The classic SU fuel pump

Published in the August edition (73) of Totally T-Type 2 was an interesting article from Peter Cole, an obvious enthusiast who has equally sought a solution to modernising his own pump. His article inspired me to make contact with him and share my experiences and my own solution that has served me very well.

An introduction and overview of the process leading to an alternative SU pump switch design

The original electric fuel pump fitted to many models of British classic cars has for many been a source of frustration, with the pump either not wanting to spring into life following a long period of dormancy or simply stopping operation when least convenient.

The principle by which these pumps operate is by repeated strokes of a solenoid, the one end of which is connected to a rubber diaphragm.  At the other end is an electrical assembly made up of sets of points that are operated by the action of the solenoid’s movement. In the main direction when the solenoid is activated, points closed, current flows through the solenoid coil, this causes the armature of the solenoid to be drawn into the body of the pump housing, and in so doing sucks fuel via an inlet valve from the fuel tank.  During the course of the stroke the points open, current now stops following in the coil and the internal spring fitted around the pump diaphragm now starts to push the fuel out of the pump via the outlet valve towards the carburetters.

Each operation of the pump causes a certain amount of sparking across the points, which eventually leads to wear; this wear not only inhibits the ability of the electrical points to carry the current to the pump coil but eventually upsets the mechanical settings of the pump.

The design of the pump coil included with its winding, a resistive coil that was included to help limit the back EMF generated when the operating current to the coil was interrupted. This resistive winding used very fine wire and I suspect that in many cases this winding has become damaged, either electrically or mechanically.  The most likely cause being as a result of an overload condition or damage caused by poor handling of the coil connections during servicing of the pump. Beside the electrical challenges associated with the points themselves, there are also the mechanical challenges of the switch rocker mechanism that requires careful setting.  These settings can only be achieved when the pump is pulled apart so that the pump diaphragm can be manually operated during the setting process. The process itself is well documented, but still presents a challenge.  There are the levers themselves, with their adjustments, the pin that secures the mechanism to the pedestal and the springs that influence the throw over of the levers as the diaphragm shaft oscillates to and fro; quite a clever arrangement considering when it was first conceived.

The switch rocker mechanism

My own experience with pump problems goes back many years, and first became an issue when I owned an MGBGT V8.  The pump in this car had plenty to do keeping up with the demands of a much younger throttle foot and the V8 engine.

At the time, alternatives to the mechanical switching mechanism within the pump were limited; there was an offering from a gentleman who was based in Liskeard in Cornwall, England, the place of my birth as it happens.  His offering was based on replacing the electrical points with an optically triggered transistorised switch that utilised the switch rocker mechanism as its operating platform.  The advantage of this solution overcame the problems of arcing points but still relied on the carefully setting up of the optical vane on rocker mechanism. My own opinion is that the pump’s reliability is a two fold problem, the points together with the intricacies of the switch rocker mechanism.

Alternative ways of extending the life of the points have included components that help to suppress the sparking at the points; devices ranged from transient diodes, metal oxides varistor (MOV), to zener diodes, all to limit the voltage being exposed to the points.

My own solution back then was the use of a transistor to bypass the bulk of the current away from the switch contacts, whilst still maintaining a small current to keep the point contacts clean. The circuit below certainly helped to resolve my issues with the MGBGT V8, but still relied on the mechanical rocker mechanism.

My next encounter with a stubborn fuel pump was on the inaugural run of my fully restored MG TD in 1996. The restoration had started with a basket case of an MG that I purchased 1986.  As part of the restoration the fuel pump was completely overhauled, set up according to the workshop manual and tested.  However, on the very first test drive the TD came to a stop, resulting in the car being pushed back to its garage.  Being the first run of the car, I was not sure what had gone wrong. Only once back at the garage was the fault traced to the fuel pump.  The pump was removed, the mechanical settings readjusted, together with the fitting of the contact bypass arrangement as illustrated above.  The TD was no longer 100% authentic, but I had piece of mind and IT operated well for many years.

Pictured is the simpler approach fitted to my TD in 1996.

Fast forward to 2003……

In 2003 on a long journey home, my very trusty TD started to falter. Then came a challenging mountain pass that took a number stops while ascending the hill to attend to the fuel pump. By the third stop a replacement pump was offered, but was rather tired and more fuel leaked from it than went to the engine. A second replacement from a very dear friend who carries with her many spares, included a new fuel pump, which got me home. The delays and inconvenience to others had me thinking all the way home, there has to be a better way. 

The problem was that the switch rocker mechanism had worn to the point that normal adjustment no longer provided reliable operation.  What was needed was a method of switching the pump solenoid, simple enough with a solid-state switch, but also a method of triggering it that did not utilise the switch rocker mechanism. What I designed is the arrangement in the picture below. It is a thin printed circuit board that lives under the original pedestal. It switches the pump’s coil without the need of complicated or sensitive mechanisms to determine when the solenoid should operate. On the end of the diaphragm shaft is a boss which maintains its alignment, as well as providing a means of knowing when the pump is to be energised. The rest of the circuitry supports the sensing and protection.  The behaviour of the pump is very similar to an original pump, it ticks fast until fuel pressure is achieved and then when required to maintain fuel delivery.

First unit fitted to my TD in 2004

The unit has been in operation now for more than fifteen years, without need of any attention, other than to take pictures. The pedestal and connections have been removed in order to visualise what is buried beneath. 

My own TD also benefits from other innovations such as a solid-state voltage regulator that lives neatly inside the original cover. Again, it has given many years of dependable service and it compensates for added loads without resulting in battery voltage degradation, even when driving long distances at night. With the original two pole regulator, getting the balance right was quite a challenge.

Roger Lewis, Cape Town, South Africa



2 thoughts on “The classic SU fuel pump

  1. John Hall says:

    Roger, I marvel at your ingenuity and tenacity. My wife and I try to use TD and YA as daily drivers – is there a chance you might be offering an exchange/sale arrangement to fellow enthusiasts with less skill?? Yours, John Hall (Queensland, Aus.).

Leave a Reply

Your email address will not be published. Required fields are marked *