Restoration of 1952 MG TD 2

Author: Bob McCluskey
First posted: 1 Sept 2000
Last amended: Dec 2015
Please email Bob McCluskey
Car No TD/11935
Engine No XPAG/TD2/12333
Body Type 22381
Body No 11301/78948

B carburettor ell, first of all, I should emphasise that I am no expert on carburettors of any description, far less SU carburettors. Almost everything I say here has come from the web (1 and 2, and from the manufacturers themselves, 3 and 4).

Having said that, I must now say that from the experience of several years as a home mechanic, SU Carburettors are far and away the simplest I have ever had to work on. They have one large jet which never blocks, and a simple and reliable method to meter the fuel/air mixture.

There are only two difficulties to struggle with. The first is spelling, and although I have seen it spelt with two r's, one t, -or and -er and any combination, the spelling I have used here is the preferred spelling used by Oxford English Dictionary (although even they give -er as an alternative). The second is nomenclature. TDs (and other T-types) use H2 carburettors. H obviously stands for semi-downdraft, and by a wonderful piece of logic, 2 stands for 1 1/4 inches: the 1 is taken as given, and the 2 stands for two eighths, or one quarter. I don't think they made a 2 1/4" carburettor, but if they did, it would have to be either H110 or H20.

SU carburettors were invented by George Herbert Skinner ("Herbert"), eldest son of Banks Skinner of Lilley and Skinner fame. Herbert was an outstanding business man, and car enthusiast and successful inventor, and followed his father into the business of Lilley and Skinner. But his enthusiasm lay in the development of the petrol engine, and with his brother Thomas Carlisle Skinner ("Carl") developed the new form of carburettor. In 1906 they were awarded a patent on their carburettor. Initially it was manufactured for them under the brand name Union Carburettor but in 1910 they formed a limited liability company and changed its name to S.U. Company Limited. His brother Carl sold out of Lilley and Skinner to take over the carburettor business, but Herbert remained with Lilley and Skinner and continued to develop his carburettor with patented improvements. The carburettor company struggled initially after the First World War, and wasn't profiable, so in 1918 Carl Skinner sold the business to his customer William Morris. In 1936 Morris sold SU Carburettors along with many of his other interests into Morris Motors. Carl Skinner himself became a director of Morris's business and remained MD of SU Carburettors until he retired aged 65 in 1948 - in this he may have been just in time to avoid the fate of others (eg Cecil Kimber himself in 1940 and Victor Riley in 1947) whose businesses were incorporated into Morris's empire and who subsequently lost control of their companies as a result.

SU carburettors were fitted to a wide variety of high performance sports cars and quality cars such as Rolls Royce, Bentley, Rover, Riley, Austin, Jaguar E-type, Triumph Spitfire and others. It must have been galling to Morris's competitors to have to source their components from him, and there would have been considerable incentive to develop an alternative. In the 1960s Zenith and Triumph collaborated to develop the Zenith-Stromberg carburettor, specifically to break SU's patents. It was almost identical in appearance, but instead of a piston rising in the dashpot, it used a rubber diaphragm. This alternative was readily embraced by, specifically, Jaguar and Triumph, and later E-Types almost exclusively used Strombergs. In the 1980s, due to the competition, and also because of development of fuel injection, SU became less and less profitable, and in 1994 went into voluntary liquidation. In 1996 the name and rights were acquired by the Burlen Fuel Systems Limited which continues to manufacture parts and systems, largely for the classic car market.

All carburettors (but not fuel injectors) work essentially the same way. Their job is to deliver a stream of air mixed with vapourized or finely atomised fuel in the right proportions to meet the varying demands of the engine. As air is drawn into the cylinders it passes through a choke of fixed or variable size (a "venturi") and its velocity is increased. From Bernoulli's Principle (but see footnote), the pressure in the accelerated air flow is reduced, leading to a depression (partial vacuum). The carburettor is arranged so that a jet (a small orifice supplied with fuel) is located at this depression, and fuel is drawn into the air stream and vapourizes. The faster the air flow, the greater the depression, and the more fuel is drawn in. It is difficult to arrange that the mixture is absolutely correct at all engine settings, and there are always a series of compromises which varies from carburettor to carburettor Most usual are so-called fixed venturi carburettors which were used on most American and European cars, which require considerable compromises to approximate correct mixture under most operating conditions. They have a series of jets and venturis of different sizes to meter the fuel at different engine speeds. But with enough compromises, they can work surprisingly well over a very wide range of demands. For example Webers, which worked well enough to be used almost exclusively on normally aspirated F1 cars for many years, have a main jet fitted into an emulsioning tube with an air corrector jet, an idling jet with two holes in an idling jet holder with four holes, an idling mixture adjustment screw and three progression jets, an appropriately named stuffing ball with its stuffing screw, an accelerator pump to cater for sudden demands for acceleration, and about a kilometre of ducts drilled into the body casting, any of which can and do get blocked with the result that the car doesn't work. Although they sound complicated, they have the advantage that, in principle, there is only one moving part, that is the butterfly valve. In practice they also have a cold-start device (the choke) and an accelerator pump for sudden changes in demand. Although these carburettors are difficult to set up initially, and in my experience they can be temperamental, yet once the jets have been selected during engine development, they never need to be changed. Tuning is limited to cleaning blocked jets and to slow running adjustment; no adjustment is needed (or possible) to the main jet.

An alternative is the so-called variable venturi, constant depression carburettor.

Typical of this sort of carburettor are several slide-action carburettors, for example Amal, such as those fitted to my early motorcycles. In these the size of the venturi was adjusted for different demands by a slide, which was actuated by a cable directly from the throttle. These slide carburettors were pretty good approximations, and had the advantage of simplicity. They were suitable for motorcycles, because with their enormous power-to-weight ratios and excellent fuel economy, a little poor adjustment wasn't so critical.

carburettorSUs are of this variable venturi, constant depression class of carburettor, but in this case the size of the venturi is moderated indirectly by the airflow in the venturi itself. The diagram on the left shows how this works. (It also shows the butterfly valve working in the opposite sense to T-Types, and the fuel is supplied from the float chamber differently, but apart from those features it does show the essential working of H2s.) The essential feature of the carburettor is a piston which is free to move in a dashpot. In my carburettors the pistons are made of brass and are quite heavy. In later carburettors the piston is lighter, and a light spring is fitted to supplement the weight of the piston. If fitted, the spring is arranged so that it operates over only a small part of its possible extension, so its force is almost independent of deflection; its force is almost constant.

As air is drawn into the engine it is accelerated through the venturi. Following Bernoulli, the pressure in the accelerated airflow is reduced. The reduced pressure (depression) is transmitted to the top of the piston via a channel in the piston, and I have tried to suggest this by the blue shading.

There are four forces acting on the piston: atmospheric pressure is acting to lift the piston; the weight of the piston and the force of the spring (if fitted), along with the reduced pressure above the piston, are acting to move the piston downwards. When these four forces are in equilibrium, that is, when the upwards force is exactly equal to the downwards force, the piston is stationary, moving neither upwards nor downwards.

Now consider what happens as you open the throttle. As the engine demands more air, the air flow over the venturi has to be accelerated even more. Again following Bernoulli, the pressure in the venturi - and in the dashpot - is further reduced. Now the forces on the piston are no longer in equilibrium; because the pressure above the piston is reduced, the force acting to raise the piston is greater than the sum of the forces acting to lower it, and the piston will rise in the dashpot. As it rises, the area across the venturi increases, so the velocity of the air decreases and the depression at the venturi reduces (ie the pressure increases). The piston will continue to rise until a new equilibrium is reached, and the piston is stationary once again. But here's the thing: at that point, where the four forces are in equilibrium, the weight of the piston hasn't changed, the spring deflection is what it was (to a first approximation), atmospheric pressure hasn't changed: the only thing that had changed was the reduced pressure in the dashpot, and because the system is once again in equilibrium, the pressure in the dashpot must be exactly what it was before. And because the pressure in the dashpot is the same as the pressure in the venturi, that means that the system must have changed to preserve the nominal value of venturi pressure. And from Bernoulli's theorem, that means that the velocity of air through the venturi is also constant. That is what is meant by variable venturi, constant depression carburettor - unique until Stromberg copied its essential feature.

Well, that would be about half of the story. The other half involves metering of the fuel as engine demands vary.

Fuel is metered through a jet more or less level with the venturi. Fuel is supplied to it from the float chamber, and the level in the float chamber is adjusted so that it is about 3/8 inch below the jet head. Obviously if the jet were a fixed aperture, fuel would be drawn into the constant depression at a constant rate, and would not match engine demand: the mixture would become excessively lean as a constant amount of fuel would be drawn into a much larger airflow. Fixed jet carburettors get over this by having a series of jets which are progressively uncovered as the butterfly opens; in SUs, fuel is metered by way of a tapered needle fixed to the piston and inserted into the jet. As engine demand increases and the piston rises in the dashpot, the tapered needle rises with the piston and increases the effective area of the jet to admit more fuel. The taper on the needle is carefully selected during engine development, and should not be changed.

The dashpot is filled with light oil, and a damper immersed in the oil moderates the rise and fall of the piston. A one-way valve in the damper stops the piston from rising too fast, but allows it to fall unimpeded. This has the effect of increasing the richness temporarily when the throttle is suddenly opened; it fills the same role as an accelerator pump in fixed venturi carburettors.

A jet adjuster allows the mixture to be set leaner or richer by moving the jet up or down, one flat at a time. Lowering the jet makes the aperture greater, it admits more fuel into the air flow, creating a richer mixture. The jet adjuster has a fine thread (26 tpi) so one flat adjusts the jet up or down by 6.4 thou, allowing for very fine adjustment of the mixture during tuning. Once set, it should never need further adjustment.


Assembly is a little more complicated than fixed jet carburettors, where you simply have to work out which jet goes into which orifice, but not anything to worry about. Adjustment and tuning is about as simple as it could be.

This diagram of the jet assembly came from MG Service Manual published by Scientific Publishing Co Pty Ltd by arrangement with and in association with British Motor Corporation (Aust.) Pty. Ltd and copyrighted 1957. I also found it, admittedly with the names of some components altered but otherwise identical and without attribution, in "The T Series Handbook", edited by Dick Knudson, published by The New England "T" Register, Ltd, who asserted their copyright over the drawing some 24 years later, in 1981!
The first task is to put the jet assembly together. First assemble the jet adjusting nut, jet lock spring and jet screw with the lower jet bearing and brass jet washer, screwing the jet bearing into the jet adjusting nut (confusingly, the jet washer is shown too small in the diagram; the photo on the right shows clearly that it fits over the jet bearing). Now push the jet itself through the jet holder, and assemble the lower gland washer with the brass gland washer onto the jet, and push them into place at the bottom of the jet holder with the jet gland spring. (Earlier gland washers were made of cork; later iterations used synthetic materials, alkathene for the gland washers, and langite for the packing washer. In the pictures on the right the washers are cork, and have mostly crumbled away; they need to be replaced with the modern synthetic materials). The alkathene washer goes first, then the brass washer, with its dished side facing towards the alkathene washer. Now you can assemble the upper gland washers, in this case the brass washer first and then the alkathene washer; as before, the dished side of the brass washer faces towards the alkathene washer. Finally the upper jet bearing with its jet washer. The jet sealing washer and the langite sealing washer fit over the jet screw, and the whole assembly fits into the base of the carburettor; it must be screwed tightly so that the langite washer is fully compressed. When the jet screw is fully tightened, the jet gland spring compresses the two alkathene washers onto the jet itself, and presses the two jet washers firmly into position against their lands inside the jet screw and carburettor body respectively. I found that the alkathene and langite were both too hard to compress easily and make a good seal: the glands persisted in leaking. Eventually I softened them by soaking in hot oil which I heated in the microwave, which allowed them to compress fully and stopped the leaks.

In our H2 carburettors, fuel gets into the carburettor from the float chamber through a union on the bottom and via a duct in the side of the body which leads into the area of the jet assembly near the cutout in the top jet bearing, and from there into the jet itself via the two holes in the side of the jet. Without the various washers there would be several paths for leaks: along the sides of the jet, which would leak between the jethead and the jet adjusting nut; between the lower jet bearing and the jet screw, which would leak underneath the jet screw; along the threads of the jet screw itself; and finally, and invisibly, between the upper jet bearing and the body of the carburettor, allowing fuel to bypass the metering system and enter the carburettor directly, upsetting the mixture control at all settings.

By being fully compressed against the sides of the jet screw and the body of the carburettor, the langite washer looks after leaks down the threads of the jetscrew. The two alkathene washers are compressed by the jet gland spring against the sides of the jet and the top and bottom halves of the jet bearings, and prevent leaks along the sides of the jet while allowing the jet itself to move up or down. The bottom jet washer prevents leaks along the sides of the jet bearing, and the top jet washer prevents fuel from leaking into the carburettor at the top; so it is important that these washers, and their seatings inside the jet screw and inside the body of the carburettor, be in good condition.

The final step in assembly is centring the jet. Before the jet screw is fully tightened, there can be a little sideways movement of the jet assembly in the carburettor body. It is important that the jet be correctly centred to allow the piston to rise and fall freely. So before final assembly, slacken off the jet screw and observe the jet as you manipulate the jet assembly while tightening the jet screw. (The human eye has evolved to be very efficient at judging concentric circles, and it has been suggested that this is in order to help avoid predators. It has also evolved to be very efficient at judging horizontal and vertical lines, and it is hard to see a reason for this, since a need to straighten picture frames wouldn't exist for several millenia.) Alternatively, you can use the jet needle itself to do the centring: slacken off the jet nut, put the piston in place and hold it down, then fit the dashpot and tighten the jet screw. However you've centred the jet, to check if it's correct, assemble the piston and dashpot, then lift the piston and allow it to fall, when it should fall with a gentle click. If it doesn't make the gentle sound, then something is wrong and you must repeat the process. In the photograph on the right, the jet is in fact perfectly centred; it appears slightly off-centre due to the shadow cast by the camera flash. Also in this photograph you can see the two ducts which lead outside the carburettor body, by which atmospheric pressure is allowed underneath the piston.

And the final final step is to insert the piston and dashpot. Note that the piston fits in only one position: the channel to the top of the piston is always to the rear, the low pressure side of the venturi. The dashpot should be tightened slowly, a bit at a time, making sure that the piston is always free to move. Finally, when the dashpot is fully tightened, you should be able to lift the piston fully, and when released it should fall with a soft metallic click. If it does not fall, or if it falls but doesn't make the soft click, try loosening and retightening the dashpot or - if the dashpot has the two-screw mounting - try rotating it by 180degrees, and if this fails, you may need to try recentring the jet.

Adjustment and Tuning

The first step is to symchronise the carburettors. Although simple, it does involve a bit of fiddling. First you must take off the aircleaner(s) if fitted, and undo the linkage joining the carburettors and the choke control on the rear carburettor. On each carburettor, wind the jet adjusting nut fully up, keeping the jet head pushed tightly up against the nut. Then wind the jet head down by six flats (one compete turn). This gives a reasonable starting point for tuning.

Set the idling to about 1000-1200 rpm, and, using a piece of plastic or rubber tube as a stethoscope, adjust the slow running controls on both carburettors until both sound the same. Retighten the linkage between them. Done.

Tuning is a little more complicated, but not much. Using the lifting pin underneath the carburettor, if there is one, or by using a piece of wire through the air ducts, or by using a screwdriver directly onto the piston, lift the piston by about 1/8 inch. Now, here's a challenge: the velocity of air over the jet will have decreased, so the depression is less and you might expect less fuel to be drawn in, but on the other hand, by lifting the needle, the effective area of the jet is increased so you might expect more fuel. So what happens? Well, in fact the mixture is weakened.

Lifting the piston reduces the impedance to air flow, and the engine breaths a little more freely. If the carbs are correctly tuned, the engine will speed up a little as it uses the mixture already in the manifold, but then, a few seconds later, the weakened mixture reaches the spark plugs, and the engine slows down. If it doesn't slow down, or possibly even continues to speed up further, the engine likes the weaker mixture, meaning it was too rich. Weaken it slightly by screwing the jet adjusting nut upwards by one or two flats, always keeping the jet head firmly against the nut, wait a minute or so, and try again. If it doesn't speed up, but does slow down and possibly even stalls, then the mixture was much too weak, and you should enrich it by screwing the jet adjusting nut down by one or two flats, waiting again before trying again. This is an iterative process: after each adjustment, adjust the other carburettor.

When you think they're both properly tuned, use the stethoscope again to ensure they're both matched, make sure you can still raise and drop the piston, and then put the air filter back. The air filter may upset the tuning a little by impeding the air flow; the mixture may become a little richer, and you should verify again the correct mixture by lifting the piston and listening for speeding up/slowing down, and adjusting the jet adjusting nut if necessary. But this is easy to do now.

Finally, reconnect all the linkage and starting choke cable, and adjust the slow running to your taste.

Close the bonnet and have a beer.

Footnote: Bernoulli's Theorem: Actually, I've always had a bit of a problem with Bernoulli's Principle. Every proof of Bernoulli that I have seen applies to non-viscous, incompressible fluids. Importantly the proofs of Bernouilli start with a stream of fluid flowing through a pipe of varying diameter, consider what happens when the diameter is reduced and the fluid is accelerated, and show how the pressure in the flow decreases. But if you use the same assumptions and run the proof backwards , starting from a narrow streamline and allowing it to become wider, the same maths shows that there should be a pressure increase. Well, in the case of carburettors, this simply does not happen: the pressure downstream of the venturi is not higher than the pressure at the venturi. Otherwise there would not be a reduced pressure to cause the piston to rise and the carburettor would not work.

Furthermore, it has always seemed to me that whenever Bernoulli is invoked - for example sailing boats, aeroplanes, flapping shower curtains, supertankers bottoming in the English Channel - a simpler explanation is usually available from Newton's Laws of Motion, and according to Occam's Razor, you should always choose the simpler explanation.

In this case also, a simpler explanation appears to be available from consideration of Newton. As air is drawn into the engine by the descending piston on its inlet stroke, air/mixture is forced by the atmospheric pressure on the inlet side of the carburettor past the obstructions caused by the venturi, the butterfly valve, and the inlet valve. From Newtons Laws, there must be a pressure drop across each of these obstructions, to force the air past them (alternatively, and equivalently, as air passes each of the obstructions, a pressure differential is formed across the obstruction). The magnitude of the pressure drop will depend on the impedance at each obstruction, relative to the other obstructions. So downstream of the venturi, the pressure will be less than the pressure upstream (atmospheric), and this depression will be communicated to the top of the piston in the dashpot causing it to rise until equilibrium is reached. When the butterfly is opened, the impedance at the butterfly is reduced, so the overall impedance to air flow is reduced and air flow increases. Because of the increased air flow, the pressure drop across the impedance at the venturi is increased, the pressure downstream of the venturi is further reduced and the piston will rise, until a new equilibrium is reached with the four forces again in equilibrium. So once again, what has happened here? - first the velocity of the air flow increased, so the pressure drop across the venturi increased (ie the pressure downstream was reduced), the reduced presure was communicated to the top of the piston, the piston rose in the dashpot, the impedance at the obstruction at the venturi was thereby reduced and downstream pressure was restored to its original value, just as predicted above. As above, fuel is forced by atmospheric pressure from the float chamber into the venturi, and the amount of fuel is metered by the shape of the needle, which is determined empirically by experiment, not from theory.

Well, why wasn't Bernoulli adequate? I'm not a fluid physicist, so I really don't know. But as I've been pondering it, I think it's because of the assumption that the mixture is non-viscous and incompressible. Non-viscous means there are no viscous losses: once a stream of ideal nonviscous fluid has started moving in a pipe, it will carry on moving, not slowing down or losing energy due to viscosity. When the pipe gets narrow, the fluid is speeded up. Some force must have been applied to make it speed up; the force is the reduced pressure at the front, and because it's incompressible, the pressure is everywhere the same in the narrow section, ie the pressure in the fluid is reduced. When the pipe widens out again, downstream of the narrow section, the fluid slows down again. Some force must have been applied to make it slow down: the force is the increased pressure at the front of the new section of pipe, and again, because it's incompressible, the pressure in the fluid in the widened section must have increased. So if it started at atmospheric pressure, the pressure in the fluid would be reduced at the constriction, but would increase again to atmospheric when the pipe widened. Real fluids don't behave like that. They move in a pipe because of a pressure difference between the inlet and the outlet; as soon as the pressure difference is removed, the fluid stops moving, due to viscous losses. There is a pressure drop all the way along the pipe, greater at the constriction, lesser when the constriction widens out, but still a pressure drop.

But in this chapter, I have been lazy, and have used the most common explanation, ie Bernoulli.

send me an email

Top of Page