-iron alloy, the oxide of which is not reduced in the reducing atmosphere. In this way, the liquid silver solder is prevented from sticking to the endplates. As a result, it prevents the block would have to be soldered to end plates and could not be removed after cooling.
The three spouts are arranged using attachments mounted in housing B in such a way that they are perpendicular and diametrical to the anode block; they are held in that position during soldering.
Before the placement of each nozzle in the anode block, a molybdenum wire and a ring of silver solder are inserted. The silver solder ring must be pickled and reduced in advance.
The photos above show the arrangement of the spouts in the anode block with the attachments that hold them in place. After the spouts are arranged, the two glass cylinders are placed in the correct position as well as lid D. Thereafter, a gas mixture flows for 5 minutes after which the two burners will be ignited. The generated heat will now pass through the endplates A in the anode block. After the silver solder has flowed, both burners are extinguished.
An advantage over the HF heating method is that compressed air can now be blown into socket A, as a result of which these two sockets and the anode block are forced to cool, something that is not possible with HF heating. A time saving of +/- 50% is achieved by this type of forced cooling.
The gas mixture is supplied inside the anode block to get the gas mixture in direct contact with the surfaces, which are to be reduced. Due to the mixture – the introduction of gas at the top of the soldering apparatus, which was initially done, the gas mixture did not penetrate sufficiently enough to the inside of the block; therefore, the internal surfaces were insufficiently reduced.
Spot welding and adjustments
After the block has cooled down and has been removed from the soldering device, the cathode is welded to it. For spot welding, the cathode is placed with the aid of two moulds in the correct position in the central bore, and then spot welded to the two molybdenum wires.
The next action is to set the correct frequency at which the magnetron (radar) must work. This is performed by bending the straps slightly while using a suitable measuring device. It has been found necessary to clamp the two sides to the anode block during measurements; otherwise, an incorrect adjustment will be obtained. After this step, the covers are stained and the edges of the block are sanded blank.
A pre-prepared gold ring with a lid on top is now placed on one edge of the block. Then the whole assembly is carefully turned. Then the second lid is fitted after which the whole is clamped between a pair of sturdy iron plates with four bolts. When fitting the covers, care should be taken that the cover labelled cathode, is placed on the uninsulated side of the cathode. The four bolts must be tightened sufficiently to prevent the air between the gold ring and the cover, respectively, the block, can leaking inside during the vacuum pumping.
The whole assembly is now suspended from a tripod. The glass tube, which is connected to one of the spouts, is melted into the inlet tube of the high-vacuum pump device.
The pump device and magnetron are now pre-emptied with the pre-vacuum pump, after which the high-vacuum pump is turned on. When the prescribed vacuum level is reached, which is a sign that the gold rings seal sufficiently, the oven is applied as a cover over the magnetron. The temperature is gradually raised to 470 °C while the high-vacuum pump continues pumping. During this operation, the block and the covers are firstly degassed, which can be seen from the initial rise of the manometer. However, the actual purpose of this treatment is the soldering of the lid to the anode block, thus obtaining a good high-vacuum sealing. This occurs because, at this temperature and pressure, diffusion takes place between the gold and the copper. In other words: gold and copper formed an alloy. Because of the difference in expansion coefficients of the copper and the steel of the bolts, the pressure between the gold and the copper will increase even more.
Making of the cathode
A second construction step is the creation of the cathode. That requires that, before the oven is placed over the magnetron, the two cathode connections are connected to a voltage transformer. When the oven has reached the right temperature, the voltage is carefully increased. It is now observed on the Penning manometer that large gas quantities are released, viz. Carbon dioxide, because the barium-strontium-carbonate is converted into oxide, thereby releasing CO2 gas. After all the procedures have been performed according to the specifications, the cathode is complete.
To accelerate the disposal of the C02 gas, liquid nitrogen is applied around the condenser of the high-vacuum unit instead of liquid air. As a result, the carbon dioxide gas does not have to be pumped away as it is retained in the cooler due to the much lower temperature of the liquid nitrogen.
The total step takes about two hours and may be considered terminated when the vacuum has reached the prescribed value. The oven is now switched off and must at first cool down to +/- 250 °C, after which it may be removed from the magnetron.
When the magnetron has cooled down sufficiently, it is operated as a diode with 300 V=; the anode current must then have the prescribed value. This is followed by high-voltage pulsing. The voltage is increased very gradually during this step to prevent very violent discharges which could damage the cathode. The vacuum must be carefully monitored since large quantities of gas can be released during discharges.
This treatment is terminated when the voltage has been increased to +/- 14 kV and there are practically no more discharges. Then the 300 V= is switched on for a while. Finally, the filament is overloaded for a few seconds to expel any gas residues.
When the vacuum has reached the final required value, the magnetron device can be detached from the vacuum pump by melting the glass tube.
Due to the heating in the oven, the outside of the magnetron is heavily oxidized and browsed. This oxide layer must be removed before assembling the cooling fins. Therefore, the entire magnetron is first pickled in a stain bath, so that it becomes white again. However, during this pickling, care must be taken that the Litz wires, which are located at the cathode connections, are not kept in the pickling bath. The acid, which is sucked up during the clamping, is no longer removed after rinsing in water, as a result of which the threads would disintegrate within a few hours.
According to the [Philips] procedures, the block should be electrolytically tinned after pickling. However, we do not do that [at TNO] because we do not have such a bath. However, if the magnetron is no longer contaminated after the pickling, the soldering tin will certainly flow well on the copper. When the two half-cooling fins with the three parts are then arranged in the correct position around the outlet spout, the entire assembly is soldered together with tin. Thereafter, the magnetron is cleaned in warm water to remove the soldering water. After drying, the magnetron is sprayed with varnish, after which the glass clock is placed to protect the two glass spouts of the cathode. Finally, after both litze wires are soldered to the connections, the magnetron is ready to be put into operation.
From the mid-1950s, commercial magnetrons were purchased from Canadian, English and French manufacturers alongside Philips magnetrons. Our Laboratory’s manufacturing process stopped.
The radar magnetrons shown above can be seen at Museum Waalsdorp.