© Chris Starr 2023

Preamble, by C. Douglas

The story of the carburettor of the Merlin in the Second World War, more specifically the Battle of Britain, and the German fuel injected engines has been an ongoing saga since 1940. For a variety of reasons it has not been properly studied until relatively recently, the technical nature of the issue is one reason for the lack of clarity brought to the topic by popular historians. The other being that a large number of documents revealing the severity of the problems using a gravity carburettor caused the RAF, were not declassified until the 1970`s, rendering virtually everything written for three decades after the Battle of Britain as little more than guesswork or elaborations on established narratives on the matter (The Air Ministry Merlin development files were closed until 1972, for example).

Although my first book was I think the first to break the story of how far advanced Britain really was in fuel injection development in the 1930 (indeed even in the 1920), it has been left to Chris to really nail down the key details not on the failure to adopt injection, but on what DID happen with the carburettors used in the Merlin, and when. This he has done with exemplary detail, in a way I was not able to do in my own book.

The article Chris has written is (as it should be) fairly technical, so a brief overview of some of the terms used will help those less familiar with the esoteric study of carburettors to appreciate it. The float carburettor is essentially just a little bathtub of fuel, kept at atmospheric pressure which is used as a resovoir of fuel which the rest of the carburettor allows to be drawn into the engine by suction at a pre-determined rate, depending on how much fuel the engine needs relative to the air it ingests. To create this “suction”, the carburettor employs one or more “chokes” which are essentially just a constriction in the pipe through which air must pass to enter the carburettor, and later the engine cylinders. The flow through these chokes (which DO quite literally choke the engine of air and reduce its potential power) must accelerate through the narrowest point, and this creates a lowering of the air pressure in that locality, which in the case of a float carburettor, is used to locate the “jets” which are for all practical purposes just very precisely made holes, such that the pressure where the fuel issues forth from the jets, is at a lower pressure than the fuel in the float chamber. Otherwise it could not emerge from it and the engine would not start, let alone run properly.

To maintain this carefully controlled pressure difference, between where the fuel is temporarily stored in the carburettor body, and where it emerges into the engine inlet pipe, the fuel cannot be allowed to be pressurized in the float chamber, or it would ignore the effect of the carefully contrived holes, and merely spray into the engine at any old rate it saw fit (certainly at a much higher a rate than the engine could burn). Therefore the chamber has a float, which much like a toilet cistern, moves up and down and when the chamber is full, it closes a stop valve which prevents the bowl receiving any more fuel from a very low pressure fuel pump which draws from the main tanks of the aircraft. The behavior of this float, is, sadly, of course amenable to change from the external forces acting upon it, from gravity (and the direction that gravity applies itself), and the accelerations of the aircraft to which it is mounted. Chris will elaborate on that further, but suffice to say that wartime tests showed that even quite moderate banking of the wings threw off the ideal carburettor fueling quite significantly, not just more extreme events like negative-g or inversion.

The float carburettor has several inherent problems which are essentially insoluble, but can be made manageable with all manner of correcting gizmos and widgets, such that although the carburettor is “simple”, it is so purely in basic principle, the aviation carburettor in practice, is an immensely sophisticated piece of equipment, to my mind, only less so than fuel injection on a surface level of investigation. For example, the flow through the carburettor changes in density as the aircraft climbs, but the density of the fuel does not, this is an intractable fundamental problem with physics, and can only be fudged by various correcting measures, none of which produce the ideal fuel metering at all altitudes.

The reader may at this point be wondering how any carburettor was ever used in an aircraft at all, it should be remembered that carburettors had been developed on the ground, where most of the very serious problems present in aircraft applications simply do not exist, in that they remain broadly parallel to the ground, and are used under a tiny range of variations in pressure and temperature compared to an aircraft. Aircraft in the First World War had engines of such appalling lifespan and reliability that the behaviour of the carburettor was the least of the worry, and the development of very high altitude flight happened in a very short space of time and was possible as the development of the supercharger drastically outstripped the development of the fuel delivery technology.

I cannot emphasize enough what a serious issue this matter was for the RAF, or to drive home that not only was the “orifice” not fitted during the Battle-of-Britain, but even once it was, the result was not a complete solution, but hopefully enough to prevent any more unnecessary deaths of RAF pilots, which without any question happened on many occasions directly due to loss of flight performance or indirectly due to falling victim to a German fighter, who`s pilot had perhaps just the day before, escaped back to France by exploiting the stick forward, nose-down dive & escape advantage of the Bf109 – and thus returned to the fray when, had the RAF fighter been able to follow him the previous day, he would not have done so.

This story is far more than just an academic discussion of the fuelling technologies used in fighter aircraft of the day.

Chris`s Article

Figure 1: Claudel-Hobson Carburettor

When Rolls Royce designed the new 1000hp class Merlin engine in the early 1930’s they chose to equip the engine with a development of a contemporary aircraft carburettor. Their initial options at that time were probably limited to the use of a float carburettor, a type which had been used quite successfully in many of their preceding aero-engines. The Merlin was conceived as a supercharged V12 engine, laid out similarly to the R Type racing engines, with the carburettor placed before the supercharger air intake, albeit offset slightly below the intake to reduce engine length, and thus being an updraught carburettor. This layout also closely resembled that of the successful supercharged Kestrel engines. The Kestrel carburettors had generally used design features of the Claudel-Hobson type, where fuel jetting was largely adjusted for altitude and power levels by the physical adjustment of the carburettor fixed jets, mostly operated by the pilot controlled mixture lever and throttle position.

A contemporary Claudel-Hobson instrument is shown at figure 1.

Although this Claudel-Hobson carburettor is not a prototype Merlin version, the fundamentals of the Claudel-Hobson type; throttle plates, acceleration pump, fuel in a float regulated chamber and calibrated fuel jets, etc are apparent. It can be seen in the diagrammatic sketches on the right side of the illustration that the fuel jets for altitude compensation and boost enrichment were mechanically controlled. Rolls Royce designer A.A. Rubbra stated that Rolls Royce designed the carburettors on the development Merlin engines following (supercharged) Kestrel practice and Rolls-Royce certainly seem to have had considerable design freedom in adapting or redesigning carburettors for their engines.

There was also a requirement to ease the pilot workload and automatic carburettor control was desirable where possible. In a design competition for the initial Merlin production carburettor, a Skinners Union company variable jet carburettor was chosen over a Claudel-Hobson contender and so, a Rolls Royce/SU type carburettor was used when the early Merlin engine went into production. However, the carburettor controls still included a pilot operated mixture control lever, with a certain amount of automation in the operation of other carburettor functions.

Despite the obvious complexity of these float type carburettors, their weaknesses in operation with a vulnerability to malfunction during upset in flight and acceleration or negative G effects were recognised but unresolved. The attitude and acceleration effects on the float type carburettor are principally due to the use of a simple float regulated fuel supply in a float chamber. The various fuel feeds to the intake venturi from the float chamber rely on the upright and unaccelerated condition of the carburettor, in any deviation from the straight and level attitude a simple float type carburettor will suffer from effects upon its fuel metering ability.

In Germany, development work on petrol fuel injection had begun in 1928 and subsequently the RLM had directed in 1933 that large aero engines should be developed with fuel injection rather than carburettors. The primary aims of this fuel injection were; to optimise the air/fuel ratio for every engine cylinder, to optimise fuel consumption with large valve overlap, to remove the restricting venturi’s from the engine air intake, to simplify pilot workload, to prevent carburettor icing and to prevent fuel mixture upset due to inverted flying or G effects. Unfortunately, these issues remained unresolved with the float type carburettors fitted to the Merlin until a partial remedy in 1941 and further resolution on new engines from approximately 1943 onwards with the introduction of pressure carburettors and fuel injection.

Figure 2: Merlin SU AVT 32/132 Carburettor

The initial Merlin production Carburettor type SU AVT 32/135 was a twin choke updraught type with twin venturi, each 3.3 inches in diameter. The left choke incorporated the aneroid controlled altitude needle/jet and the right choke had the aneroid controlled boost needle/jet enrichment with a manual Rich/Weak datum control. The carburettor had an automatic accelerator pump and idle running circuits. The pilot controlled the throttle position. However, the Rolls Royce Boost controller fitted to the production Merlin could limit throttle plate movement, via a linkage and differential, to control the throttle plate position and so impose the engine boost limits.

The boost controller also had various styles of over-ride cut-out to allow overboosting in emergency by disabling the boost control influence on throttle plate position when the pilot operated the over-ride. Unfortunately, this carburettor was vulnerable to adverse function in several respects. Due to the fact that float type carburettors rely on a relatively weak air pressure effect to raise the fuel into the venturi, they are badly effected by any positional deviation from the straight and level datum that thereby alters the float chamber fuel level relative to the outlet into the venturi.
The internal operating levers of the carburettor may also suffer from influence under G effects, if they are not designed to avoid it. Perhaps the worst problem is that under negative G the fuel supply held in the carburettor float chamber can become totally inverted in the chamber, with uncontrolled effects on fuel mixture to the engine. These weaknesses in the operation of float type carburettors in manoeuvring aircraft were known at the time. The positive G effects and short periods of negative G flight were common in fighter type aircraft of the early 1930’s but little effective work seems to have been done in the UK to solve these weaknesses in the aircraft float type carburettor. It should be realised that at high speeds, an aircraft does not need to be inverted to experience negative G. Simply pushing the nose down in manoeuvres at high speed will generate negative G forces. Action to prevent the worst of these negative G effects on the RR/SU float type carbs was not effective until 1941.
A further issue is carburettor icing. The temperature drop due to evaporation of the fuel causes ice formation from any water vapour in the air passing through the carburettor. This carb icing presents a dire threat to the aircraft even in warm climates and the ice formed inside the carburettor can restrict airflow sufficiently to stop the engine.

The SU carburettors prevented carb icing by having hot engine oil fed inside the carb throttle plates and hot engine coolant was circulated around the choke tubes. These arrangements were effective to prevent carb icing but, incurred wasteful heating of the intake charge and the additional oil and coolant circuits were unwanted vulnerabilities that could contribute to engine failure.
Notwithstanding these issues, and other considerations like backfires and unequal cylinder mixture strengths, Rolls Royce used these SU carburettors on Merlins I, II, III, IV, V, VIII, X, XII and 30. The function of this SU AVT 32/135 carburettor is easy to understand from Fig 2. The fuel supply is controlled by the float to maintain a level slightly lower than the diffuser tubes, one in each venturi. The fuel is available to the idle jets, the accelerator pump and the main jets (both aneroid regulated), one for altitude the other for boost enrichment, with a selectable rich/weak datum bias.

The float chamber air vent (43) is important as it balances the pressure in the float chamber with that in the carb intake at the entrance to the venturi. At Idle, the closed throttle valve (38) has generated a higher depression at the slow running nozzles (9), enough to raise the fuel up to them from the idle jets (5). As the throttle is opened airflow increases and the depression in the venturi also increases until it is strong enough to lift fuel through the two variable main jets and into the diffuser tubes (30) as an emulsion. When the throttle is moved in an open direction, the accelerator pump also pushes a linked volume of fuel up the passages to the delivery ducts (25) and into the airflow, thus preventing hesitation or weak cutting that would occur if the throttle is opened quickly.

Problems with unequal fuel distribution to the cylinders were partially cured by careful intake design but, the general solution to poor fuel distribution was to richen the overall mixture setting until the weakest running cylinder was at an acceptable setting. Engine damaging backfires at increasing boost pressures were controlled by fitting flametraps in the inlet manifolds at the inlet ports. These traps consisted of a fine matrix of nickel-silver plates that quenched a developing flame front in the intake manifold and prevented engine damage or failure. However, there was a considerable maintenance penalty and the flametraps also robbed power.

A curious omission from the initial standard Merlin AVT 32 carburettor was the lack of a slow- running cut-off in the Idle circuits. The SRCO was present on Claudel-Hobson carburettors and it functioned to allow a clean cut-off of engine running by stopping the fuel supply to the idle passages in the carburettor. Without this SRCO, early Merlins were stopped by shutting the main fuel cocks and waiting till the engine hesitated as the fuel level dropped in the floatbowl, then smartly turning off the magneto’s. This was an irksome process for stopping the engine that, in addition to being time wasting, left the likelihood of a dangerous combustible mixture in the induction system after shutdown and would regularly result in the engine running-on or kicking back and running backwards after the mags were turned off.

At about the time of the introduction of the Merlin III, SRCO valves were added to the idle fuel circuits, operated by cable from a pull ring on the instrument panel. The Merlin then became quite civilised on shutdown and the risks from leaving ignitable mixture in the induction system after shutdown were removed.

This then, was the approximate in-service standard of the Merlin carburettor and induction system during the Battle of Britain. However, since the time when the Merlin was conceived with a float carburettor in 1933, aircraft carburettors and fuel injection systems had advanced greatly in some other countries. Fuel injection was not favoured by the RAE in Great Britain and other advances in carburettor technology seem to have been ignored, notably some from the USA.

Other than fuel injection, the greatest advance was the development by Bendix-Stromberg of their pressure carburettor system, which replaced the vented float chamber, float valve and the diffuser into the venturi, with a pressure regulated and metered fuel flow that was sprayed into the airstream just before the supercharger. At a stroke, this new fuel system removed virtually all of the difficulties that float carburettors had suffered with G effects and inverted flight. Also, the fact that the fuel was sprayed into the airstream after the pressure sensing venturi and the throttle plates, straight into the supercharger intake, removed almost all risk of carburettor icing. This new Bendix PD12 type of carburettor was successfully tested and fitted to the Allison V1710 C10 in 1938.

At the time of the Battle of Britain, the Spitfire and Bf109 were closely matched in performance. The Bf109’s DB601A engine was fitted with direct fuel injection that provided uninterrupted engine running in all attitudes, even in fully negative G conditions for short periods.

The Merlin engine with the float carburettor suffered various carburation faults in unusual attitudes and would cut completely under negative G. As the desperate fighter vs fighter combat tactics developed in the Battle of Britain, the tactical disadvantage of the carburettor problems in Merlin engines became a high priority for improvement. German pilots could push their aircraft into negative G with their engines running normally but, the Spitfire and Hurricane would fluff and rich- cut if they tried to do the same. This problem was more than a nuisance, lives were lost in combat as a result.

Understanding this problem can be helped by studying the AVT 32 carburettor in Fig 2. The fuel (in yellow) will move to the top of the float chamber under negative G. This displaced fuel can then flow unrestricted back down the float chamber air vent lines(43) into the carburettor intake and initiate a rich-cut of the engine. In zero or negative G conditions, the fuel float can then rise in the fuel chamber, in the inverted fuel, and open the fuel valve (33) allowing pressurised fuel to flood the float chamber and continue flooding all the passages (air vents, air ducts, jets and diffusers) into the inlet airflow.

These conditions inside the carburettor were chaotic, the exact process of flooding or starving the engine of fuel would depend somewhat on the way the aircraft manoeuvred into negative G conditions, and the time period that the negative G condition was maintained.

Nonetheless, the engine stopped producing power virtually as soon as negative G was applied, instead becoming an airbrake through the effect of the constant speed propeller having to drive the dead engine throughout the period of the cut-out, only recovering normal running several seconds after regaining positive G conditions. Unfortunately for the RAF, there was little that could be done quickly to cure this problem during the Battle of Britain, other than for pilots to be briefed on enemy tactics that used negative G and some limited counters to that, where positive G could be maintained, and by rolling and pulling to maintain positive G. However, this was no panacea. Spitfire and Hurricane pilots found themselves at a disadvantage when involved in fighter vs fighter combat where negative G manoeuvres were used by enemy pilots.
At about this time Rolls Royce were also bringing the Merlin XX into production. The improved engine featured a new version of the RR/SU carburettor, the AVT 40/193. This carburettor featured larger chokes with 3.75 inch diameter venturi for greater airflow in the more powerful engine and twin float chambers that made some attempt to reduce the fuel flow problems that these float type carburettors suffered when angled away from straight and level. However, this carburettor was still unable to function in negative G. The AVT 40/193 carburettor was designed for use on the Merlin XX, 21, 45, 46 and 47 and is shown in Fig 3.

Figure 3: AVT 40/193 – Merlin Series XX

RAE scientist and engineer Beatrice Shilling is credited with successful work to cure some of the faults with the float carburettors. Well aware of the negative G problems, she developed a fuel line restrictor to limit the maximum fuel flow to the carburettor to equal the maximum fuel flow that the engine required. This flow limit only restricted the fuel flow when the false excess demand from a flooding carburettor in negative G would otherwise cause a rich cut-out. The restrictor was tested and provided a partial cure, so a crash modification program began in the beginning of 1941. The restrictor was fitted in the fuel inlet of the carburettor, and was correctly named the “RAE Restrictor”.
The initial version of restrictor for the AVT 32/135 was a thimble-like pressing with a hole at the end that was simply fitted like an olive between the carburettor fuel inlet union and the fuel feed pipe of the Merlin II, III, IV, V, VIII, X, XII and 30. In later versions, the restrictor was formed as a disc with a central hole and was incorporated into the actual fuel inlet union fitting of subsequent Merlins with standard float carburettors. Rolls Royce Merlin engines were being developed at great pace and flow restrictors to suit the various versions of the engine were designed. A comprehensive guide to the versions of RAE restrictor and their applicability is provided by the R R Technical instruction sheet By.84, copied from RAF AP 2308 Feb ‘43 that should be read at this point in Appendix 1.
Rolls Royce were keen to further improve the unusual attitude and negative G performance of the Merlin fitted with the AVT 40/193 carburettor. A redesign was performed by Rolls Royce and SU that replaced the simple float controlled inlet fuel valve with a weight balanced diaphragm controlled fuel inlet. This considerable work resulted in a new carburettor, the RR/SU AVT 40/213. This was designed for the Merlin 45, 46, 47, 50 and 56 engines and was intended to perform correctly under positive and negative G conditions. The balanced diaphragm was intended to provide the correct basic fuel flow into the carburettor in positive and negative G. Unfortunately, service trials with two Squadrons of Spitfire MkV using this new carburettor were unsatisfactory and in a major upset to Merlin development, the AVT 40/213 carburettor was withdrawn from use. A diagram of the AVT 40/213 carburettor is at Fig 4.
The RAF AP 1590J and L covering the AVT 40/193 and the AVT 40/213 is at Appendix 2 and this comprehensively describes both the 40/193 and the unsuccessful 40/213 carburettors.

Figure 4: Merlin AVT 40/193 Carburettor

Further development to resolve the negative G problems with the Merlin AVT 40/193 carburettors had been ongoing at the RAE. This type of carburettor was to remain important and in heavy use as it was used in many aircraft and their Merlin engines. It seems that the RAE chose to attack the specific problems that this float type carburettor suffered under negative G, while retaining the basic layout and functions of the float design.The modifications were designed to cure three internal problems during negative G; the flooding flow of fuel in the air balance and vent lines, the interrupted flow of fuel to the main jets and the excessive flow of fuel through the fuel inlet valves due to float malfunction in negative G which caused continued flooding.

The air balance and vent lines were prevented from allowing fuel back flowing by fitting ball valves. The flow of fuel to the main jets was maintained by isolating the jet orifice in a chamber formed by a plate and fed with fuel from shroud tubes that acted as standpipes, feeding fuel when either upright or inverted. Excessive fuel flow, due to the floats rising when inverted in the displaced fuel, was limited to maximum engine demand flow, by fitting pintles on the float valves and maximum flow rate stops on the levers, that limited the maximum travel of the floats and hence maximum fuel flow, in a similar but more complicated way than the RAE Restrictor.

These combined modifications prevented the worst combinations of rich cutting and/or weak cutting that unmodified carburettors had suffered. The new carburettor was identified as the AVT40 anti-G. It also seems that carburettors modified in this way are referred to as an “RAE Anti-g carburettor”.

The author has examined such an anti-G carburettor with the stamping, AVT40 SUX 216 and also, an example of the earlier non anti-G version stamped AVT40 SUX 193.

The comprehensive diagrammatic illustration of the AVT 40 Anti-G is at Fig 5.

Figure 5: AVT 40 Carburettor (AVT 40/216)

Merlin engine development continued at pace and the upcoming 2 Speed, 2 Stage 60 series Merlin engines required a larger carburettor for their greater airflow and power.
The new carburettor was the RR/SU AVT 44/199/1 and this was listed for the Merlin 60, 61 and 62 in the diagrams, although the Rolls Royce tech note lists the version of the restrictor for the Merlin 61, 63 and 64 with AVT 44/201 and AVT 44/206 versions of this carburettor. The Merlin 61, 63 and 64 were fighter engines and the RAF were, no doubt, desperate for the improvement of the fully modified anti G carburettor. The fully modified Anti-G type of carburettor for these engines was introduced as the AVT 44/208 Anti-G.
Spitfire IX trials with these different carburettors illustrate the importance of the carburettor function under negative G. A Service trial of AB505 in April 1942 notes that the “Merlin 61 fitted with the latest Anti-G carburettor” was able to “change rapidly from climb to dive without the engine cutting”. Conversely, a July 1942 a trial of a Spitfire IX without an Anti-g carburettor flown against a Fw 190 had “great difficulty following” a diving turn as, “this type of engine with ordinary carburettor cuts out very easily”.

An illustration of the AVT 44/199/1 is shown at Figure 6 and a diagram in Figure 7.

Figure 6: AVT 44/199/1 Carburettor Illustration
Figure 7: AVT 44/199/1 Carburettor Diagram

For the next important and powerful Merlin 66 engine, Rolls Royce finally decided to use the Bendix–Stromberg Injection carburettor. The American Bendix-Stromberg pressure carburettor was developed in the mid 1930’s and was in production from 1938. This carburettor was designed to operate as a fully pressurised fuel system that dispensed with the problematic float controlled fuel level with its emulsion tubes and diffusers. Negative G had no effect on fuel flow or carburettor function. The pressurised and metered fuel flow was delivered as a spray into the inlet air stream just in front of the supercharger inlet. This feature virtually removed the risk of carburettor icing, in fact the throttles and chokes of the injection carburettor did not need heating by hot oil or coolant circulation at all and their deletion removed several other problems associated with the previous provision of those heating circuits.

Rolls Royce had been aware of the Bendix-Stromberg Pressure type of carburettor for several years and versions of the carburettor were used on many American engines including the Allison V-1710. Notably, Packard built their Merlins in the USA with a version of the Bendix PD16 from the very start of Packard Merlin production.

For the high power Merlin 66, Rolls Royce needed an even larger choke carburettor and decided to modify the Bendix PD18 type Pressure carburettor to suit. Suitably modified, this carburettor was fitted as the Stromberg 8D44/1. There were some problems with the Bendix pressure carburettor, particularly if excess air was allowed into the fuel system, but this was resolved with internal bleed venting of the D chamber of the carburettor and, by using a vane type fuel pump that could clear the airlock.

A diagrammatic illustration of the 8D44/1 carburettor is shown at Figure 8.

Figure 8: 8D/44/ Bendix Stromberg Injection (a.k.a. ‘Pressure’) Carburettor

In the USA, Packard built their versions of the Merlin 66, the 266 and V-1650-7, with a very similar Bendix PD18. These Bendix-Stromberg PD pressure carburettors were a step towards fuel injection, and were sometimes referred to as injection carburettors.

In further UK developments of the Merlin engine, the Bendix pressure carburettor was replaced with an SU developed single point fuel injector with a variable output swashplate type injection pump. Finally, in the UK, Rolls Royce developed their own simplified gear pump single point fuel injection that was cheaper to produce. This system missed the last Merlin engines but, was featured on late Griffon engines-serving operationally with the RAF until 1991 when the Avro Shackleton was retired from service.

The author wishes to thank Peter Grieve for his generous assistance with source documents and Dave Piggott RRHT for correspondence on Rolls Royce carburettors.