13: Liquid Cooling the Eimac 4CX250 Valves
The Eimac 4CX250 and the numerous variants made by other manufacturers is
a popular tube for VHF / UHF amplifier duty. It does, however, suffer from one major disadvantage when the cooling aspect is considered. The tube is designed to fit into a special airflow socket and this inherently causes some obstruction to the airflow, necessitating a large blower if the tube is going to be pushed to its limit. Blowers that can pass sufficient air at the back-pressure produced in the socket are very noisy. When inserted in the airflow socket the valve occupies a large portion of the space, leaving very little room for the air to pass. The picture shows how much of the area is blocked by the valve – about 80%. This causes a high restriction to the free flow of air. The ceramic chimney fitted on the top of the airflow base restricts the airflow even more.
View showing a 4CX250 valve inserted in the SK-600A airflow socket. The white areas are the only parts of the socket that allow air to flow.
The SK-600 series airflow sockets are provided with a metal tube on the bottom that Eimac intended to be used to attach a tube to direct the air through the socket. Four holes are visible in the picture, punched in the thin metal skirt tube, for screws to attach the inlet tube. Many amateur designs ignore this method and simply flood the lower compartment with air. If there are any shafts and other cut outs in the metalwork, air will leak from the compartment. Air will also leak more readily through a low resistance opening rather than go through the high restriction airflow socket.
The K2RIW method was used in the writer’s 70cm amplifier, dispensing with the normal ceramic chimneys and pressurising the anode box, rather than the grid box, which the normal airflow socket was designed for. This requires additional chimneys above the valve to duct the hot air out of the anode box. The noise reduction using this technique is quite significant and a smaller blower can often be used with greater airflow and less backpressure.
The 2m linear was a design by Joe Ludlow, GW3ZTH, published in Short Wave Magazine  many years ago. It was commonly referred to as ‘the plumber’s delight’ as it uses large bore copper tubes in a push-pull design. Joe and the writer both worked for a company in Durban, South Africa. This amplifier was obtained from Joe when he returned to the UK from South Africa and it came with a massive wooden box in which resided a monster double ended centrifugal fan and two long four-inch diameter flexible air hoses which connected to the grid box. This was very noisy, but it did supply plenty of air – so much so that with the anode copper tubes removed it would push the tubes out of the sockets! I discarded the ceramic chimneys, pressurised the anode box with a much smaller blower, and obtained not only far better airflow but also much less blower noise. The air being blown into the anode box is directed on to the copper tubes and just this flow pulls a lot of heat out of the 4CX250B tubes and ample air flows through the finned cooler on the 4CX250B tubes. The backpressure is close to zero when using this method.
Having built several liquid-cooled amplifiers over the years, including a 6m amplifier using Russian GI-7BT triodes, I wondered if a 4CX250 could be liquid cooled and decided, as I had a few dud tubes, to experiment. Amplifiers such as the GW3ZTH using large copper tubes for the anode lines are not so suitable for liquid cooling as the pipes for the liquid are not easy to arrange to clear the anode tubes. The K2RIW 70cm amplifier and the W1SL 2m amplifier, however, are both very suitable. In both these designs, the tubes are mounted upright, and an extension on the length of the tube is easily accommodated, as there is sufficient headroom in the anode box.
Liquid cooled valves
Eimac and other manufacturers also manufactured valves intended for liquid cooling, although these are not very common in the power ratings used by radio amateurs. The smallest version was in fact a 4CX250B and used the same eight-pin base, with a 6V heater. It is the 4W300B / 8249. In every respect, it has the same performance as the 4CX250B but with an anode dissipation of 300W.
The original 4W300 valve entered production in about 1955 and it was based on the 4X150 having glass construction. The later version utilised the 4CX250B internal parts and had ceramic construction, although the old part number appears to have been kept, despite the new ceramic seals. It was phased out in the mid-1970s, leaving only the higher power versions. The ‘W’ in the part number denotes water-cooling.
The larger 4CW800B and F valves are different, using another valve base, and are available in 6V and 26.5V heater versions. Both of these valves also have
¼-inch pipe unions with supplied nuts and compression olives to suit standard copper pipe.
Original 4W300B valve
Later 4W300B valve with ¼-inch fittings.
Although replacing the anode cooler improves the overall cooling, the need is for some additional airflow over the grid pin and the bottom part of the socket to keep these parts cool. This does not need to be a great deal of airflow. A simple method is a small fan positioned in the grid compartment directed on to the bottom of the socket. A 12V DC computer fan of about 60mm or 80mm diameter is more than sufficient for two valves and this can be mounted either inside or outside the grid box with suitable ducting.
Thermal characteristics of liquids
Liquids transfer heat by two distinctly different mechanisms, one being conduction and the other convection. Water has a specific heat of 1.00; it is the benchmark for all other liquids. Water, however, is a very poor conductor of heat. The classic physics experiment with a piece of ice in the bottom of a glass test tube and the water at the neck of the tube being boiled, shows it takes a very long time before the ice melts. However, if a convection current is established any liquid will conduct heat away from one part to another more rapidly. Lubricating oils in internal combustion engines have a pump with a high flow rate; their primary function is to conduct heat away and the lubrication function is less important. The specific heat of most common oils is about half that of water, so to obtain the same cooling effect as water it requires twice the flow to move the same amount of heat away. Even some of the exotic heat-transfer fluids, such as Shell S2, are about the same as common oils, but they can operate at much higher temperature. These special oils are primarily intended for heating things up, for example plastic injection moulding dies. Automatic transmission fluid (ATF) and heat transfer oils are both typically rated at +350ºC maximum safe operation, but ATF is more freely available and far lower in cost.
Most cooling systems rely on convection currents to transfer the heat from a hotter part to a cooler part. In early automotive cooling systems they simply relied upon a thermo-syphon method to produce a flow of coolant. Using a pump to establish strong convection currents (high flow rates) allows large amounts of heat to be transferred, irrespective of the liquid being used. Air is another type of liquid, though not usually regarded as such; it is normally regarded as a gas. With forced air cooling it is the convection mechanism being utilised, as air is an even worse conductor of heat than water. Static air is the thermal insulator of choice, as it has very poor conductivity. When the object to be cooled is at a very high temperature, then radiation transfers the heat, even across a vacuum. It is how the infrared radiation from the Sun manages to heat things up after having travelled through the vacuum of outer space and how valves such as the 811 and 813 dissipate the heat from the graphite anode.
Removing the air cooler on the 4CX250 valve
To remove the air cooler on valves such as the 2C39 variety is easy. In most cases they unscrew or have two socket cap screws holding the cooler. The 4CX250 valves need a bit more effort but it can be done without damaging the valve if care is taken. If you examine the 4CX250 cooler it is very similar to a car radiator and it has many thin bent pieces of metal spot welded to the anode and the outer ring. You will not be able to see exactly how many layers of fins there are but it is normally three, running from top to bottom, and each layer is slightly offset to get as much surface area as possible into the small space.
Before attempting to de-fin a tube a few essential items are needed to protect the delicate metal to ceramic seals. Any excessive stress on these is likely to junk the valve. Use a piece of flat wood about 100mm square and about 38mm thick. Preferably select a hard wood such as mahogany so it doesn’t crumble during the detachment process. Find the centre of the wood and bore a hole so that the screen ring portion of the tube is a good fit into the wood. If you have access to a metal cutting lathe, a better method is a piece of aluminium with a hole bored to be a nice sliding fit on the screen ring. This is 36mm in diameter nominally, check with the tube before finishing the hole. If no lathe is available then proceed as follows.
2C39 valve used in liquid cooled 1.3GHz amplifier designed by N6CA.
Place the tube into the hole in the piece of wood and ensure the pins on the bottom are clear of the work surface. Should the grid pin stick too far below the wood place another couple of pieces of wood underneath so it is clear of the worktable. We need the outer ring of the cooler to be sitting flat on the wood top face. This outer ring takes all the pressure and the tube is attached by the slightly flexible cooling fins so it does not experience very much shock during the next operation.
A typical 4CX250B valve made by Siemens.
The anode cooler outer ring is nominally 41mm diameter and it is 19mm in height. Ensure it is sitting squarely in the hole. Hold the anode ring with one hand and push a small screwdriver between two of the fins as far down as you can get to spread the fins to open up a gap. Use a small, very sharp, wood chisel and gently tap the chisel with a small hammer – no more than four ounces is needed. Cut through the fins close to the anode tube until you reach the bottom of the cooler ring. I used a chisel of 10mm width (3/8-inch) and a four-ounce hammer. You will be surprised how little effort it takes to shear the thin copper fins with a razor sharp chisel. Move around about 45º and repeat the chiselling until all fins have been cut through.
4CX250 valve in a wooden block for de-finning the anode cooler.
When the outer ring is free, the anode tube and the remaining bits of fins will look like the picture shown here.
I experimented with a dead valve so that if anything went wrong I was not wasting a good one. I was surprised how easy it was to de-fin the tube and there was no stress placed on the valve during the chiselling process.
The valve is shown being held very gently in a lathe three-jaw chuck to allow the operation of removing the rest of the fins from the anode. This was simply a convenient way to hold the tube whilst I attacked the remainder of the fins with the chisel, working closer to the anode pipe to get as many of the fins off as possible. The protective top hat pressed metal cover will fall off as the fins are gradually cut away exposing the anode pinch-off tube. This was used when the valve was evacuated in manufacture. Do not touch this part with a file or any sharp tool because if the vacuum is destroyed, so is the valve!
To remove the remnants of the copper shim fins you will need either a small pair of sharp side cutters or a small pair of needle nose pliers. Work around the anode pipe cutting or waggling the fins until all are removed. As you remove more of the fins you will see that the bottom of the fins were spot welded on to the anode pipe. It is not necessary to smooth this pipe as it helps with the turbulence of the cooling liquid, so you can leave the last 2mm or so of the fins if you desire. I wished to get a smoother finish so I made a split collet to hold the tube safely in the lathe chuck. This can be any soft material, either aluminium or a bit of hard wood.
The problem with trying to grip the screen ring in a three-jaw chuck is that it is only held in three places and excessive pressure from the chuck jaws could crack the metal to ceramic seal and then the valve would be junk.
De-finned 4CX250 valve. There are three layers of fins.
A short piece of thick walled aluminium tube was found in the off-cut box and the inner diameter bored to be an easy, but not sloppy, fit on the screen grid ring of 36mm diameter. When the hole had been bored I made a mark on the ring adjacent to #1 jaw of the three-jaw chuck so I could always replace it in the correct way. With a thin hacksaw on the side opposite the mark for #1 jaw; saw through the wall to make a giant version of a piston ring. Clean up any burrs and put the tube into the split bush and replace it in the three-jaw chuck and it will run perfectly truly and without any danger of crushing the valve when you tighten the chuck jaws – but don’t over tighten. It should need very little pressure to hold it firmly as it is being gripped all the way around the periphery of the screen ring.
The liquid cooler may need to be a specific diameter, such as the original 41mm diameter air cooler if the valves are being used to replace existing air-cooled valves. In this case, a suitable brass or copper tube will need to be obtained. However, if it is a new amplifier we have more freedom in the diameter of the cooler and its length. A suitable tube is 38mm outer diameter which is a standard imperial size of 1.5in diameter and is only 3mm smaller in diameter than the air cooler. If the anode is held by finger-stock it probably will still fit tightly enough with this slightly smaller diameter.
Split bush collet to hold the tube safely in the three-jaw chuck.
A circular washer is needed that is a close fit on the bottom metal ring of the anode and made from a material that can be soft soldered. This needs a hole of 27mm diameter (check from the tube you are using) so it is an easy fit on the raised ring on this part of the valve. The picture shows a brass disc machined to suit the valve. This will form the electrical connection to the anode, and the seal for the liquid, and it will be soft soldered in place at the final assembly.
Bottom plate of the cooler showing how it locates on the anode ring.
The inlet and outlet pipes can be either let into the side of the cooler or on the top face (Fig 13.1). It can either be a brass or copper disc with two holes for the stub pipes. If the cooler is made shorter, the disc will require a central hole just large enough to take the top of the anode pipe. It will be sealed with soft solder leaving the pinch-off portion proud of the surface. For the W1SL and K2RIW amplifiers, the stub pipes may require right angle bends to clear the top cover.
Fig 13.1: Cross section of a typical top and side feed cooling jackets.
No matter which version is made, the length of the stub pipes is important. The inlet pipe, if the pipes are on top, should go almost to the bottom of the cooler space and the outlet pipe just below the top plate. This is so the coolant liquid is directed on to the hottest part of the anode. For side-mounted stub pipes, the inlet pipe should be close to the anode pipe and the outlet pipe closer to the wall. When making the stub pipes the inlet stub pipe can be made a smaller diameter so we can tell which one is which. The bore diameter should not need to be more than about 5mm.
The bottom plate and the outer tube form a bucket shape and they should be silver soldered together so that when the final assembly using soft solder is used they don’t fall apart under the heating. The top similarly is soft soldered into the top of the bucket on final assembly and the two stub pipes should be silver soldered in place beforehand.
An alternative to a liquid jacket may be a coiled helix of small bore tube wrapped tightly around the anode pipe and soft soldered in place. The writer has used this method but it leaves very little room to attach the tank circuit components and the bore of pipe needs to be small to allow enough turns. Using 6mm OD pipe only allows three complete turns on a 4CX250 anode pipe. Anneal a length of copper tube by heating to dull red with a blowtorch and then drop it into a bucket of cold water. It will now be very soft and can be bent with the fingers easily. Wind the helix on a wooden dowel slightly smaller than the anode so the turns are adjacent and then spring the coil on to the anode where it should fit tightly. Smear some flux over the coils and the anode pipe, gently heat with a small blowtorch and run some plumbing acid-core soft solder all over the coil and the anode pipe and allow to cool. When cool enough to touch with the fingers, wash well in water to remove the acid flux and clean with some steel wool or scouring powder.
When doing the final assembly thoroughly clean the two parts to remove any dirt or grease, anoint them with some solder flux, heat until plumber’s acid cored soft solder can be melted and tin them well. Then put the bucket on to the valve and sweat them together using a small blowtorch to make the solder run into all the crevices. Put the top cap in place and sweat that at the same heating. Allow it to cool naturally, do not drop them into water until thoroughly cool to avoid thermal shock to the ceramic seals. Wash and scrub off any excess flux and then dry. If you have compressed air, blow it through one of the pipes to remove any water and then dry in a warm place.
When silver soldering brass and copper the metal grows a thin copper oxide surface. To remove this, the best way is to pickle the parts in very dilute sulphuric acid for about 20 minutes. Mix one volume of normal battery acid with 15 to 20 volumes of water in a suitable container, an old glass jar works well or a plastic bucket for larger pieces. The weaker the solution the better it works, it just takes a little longer. Let the part cool down so it is black and gently immerse it into the pickle solution, suspended on a piece of copper wire, and watch out for any splashes of hot acid! This not only strips the copper oxide off the parts but also cracks off and dissolves any flux remaining. The parts after washing in water look salmon pink and are easy to polish to a nice finish if you so desire.
The writer normally uses LM-35 temperature sensing integrated circuits to monitor the temperature. They operate on any DC voltage between 4V and 35V and draw virtually no current, a few mA. The nice thing about them is that the output voltage is very accurately scaled at 10mV per degree Celsius. So interpreting the reading is simple. At +100ºC they output 1.00V and will work up to +150ºC on a single DC supply rail. If you want to measure negative temperatures down to –55ºC you will need a split rail supply with a positive and negative supply; similar to ground referenced op-amps. They are not too fussy about ripple on the supply rail so the supply can be quite crude, just a half-wave rectifier and a small capacitor will work fine.
The LM-35 comes in various packages, metal can and plastic, and the lowest cost plastic package is the three legged TO-92. One pin is ground, one the positive supply and the other pin the output voltage. I use them all over the place, as they are not expensive and accurate enough for most applications. I normally make a
T-piece with the fluid flowing through and the LM-35 in the third pipe immersed in the fluid and sealed with epoxy glue. When high power RF is used they do need some RF filtering, but it is normally minimal: a suitable capacitor between the positive supply pin and ground and one across the output to ground is usually sufficient. The output voltage can be fed into a simple comparator circuit to trip an alarm if the temperature is too high.
Plumbing the system
The pipes connecting the various parts need to be electrically non-conductive, so avoid black rubber hose as they may contain carbon-black filler. At the temperatures to be expected clear vinyl tubing is normally acceptable, and you can see the liquid. All joints should be a tight push fit or clamped with miniature hose clamps so that no leakage occurs. For multiple tube applications some manifolds or Y-pieces will be required that take the inlet and outlet hoses from the individual pipes and split and recombine the fluids. Below is a picture of the 6m amplifier system. This system had three holes of 6mm diameter running through the cooler block which the two GI-7BT triodes were bolted to and 6mm bore vinyl hose was used to run from the manifolds to the cooler block.
Long pipe runs will require a larger bore hose and 8mm or 10mm high-pressure automotive fuel hose is a good choice. I standardised on 10mm bore fuel hose and as this is rated to at least 6-bar (90 psi) it is unlikely to cause a problem at the pressure we will require in practice.
Anode cooler used with the GI-7BT cooling system.
An example of the fluid system is shown in Fig 13.2, the precise parts will differ from system to system. Note that the return pipe for the coolant enters the reservoir tank below the coolant level. If the return line is above the coolant level a great deal of splashing occurs which entrains air into the fluid. Apart from the frothing caused, the air entering the fluid can cause problems with air locks when the fluid heats up in the pipes and the tube cooler, and should be avoided. Ensure you have a close fitting lid with a small air vent hole to allow the heated air to escape.
Fig 13.2: Typical coolant system showing the major parts.
The preferred coolant is automotive transmission fluid (ATF) as this is essentially non-conducting at several kilo-volts. (The breakdown voltage for ATF is ~15MV/m). The more exotic types are made from silicon oils that do not absorb moisture. If water is used the coolant will require regular changes as the water becomes contaminated by the metals it is in contact with. The addition of 20% antifreeze (glycol) will slow the effect, but it will still need changing when the leakage current becomes excessive. ATF does not suffer from this problem. ATF is used in automatic transmissions and routinely experiences temperatures as high as 350ºC without deteriorating.
Using ATF allows a Bosch fuel injection pump to be used. These can pump many litres per minute against high backpressure. In a typical petrol injection system, the fuel rail pressure is about 3-bar (45 psi) and a pump will deliver about 200 litres per hour. Into a much lower backpressure the flow rate increases, to about 1000 litres per hour for zero backpressure for the larger pumps. The one liquid they must never be used with is water, as this causes rapid corrosion of the internal parts.
In automotive fuel systems very often the fuel pump is immersed in the fuel tank and the fuel is fed via a pipe up to the engine compartment. The fuel then enters a manifold, called the fuel rail, where it is distributed to the various injectors. To maintain a constant pressure a pressure regulator adjusts the flow rate and any excess fuel is returned to the fuel tank. Hence, although the engine does not require a high rate of flow the fuel is continuously circulating in the system. This keeps the fuel rail cool, so preventing vapour locks, and purging any air from the system.
In a petrol or diesel vehicle, the fuel flows not only through the pump portion but also through the DC motor. This keeps the motor cool and lubricates the internal scroll pump. It is the only lubricant the very fine clearance of the scroll pump has (about 0.05mm), as a way of resisting wear. When used on diesel fuel the pumps last about three times longer than with petrol, because diesel is basically kerosene with an added wax as a lubricant for the high pressure fuel injection pump that the Bosch pump supplies. When a pump becomes too worn for normal use it can still be used with a more viscous liquid such as oil or ATF. Another suitable liquid is paraffin / kerosene with about 20% ATF added as an extra lubricant. This fluid is lower viscosity than plain ATF, which is about SAE-20 at room temperature, and for long runs of pipe reduces the backpressure on the pump. Diesel is also a good substitute as it contains wax as the lubricant. The temperature used will be much lower than the flash point of these two liquids. If the use of oils does not appeal to you, then resort to water and glycol, but you need to find a suitable pump and to change the water regularly.
If water is used the plumbing becomes more complex. Because water does conduct electricity to some extent, grounded ferrules must be fitted in both the inlet and outlet hoses a short distance from the anode cooler in the pipes. Also, the outlet temperature must be less than 100ºC for obvious reasons! The maximum safe anode temperature of the 4CX250 is 200ºC and ATF will handle this without any problem. The higher quality grades of silicon hose are usable up to 150ºC.
Heat exchanger basics
Heat can only flow from a hotter body to a cooler body and the rate of heat flow is purely determined by the temperature difference, DT. Temperature is an indication of the amount of heat energy contained in the body: higher temperatures indicate higher energy contained. These are the laws of thermo-dynamics.
If the heat exchanger is a liquid-to-air unit, clearly the liquid leaving the exchanger cannot be cooler than the air flowing through the cooling fins. When the liquid entering the exchanger and the air flowing through the fins are the same temperature no heat flow occurs. The thermal characteristics of common oils, compared with water, are shown in Table 13.1.
Specific Heat Ratio (vs Water)
Boiling Point ºC at atmospheric pressure
Kerosene / Diesel
SAE – 30
Shell Transfer Oil
Sunflower / Canola Oil
Petrol / Gasoline
Table 13.1: Thermal characteristics of common oils
The parameter of most interest is the specific heat value. Water is the benchmark and is assigned the value of 1.00, all other liquids are lower. This means that if using water with a particular flow rate the temperature gradient across the cooler is, say, 10ºC then when using ATF at the same flow rate it will be about 20ºC. As we saw earlier, this means that the heat exchanger shown in Fig 13.2 will be able to dissipate this heat faster than with water. All the common oils fall within the 0.4 to 0.5 region.
The best liquid is ethyl glycol, which is antifreeze, but it is a toxic liquid and absorbs water very enthusiastically from the air, as does brake fluid which is also a glycol. In automotive water cooling systems this isn’t an issue as the antifreeze is diluted in the water, but trying to use it neat means we have to have a completely closed system to exclude air.
Although the system shown in Fig 13.2 is the bare bones we can add extra features that will improve the operation. One useful addition would be to supply the pump with a PWM waveform to control the speed and hence the flow rates. The temperature sensors allow monitoring the outlet temperature of the anode cooler and the ∆T of the fluid. When the value of both is low, for example when on long receive periods, the flow can be reduced. On transmit there will be some time lag as the fluid heats up, but once the temperature starts to rise the pump can be driven faster to increase the flow.
A suitable heat exchanger can be an automatic transmission oil cooler or a heater matrix. Both of these can be obtained from car breakers for a small amount of money. A saloon car heater unit, with a multiple speed 12V DC fan, can be purchased, making the assembly of a system simpler. A car heater matrix can normally dissipate about 3kW to 5kW when the inlet temperature is over 80ºC. The writer has a 2500 litre rain water tank outside the shack with a coil of copper tubing immersed in it. This forms almost an infinite heatsink and can dissipate many kW of heat continuously.
Despite the fact that a liquid is being used to transfer the heat it ultimately needs to be dissipated in the ambient air. All liquid heat transfer cooling systems are in reality just part of a bigger air cooled system. The automotive liquid cooled engine is a classic example. No matter which liquid is used to transfer the heat from the device to the heat exchanger ultimately it is always finally transferred to the ambient air as the final part of the cooling system. The liquid portion is simply a convenient method of moving the heat.
Microwave Oven Testing
Placing a known volume of water into a microwave oven and measuring the temperature rise over a length of time can yield surprisingly accurate results. To test this method out I used the microwave oven in the kitchen and placed a 500ml glass measuring jug filled to the 500ml mark. The water was drawn from the cold tap and after standing for a few minutes to reach a stable temperature it was measured with a digital thermometer. The oven was turned on for a timed period of two minutes and the temperature was measured again.
Note: It is necessary to stir the water well before taking the end temperature as the surface of the water was much hotter than the water near the bottom. This is to be expected as the glass forms an attenuation path for the microwave energy and the exposed surface water absorbs energy more readily.
The results I obtained were:
Start temperature = 12.3ºC
End temperature = 49.5ºC
DT = 37.2ºC in 2 minutes.
If the power output were 1kW, 500ml of water should rise by (14.43 x 2) = 28.86ºC per minute, the end temperature should be 70.0ºC. The measured rise was (37.2 / 2) = 18.6ºC per minute. Hence, the power absorbed was (18.6 / 28.86) x 1000 watts = 644 watts.
The microwave oven is 650W output, according to the manufacturer, so the kitchen experiment agrees very well with the claimed power output!
Simple pump speed controller
A basic pump control circuit is shown in Fig 13.3. It uses the LM-35 temperature sensor measuring the output of the tube cooler and varies the pump supply voltage, speeding up the pump as the temperature rises. It is a series regulated power supply with a variable reference voltage. The LM-358 is a dual op-amp with a common supply.
Fig 13.3: Variable speed pump schematic.
The pump driver transistor TR2 will require selecting for the maximum pump current. A suitable device for TR2 would be an NPN Darlington such as the BDX-33 or TIP-142 as these are both rated at 10A at 60V. It should be mounted on a suitably rated heatsink as the Bosch pumps draw several amps when pumping against high back pressure. TR1 can be a small signal device such as 2N2222 as the current is low.
The first op-amp amplifies the LM-35 output voltage by four and this is applied to the second op-amp. The reference input comprises the output of the first op-amp divided by two, by R5 / R6, and a variable reference voltage provided by VR1. The maximum reference voltage is set by D1 at 3.9V. The second op-amp also has a gain of four and the output of TR2 is fed to the sensing resistors R7 / R8. The circuit is a linear power supply and as the reference voltage varies with temperature so does the voltage applied to the pump. Set VR1 to give a voltage across D1 / R6 of ~2V with no input from the LM-35. This will cause the output voltage to be ~8V and the pump will run at the lowest speed. When the LM-35 senses a temperature of +100ºC it raises the reference voltage and causes the pump to run at its fastest speed. Zener diode D1 prevents the reference voltage exceeding 3.9V, hence the maximum pump supply voltage is (4 x 3.9) = 15.6V. The raw DC supply can be quite simple and the ripple on the supply rail is not important as long as the trough of the ripple does not sink too low.
Component parts of a cooler using plumbing fittings.
Cooler made from plumbing fittings. The reduced height version is on the right.
Alternative anode cooling jackets
Searching amongst the myriad of copper fittings in the local hardware store, some pipe reducers were found that fitted the anode of the de-finned 4CX250 perfectly. The larger end fitted on to the anode ring where the raised ridge is and this can be soldered in place. This end accepts a 28mm OD tube; it being for a nominal 1in bore pipe. The other end accepts a standard 25mm OD pipe (¾in bore) and a stop end also can be used to close off the top of the cooling adapter. A short piece of copper pipe cut to fit both the adapter and the stop end can be soldered in place to seal the jacket. Holes for the inlet and outlet pipes can be drilled in suitable places to suit the application. When soldering the jacket on to the valve ensure these stub pipes are correctly aligned to suit the amplifier layout. If the jacket is too tall the smaller diameter portion of the adapter can be cut down in height, as shown in the photograph above right.
Cooler installed on 4CX250 valve.
 ‘A 144MHz Linear Amplifier’, by J D V Ludlow, GW3ZTH, and C J S Dunbar, GW8EHK, Short Wave Magazine, Vol 31, editions 5 and 6 (July – August 1973).