The SU Pump Revisited 

This year marks the ninetieth anniversary of the invention of SU’s Type ‘L’ Fuel pump which was used through the T Series production cycle, including early TFs up until December 1953, when from TF1501 the pump was replaced by a high-pressure variant located at the back of the car over the rear axle.  This change was probably motivated by the higher under-bonnet temperatures of the TF and the tendency for fuel to boil when moved by the negative pressure pump.  The Type L pump was patented in the names of R. L. Kent and J. N. Morris with no mention of The SU Carburettor Company.  The patent was applied for in October 1932 and finally granted in December 1933 as GB Patent 403209.  It was followed in 1935 by US Patent 1987257. 

Fig 1 – a cross sectional drawing of the pump taken from GB Patent 403209.

These patents describe how the internal shape of the coil housing is contoured to linearise the magnetic pull on the diaphragm over its range of travel and how the diaphragm can be centralised in the coil housing by the use of “a plurality of radial thrust pieces having spherical peripheries”.  Pretty much everything else is described as being accomplished by “convenient means” but the patent goes on to mention an adaption to pump liquid concrete.  The cross-sectional drawing of the Type L pump, (previously shown) is taken from GB Patent 403209.  Note there was originally a spring holding the lower valve shut.  This spring is not present today.

Prior to the invention of the Type L pump, earlier MGs were dependent on the Moriscott ‘Petrolift’.  The name Moriscott being a contraction of the names of its inventors, the same J. N. Morris but this time with co-inventor M. D. Scott. The Petrolift was a ‘Heath-Robinson’ affair that featured a vent, which under certain conditions was found to spray fuel over the hot engine.  So concerned were MG about this they issued Service Information Sheet No 7 in January 1932, reassuring owners “that the matter has been taken up with the SU Carburettor Company, who are investigating”.  Meanwhile, they advised owners that it was imperative to ensure the vent hole was oriented away from the engine towards the dash! Ed’s note: I think that the firewall was originally referred to as the dash.

Fig 2 – M.G. Car Company Service Information No. 7 January 1932.

The SU part number for the Type L variant specified for most of the T Series is AUA25, and for the later TFs is AUA54.  The numbers were originally punched into an aluminium label held by two of the coil housing fixing screws but few of these remain intact today.  There were many variants of the pump with only minor tweaks to suit the requirements of different car manufacturers. For instance, the Morris Minor used an AUA66 which is identical to the AUA25, except that the familiar ¼ BSP threaded output connector is replaced with a plain spigot.  Later, there were higher capacity variants with a coil at each end of the pump chamber used in Rolls Royce and Bentley cars and a high capacity single ended variant, the LCS.  Pumps were available in 6V and 12V versions. Originally the 6V pumps had a brown cap and the 12V pumps had a black cap.  Unfortunately, these caps no longer provide an accurate guide to the pump’s operating voltage since the first attempt to fix a dead pump was usually to hit it with a large spanner, thus many caps were broken and replaced with whatever colour cap was to hand.

So, what is the difference between a Low Pressure (LP) and a High Pressure (HP) pump?  Originally the answer was not very much, now the answer is even less.  Both pumps have an active suction stroke when the coil is energised, creating a magnetic field that attracts the diaphragm, drawing fuel into the pump base.  This is followed by a passive delivery stroke when an internal spring returns the diaphragm to its starting position whilst expelling fuel to the carburettor float chambers.  Both pumps have two valves; one which closes during the suction stroke preventing fuel from being drawn back from the float chambers and the other which closes during the delivery stroke, preventing fuel from being returned to the tank.

During the suction stroke the magnetic field created by the energised coil must attract the diaphragm, compress the return spring and lift fuel from the tank.  The LP pump has a relatively weak return spring, so cannot expel fuel to a height much above itself during the passive delivery stroke.  For this reason, LP pumps are usually found on the bulkhead of the car at or around carburettor height. 

The HP pump has a stronger return spring so it can deliver fuel to a much greater height above itself, as required following the change on later TFs when the pump was relocated to low down at the rear of the car.  To compress the stronger return spring during the suction stroke, the HP pump requires a more powerful magnet.  The very first HP pumps for the TF achieved this by using a taller 3 inch coil housing compared with the shorter 2 ⅜ inch coil housing of the LP pump.  Later HP pumps reverted to the shorter coil housing and this shorter coil housing with its stronger magnet is now common to both HP and ‘Post 85’ LP pumps.  Thus, today, the LP and HP pumps essentially only differ in the choice of return spring.  

The Low Pressure pump started life with a brass base.  This continued until 1948 when it was changed to the current two-part die cast alloy base.  As far as I have been able to establish, there were no original brass based HP pumps but some may exist today because most of the pump’s components are interchangeable.  Spare parts are readily available from Burlen Fuel Systems, making the pumps eminently restorable, but because of the price of spare parts, it is probably only worth restoring the early brass based pumps.  Sadly, for the later alloy based pumps, and for the environment too, because of Burlen’s pricing policy, it’s almost cheaper to buy a whole new pump and throw away the old one.  As an example of just how restorable pumps are, Fig 3 shows the same brass based pump, which appears to have spent most of its life at the bottom of a lake, before and after restoration.

Fig 3 – the same brass based pump, before and after restoration.

The SU pump was fitted to many British cars up until the mid-1970s and possibly beyond that, but reliability has always been an issue for a number of reasons.  The first was the earth terminal used on very early brass based pumps.  This consisted of a 2BA post that replaced one of the six screws around the periphery of the coil housing attaching it to the base.  Over time, corrosion could prevent a good electrical contact between the terminal post and the coil housing, causing the pump to fail.  SU resolved this by replacing the earth post with an earth screw tapped directly into the side of the coil housing.  LP pumps used a 2BA earth screw and HP pumps used a 4BA screw.  The earth screw provided a reliable earth connection and for a while also provided a useful means of differentiating between the pump types.  The exception to the rule however was the AUA54 HP pump for later TFs which also had a 2BA earth screw.  Today, both ‘post 1985’ LP and all HP pumps feature a 4BA earth screw in the side of the coil housing, so that means of determining which type of pump you are looking at has gone. 

The more serious ‘Achillies’ Heel’ of the pump are the electrical contacts used to energise the coil.  As these contacts open, current continues to flow in the form of a tiny arc vaporising a minute amount of contact material each time they open.  The result is the contacts don’t last very long, again causing the pump to fail.  An arc can result whenever, current flowing in a coil is interrupted.  Energy stored in the coil develops whatever voltage is necessary to jump across the barely-opened contacts.  Unless a means is found to prevent this arcing, the pump coil can produce several hundred volts from the 12V battery supply.  This phenomenon is of course put to beneficial effect in our Kettering ignition system where the 12V supply is used to store energy in the ignition coil primary winding, which is then transformed to several thousand volts at the spark plugs as the distributor points open. 

SU made several attempts to reduce the arc and thereby extend the life of the contacts.  Very early pumps used a second winding of resistance wire wound on top of the main coil winding and connected in parallel with it.  This so called ‘burden’ resistor was intended to absorb energy from the coil to protect the contacts as they opened, but sufficient energy remained for the contacts to arc.  Later, SU tried adding a capacitor in parallel with the contacts, at which point the pump circuit became exactly like the ignition circuit, where the capacitor fitted inside the distributor slows the rise of the voltage across the just-opening contacts until they are far enough apart not to arc.  The capacitor was fitted inside a modified pump cap provided with a familiar bulge.  In terms of improved reliability, it was only a marginal improvement as capacitors of the day were notoriously unreliable so the quest for a better solution continued.

Somewhere along the way the single contact was replaced with twin contacts, but in terms of extending contact life twice ‘not very long’ was still ‘not very long’.  SU then looked for more exotic solutions and tried a ‘Varistor’.  This is a device that does not conduct at low voltages, but as the voltage across it rises it clamps the voltage to a predetermined level.   Unfortunately, the Varistor did not really clamp the voltage across the contacts to a sufficiently low level to prevent them arcing.  Next, they tried a semiconductor diode.  This proved to be a very effect solution which clamped the voltage across the contacts to less than a volt but immediately made the pump polarity conscious.  Whilst a diode could be fitted to suit either battery polarity, it meant care was needed to ensure the correct polarity of pump was fitted to each car or again another failure resulted. 

About thirty years ago I introduced the idea of fitting a ‘Transil’ across the contacts which provides a perfect solution and extends points life significantly.  A Transil is another semiconductor device which limits the voltage across the opening points to a low level, thus preventing them arcing completely.  It is not polarity conscious like a simple diode, so any pump fitted with a Transil can be fitted to any car irrespective of battery polarity.  Some time later I offered this solution to Burlen but they wrote back saying my solution was flawed and they didn’t want to lower their standards.  (I still have their letter.) It seemed they preferred to continue with even more attempts to solve the problem including one with multiple diodes and a capacitor bundled up in a piece of heatshrink sleeving.  Fast forward to today; guess what?  They now supply Transils with their pump repair kits.  I guess they feel enough time has elapsed since my offer that they don’t even feel the need to acknowledge my contribution. 

Whilst the Transil provides the perfect solution to prevent the contacts arcing as they open, there is still another problem associated with the contacts. During storage the contact material has a tendency to oxidise and become coated with an insulating film.  So just when a spare pump is called upon to rescue a stranded car the spare pump may also fail to operate.  It can be no coincidence that new contacts are supplied in a sealed airtight polythene bag.  To help prevent this happening, pumps stored under the bench or in the car for just such an emergency should be stored in an airtight bag, ideally with a desiccator pouch. 

Clearly a solution that replaces the points is required.

To this end Burlen offer a fully electronic conversion.  This consists of a Hall-effect device to monitor the diaphragm position and a semiconductor switch to replace the contacts.  It’s a solution, but it needs careful adjustment and is eye-wateringly expensive.  Surely there must be a better solution?

It occurred to me at about the time I started fitting Transils to pumps that a feature of every pump variant so far, both those using contacts and the electronic version, is that the pumping capacity is limited by the hysteresis of the contacts or the Hall effect device.  That means the diaphragm only travels far enough to flip between the hysteresis levels, it does not fully expel all of the fuel in the pump chamber on every cycle.  I felt it should be possible to design an electronic circuit not limited by hysteresis but still able to mimic the rapid ‘ticks’ of the pump at turn on and then settle down to regular slower ticks to keep the float chambers topped up.  Such a system would then utilise the full travel of the diaphragm and hence increase the amount of fuel delivered in each pump cycle.  I made a few half-hearted attempts to design such a circuit at the time, using the discrete electronic components of the day, but I quickly realised they would never all fit inside the pump cap.  However, today in an era when electronic components are significantly smaller and have never been cheaper, it can now become a reality.  Normally, the pump actually has three operating phases.  Firstly, as the ignition is turned on, it makes a series of rapid ticks when it is just pumping vapour raising the fuel from the level in the tank to the pump inlet.  Next, it makes a series of less rapid ticks as fuel is pumped to fill the float chambers.  Then finally, it either stops, or if the engine is running, it continues to tick slowly to supply fuel on an ‘as required’ basis.  This can be simplified into two phases, rapid ticks at turn on and regular ticks to keep the float chambers full.  This ‘signature’ is now very easily mimicked by a tiny microprocessor and less than a dozen lines of code*.  Such a processor today costs a few pence and is smaller than the head of a match.   The processor can then drive an electronic switch to control the pump coil.  So, it has taken ninety years to perfect, but we may now finally have it.  A pump with no contacts to oxidise, no pedestal, and no finicky adjustments.  It just works!

Peter Cole

Fig 4 – a working prototype of a new electronic pump.

* The processor is programmed using ‘C‘ programming language which was provided by my son Richard.

Availability of Transils

The editor is asked from time to time about the availability of transils. They are supplied by the author of this article (Peter Cole). pcoleuk(at) [please substitute @ for (at)].

Fitting the transil Fitting is simplicity itself. The transil is supplied with ready-made solder tag connections.  All that is required to fit it is a screwdriver.

Photo shows a transil with ready-made solder tag connections along with transils awaiting fitment of the connections.

Photo shows a transil fitted to an SU pump

3 thoughts on “The SU Pump Revisited 

  1. Eric Worpe says:

    Hi Peter, Another well researched article. However, Occam’s razor says keep things simple, do we really need a microprocessor based electronic control system when a sharp knock on the pump’s body usually gets the points working?

  2. Ray White says:

    A very interesting article!

    I have rather limited understanding of electrical matters but on the advice of someone who does I have fitted a ‘varistor’ to my S.U. petrol pump. I understand it to be a transient voltage suppression diode which has the ability to stop electrical “noise” that could damage sensitive electronics. As I have fitted a (rather expensive) electronic distributor and an alternator I am hoping the Varistor I have fitted will have the same protective effect as a transil ??

    You seem to suggest it is likely to be ineffective.!


Comments are closed.