“ALL THAT GLITTERS IS NOT GOLD” – A day on a rolling road (Part 3)

In the first two articles I spoke of the original problem with my XPEG engine, being poor cold starting and inconsistent tick over, especially when cold, and that I did not have a hot restart problem. I previously reported a number of problems, now resolved: twin carburettor mixtures different, wrong needle profile, insufficient advance at tick over, after-market air filters too small, weak valve springs, leaking head gasket and, significantly, a worn A2D4 distributor. The needles were GJ and were changed to #6 that has a slightly richer profile at station 3 (working rpm range). The distributor was replaced with a CSI electronic with 16 selectable advance curves. Curve 7 was selected initially and equated to 100 advance at tick over, 260 at 2000 rpm and 380 at 4000 rpm or higher. We still had an inconsistent tick over, but otherwise on-road performance was good. The spark plugs at this point were NGK BP6ES.

Subsequently I changed the advance curve to 4, which equated to 100 at tick over, 260 at 2000 rpm and 360 at 4000 rpm or higher. This seemed to improve tick over a bit, but otherwise no obvious road performance changes were discernible. It was now time to check out the other curve settings on a rolling road, and to measure torque and bhp (at the wheels in 3rd gear). I kept in mind the research done at Manchester (UK) University on a XPAG engine in laboratory conditions and wanted to see how I could use the published data to good effect on my engine without having sufficient engineering skills to do other than make improvements of the kind available to most historic car owners.

Obviously, the engine has to be in good condition and settings such as carburettors air intake & mixture, tappet clearances and plug gaps are correct (see below). My engine has +20 thou bores (suggesting one previous re-bore) with no discernible bore wear, piston rings and main and gudgeon bearings free from play, & cylinder pressures in the range 160-170 lb/in2. The profile of the camshaft is not known butlobe wear was negligible. Tappet clearance is 14 thou cold, static advance is 100, fuel is Shell V-Power (low ethanol), plugs are Bosch W7DC at 12 thou (oops!) and later 22 thou, a 5-speed gearbox and a standard 4.875:1 differential.

Curves 1, 3, 5, 9-16 were not tested. Max advance of curves 9-16 were 40 or 42 degrees & thought too high for standard XPEG engine; “wot” is wide open throttle at 4500 rpm. Hydrocarbon emissions were measured at the exhaust pipe exit. The key results can best be summarised in table form, below:

Key features of the results are summarised below.

  • Spark plug gap has a significant impact on emissions. The correct gap is important for fuel efficiency but not necessarily for bhp output – surprisingly.
  • Adrian of A B Garage recommends returning to NGK plugs, which I will do in due course.
  • Curve 2 returned a substantial loss of bhp (17) & torque (63) above 3200 rpm.
  • Curve 4 with the wrong plug gap shows slightly higher bhp & torque, but emissions are high at idle.
  • Curve 4 with correct plug gap surprised us with slightly lower bhp & torque throughout the rev range, but much improved emissions! That is to say, more of the fuel was used in the combustion process.
  • Curve 6 shows a substantial loss of bhp & torque after 4000 rpm, but also spoilt by wrong plug gap.
  • Curve 7 shows a generally good bhp & torque profile.
  • Curve 8 shows a steady increase in bhp throughout, but there is an anomalous result at 3600 rpm, which we cannot explain.
  • An ignition advance of 260 or 280 from 2000 rpm proves beneficial.
  • The engine performs well at higher revs when set to 380 advance.
  • With regard to tick over, curves 4, 7 & 8 produced at fairly consistent tick over at 900 rpm and this was judged to be satisfactory. We obviously could not test a cold start, which was carried out the following day in my garage.
  • Power output at 4000 rpm seemed a little low, although the original 63 bhp at the flywheel would have translated to about 50-52 bhp at the wheels. Probably got a few losses in an ageing transmission!
  • Torque is not as published, largely because we are measuring at the wheels in 3rd gear, and is an arithmetical calculation based on bhp and engine revs rather than being measured correctly. It is the relationship between different curves that is of interest in this exercise.
  • We also noted the car’s rev counter was inaccurate: a displayed 1100 rpm was actually 900 rpm, and a displayed 3000 rpm was actually 2600 rpm, an average error of about 15%.

The final decision as to which curve to use came down to either 4, 7 or 8. We decided that the anomaly in curve 8 results could be a computer glitch or, indeed, a CSI ignition programme glitch, so on balance curve 8 gave a slightly better performance throughout the useable rpm range than curve 4, albeit with slightly higher emissions at 4000 rpm. In practice on the road the useable revs tend to be in the range 1500 – 3000 rpm with a 5-speed gearbox, so the anomaly in curve 8 readings is not quite so important. We decided to set the distributor at curve 8.

I started with two problems: poor starting when cold and inconsistent tick over (cyclic variability). Were these now resolved? Yes, is the short answer; cold starting has improved but still requires several engine rotations, which may just be how I set the choke initially, and tick over is now smooth at approximately 900 rpm. I am well satisfied with the outcome and along the way have improved my knowledge of the combustion process in XPAG/XPEG engines running on modern ethanol fuels.

A combined set of graphs for curves 4, 7 & 8 follows.

Neil Wallace
March 2018

(Click image for bigger)

Pictured here is the interior of Paul Mellor’s TC0980. Paul has now completed a chassis up rebuild on the car to a very high standard. The Dashboard is the original that was walnut veneered and polished by a friend of Paul’s who has a furniture making business. The Leather hides came from UK Hides near Brooklands and Paul’s friend and mechanic Mr John Smith completed the trim. He used the old trim as templates. The rebuild took 13 months and is complete except for the hood, which is expected any day.